Repository navigation
Conversation
…yment No heating-fuel proxy lands near DESNZ's 1,842,760 delivered payments on a dataset with a 2023 year, so the model keeps paying nothing in Great Britain. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
State the size of the omission, quote the direction's account-holder test on 1 February 2023, and say what the 2023 datasets lack rather than what would fix it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Suppliers paid on the meters the government listed, whatever fuel the home burned, so the bullet now matches the section below it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…iable The formula stays the same. Only the description changes, to match the docs. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Hand-off to the UK hub at head edffb7e.
|
Aggregate impact on certified data: passBase Checks: 0 broken by this change, 1 failing on both sides, 0 fixed. Moves flagged for review: 0 (thresholds in Notes
Moves flagged for reviewNone. Spending by benefit and service
Caseloads
Standard reform costings (change in gov_balance)
Standard reforms: change in each reform's target
Household net income by decileNo household's net income changes in any year. No change in any year: receipts by tax head; government balance, each component counted once; model totals; household income totals; population; poverty rates; people in poverty; inequality; mean household net income in each side's own deciles. Checks failing on both base and head (1)
Counted-once checkEach of 92 components was raised by £1 per unit in 2026-27, and every head of its totals was read household by household: its own head must rise by exactly £1, no other tax head may move, no other benefit may rise, and the model's own total must move by what its heads moved. A head that breaks a rule is measured again twice: with every other head held at its baseline value, and with every other head held at its value after the first £1 while the component rises again. It fails where either still shows the response. 0 failed; 0 were skipped.
Provenance
Dataset: Receipts by tax head: head levels
Spending by benefit and service: head levels
Caseloads: head levels
Government balance, each component counted once: head levels
Model totals: head levels
Household income totals: head levels
Population: head levels
Poverty rates: head levels
People in poverty: head levels
Inequality: head levels
Mean household net income in each side's own deciles: head levels
Standard reform costings (change in gov_balance): head levels
Standard reforms: change in each reform's target: head levels
|
Fixes #2767
Nobody merges this until Vahid confirms the Budget-week freeze.
What and why
This PR adds no rule. On the docs page and in the description of
monthly_ebr_alternative_fuel_payment, it documents why the model leaves out Great Britain's £200 Alternative Fuel Payment: none of the proxies these datasets support lands near DESNZ's 1,842,760 delivered payments on a dataset with a 2023 year.DESNZ's end-of-scheme workbook records 1,842,760 delivered payments worth £368,551,840. Local authorities paid 89,750 more through the Alternative Fund.
Spending on the datasets
The enhanced FRS 2023-24 and Populace 2023 have no
liquid_fuel_consumptionorsolid_fuel_consumptioncolumn. Microcosm carries both but starts in 2024: a 2023 simulation on it returns a weight of 1 per household, 78,255 in all, every one in England. At 2024, Microcosm's spenders fall 83% short of DESNZ in the South West (63,000 against 366,280), 74% short in Scotland (74,000 against 286,120) and 77% short in Wales (44,000 against 190,380).How DESNZ chose the meters
Suppliers paid whoever held the account for each meter the Secretary of State named (direction, paragraph 6). DESNZ built that list from postcodes off the gas grid, minus areas with high use of electric heating (interim evaluation, page 54).
Spending in the donor survey
The LCFS finds spenders at 83% to 86% of DESNZ's count across 2021-22, 2023-24 and 2024-25, but partly the wrong households. About a third of spenders have gas central heating, and more than a third of households with alternative central heating report no purchase. Microcosm's draw then keeps 54% of the LCFS 2024-25 spenders.
Housing filters
Four no-gas filters by dwelling type, tenure or electricity use reach 86% to 118% of DESNZ's count on Microcosm, inside the bands I set after seeing the first counts: count within 20% of DESNZ, and a regional dissimilarity index of at most 0.20. None of them ships:
DESNZ's local authority table
The enhanced FRS carries area codes, but no variable reads them, and
local_authorityreturns Maidstone for all 53,508 records at 2023. A rule that read them could allocate DESNZ's table, and capping each authority at the file's households would lose 1.9% of payments. That allocation would still pay at random within each authority, a stand-in calibrated to DESNZ's own total, with nothing left to compare.Heating fuel data
In the LCFS, central-heating flags reach 79% to 91% of DESNZ's count, about where spending lands, and they record the fuel itself. Census 2021 counts 1,211,223 households in England and Wales on these fuels alone, 78% of DESNZ's count there, with a regional dissimilarity index of 0.06. DESNZ chose recipients by postcode, so a household fuel flag stays a proxy. A name search of the raw FRS 2023-24 tables finds no heating-fuel column, so policyengine-uk-data (or a 2023 Microcosm) would have to draw one from the LCFS or the census, and
monthly_ebr_alternative_fuel_paymentwould then gain a Great Britain branch. That draw could lose what Microcosm's lost: it kept 54% of LCFS spenders.Impact
None on model results. The model already pays nothing in Great Britain, and
alternative_fuel_payment.yamland the Hypothesis properties from #2735 pin that. The omission understates income and government spending by about £387m in 2023, or £369m on supplier payments alone, and in no other year.Limits of the evidence
Test plan
make format.test_energy_bills_rebate_northern_ireland_properties.py(4 passed) and theenergy_bills_rebateYAML tests (33 passed).changelog.d/2767.changed.md.Reproducer: n/a: documentation text only, no change in behaviour
axiom: n/a: documentation text only, no rule changes
🤖 Generated with Claude Code