Model rent from boarders, lodgers and sub-tenants for the householder

[MaxGhenis]

Summary

Stacked on #1995, which stops counting income from capital in the legacy means tests and adds legacy_benefits_home_letting_income for rent from letting part of the home.

This PR models the rent a householder receives from boarders and lodgers who live in the household. The Family Resources Survey records that rent only on the payer, as CVPAY (PolicyEngine/policyengine-uk-data#506 adds it as rent_paid_as_boarder and rent_paid_as_lodger). It also routes sublet_income (rent from a sub-tenant outside the household) through income tax.

What changes:

• Who receives it. rent_from_boarders_and_lodgers gives the household head the rent paid by household members outside the head's benefit unit.
• Income tax: rent-a-room relief. Receipts up to £7,500 a year are not taxed. Above that, the excess is taxed as property income, assuming the s. 800 election. The limit is halved where someone else in the household also receives rent from the home. The receipts get no £1,000 property allowance.
• Legacy means tests. Income Support, Housing Benefit, Pension Credit and council tax reduction count the rent less the statutory disregards:
  • boarders: £20 a week plus half the excess, for each boarder;
  • lodgers: £20 a week, per person at pension age and per lodger family at working age (£4 before April 2008).
    The tax on counted letting income is deducted, like the tax on other counted income.
• Tax on rent a programme does not count is not deducted. UC earned income no longer deducts the tax on rent-a-room income (UC Regs reg 55(5)(b) allows only tax in respect of the employment). Council tax reduction does the same where it excludes the rent: the Scottish working-age scheme (Treat income derived from capital as capital in the legacy means tests #1995) and, new here, Welsh applicants with a UC award, whose income is the UC assessment (WSI 2013/3029 Sch 6 para 9; anyone with a UC award is not a pensioner, reg 3).
• Tax credits count the taxable rent-a-room income.
• Universal Credit does not count it. Nothing changes there.
• Non-dependants. A boarder or lodger pays the householder on a commercial basis, so is no longer a non-dependant for UC, HB or CTR (all local CTR schemes take this flag).
• Household income. Rent paid between members of a household is not added to household market or net income. sublet_income, which comes from outside the household, is now counted in market income, household market income and HBAI net income, alongside property_income.

The law

Read from the revised text on legislation.gov.uk on 1 October 2026. An independent legal memo, which checked 284 quotations against saved sources, agrees with every rule below.

Regime	Rule	Provisions
Income tax	Rent-a-room receipts are "in respect of the use of furnished accommodation in a residence ... or in respect of goods or services supplied in connection with that use", in the individual's only or main residence. Not charged up to the limit. Above it, on an election, the receipts less the limit are charged, with no expenses deducted. Limit: £3,250 from 1992-93, £4,250 from 1997-98, £7,500 from 2016-17; halved where the exclusive receipts condition fails. The receipts are not relievable receipts for the property allowance.	ITTOIA 2005 ss. 784-802, 783BB(2); SI 2015/1539 arts. 1(2), 2; SI 1996/2953 art. 2; F(No. 2)A 1992 Sch. 10
Universal Credit	Rent is not unearned income. The list in reg 66(1) is closed, and (m) covers only income taxable under ITTOIA Part 5. ADM H5112: rent from "a sub-tenant or boarder is not defined as income for the purposes of UC".	SI 2013/376 regs 66, 72(3)
Housing Benefit, working age	Counted as income other than earnings, less: per boarder, £20 plus 50% of the excess (Sch 5 para 42); other occupier, £20 on the aggregate paid by "that person or a member of his family" (para 22, £20 from 7 April 2008, SI 2007/2618 reg 11(12)). Tax on counted income is disregarded (para 1).	SI 2006/213 reg 40, Sch 5 paras 1, 21, 22, 42
Housing Benefit, pension age	Board and lodging, and "any payment of rent made to a claimant who ... occupies part of the property", are income. Disregards: per boarder £20 plus 50%; £20 of "the amount paid by that person".	SI 2006/214 reg 29(1)(p), (v), Sch 5 paras 9, 10
Pension Credit	Prescribed income, with the same disregards as pension-age HB.	SPCA 2002 s. 15(1)(j); SI 2002/1792 reg 15(5)(e), (i), reg 17, Sch IV paras 8, 9
Council tax reduction	The same words in England (pensioners, and the working-age default), Wales and Scotland (pension age). The Scottish working-age scheme's closed income list has no such income.	SI 2012/2885 Sch 1 para 16(1)(p), (v), Sch 5 paras 9, 10; SI 2012/2886; WSI 2013/3029; SSI 2012/319; SSI 2021/249 reg 57
Income Support	£20 plus 50% per boarder since 9 April 1990 (SI 1990/547 reg 22(c); current words SI 1994/527 reg 9(4)); £20 per occupier and family (£4 before April 2008).	SI 1987/1967 Sch 9 paras 1, 19, 20
Tax credits	Property income disregards profits sheltered by rent-a-room relief.	SI 2002/2006 reg 11(2)
UC earned income	Deducts "income tax ... paid in respect of the employment", not tax on property receipts.	SI 2013/376 reg 55(5)(b)
CTR, Wales, UC recipients	Income is the Secretary of State's UC assessment; a person with a UC award is not a pensioner.	WSI 2013/3029 reg 3, Sch 6 para 9
Non-dependants	"a person who is liable to make payments on a commercial basis in respect of the person's occupation of the accommodation" is not a non-dependant.	SI 2013/376 Sch 4 para 9(2)(d); SI 2006/213 reg 3(2)(e)(i), (3); SI 2012/2885 reg 9(2)(e), (3)

