[MaxGhenis]

Addresses #2702
Reproducer: n/a: this changes no result
axiom: n/a: this changes how the package lists programmes and changes no rule or parameter value

What the code does now

policyengine_uk/programs.yaml defines the budget totals as well as model coverage.

• Each programs entry whose variable a total adds carries budget: tax or budget: benefit, and an optional group for terms that move together (for example the five childcare entitlements, or ESA's two parts).
• Terms that no programme entry carries sit in a new budget_terms section. Each has a kind: programme_measure (the amount a total counts for a listed programme, such as esa_income for ESA, with program naming it), programme (a programme the coverage list does not carry yet), reform_measure (not a programme: a hypothetical tax or benefit, or a change from baseline such as vat_change and the employer NI incidence terms) or public_service (public services in kind). These entries stay out of programs, so the coverage page never lists them.
• budget_dropped_when: gov.contrib.abolish_council_tax sits on council tax, the High Value Council Tax Surcharge and Council Tax Reduction. budget_uprating sits on the State Pension ([all]) and basic income ([]); every other benefit term takes both gov.contrib.benefit_uprating rates.
• budget_order gives each total's summation order. The loader refuses a registry whose flags and order disagree, a misspelt field whose name starts with budget, and any budget_terms field outside the documented ones.

policyengine_uk/programs.py parses the YAML (with PyYAML's C parser where the environment has it) with no import from the package's top level. budget_programs() returns one BudgetProgram per term (variable, label, tax or benefit, group, kind, programme id, dropped-when parameter, uprating rates), taxes first, each in summation order. household_tax and household_benefits build their lists from it, and HOUSEHOLD_TAX_VARIABLES and HOUSEHOLD_BENEFIT_VARIABLES stay in their modules with the same contents in the same order. build_metadata.DATA_BUILD_SURFACE now includes programs.py.

policyengine-core loads the whole of programs.yaml as the system's modelled_policies, so that attribute, and any metadata a consumer builds from it, now carries budget_terms, budget_order and the new fields on each programs entry.

The branch merges main at 46a6067, which brings #2757's corrections to programme agencies and national coverage in programs.yaml and test_program_registry.py, and #2751's rounding of projected State Pension weekly rates to the nearest 5p. programs.yaml keeps main's corrections alongside the budget fields; the merge needed no conflict resolution.

Invariants

• HOUSEHOLD_TAX_VARIABLES and HOUSEHOLD_BENEFIT_VARIABLES equal main's lists, element by element.
• Each total's formula reads the flagged terms and no others, less those whose budget_dropped_when parameter is true. household_tax reads them in budget_order; household_benefits first reads them in the order its three add calls imply (the terms the all rate scales, then those the non_sp rate scales, then the rest), each in budget_order.
• budget_variables_in_effect and uprated_benefit_variables return subsequences of budget_order, in order.
• For each total and each switch, fiscal_partition.yaml's dropped_when names the same terms as the registry's budget_dropped_when.
• Every budget_programs() variable exists in the system and appears once, and its sign agrees with the list it came from.
• For any registry, the loader returns the terms in budget_order and refuses one whose order drops or repeats a flagged term.
• household_tax, household_benefits and household_net_income equal main's values exactly: the totals add the same float32 terms in the same order.

Tests

New tests:

• tests/test_budget_programs.py (124 tests): drift between the registry and what each total's formula reads (traced, at current law and with council tax abolished, including the benefit total's first-read order); the partition's dropped_when against the registry's; the differential test against main's two lists, which the file holds as ordered literals, and its council tax and uprating rules; budget_programs() from a fresh interpreter, importing policyengine_uk.programs, the household_tax module or the package first; the existence check parametrized over all 86 terms; a Hypothesis property that the in-effect and uprated selections keep budget_order; Hypothesis properties on generated registries; and 19 negative loader cases, one for each condition the loader refuses. We count its 16 refusals as 19 conditions: the two field-name checks once for each section, and the variable and budget test once for each field. That test already refuses a budget_terms entry with no flag, so the flag check's case for budget_terms tests its message.
• tests/test_budget_totals_pinned.py (17 tests) pins household_tax, household_benefits and household_net_income with exact equality for 13 hand-built households in 2023, 2025, 2026 and 2028 under four reforms (current law; abolish_council_tax; both uprating rates; both together with a flat basic income): 208 cells. One household, a London owner, pays the High Value Council Tax Surcharge from 2028. Every value comes from a run of main at 46a6067 with this file copied in. Six of the 208 differ from the pins the branch took earlier at 2da212a, all in 2028: the pensioner couple under current law and abolish_council_tax, and the East of England household under all four reforms. Round projected State Pension weekly rates to the nearest 5p each April #2751 rounds each projected year's weekly rate to the nearest 5p, starting from the previous year's rounded rate, so the 2028 new State Pension rate comes to £259.20 in place of the unrounded £259.2357. That moves each reported 2028 new State Pension in these households by one or two float32 units in the last place: £11,000.0010 becomes £10,999.9990, £9,000 becomes £8,999.9990 and £10,500.0010 becomes £10,500. The branch, main at 46a6067 and main at 6c6e229 all reproduce the 208 pins exactly.

