UK private rent at the assigned area (#1123 item 4, "PR B"): what is in place and the pieces still missing #1262

Context

#1123 item 4: a household placed in a local authority keeps the rent of its source FRS household, which describes its region and not its district. No reweighting inside a region's FRS rents reaches the mean rent of the dearest and cheapest districts.

#1198 ("PR A", merged 2026-10-12 as 977ac983a) fixed items 1, 2, 6 and 7 of #1123 on the target side. It left rent to a second change, "PR B". PR B has no branch and no code. This issue lists what exists and what is missing, so the work can be cut into pieces.

What is in place

• The inputs are in the pinned feed. Add UK 2025 calibration publisher inputs chronicle#314 (which closed UK local-surface facts for the 2025 calibration: household counts by nation and area, communal-establishment residents by age, mid-2025 population estimates, UC households by area for calendar 2025, and BRMA geography and rent inputs (PolicyEngine/microcosm#1123, PolicyEngine/microcosm#1090) chronicle#313) is the commit uk/chronicle_feed.json pins since One consistent UK target surface: enforced cross-grain reconciliation, declared uprating holds, nation household controls and the 942a1fa re-pin (#1123 PR A) #1198. It holds:
  • Raw-only registrations: BRMA boundaries for England (VOA, May 2020), Wales (Rent Officers Wales, May 2012) and Scotland (2009, revised 2015); Rent Service Scotland's postcode-to-BRMA lookup; the population-weighted centroids of the 2021 LSOAs (England and Wales) and of Scotland's 2022 output areas; and three lists of rents: England (VOA shadow list April 2026, 502,239 rents, October 2024 to September 2025), Wales (ATISN 25142, October 2022 to September 2023) and Scotland (market evidence, 2016 to 2021).
  • Facts: PIPR September 2026 edition (January 2025 to August 2026) for all dwellings, four bedroom categories and four property types, for the 316 districts of England and Wales, the 18 Scottish BRMAs, the nations and the nine regions; and the Scottish Government's private sector rent statistics 2025 (mean, median and quartiles by BRMA and bedrooms).
• The BRMA draw uses census private renters (Weight UK BRMA assignment by census private-rented households #1087, merged 2026-10-09): frs_brma draws each benefit unit's BRMA within its region in proportion to census private-rented households of the matching size.
• The target is consistent with the reconciled surface (One consistent UK target surface: enforced cross-grain reconciliation, declared uprating holds, nation household controls and the 942a1fa re-pin (#1123 PR A) #1198): each of the 316 bound cells of ons.rent.private_rent is 12 × the calendar-2025 PIPR mean × the authority's reconciled private-renter households. Its route is no_higher_control, and the declaration names this work.
• Scotland's BRMA × council overlap is measured (Measure Scotland's BRMA × council overlap; keep BRMA rents out of council targets #1092, open): the weighted-mean rule cannot turn BRMA rents into council targets, so the 32 Scottish cells stay deferred (Scottish private-rent targets: PIPR is BRMA-grain, so council cells cannot bind by translation #1090).

What has not changed

• Nothing writes rent after a household is placed. Two stages write it: the FRS spine (hhrent × 52, in frs_spine.py) and the SPI housing shell (a QRF draw from FRS donors, in spi_housing_shell.py). Both run before the geographic clones exist. regional_uprating moves property values only.
• A household's BRMA is not derived from its area. frs_brma is a spine stage: it draws within the region, before cloning, so every clone of a household carries the same BRMA, whatever district it is placed in. The LHA rate therefore describes a different place from the household's local authority and constituency.
• The size of the gap, from the publisher's side. On the calendar-2025 PIPR means, a district's mean rent divided by its region's published mean spans 0.63 to 1.61 across the 316 bound districts, and 49% of them are more than 15% from their region's mean. The two PIPR editions in the feed give the same 2025 means for all 316.
• The model side is item 4 of UK local build (K=25, 2026-10-06): data-quality defects on the target surface and the spine #1123 (dense K=25 solve of 2026-10-06): private rent is the weakest local family (91% of its cells within 10%), and 13 English districts fail after selection. There is no local build on main since One consistent UK target surface: enforced cross-grain reconciliation, declared uprating holds, nation household controls and the 942a1fa re-pin (#1123 PR A) #1198, so there is no current control.