Modelling choices

• The receipt goes to the household head (the FRS household reference person in the data), split equally if more than one person is flagged as head. The FRS asks the CVPAY questions only about people not related to the HRP, so payers are unrelated to the householder. The model treats every such arrangement as commercial.
• Rent-a-room above the limit assumes the s. 800 election. The alternative is receipts less actual expenses, which the data cannot measure. All receipts are treated as furnished accommodation.
• The exclusive receipts condition is taken to fail when two members of a household both receive rent-a-room receipts.
• Tax on counted letting income (rent_a_room_income_tax) is the share of property income tax that falls on taxable rent-a-room income, pro rata, because nothing in the law orders the two. legacy_means_test_income_tax (Treat income derived from capital as capital in the legacy means tests #1995) shares Step 6 reductions pro rata between tax on counted income and tax on income from capital; this PR moves the rent-a-room share from the capital side to the counted side.
• Uprating. The two inputs use per-capita GDP, as property_income and sublet_income do. No national private rent index exists as a parameter.
• Tax after reductions. rent_a_room_income_tax_after_reductions is that tax less its pro-rata share of Step 6 reductions; UC and the CTR routes that exclude the rent deduct income tax less this amount.
• Weeks per year. The inputs are annual; the weekly disregards divide them by 52. policyengine-uk-data annualises weekly FRS amounts with 365.25 / 7, so a weekly payment reaches the disregards about 0.34% higher. The same gap applies to every weekly-tested FRS amount (Weekly FRS amounts are annualised with 365.25/7 but policyengine-uk's weekly rules divide by 52 policyengine-uk-data#531).
• Sub-tenants. sublet_income keeps Treat income derived from capital as capital in the legacy means tests #1995's single-occupier disregard. The policyengine-uk-data change that moves SUBRENT out of property_income and into sublet_income is separate work.

Invariants and tests

test_boarder_lodger_rent_properties.py (Hypothesis) generates households with one or two members in the head's unit and up to two paying units of one or two boarders or lodgers. It varies working and pension age, sublet income, tenure and region, for 2025 and 2026. It checks:

1. Conservation. The head receives exactly what members outside the head's unit pay, and nobody else receives anything.
2. Rent-a-room bounds. 0 ≤ relief ≤ receipts; the limit is £3,750 exactly when the person and someone else in the household both have receipts, and £7,500 otherwise; taxable income = max(0, receipts − limit).
3. Closed form for the disregards. Counted home-letting income equals the statutory arithmetic, computed independently in the test, and lies between nil and the rent received.
4. Monotone. More rent from an existing payer never lowers taxable rent-a-room income or counted home-letting income, and never raises IS, HB, Pension Credit or CTR.
5. Boarders keep more. At pension age, the same amount counts no more as board and lodging than as lodging only.
6. Invariance. UC unearned income and household market income do not depend on rent paid between household members.
7. No tax on uncounted rent. UC earned income, and CTR income under the Scottish working-age scheme and the Welsh UC route, do not change when the rent changes.
8. Tax bounds. 0 ≤ legacy_means_test_income_tax ≤ income tax.