Run	Result
test_budget_programs.py	124 passed
test_budget_totals_pinned.py on this branch, on main's code at 46a6067 and on main's code at 6c6e229	17 passed each
test_program_registry.py	119 passed
The new files with the related suites: test_program_registry.py, test_fiscal_partition.py, test_state_pension_rate_rounding.py, test_household_benefits_properties.py, test_gov_spending_properties.py, test_uc_statutory_pay_properties.py, tests/aggregate_impact, test_build_metadata.py	716 passed, 10 skipped (9 need the certified dataset, 1 needs policyengine_bundles)
The new files, test_program_registry.py, test_fiscal_partition.py and test_state_pension_rate_rounding.py on this branch merged with main at 6c6e229	337 passed
tests/code_health	3090 passed
ruff 0.17.2 format --check, check, check --select F811,F821; git diff --check	clean
Wheel from uv build of git archive HEAD	holds policyengine_uk/programs.py and policyengine_uk/programs.yaml

We ran each mutation below alone, in a scratch copy, against the two new test files:

Change	Result
Swap income_tax and national_insurance in budget_order.tax	1 failed: the differential test (the pinned sums happen not to move)
Drop the State Pension's budget_uprating: [all]	9 failed: the rules test and every uprating cell of the pinned test
Drop budget_dropped_when from the High Value Council Tax Surcharge	5 failed: the rules test, the partition test for tax, the traced drift test with council tax abolished, and the 2028 abolish_council_tax and abolish_uprating_basic_income cells, where the surcharge payer's household_tax gives 76,493.6015625 against 72,993.6015625
Make household_tax ignore budget_dropped_when	9 failed: the traced drift test and every council tax abolition cell
Sort the output of uprated_benefit_variables (test_budget_programs.py only)	3 failed: the subsequence property and both benefit drift cases
Reverse the benefit list in household_benefits (test_budget_programs.py only)	2 failed: both benefit drift cases
Disable each of the loader's 19 refused conditions in turn (the loader tests only)	each fails its own negative case; disabling either order check also fails the generated-registry property
Remove budget: benefit from Scottish Child Payment, order unchanged	the package refuses to load: budget_order.benefit and the entries flagged budget: benefit differ: only ordered ['scottish_child_payment'], only flagged []
Restore main's two formula modules	141 passed (124 + 17): main's formulas give the same totals

🤖 Generated with Claude Code

[github-actions]

Aggregate impact on certified data: pass

Base 0080274d78 against head 3730960f48, both on enhanced_frs_2024_25.h5 (policyengine-uk-data 1.58.1, sha256 056e80fd8e91), fiscal years 2025-26 to 2031-32. Each year is a full microsimulation run on each side.

Checks: 0 broken by this change, 1 failing on both sides, 0 fixed. Moves flagged for review: 0 (thresholds in report_spec.yaml; a flag does not fail the job).

Workflow run and full report

Notes

• In the years a figure from fewer than 10 records is suppressed (spending.disabled_students_allowance), every total containing it, and the decile incomes, are rebuilt from their other components, so no published total includes it. Those totals are marked †. A change is withheld where the two sides leave out different figures and can't be lined up.

Moves flagged for review

None.

Spending by benefit and service

Measure	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Disabled Students' Allowance	suppressed	suppressed	suppressed	suppressed	suppressed	suppressed	suppressed

Caseloads

Measure	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Disabled Students' Allowance	suppressed	suppressed	suppressed	suppressed	suppressed	suppressed	suppressed

Standard reform costings (change in gov_balance)

Measure	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Raise Class 1 main employee NICs rate to 10%	0	0	0	0	0	suppressed	suppressed
Raise additional rate by 3pp	0	0	0	0	0	suppressed	suppressed

The report suppresses a reform's costing and target change in these years, for the reasons below (privacy.py):

• Raise Class 1 main employee NICs rate to 10%, 2030-31 to 2031-32 (base and head): the reform changes income_tax in fewer than 10 records.
• Raise additional rate by 3pp, 2030-31 to 2031-32 (base and head): the reform changes universal_credit in fewer than 10 records.

Standard reforms: change in each reform's target

Measure	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Raise Class 1 main employee NICs rate to 10%: change in national_insurance	0	0	0	0	0	suppressed	suppressed
Raise additional rate by 3pp: change in income_tax	0	0	0	0	0	suppressed	suppressed

Household net income by decile

No 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)

• VAT excluding private school fees is never negative (sign:vat): 2025-26: 76 units below zero, a weighted £-53,888,684 in total; 2026-27: 76 units below zero, a weighted £-55,777,260 in total; 2027-28: 76 units below zero, a weighted £-57,727,917 in total

Counted-once check

Each 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.

• widowed_parents_allowance: pension_credit responds only through other heads in fewer than 10 households (other heads that moved there: carers_allowance; gov_spending; not a failure)
• widowed_parents_allowance: pension_credit responds only through other heads in fewer than 10 households (other heads that moved there: carers_allowance; household_benefits; not a failure)

Provenance

Side	Commit	Runtime	Environment hash	Partition hash	Time
base	main 0080274d78	uk 2.171.0 (files d72b65d494db), core 3.33.12, py 3.13.18 aarch64	c07a637ca255	7cb61324a46b	2133.2s
head	#2809 merged into main 3730960f48	uk 2.171.0 (files b3b28853ec8d), core 3.33.12, py 3.13.18 aarch64	c07a637ca255	7cb61324a46b	2131.45s

Dataset: hf://policyengine/policyengine-uk-data-private/enhanced_frs_2024_25.h5@2526edbbb687a3b54dbb11272e5389e4b62d2ac2, sha256 056e80fd8e9151d78f432a25cde863340baee81304c6216a37fb575ce8563784 (verified on both sides), policyengine-uk-data 1.58.1.

