[MaxGhenis]

Fixes #2008

Stacked on #2006 (base benefit-unit-rent-liability, at 7769fa3, which includes main and #1896). Retarget to main once #2006 merges.

Problem

The LHA size criteria gave no bedroom to a member of the benefit unit aged 16 or over who is not the claimant or partner. The commonest case is a young person at school or college living with a parent.

LHA_allowed_bedrooms (Universal Credit) counted:

• the renter or couple;
• non-dependants aged 16 or over from other families;
• children's rooms for members under 16.

So a lone parent with a 17-year-old at college was assessed for one bedroom (category B) instead of two (C). main subtracts every benefit-unit member aged 16 or over from the household count, which has the same effect. Housing Benefit had the same gap until #2006's rewrite of housing_benefit_LHA_allowed_bedrooms over the claim's occupiers, which this PR now builds on.

The law

Universal Credit. UC Regs 2013 Sch 4 para 10(1) says "A renter is entitled to one bedroom for each of the following categories of persons in their extended benefit unit". The categories are:

• (a) the renter (or joint renters);
• (b) "a qualifying young person for whom the renter or either joint renter is responsible";
• (c) "a non-dependant who is not a child";
• (d)-(f) children.

Para 9(2) makes anyone else aged 16 or over who normally lives with the renter a non-dependant. The exception relevant here is (g): "a child or qualifying young person for whom no-one in the renter's extended benefit unit is responsible", such as a young person looked after by a local authority (reg 4(6)). It applies across the whole extended benefit unit, so it covers a looked-after young person in a non-dependant's family too.

A child is a person under 16 (WRA 2012 s.40). A qualifying young person is defined in reg 5.

Housing Benefit. HB Regs 2006 and HB (SPC) Regs 2006 reg 13D(3) (SI 2006/213, SI 2006/214) give "one bedroom for each of the following categories of occupier", including "(a) a couple" and "(b) a person who is not a child", where a child is "a person under the age of 16" (reg 2(1)). An ordinary couple shares one bedroom; DWP circular HB A3/2017 confirms that only a couple who cannot share because of disability has an extra one.

Change

UC (LHA_allowed_bedrooms). Each member of the benefit unit aged 16 or over who is not the claimant or partner gets a bedroom:

• a qualifying young person the renter is responsible for (is_child_or_qualifying_young_person_for_universal_credit), under para 10(1)(b);
• any other such member, as a non-dependant under para 10(1)(c).

A qualifying young person no one is responsible for is not a non-dependant (para 9(2)(g)). This exclusion also applies to the count of non-dependants from other families, which now goes to the claim in #2006's uc_non_dependants_counted.

The exclusion is in the UC formula, not in is_non_dependant_of_household_head, because that variable also feeds Housing Benefit's non-dependant tests and deductions, where para 9(2)(g) does not apply. (Council Tax Reduction classifies non-dependants separately.)

HB. No code change here: #2006's rewrite already gives the benefit unit's own members aged 16 or over a bedroom under reg 13D(3)(b), and gives an occupier couple one bedroom under (a). This PR adds HB test cases and properties for both.

Why the UC rule covers non-dependants as well as qualifying young people. The issue's UC example is the qualifying young person. A benefit-unit member aged 16 or over who is not a qualifying young person is, in law, a non-dependant with their own bedroom. On the enhanced FRS that group is 19k people in 20 survey households, against 1.95m qualifying young people.

Invariants

The Hypothesis properties are in test_lha_bedroom_young_people_properties.py. Each runs 25 derandomized examples over populations of up to 4 households with sharers, boarders, lodgers and non-dependant families.

1. Monotonicity. Adding a person aged 16 to 19 to any family never lowers any family's UC or HB bedrooms. This holds whether the claimant and partner are supplied or inferred.

2. Exact effect. When the added person is not the claimant or partner, these changes happen and nothing else:

   • Their family gains one HB bedroom, and one UC bedroom unless they are a qualifying young person no one is responsible for.
   • If they join a non-dependant's family, the claim counting the household's non-dependants gains one UC bedroom (with the same exception), and every family liable for the rent gains one HB bedroom.
   • If they join a boarder's or lodger's family, every family liable for the rent gains one HB bedroom.