Missing pieces

A. Decisions before design

1. Method. Preferred on 2026-10-07: re-draw each private renter's rent from the list of rents of the BRMA that holds its assigned area. The alternative is to keep the FRS rent and scale it by the district-to-region ratio. Record the choice here.
2. Direction, to agree with @MaxGhenis. Draw private renters' BRMAs given their rent and bedrooms policyengine-uk-data#521 (draft) goes the other way: it keeps the reported rent and draws the BRMA given rent and bedrooms. That fits a file with no local geography. The local line places the household first, so the BRMA follows from the area and the rent must follow from the BRMA. The branch uk-brma-clone-spread (no PR) spreads BRMAs across clones by LHA rate order, independent of the area; a BRMA derived from the area would replace it for placed households. UK: track the defects uk-data is fixing that microcosm still has #1095 lists the port of the rent-conditioned draw with a ruling still needed.
3. Level anchor. The lists are rent-officer evidence for LHA (lets in payment). Outside London their BRMA means are below the PIPR means of the main district in the same market: Brighton and Hove £1,534 against £1,822 a month, Bristol £1,451 against £1,786, Central Greater Manchester £1,196 against Manchester's £1,316. In Central London the BRMA mean (£3,919) is above Kensington and Chelsea's PIPR mean (£3,633). The list means are for self-contained homes (categories B to E; weekly rent × 52 ÷ 12) on the England list of 2024-25; the PIPR means are calendar 2025. The BRMA and the district are different areas, so these are indicative. Options: draw the list rents as they are; take the list's shape and set the level from PIPR by district and bedrooms; or take the list's shape and keep the FRS level by region.
4. Circularity. PIPR, the lists and the Scottish statistics share one evidence base. If PIPR sets the level, ons.rent.private_rent fits by construction. Decide: keep it bound and say so, or make it a validation row.
5. Scope by nation.
   • England: list of 2024-25.
   • Wales: the list is two years older than England's; it needs a declared carry-forward (PIPR for Wales is the candidate).
   • Scotland: letting-level evidence to 2021, plus the 2025 quartiles by BRMA and bedrooms.
   • Northern Ireland: no BRMA boundaries under an open licence and no list of rents. It stays on FRS rents and the current draw unless a source appears.
6. Who is re-drawn. Private renters with a positive rent only, or also the SPI-channel households whose rent comes from the QRF. Social renters are out of scope. The lists use LHA categories A to E; Draw private renters' BRMAs given their rent and bedrooms policyengine-uk-data#521 maps B to E to the home's bedrooms (num_bedrooms). Category A (a room in shared accommodation) needs its own rule.
7. Joint distribution. A free draw breaks the link between a household's rent and its income, benefits and bedrooms. A draw that keeps each household's rank inside its source cell (region × bedrooms) preserves it. Choose one; the second is my recommendation.
8. The national line. A build with no geography keeps FRS rents, so the national and local files would carry different rents for the same source household. Accept that and state it, or define the national counterpart.

B. Build pieces

1. An output-area-to-BRMA layer.
   • A tool that builds it from Chronicle's raw registrations: England and Wales by the LSOA centroid in the BRMA polygons, Scotland by the output-area centroid corrected with the postcode lookup.
   • A packaged resource with sidecar provenance, like the atomic-area supports.
   • One method and one copy: it must agree with scotland_brma_la_overlap.json (Measure Scotland's BRMA × council overlap; keep BRMA rents out of council targets #1092) and with the BRMA table of Weight FRS BRMA draws by census private-rented households policyengine-uk-data#516. Which copy each consumer reads is still to agree with Max, as UK local-surface facts for the 2025 calibration: household counts by nation and area, communal-establishment residents by age, mid-2025 population estimates, UC households by area for calendar 2025, and BRMA geography and rent inputs (PolicyEngine/microcosm#1123, PolicyEngine/microcosm#1090) chronicle#313 section 5 says.
   • A check that every BRMA name it emits is one the engine's LHA rate table knows.
2. brma derived from the assigned area. For a placed household, brma becomes a layer of its output area, like its local authority and constituency. BRMA, rent and the LHA rate then describe one place. Northern Ireland keeps the regional draw.
3. Rent distributions by BRMA and category.
   • Built in microcosm from the raw lists: Chronicle emits no facts from record-level rent evidence.
   • Committed as a derived table (quantiles by BRMA × category) with provenance and a rebuild check, with a pooling rule for thin cells.
   • Each list moved to the calibration year by a declared index.
4. The stage. A node after uk.full.geography.local_authority that owns rent as a rewrite:
   • identity-keyed draws, so the result does not depend on row order or K;
   • declared cells and dtypes, the kernel hash, the stage roster's derived surfaces, the spine graph parity fixture and the contract mirror in microcosm-data.
5. Columns that depend on rent. List every stored column that was drawn or reported together with rent (housing benefit reports, service charges, the boarder and lodger rents of UK: port the remaining uk-data fixes tracked in #1095, engine items included #1121, subrent) and state for each whether it moves with the new rent or stays.