Receipts by tax head: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Final charge on cash interest in non-cash ISAs	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Lifetime ISA withdrawal charge	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Income tax	£294.46bn	£315.23bn	£335.82bn	£351.28bn	£366.96bn	£385.36bn	£400.00bn
National Insurance (employee and self-employed)	£51.70bn	£54.14bn	£56.13bn	£57.73bn	£59.76bn	£61.64bn	£63.51bn
National Insurance (employer)	£148.71bn	£155.37bn	£160.88bn	£165.09bn	£172.06bn	£177.36bn	£182.57bn
Employer NI change borne by capital	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Employer NI change borne by consumers	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Capital gains tax	£12.60bn	£13.15bn	£13.60bn	£14.11bn	£14.67bn	£15.24bn	£15.80bn
Student loan repayments	£7.82bn	£8.13bn	£8.39bn	£8.58bn	£8.79bn	£9.04bn	£9.29bn
VAT excluding private school fees	£249.37bn	£256.87bn	£266.33bn	£272.70bn	£278.97bn	£285.24bn	£291.59bn
Change in VAT from its baseline	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
VAT on private school fees	£1.95bn	£2.01bn	£2.09bn	£2.14bn	£2.20bn	£2.26bn	£2.31bn
Change in private school VAT from its baseline	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Fuel duties	£24.71bn	£24.13bn	£26.00bn	£26.83bn	£26.51bn	£25.78bn	£24.63bn
Alcohol duties	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Tobacco duties	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Vehicle excise duties	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Betting and gaming duties	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
TV licence fee	£4.26bn	£4.40bn	£4.41bn	£4.42bn	£4.43bn	£4.44bn	£4.45bn
Council tax	£49.21bn	£52.16bn	£54.89bn	£57.95bn	£61.16bn	£64.47bn	£67.91bn
High value council tax surcharge	£0.00bn	£0.00bn	£0.00bn	£0.48bn	£0.49bn	£0.50bn	£0.51bn
Northern Ireland domestic rates	£0.55bn	£0.58bn	£0.60bn	£0.63bn	£0.66bn	£0.69bn	£0.72bn
Business rates	£34.68bn	£36.20bn	£38.00bn	£39.46bn	£42.27bn	£43.25bn	£44.29bn
Stamp duty land tax on the household's own purchases	£6.88bn	£7.32bn	£7.66bn	£8.06bn	£8.47bn	£8.91bn	£9.35bn
Stamp duty land tax paid by companies	£4.72bn	£4.57bn	£5.29bn	£6.03bn	£6.37bn	£6.92bn	£7.45bn
Land transaction tax (Wales)	£0.07bn	£0.08bn	£0.08bn	£0.09bn	£0.10bn	£0.10bn	£0.11bn
Land and buildings transaction tax (Scotland)	£0.31bn	£0.32bn	£0.33bn	£0.35bn	£0.36bn	£0.38bn	£0.40bn
Unspecified tax change borne by capital	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Unspecified tax change borne by consumers	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Wealth tax	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Wealth tax excluding the main residence	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Land value tax	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Carbon tax	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn

Spending by benefit and service: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Lifetime ISA bonus	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
State Pension	£121.68bn	£125.92bn	£126.72bn	£129.49bn	£132.99bn	£136.66bn	£140.42bn
Pension Credit	£6.66bn	£7.03bn	£7.29bn	£7.48bn	£7.60bn	£7.84bn	£8.03bn
Winter Fuel Payment	£1.86bn	£1.85bn	£1.79bn	£1.77bn	£1.78bn	£1.78bn	£1.78bn
Pension Age Winter Heating Payment (Scotland)	£0.19bn	£0.20bn	£0.20bn	£0.21bn	£0.21bn	£0.22bn	£0.22bn
Universal Credit	£75.48bn	£80.52bn	£82.95bn	£84.96bn	£86.11bn	£87.61bn	£88.21bn
Housing Benefit	£8.56bn	£8.89bn	£9.23bn	£9.46bn	£9.69bn	£9.99bn	£10.27bn
Council Tax Reduction	£2.07bn	£2.20bn	£2.31bn	£2.45bn	£2.58bn	£2.72bn	£2.86bn
Employment and Support Allowance (income-related)	£0.79bn	£0.82bn	£0.84bn	£0.86bn	£0.88bn	£0.90bn	£0.92bn
Employment and Support Allowance (contribution-based)	£5.20bn	£5.37bn	£5.56bn	£5.69bn	£5.82bn	£5.95bn	£6.09bn
Jobseeker's Allowance (income-based)	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Jobseeker's Allowance (contribution-based)	£0.29bn	£0.30bn	£0.31bn	£0.31bn	£0.32bn	£0.33bn	£0.34bn
Income Support	£0.11bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Working Tax Credit	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Child Tax Credit	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Incapacity Benefit	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Personal Independence Payment	£26.95bn	£28.03bn	£29.05bn	£30.09bn	£30.75bn	£31.51bn	£32.21bn
Disability Living Allowance	£7.49bn	£7.79bn	£8.07bn	£8.36bn	£8.55bn	£8.76bn	£8.95bn
Attendance Allowance	£8.68bn	£9.03bn	£9.36bn	£9.69bn	£9.90bn	£10.15bn	£10.38bn
Carer's Allowance	£4.48bn	£4.66bn	£4.83bn	£5.00bn	£5.11bn	£5.27bn	£5.42bn
Carer Support Payment (Scotland)	£0.31bn	£0.32bn	£0.33bn	£0.34bn	£0.35bn	£0.36bn	£0.37bn
Carer's Allowance Supplement (Scotland)	£0.00bn	£0.04bn	£0.04bn	£0.05bn	£0.05bn	£0.05bn	£0.05bn
Severe Disablement Allowance	£0.03bn	£0.03bn	£0.03bn	£0.03bn	£0.03bn	£0.03bn	£0.03bn
Industrial Injuries Disablement Benefit	£0.50bn	£0.52bn	£0.54bn	£0.55bn	£0.56bn	£0.58bn	£0.59bn
Armed Forces Compensation Scheme	£0.61bn	£0.63bn	£0.65bn	£0.66bn	£0.68bn	£0.69bn	£0.71bn
War Widow's or Widower's Pension	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Child Benefit	£16.19bn	£16.82bn	£17.39bn	£18.00bn	£18.39bn	£18.84bn	£19.25bn
Scottish Child Payment	£0.49bn	£0.52bn	£0.53bn	£0.55bn	£0.56bn	£0.57bn	£0.56bn
Two Child Limit Payment (Scotland, counterfactual)	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Healthy Start vouchers	£0.05bn	£0.05bn	£0.05bn	£0.05bn	£0.05bn	£0.05bn	£0.05bn
Maternity Allowance	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Statutory Maternity Pay	£3.30bn	£3.41bn	£3.54bn	£3.62bn	£3.70bn	£3.78bn	£3.87bn
Statutory Paternity Pay	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Statutory Sick Pay	£1.12bn	£1.16bn	£1.20bn	£1.23bn	£1.26bn	£1.29bn	£1.32bn
Sure Start Maternity Grant	£0.01bn	£0.01bn	£0.01bn	£0.01bn	£0.01bn	£0.01bn	£0.01bn
Bereavement Support Payment	£0.67bn	£0.69bn	£0.71bn	£0.73bn	£0.75bn	£0.76bn	£0.78bn
Widowed Parent's Allowance	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Tax-Free Childcare	£0.56bn	£0.57bn	£0.57bn	£0.58bn	£0.60bn	£0.61bn	£0.59bn
Extended childcare entitlement	£2.99bn	£3.50bn	£3.59bn	£3.71bn	£3.78bn	£3.86bn	£3.93bn
Universal childcare entitlement	£2.06bn	£2.22bn	£2.29bn	£2.37bn	£2.43bn	£2.48bn	£2.53bn
Targeted childcare entitlement	£0.50bn	£0.51bn	£0.52bn	£0.54bn	£0.55bn	£0.57bn	£0.56bn
Childcare Offer for Wales	£0.26bn	£0.28bn	£0.28bn	£0.29bn	£0.30bn	£0.31bn	£0.32bn
Care to Learn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Childcare Grant	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Parents' Learning Allowance	£0.20bn	£0.20bn	£0.20bn	£0.20bn	£0.20bn	£0.20bn	£0.20bn
Adult Dependants' Grant	£0.06bn	£0.06bn	£0.04bn	£0.04bn	£0.04bn	£0.04bn	£0.04bn
Travel Grant	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Disabled Students' Allowance	suppressed	suppressed	suppressed	suppressed	suppressed	suppressed	suppressed
16 to 19 Bursary Fund	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
NHS spending	£203.89bn	£204.30bn	£204.56bn	£204.91bn	£205.34bn	£205.85bn	£206.35bn
State education spending	£0.02bn	£0.02bn	£0.02bn	£0.02bn	£0.02bn	£0.02bn	£0.02bn
Transport subsidy spending	£25.05bn	£25.10bn	£25.14bn	£25.18bn	£25.23bn	£25.30bn	£25.36bn
Cost-of-living support payment	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Energy Bills Rebate	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Household Energy Payment	£0.00bn	£2.11bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Energy Price Guarantee subsidy	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Basic income	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Disability basic income	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn
Other public spending budget change	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn

Caseloads: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Lifetime ISA bonus	0k	0k	0k	0k	0k	0k	0k
State Pension	11,264k	11,145k	10,819k	10,700k	10,723k	10,750k	10,776k
Pension Credit	1,486k	1,480k	1,472k	1,452k	1,450k	1,455k	1,464k
Winter Fuel Payment	7,844k	7,783k	7,473k	7,375k	7,391k	7,409k	7,427k
Pension Age Winter Heating Payment (Scotland)	803k	802k	800k	797k	799k	801k	802k
Universal Credit	6,342k	6,347k	6,336k	6,297k	6,292k	6,271k	6,223k
Housing Benefit	1,249k	1,245k	1,244k	1,227k	1,227k	1,229k	1,232k
Council Tax Reduction	3,176k	3,195k	3,200k	3,214k	3,224k	3,235k	3,242k
Employment and Support Allowance (income-related)	71k	71k	71k	71k	71k	71k	71k
Employment and Support Allowance (contribution-based)	719k	720k	721k	722k	724k	726k	728k
Jobseeker's Allowance (income-based)	0k	0k	0k	0k	0k	0k	0k
Jobseeker's Allowance (contribution-based)	67k	67k	67k	67k	67k	68k	68k
Income Support	25k	0k	0k	0k	0k	0k	0k
Working Tax Credit	0k	0k	0k	0k	0k	0k	0k
Child Tax Credit	0k	0k	0k	0k	0k	0k	0k
Incapacity Benefit	0k	0k	0k	0k	0k	0k	0k
Personal Independence Payment	3,567k	3,575k	3,579k	3,585k	3,593k	3,602k	3,611k
Disability Living Allowance	1,291k	1,294k	1,296k	1,298k	1,300k	1,304k	1,307k
Attendance Allowance	1,695k	1,698k	1,700k	1,703k	1,707k	1,711k	1,715k
Carer's Allowance	1,034k	1,037k	1,038k	1,040k	1,042k	1,060k	1,062k
Carer Support Payment (Scotland)	71k	71k	71k	72k	72k	72k	72k
Carer's Allowance Supplement (Scotland)	0k	71k	71k	72k	72k	72k	72k
Severe Disablement Allowance	4k	4k	4k	4k	4k	4k	4k
Industrial Injuries Disablement Benefit	137k	137k	137k	138k	138k	138k	139k
Armed Forces Compensation Scheme	93k	93k	94k	94k	94k	94k	94k
War Widow's or Widower's Pension	0k	0k	0k	0k	0k	0k	0k
Child Benefit	8,123k	8,120k	8,120k	8,130k	8,145k	8,161k	8,178k
Scottish Child Payment	351k	352k	352k	353k	350k	348k	337k
Two Child Limit Payment (Scotland, counterfactual)	0k	0k	0k	0k	0k	0k	0k
Healthy Start vouchers	200k	200k	200k	201k	201k	201k	202k
Maternity Allowance	0k	0k	0k	0k	0k	0k	0k
Statutory Maternity Pay	333k	333k	334k	334k	335k	336k	337k
Statutory Paternity Pay	0k	0k	0k	0k	0k	0k	0k
Statutory Sick Pay	181k	182k	182k	182k	182k	183k	183k
Sure Start Maternity Grant	26k	26k	26k	26k	26k	26k	26k
Bereavement Support Payment	98k	98k	98k	98k	98k	99k	99k
Widowed Parent's Allowance	0k	0k	0k	0k	0k	0k	0k
Tax-Free Childcare	1,043k	1,027k	1,014k	1,009k	1,011k	1,009k	993k
Extended childcare entitlement	795k	800k	796k	797k	798k	798k	792k
Universal childcare entitlement	1,071k	1,072k	1,072k	1,073k	1,076k	1,079k	1,074k
Targeted childcare entitlement	165k	156k	154k	154k	155k	155k	150k
Childcare Offer for Wales	39k	39k	39k	39k	39k	39k	39k
Care to Learn	0k	0k	0k	0k	0k	0k	0k
Childcare Grant	0k	0k	0k	0k	0k	0k	0k
Parents' Learning Allowance	98k	98k	98k	99k	99k	99k	99k
Adult Dependants' Grant	18k	18k	12k	12k	12k	12k	12k
Travel Grant	0k	0k	0k	0k	0k	0k	0k
Disabled Students' Allowance	suppressed	suppressed	suppressed	suppressed	suppressed	suppressed	suppressed
16 to 19 Bursary Fund	0k	0k	0k	0k	0k	0k	0k
Cost-of-living support payment	0k	0k	0k	0k	0k	0k	0k
Energy Bills Rebate	0k	0k	0k	0k	0k	0k	0k
Household Energy Payment	0k	21,065k	0k	0k	0k	0k	0k
Energy Price Guarantee subsidy	0k	0k	0k	0k	0k	0k	0k
Basic income	0k	0k	0k	0k	0k	0k	0k
Disability basic income	0k	0k	0k	0k	0k	0k	0k

Government balance, each component counted once: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Government receipts, each component once	£892.00bn	£934.66bn	£980.50bn	£1,015.93bn	£1,054.24bn	£1,091.56bn	£1,124.89bn
Government spending	£529.38bn †	£545.66bn †	£550.77bn †	£559.51bn †	£567.20bn †	£575.95bn †	£583.60bn †
Government balance, each receipt once	£362.62bn †	£389.00bn †	£429.74bn †	£456.42bn †	£487.04bn †	£515.61bn †	£541.29bn †
Benefits in household_benefits but not in gov_spending	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn	£0.00bn

Model totals: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
gov_tax	£892.00bn	£934.66bn	£980.50bn	£1,015.93bn	£1,054.24bn	£1,091.56bn	£1,124.89bn
household_tax	£491.97bn	£520.41bn	£551.21bn	£576.01bn	£601.00bn	£626.70bn	£648.42bn
pre_budget_change_household_tax	£484.15bn	£512.28bn	£542.82bn	£566.95bn	£591.72bn	£617.16bn	£638.61bn
gov_spending	£529.38bn †	£545.66bn †	£550.77bn †	£559.51bn †	£567.20bn †	£575.95bn †	£583.60bn †
household_benefits	£529.38bn †	£545.66bn †	£550.77bn †	£559.51bn †	£567.20bn †	£575.95bn †	£583.60bn †
gov_balance	£362.62bn †	£389.00bn †	£429.74bn †	£456.42bn †	£487.04bn †	£515.61bn †	£541.29bn †

Household income totals: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Household market income	£1,637.96bn	£1,702.77bn	£1,757.35bn	£1,806.41bn	£1,855.97bn	£1,912.43bn	£1,970.44bn
Household net income	£1,636.69bn †	£1,687.82bn †	£1,715.45bn †	£1,747.49bn †	£1,778.69bn †	£1,817.02bn †	£1,859.69bn †
Household net income (HBAI definition)	£1,449.57bn	£1,499.83bn	£1,530.86bn	£1,565.17bn	£1,597.89bn	£1,635.60bn	£1,676.61bn