3. Reference agreement (differential). Every family's bedrooms equal an independent count of the size criteria under both schemes:

   • a couple among the HB occupiers counts once;
   • children's rooms are the fewest that hold them two to a room (same sex, or both under 10), found by brute force, pooled across the HB occupiers.

   This also checks child_bedrooms' greedy pairing against the optimum on every generated family.

The corrected reference property fails on #2010's previous formulas, which counted an occupier couple as two HB bedrooms.

Tests

• LHA_young_person_bedroom.yaml, 9 cases at 2026:

  • qualifying young people aged 17 and 19;
  • an 18-year-old non-dependant;
  • a looked-after qualifying young person (no UC bedroom);
  • a couple with a young person, a boy of 12 and a girl of 8 (4 bedrooms, E);
  • a partner aged 19 (no extra bedroom);
  • a pension-age couple's grandchild at college under HB (SPC), where HB meets a £200-a-week rent in full;
  • a looked-after young person in a non-dependant's family (adds no UC bedroom for the household head);
  • a non-dependant couple (UC 3 bedrooms, HB 2).

  The £ expectations use a rent of £200 a week. That sits between the published Maidstone one- and two-bedroom rates (VOA LHA April 2024, held for 2025-26: B £172.60, C £208.27), so the expectations do not depend on the model's rate table.

• LHA_claimant_composition.yaml: two cases asserted "existing bedroom formula = 1: B", for a 17-year-old apprentice non-dependant and an 18-year-old qualifying young person. Both are now C, citing para 10(1)(c) and (b).

• Full YAML suite (policyengine-core test policyengine_uk/tests/policy -c policyengine_uk): 1,518 passed on 6e6234d. On 2a88024, which adds only Give sharers, boarders and lodgers their own rent, tenure and LHA category #2006's merge of main, the three bedroom and rent files (40 cases) pass, and CI runs the full suite once this PR targets main.

• Full pytest (pytest policyengine_uk/tests --ignore=policyengine_uk/tests/microsimulation) on 6e6234d: every one of the 2,069 collected tests ran, with 0 failures and 0 errors (19 skipped). It took three invocations, because the host's tool time limit cut off the first two.

• Hypothesis properties (pytest policyengine_uk/tests/test_lha_bedroom_young_people_properties.py) on 2a88024: 3 passed.

• Formatting: ruff format and ruff check on the changed files are clean.

Impact (real microsimulation)

The data is the released enhanced FRS 2024-25 (policyengine-uk-data 1.57.4, enhanced_frs_2024_25.h5, sha256 03fe15e4…), for 2026. The base is #2006's head, 7769fa3; the branch is 2a88024. Both runs use the same dataset. Earlier runs against #2006's pre-main heads gave the same bedroom counts and UC within £1m.

Every figure below has at least 10 distinct survey households contributing to it, and no suppressed cell can be recovered by subtracting published ones.

• Survey households are counted by their rent and members' ages. Every enhanced-FRS record shares these with its FRS 2024-25 source household.
• Collisions only merge households, so the counts are lower bounds.
• Money is rounded to whole £m.

	2026
Benefit-unit members aged 16+ who are not the claimant or partner	1.97m people, of whom 1.95m are UC qualifying young people and 19k are non-dependants
Benefit units whose UC bedrooms rise	1.80m; none falls
… of which privately renting	363k: B→C 56k, C→D 216k, D→E 58k, other 34k
UC housing costs element, as computed for every renting benefit unit	+£577m
Universal Credit paid (151k benefit units gain)	+£272m
Housing Benefit	no change (#2006 already gives the HB bedroom)
Household net income	+£272m
People in poverty (BHC and AHC)	changes in fewer than 10 survey households: suppressed

About half of the element's rise is in benefit units with no UC award, where it changes nothing. Within UC awards, the remaining gap reflects income means-testing for newly positive awards, the benefit cap and UC deductions.

In fewer than 10 survey households, the LHA cap falls when the bedroom count rises. This is because in 5 of 200 BRMAs the model's rate table prices a larger category below a smaller one; #2022 (published LHA rates) addresses that.