C. Target side (left by #1198)

1. Bound or validation for ons.rent.private_rent, after decision A4.
2. PIPR by bedrooms and property type is in the feed and nothing reads it (ons.pipr_september2026.*). Use it as the validation receipt of the re-draw by district and bedrooms, or as the level anchor of A3.
3. Scotland's 32 cells (Scottish private-rent targets: PIPR is BRMA-grain, so council cells cannot bind by translation #1090). The BRMA-grain route needs a third area type on the local surface and private-renter counts by BRMA. The census input for the counts is still open in Finish Scottish Census 2022 tenure inputs for BRMA rent targets chronicle#326. Measure Scotland's BRMA × council overlap; keep BRMA rents out of council targets #1092 carries a test that fails when a BRMA grain appears, so that the concern is read again.
4. Northern Ireland's deferral text is out of date. private_rent_pipr_ni_absent says the feed has no Northern Ireland row at any level. The feed carries the Northern Ireland mean for all 12 months of 2025, and the means of the UK, Great Britain, each nation and the nine regions. Correct the text, and decide whether nation and region means become higher controls. A mean is not additive, so that needs a declared relation, not a sum bridge.
5. A stale label. The receipt of uk_private_rent_mean_to_total still names the renter count "A17-uprated". Since One consistent UK target surface: enforced cross-grain reconciliation, declared uprating holds, nation household controls and the 942a1fa re-pin (#1123 PR A) #1198 the count is drifted by SPREE in England and reconciled under the nation household controls.
6. Two English authorities stay absent in PIPR (City of London, Isles of Scilly). No action; listed for completeness.

D. Measurement

1. A control on the current base. A dense local solve on main after One consistent UK target surface: enforced cross-grain reconciliation, declared uprating holds, nation household controls and the 942a1fa re-pin (#1123 PR A) #1198, before any rent change. The 2026-10-06 figures carry the effects of every merge since.
2. A distribution receipt against the pre-move build (UK local build (K=25, 2026-10-06): data-quality defects on the target surface and the spine #1123's ask, and UK graph: move geography assignment ahead of enrichment (geography-first reordering) #931 item (b)): rent by nation, region and bedrooms, before and after.
3. The national rows that move: the UC housing element rows, obr.housing_benefit and the after-housing-cost rows. The reviewed exclusion on the UC housing element expires on 2026-10-15 (UK local build (K=25, 2026-10-06): data-quality defects on the target surface and the spine #1123 item 9).
4. UC renters above their LHA rate, against DWP. Draw private renters' BRMAs given their rent and bedrooms policyengine-uk-data#521's draw widened that overstatement by about 8 points (UK: track the defects uk-data is fixing that microcosm still has #1095). The same check applies here.
5. Disclosure. Model figures for a single district are survey-derived figures below region level. They need a disclosure check before they go in an issue, a PR or a receipt.

Proposed split

• B1: the area-to-BRMA layer and the derived brma (pieces B1 and B2). It moves the LHA rate only. It is the input of every later piece and of Scottish private-rent targets: PIPR is BRMA-grain, so council cells cannot bind by translation #1090.
• B2: the rent re-draw (B3 to B5, C1, C2, D). It needs decisions A1 to A8.
• B3: Scottish rent targets at BRMA grain (Scottish private-rent targets: PIPR is BRMA-grain, so council cells cannot bind by translation #1090, C3), after Finish Scottish Census 2022 tenure inputs for BRMA rent targets chronicle#326.
• Small fixes that can go with any of them: C4 (text) and C5.

Acceptance

• A placed household's brma is the BRMA of its assigned output area in Great Britain.
• A private renter's rent comes from its assigned area's rent distribution under the recorded decisions, with a receipt against the pre-move build.
• UK local build (K=25, 2026-10-06): data-quality defects on the target surface and the spine #1123 item 4 closes: after a dense solve on the current spine, no bound private-rent row is past 25% without a signed entry.
• The moves of the UC housing element, Housing Benefit and after-housing-cost rows are measured and stated.

Related

#1123 (item 4), #1198, #1090, #1092, #1087, #931 (item (b)), #1095, #1121. PolicyEngine/chronicle#313, PolicyEngine/chronicle#314, PolicyEngine/chronicle#326. PolicyEngine/policyengine-uk-data#516, PolicyEngine/policyengine-uk-data#521.