Population: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
People	69,245k	69,383k	69,473k	69,592k	69,738k	69,912k	70,080k
Benefit units	35,110k	35,181k	35,226k	35,286k	35,360k	35,449k	35,534k
Households	31,046k	31,108k	31,149k	31,201k	31,267k	31,345k	31,420k

Poverty rates: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Absolute poverty, before housing costs: everyone	16.3%	15.5%	16.4%	16.3%	16.5%	16.5%	16.6%
Absolute poverty, before housing costs: children	24.7%	22.1%	23.3%	22.9%	23.1%	23.2%	23.5%
Absolute poverty, before housing costs: working-age adults	14.1%	13.8%	14.7%	14.7%	14.8%	14.8%	15.0%
Absolute poverty, before housing costs: pensioners	13.7%	13.4%	14.0%	13.8%	13.9%	13.9%	13.9%
Absolute poverty, after housing costs: everyone	21.6%	21.0%	22.0%	22.6%	22.9%	23.1%	23.4%
Absolute poverty, after housing costs: children	33.3%	31.3%	32.3%	32.5%	32.9%	33.4%	34.2%
Absolute poverty, after housing costs: working-age adults	19.9%	19.8%	20.8%	21.6%	21.9%	22.0%	22.3%
Absolute poverty, after housing costs: pensioners	12.7%	12.4%	13.2%	13.2%	13.3%	13.4%	13.6%
Relative poverty, before housing costs: everyone	17.0%	16.0%	16.4%	16.3%	16.4%	16.3%	16.6%
Relative poverty, before housing costs: children	25.9%	22.9%	23.3%	22.9%	23.0%	22.9%	23.5%
Relative poverty, before housing costs: working-age adults	14.5%	14.1%	14.6%	14.7%	14.7%	14.7%	15.0%
Relative poverty, before housing costs: pensioners	14.5%	14.0%	13.9%	13.8%	13.8%	13.9%	13.9%
Relative poverty, after housing costs: everyone	21.4%	21.0%	21.3%	21.3%	21.3%	21.5%	21.7%
Relative poverty, after housing costs: children	33.1%	31.1%	31.2%	31.0%	31.1%	31.2%	31.8%
Relative poverty, after housing costs: working-age adults	19.9%	19.8%	20.2%	20.4%	20.5%	20.6%	20.8%
Relative poverty, after housing costs: pensioners	12.5%	12.4%	12.2%	11.7%	11.7%	11.8%	12.0%

People in poverty: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Absolute poverty, before housing costs: everyone	11,278k	10,779k	11,402k	11,365k	11,476k	11,513k	11,651k
Absolute poverty, before housing costs: children	3,688k	3,314k	3,496k	3,445k	3,480k	3,502k	3,549k
Absolute poverty, before housing costs: working-age adults	5,965k	5,893k	6,308k	6,362k	6,420k	6,430k	6,521k
Absolute poverty, before housing costs: pensioners	1,624k	1,572k	1,598k	1,558k	1,576k	1,580k	1,581k
Absolute poverty, after housing costs: everyone	14,944k	14,600k	15,310k	15,721k	15,937k	16,141k	16,426k
Absolute poverty, after housing costs: children	4,975k	4,682k	4,837k	4,888k	4,945k	5,040k	5,179k
Absolute poverty, after housing costs: working-age adults	8,463k	8,461k	8,971k	9,344k	9,491k	9,584k	9,706k
Absolute poverty, after housing costs: pensioners	1,506k	1,457k	1,502k	1,490k	1,501k	1,517k	1,542k
Relative poverty, before housing costs: everyone	11,750k	11,084k	11,388k	11,364k	11,416k	11,415k	11,651k
Relative poverty, before housing costs: children	3,872k	3,423k	3,496k	3,445k	3,466k	3,454k	3,549k
Relative poverty, before housing costs: working-age adults	6,159k	6,013k	6,301k	6,362k	6,389k	6,389k	6,521k
Relative poverty, before housing costs: pensioners	1,720k	1,648k	1,590k	1,557k	1,562k	1,572k	1,581k
Relative poverty, after housing costs: everyone	14,849k	14,552k	14,776k	14,809k	14,874k	15,014k	15,234k
Relative poverty, after housing costs: children	4,942k	4,660k	4,673k	4,663k	4,679k	4,707k	4,806k
Relative poverty, after housing costs: working-age adults	8,426k	8,440k	8,711k	8,823k	8,873k	8,967k	9,062k
Relative poverty, after housing costs: pensioners	1,482k	1,452k	1,392k	1,323k	1,322k	1,339k	1,365k

Inequality: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Gini coefficient, equivalised HBAI income before housing costs	0.3080	0.3052	0.3058	0.3058	0.3063	0.3069	0.3085