Known gaps (other PRs)

• In-unit non-dependants now have a bedroom but no non-dependant deduction: Charge non-dependant deductions for non-dependants within the claimant's benefit unit #2017, stacked here.
• UC does not yet count children of a non-dependant's family, and foster and adoption placements and the foster carer's bedroom are not modelled: Count other families' children and add the foster and overnight-care bedrooms in the LHA size criteria #2016, stacked here. (Give sharers, boarders and lodgers their own rent, tenure and LHA category #2006 already pools other occupiers' children for HB.)
• The model's LHA rates come from its 2020 list of rents, uprated, rather than the published rates: Use published LHA determinations and apply Sch 3B paras 3 and 3A #2022.

Review

The first independent review (Subfleet, Codex lane) of 7c6f5f3 requested changes, and all five findings are fixed:

• the PR text published cells that could be recovered by subtraction from groups of fewer than 10 survey households;
• the UC property oracle counted a looked-after young person in a non-dependant's family;
• the HB oracle counted an occupier couple as two bedrooms;
• two corrections to the rulespec-uk#406 dispatch note.

The model now matches the first two oracle findings: a code fix for UC, and #2006's HB rewrite for the couple. A second review of 2a88024 is running.

Documentation

The variable's documentation field describes the new rule. The Housing Benefit docs notebook does not cover the size criteria.

axiom: uk/regulations/uksi/2013/376/schedule/4/paragraph/10 (UC Sch 4 para 10(1)(b), (c)) encoded-correct, companion cases in TheAxiomFoundation/rulespec-uk#408 (merged) | HB Regs 2006 / HB (SPC) Regs 2006 reg 13D(3)(a), (b): TheAxiomFoundation/rulespec-uk#406 queued (addendum: TheAxiomFoundation/rulespec-uk#406 (comment))

🤖 Generated with Claude Code

Reproducer: policyengine_uk/tests/policy/baseline/finance/benefit/family/LHA_claimant_composition.yaml::A 17-year-old apprentice in the renter's benefit unit is a non-dependant, policyengine_uk/tests/policy/baseline/finance/benefit/family/LHA_claimant_composition.yaml::Lone parent with an 18-year-old qualifying young person, policyengine_uk/tests/policy/baseline/finance/benefit/family/LHA_young_person_bedroom.yaml, policyengine_uk/tests/test_lha_bedroom_young_people_properties.py::test_adding_a_person_aged_16_to_19_never_lowers_bedrooms, policyengine_uk/tests/test_lha_bedroom_young_people_properties.py::test_exact_effect_of_adding_a_person_aged_16_to_19, policyengine_uk/tests/test_lha_bedroom_young_people_properties.py::test_bedrooms_match_an_independent_count_of_the_size_criteria

[MaxGhenis]

Stacking change from #2017 (in-unit non-dependant deductions). #2017 is no longer stacked on this PR. It now sits on #2007 (hb-ctr-non-dep-exemptions), because it needs #2007's HB and CTR claimant exemptions and nothing of #2010's. Merge commit 7f9ef1a restores #2010's own files to main's versions on #2017's branch.

There is one cross-PR invariant that whichever of #2010 and #2017 lands second should carry. #2017's property file had it until 7f9ef1a:

• every member that is_benefit_unit_non_dependant_for_universal_credit marks (16+, not claimant or partner, not a UC qualifying young person) has a bedroom in the size criteria (UC Sch 4 para 10(1)(c));
• so turning those members into qualifying young persons leaves LHA_allowed_bedrooms unchanged.

This ties the deduction and the bedroom to one definition. The test body (Hypothesis, population generator from test_benefit_unit_non_dependant_properties.py) is saved at ~/reviews/in-unit-non-dep-deductions-2026-10-01/handover/bedroom_invariant_for_2010.py:

in_unit = calc(sim, "is_benefit_unit_non_dependant_for_universal_credit")
rooms = calc(sim, "LHA_allowed_bedrooms")
as_dependants, *_ = run(population, {"non_dependant": "dependant"})
assert np.array_equal(rooms, calc(as_dependants, "LHA_allowed_bedrooms"))

If #2010 lands first, #2017 will add it back. If #2017 lands first, #2010 should add it, since test_benefit_unit_non_dependant_properties.py will be on main by then.