Scottish private-rent targets: PIPR is BRMA-grain, so council cells cannot bind by translation

[MaxGhenis]

The gap

ONS's Price Index of Private Rents (PIPR) publishes Scottish rent levels for the 18 Broad Rental Market Areas (codes S33000001–S33000018), not for the 32 council areas. Microcosm's ons.rent.private_rent target is bound at local-authority grain, so all 32 Scottish cells are signed deferred (private_rent_pipr_scotland_brma_grain), and the parity register routes the concern cross_grain_private_rent_scotland_brma. Its fence says to keep the cells deferred "until a source-backed BRMA-to-LA bridge with an explicit aggregation rule exists". #736 lists this gap with no issue of its own.

A census crosswalk now exists (built for PolicyEngine/policyengine-uk-data#516 and #1087), so the question is whether a rule such as "household-weighted mean of BRMA rents per council" can lift the deferral.

What the overlap looks like

Built from Scotland's Census 2022 and the official BRMA geography (2022 output-area centroids in the Scottish Government's BRMA polygons, corrected by Rent Service Scotland's postcode lookup, FOI 202300368850):

• 63 BRMA × council intersections hold 322,995 private-rented households.
• 18 of the 32 councils fall in more than one BRMA, and 16 of the 18 BRMAs cover more than one council.
• 29 councils and 17 BRMAs form one connected block; the three Ayrshire councils and the Ayrshires BRMA are the only group that separates. No finer grouping makes BRMAs a union of whole councils. Much of the joining is slivers: dropping intersections under 1% of both sides leaves 11 blocks, one of which still holds five BRMAs and seven councils.
• Four councils are the same place as one BRMA to within 1% of private-rented households in both directions: Dumfries and Galloway, Fife, West Lothian and Scottish Borders.

What the weighted-mean rule would get wrong

For a council that sits inside one BRMA, the rule returns that BRMA's mean unchanged, so every council sharing a BRMA gets the same rent. Rent Service Scotland's market evidence for the year to September 2021 (FOI 202200303624: 31,833 lettings with BRMA, rent and postcode district) shows how far that is from the councils' own rents:

Council	BRMA	Rule ÷ council's own mean
Midlothian	Lothian	+30.9%
Clackmannanshire	Forth Valley	+26.8%
East Lothian	Lothian	+23.1%
Falkirk	Forth Valley	+20.5%
Stirling	Forth Valley	−15.2%
Orkney Islands	Highland and Islands	+13.6%
Na h-Eileanan Siar	Highland and Islands	+13.3%
East Renfrewshire	Greater Glasgow / Renfrewshire and Inverclyde	−12.5%
Angus	Dundee and Angus	+11.1%

Across the 31 councils with at least 30 records, the median absolute gap is 3.3%, 12 exceed 5% and 9 exceed 10%. The four coincident councils are within 0.1%.

Proposed outcome

1. Commit the overlap as a source-backed resource with a rebuild tool and tests.
2. Record that no rule turning BRMA rents into council targets satisfies the fence: any such rule gives a council a value that only BRMA averages determine, which is the allocation the fence forbids. CrossGrainBridge cannot carry it either: it equates additive totals over a partition of legs, and a mean over overlapping geographies is neither.
3. Keep the 32 cells deferred, with the reason updated: the crosswalk exists; the missing piece is a target grain, not a translation.

Routes left open

• BRMA-grain target rows. Keep PIPR's geography on the target side (12 × BRMA mean × BRMA private renters) and aggregate the model side with the overlap shares. This uses all 18 published rents and allocates nothing, but AREA_TYPES is ("constituency", "la"), so it needs a new area type through the rowwise surface.
• A signed exception for the four coincident councils. Their BRMA rent is their own rent to within about 2% in the worst case. Binding them means letting a fact on the BRMA boundary set reach a local-authority cell, which the boundary-frame gate added in Adopt the UK local-geography and firms contracts and compile the local target surface (#708, #759) #795 refuses by design.