Mean household net income in each side's own deciles: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Decile 1: mean household_net_income per household	£17,681 †	£17,997 †	£17,915 †	£18,152 †	£18,307 †	£18,531 †	£18,717 †
Decile 2: mean household_net_income per household	£27,939 †	£29,023 †	£28,711 †	£29,013 †	£29,276 †	£29,691 †	£30,150 †
Decile 3: mean household_net_income per household	£34,320 †	£35,485 †	£36,073 †	£36,611 †	£37,197 †	£37,737 †	£38,187 †
Decile 4: mean household_net_income per household	£38,933 †	£40,687 †	£41,723 †	£42,727 †	£43,334 †	£44,132 †	£44,432 †
Decile 5: mean household_net_income per household	£43,239 †	£45,567 †	£46,014 †	£46,959 †	£47,712 †	£48,584 †	£50,282 †
Decile 6: mean household_net_income per household	£47,927 †	£49,210 †	£50,892 †	£51,939 †	£52,945 †	£53,942 †	£54,680 †
Decile 7: mean household_net_income per household	£55,662 †	£56,571 †	£57,442 †	£57,670 †	£58,729 †	£59,657 †	£61,082 †
Decile 8: mean household_net_income per household	£68,789 †	£70,803 †	£71,858 †	£73,466 †	£73,880 †	£75,320 †	£77,178 †
Decile 9: mean household_net_income per household	£78,890 †	£81,013 †	£80,861 †	£82,646 †	£85,464 †	£86,946 †	£89,509 †
Decile 10: mean household_net_income per household	£117,005 †	£120,339 †	£124,742 †	£126,632 †	£125,971 †	£128,600 †	£130,938 †

Standard reform costings (change in gov_balance): head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Raise basic rate by 1pp	£6.41bn	£6.66bn	£6.63bn	£6.85bn	£7.03bn	£7.25bn	£7.49bn
Raise higher rate by 1pp	£3.61bn	£3.87bn	£4.00bn	£4.22bn	£4.44bn	£4.71bn	£4.83bn
Raise personal allowance by ~800GBP/year	£-3.80bn	£-3.92bn	£-3.98bn	£-4.04bn	£-4.10bn	£-4.23bn	£-1.95bn
Raise child benefit by 25GBP/week per additional child	£-1.99bn	£-1.75bn	£-1.55bn	£-1.38bn	£-1.30bn	£-1.18bn	£-1.07bn
Reduce Universal Credit taper rate to 20%	£-20.77bn	£-21.32bn	£-21.75bn	£-21.99bn	£-22.34bn	£-22.75bn	£-22.94bn
Raise Class 1 main employee NICs rate to 10%	£10.89bn	£11.30bn	£11.71bn	£12.01bn	£12.31bn	suppressed	suppressed
Raise VAT standard rate by 2pp	£24.89bn	£25.71bn	£26.63bn	£27.24bn	£27.87bn	£28.50bn	£29.14bn
Raise additional rate by 3pp	£2.90bn	£3.15bn	£3.22bn	£3.41bn	£3.61bn	suppressed	suppressed
Raise the Universal Credit standard allowance by 10%	£-3.06bn	£-3.18bn	£-3.26bn	£-3.35bn	£-3.44bn	£-3.49bn	£-3.50bn
Raise the State Pension by 10%	£-7.78bn	£-8.01bn	£-8.24bn	£-8.47bn	£-8.82bn	£-8.55bn	£-8.89bn

Standard reforms: change in each reform's target: head levels

Measure (head)	2025-26	2026-27	2027-28	2028-29	2029-30	2030-31	2031-32
Raise basic rate by 1pp: change in income_tax	£6.55bn	£6.86bn	£6.79bn	£7.00bn	£7.20bn	£7.44bn	£7.66bn
Raise higher rate by 1pp: change in income_tax	£3.61bn	£3.92bn	£4.01bn	£4.23bn	£4.45bn	£4.72bn	£4.84bn
Raise personal allowance by ~800GBP/year: change in income_tax	£-3.93bn	£-4.07bn	£-4.13bn	£-4.18bn	£-4.22bn	£-4.40bn	£-1.75bn
Raise child benefit by 25GBP/week per additional child: change in child_benefit	£2.33bn	£2.14bn	£1.94bn	£1.76bn	£1.65bn	£1.52bn	£1.40bn
Reduce Universal Credit taper rate to 20%: change in universal_credit	£20.74bn	£21.31bn	£21.78bn	£22.04bn	£22.37bn	£22.78bn	£22.99bn
Raise Class 1 main employee NICs rate to 10%: change in national_insurance	£11.15bn	£11.63bn	£12.00bn	£12.29bn	£12.61bn	suppressed	suppressed
Raise VAT standard rate by 2pp: change in vat	£24.69bn	£25.51bn	£26.42bn	£27.03bn	£27.65bn	£28.28bn	£28.91bn
Raise additional rate by 3pp: change in income_tax	£2.90bn	£3.15bn	£3.22bn	£3.42bn	£3.62bn	suppressed	suppressed
Raise the Universal Credit standard allowance by 10%: change in universal_credit	£3.06bn	£3.19bn	£3.28bn	£3.37bn	£3.43bn	£3.49bn	£3.51bn
Raise the State Pension by 10%: change in state_pension	£10.68bn	£11.17bn	£11.28bn	£11.60bn	£12.03bn	£12.47bn	£12.97bn

[MaxGhenis]

Ready for hub review

Head e3308b11e6d78c9b2097bbcceef123d3652d45f1 on branch claude/budget-totals-from-registry-4rggpz, based on main at 2da212abcf2086337fc088a4a1236f4ae7b6af63. The task pinned pushes to this claude/... branch. Issue #2702 stays open for step 3.