YAML tests (hand-worked from the provisions) cover:

• Income tax: within and above the limit; the 2015-16 limit; the halved limit; the property allowance not applying; a payer inside the head's unit; tax credits.
• Disregards: a lodger, a boarder, a boarder under £20, a lodger couple at pension age (£40 disregarded) and working age (£20); a boarder and a lodger together; the 2007 £4 lodger disregard; UC unchanged; household market income unchanged; tax on counted letting income deducted, including when a tax reduction is shared pro rata with tax on capital income; UC earned income not reduced by tax on the rent; Welsh CTR for a UC recipient without the rent.
• Non-dependants: a paying lodger and a non-paying adult, under UC, HB and CTR.

#1995's tests (YAML and property) pass on the merged head, as do the CTR Pension Credit property tests.

Mutation check (run at 298a8ccba, before the merge of #1995's 89bdbeed9; the rules it mutates did not change in that merge). Each mutant below was applied to a copy of this branch, then the 32 YAML cases and the property tests were run against it (property tests with shrinking off).

Mutant	YAML cases failing	Property tests
Count all of a boarder's excess (drop the 50%)	3	caught
Lodger disregard per person at working age	2	not caught
Lodger disregard per family at pension age	1	not caught
Never halve the rent-a-room limit	1	not caught
Apply the property allowance to rent-a-room receipts	1	not run
Keep boarders and lodgers as UC non-dependants	1	not caught
Keep boarders and lodgers as HB non-dependants	1	not caught
Keep boarders and lodgers as CTR non-dependants	1	not caught
Do not deduct tax on counted letting income	1	not caught
Give the receipt to everyone in the household	7	caught
Count rent between members as household market income	1	caught
Tax credits count all rent-a-room receipts	1	not caught

The YAML cases catch all 12. At 10 examples per test, the property tests catch the three mutants that break an invariant on most households. The other nine change a rule only for particular household shapes, so the YAML cases cover those. A twelfth mutant first tried, moving the rent-a-room term inside the max in taxable_property_income, turned out to be equivalent to the original, because property_allowance_deduction never exceeds property_income. It was replaced by the property-allowance mutant above.

Measured impact

One real 2026 Microsimulation per model version, on the same dataset:

• Dataset. An enhanced FRS 2024-25 built at policyengine-uk-data#506 87a5f50 with production settings and seed 0. It is identical, in every column of every table including household weights, to a seed-0 build of its parent (Add Tax Credits statistics to validation page #503 head 0812cd1), apart from the two new inputs. Its later commit ee40d7a adds uprating rows and tests and changes nothing the locked build model reads.
• Models. Baseline: Treat income derived from capital as capital in the legacy means tests #1995 head 89bdbeed9. This PR: 6df59cdb4.

Disclosure. Each enhanced-FRS record of a paying household was matched to its survey household by its members' ages and FRS rent amounts. Every change falls inside those households. Each figure below rests on at least 10 survey households. Figures resting on fewer are not reported, and neither are figures that would reveal one by subtraction. For the same reason, no survey-household counts are given.

2026-27	Change
Rent received from boarders and lodgers	+£185m
of which relieved by rent-a-room relief	£149m
Taxable rent-a-room income	+£36m
Income tax	+£8.5m
Home-letting income counted by the legacy means tests	+£135m

• Low weights in the calibrated data. The survey-weighted FRS has £0.81bn of this rent a year. The calibrated enhanced FRS gives the paying households low weights, so it carries £0.19bn.
• No change in Housing Benefit, council tax reduction, Income Support or tax credit awards, Universal Credit unearned income, household market income, AHC poverty, or child poverty (BHC or AHC).
• Not reported. Each of these changes in fewer than 10 survey households: UC and Pension Credit awards, Guarantee Credit entitlement, total household benefits, and BHC poverty for all people. Household net income is not reported, because together with income tax it would reveal the benefit change. Non-dependant deductions are computed for every family whether or not it claims; their award effects are the ones above.

Not in this PR

• The payer's own housing costs. Benefit units after the first still have no rent in the model, so a boarder's or lodger's UC housing element and HB are not modelled. Separate work is in progress on this.
• SUBRENT. Moving SUBRENT into sublet_income in the data is separate.
• Second adult rebate and Pension Credit housing costs, where the same commercial-occupier rules apply, are not modelled here.

Known interactions

• Who claims CTR. This PR gives the rent to the household reference person (is_household_head). The CTR award gates and non-dependant rules still find the claimant as the oldest adult (benunit_contains_household_head), so where a lodger is older than the householder the two differ. That predates this PR; Read one resolved household head in every programme #2062 makes every programme read one resolved household head.
• UC tax deduction. UC earned income still deducts all of a member's income tax apart from rent-a-room tax (UC earned income deducts the whole benefit unit's tax, including tax on pensions and other unearned income #1942); Deduct only each person's own tax and NI on earnings from UC earned income #1949 limits it to tax on earnings, and its merge with this PR should drop the rent-a-room subtraction here.
• Close relatives. The non-dependant exclusion assumes every payment is commercial and the payer is not a close relative living there. That holds for the survey data, which asks about this rent only for people unrelated to the household reference person; inputs set by hand are not checked.

axiom: TheAxiomFoundation/rulespec-uk#402 queued (UC reg 66 is encoded-correct: uk/regulations/uksi/2013/376/66.test.yaml rent_from_a_lodger_in_the_home_is_not_unearned_income)

🤖 Generated with Claude Code

[MaxGhenis]

Coordination note from #2062 (issue #2059), which makes every programme read one household head.

This PR's legacy_benefits_home_letting_income, rent_from_boarders_and_lodgers and pays_rent_to_householder read the raw is_household_head input. That input can flag several members of a household, or none, and programmes reading it directly then disagree about whose household it is. #2062 adds is_resolved_household_head (exactly one per household) and defines benunit_contains_household_head from it.

Whichever of the two PRs lands second needs these readers switched:

• family level: benunit.any(person("is_household_head", period)) becomes benunit("benunit_contains_household_head", period), and person.benunit.any(person("is_household_head", period)) becomes person.benunit("benunit_contains_household_head", period);
• person level: person("is_household_head", period) becomes person("is_resolved_household_head", period).

#2062 adds a test, test_only_the_resolved_head_reads_the_flag, that fails while any module other than is_resolved_household_head reads the raw flag. If this PR lands first, #2062 makes the switch when it merges main. If #2062 lands first, the switch belongs here. Survey results do not change either way: the Enhanced FRS has exactly one flagged head per household.

[MaxGhenis]

#2062 is now on main (merge commit c4659aa). main has one resolved household head: is_resolved_household_head (Person: the eldest member the is_household_head input flags, else the eldest member) and benunit_contains_household_head (BenUnit, defined from it). No formula outside is_resolved_household_head reads is_household_head, and test_only_the_resolved_head_reads_the_flag in policyengine_uk/tests/test_household_head_agreement_properties.py enforces that.

This PR still adds formulas that read the raw flag:

• policyengine_uk/variables/gov/dwp/legacy_benefits_home_letting_income.py: head = person("is_household_head", period)
• policyengine_uk/variables/household/income/pays_rent_to_householder.py: in_head_benunit = person.benunit.any(person("is_household_head", period))
• policyengine_uk/variables/household/income/rent_from_boarders_and_lodgers.py: head = person("is_household_head", period)

Each needs to read the resolved head instead:

• family level: benunit.any(person("is_household_head", period)) becomes benunit("benunit_contains_household_head", period), and person.benunit.any(person("is_household_head", period)) becomes person.benunit("benunit_contains_household_head", period);
• person level, only where the formula needs the head as a person: person("is_household_head", period) becomes person("is_resolved_household_head", period).

test_only_the_resolved_head_reads_the_flag fails until every such read is moved. Please also add each variable you move to OUTPUTS in that test file, so its metamorphic test (flags matter only through the resolved head) computes it. Test inputs that set is_household_head are fine as they are.