Items

1. Read first. I read programs.yaml, both total modules and their lists, the loader (tax_benefit_system.py modelled_policies), test_program_registry.py, the three code-health tests, utils/aggregate_impact/partition.py (it imports both lists by module and name, and that still works), fiscal_partition.yaml, the Program registry section of .github/CONTRIBUTING.md, and Classify the Budget helper variables in the fiscal partition #2763. CLAUDE.md has no Program registry section; the section it points to lives in docs/engineering/skills/model-structure.md.
2. Registry fields (policyengine_uk/programs.yaml).
   • Header documents the fields: lines 6-34.
   • 39 programs entries gain budget (12 tax, 27 benefit), each inserted directly after the entry's variable: line. 13 gain group. council_tax gains budget_dropped_when (line 657) and state_pension gains budget_uprating: [all] (line 312).
   • The new budget_terms section (line 799) holds the 47 terms no programme entry carries, each with a kind: 15 programme_measure with program, 16 programme, 13 reform_measure, 3 public_service.
   • budget_order (line 1119) sets summation order.
   • No verified-year field was added or changed.
3. Totals from the registry.
   • policyengine_uk/programs.py reads the YAML with only yaml and the standard library. build_budget_programs (line 129) validates the registry, and the package refuses to load if flags and budget_order disagree.
   • household_tax.py:5 and household_benefits.py:9 set HOUSEHOLD_TAX_VARIABLES and HOUSEHOLD_BENEFIT_VARIABLES in their old modules: 29 and 57 names, identical in content and order to main.
   • The formulas (household_tax.py:19, household_benefits.py:25) keep main's arithmetic. Council tax abolition comes from budget_dropped_when, and uprating from budget_uprating through uprated_benefit_variables (programs.py:206).
   • __init__.py:18 imports policyengine_uk.programs explicitly. build_metadata.py:53 adds programs.py to the data-build surface.
   • uv build puts policyengine_uk/programs.py and policyengine_uk/programs.yaml in the wheel.
   • The pre-Budget lists stay unchanged; see the last section.
4. Stable function.
   • policyengine_uk.programs.budget_programs() (programs.py:118) returns a tuple of frozen BudgetProgram records: variable, label, budget (tax or benefit), group, kind, program, dropped_when and uprating, plus is_tax and is_programme.
   • Taxes come first, then benefits, each in summation order.
   • budget_variables(budget) returns a plain list. test_budget_programs_in_a_fresh_interpreter runs it in a subprocess three ways: importing programs first, the household_tax module first, or the package first.
5. Tests.
   • tests/test_budget_programs.py covers drift (lines 166 and 192, traced from the formula, with council tax abolished and without), the differential against the frozen literal lists (lines 209 and 214), the council tax and uprating rules from main (line 219), the fresh interpreter (line 250), and Hypothesis properties (lines 297, 337 and 352).
   • tests/test_budget_totals_pinned.py checks 12 households × 2023/2025/2026/2028 × 4 reforms = 192 cells. I computed the values on main before the change and compare them with exact equality.
6. Docs. .github/CONTRIBUTING.md:55 and docs/engineering/skills/model-structure.md:66. CLAUDE.md is unchanged: it holds no registry section and says it should not duplicate these rules. No other page under docs/ lists the registry fields.
7. Overlaps.
   • The diff only adds keys and appends sections, and reorders or reformats nothing. Merge trees against Correct programme agencies and national coverage #2757, Pay the 2022 £150 to Council Tax Reduction recipients in any band in Scotland and Wales #2768, Report the benefit cap reduction actually applied to Universal Credit #2647, Fix HB and national CTR non-dependant deductions #2007, Give Scotland and Wales working-age CTR their own rules #1966 and Abolish working-age Housing Benefit from 1 July 2026 (NI 1 October 2026) #1910 are not computed; I read each one's programs.yaml hunks. They change agency, coverage, status and notes lines, and I add lines only after variable: lines. Correct programme agencies and national coverage #2757 changes coverage directly above winter_fuel_payment's variable:, but that entry gets no key.
   • Classify the Budget helper variables in the fiscal partition #2763 does not touch programs.yaml or the two total modules. Its fiscal_partition.yaml heads stay a separate specification keyed by variable. test_fiscal_partition.py's differential test now compares those heads with lists derived from the registry. A new term added to the registry without a partition head will fail that test, which keeps the two in step.

Tests run

Run	Result
test_budget_programs.py	16 passed
test_budget_totals_pinned.py, on this branch and on main's code	17 passed each
The new files plus test_program_registry.py, test_fiscal_partition.py, test_household_benefits_properties.py, test_gov_spending_properties.py, test_uc_statutory_pay_properties.py, tests/aggregate_impact, test_build_metadata.py	549 passed, 10 skipped (9 certified-dataset tests without a Hugging Face token; 1 needs policyengine_bundles)
tests/code_health (includes the three named list tests, unchanged)	3088 passed
ruff 0.17.2 format --check, check, check --select F811,F821; git diff --check	clean
Pinned grid, branch against main	all 192 cells bit-identical
CI on e3308b11	20 check runs, all success

I ran the 549-test batch before commit e3308b11 added the package import, then reran test_budget_programs.py after it: 16 passed. CI covers the final head.

Mutation and revert checks (scratch worktree)

Change	Result
Swap income_tax and national_insurance in budget_order.tax	1 failed: the differential test. The pinned float sums did not move for these households
Drop the State Pension's budget_uprating: [all]	9 failed: the rules test and all 8 uprating cells
Drop budget_dropped_when from the High Value Council Tax Surcharge	2 failed: the rules test and the traced drift test with council tax abolished. The pinned grid has no surcharge payer, so it does not catch this
household_tax ignores budget_dropped_when	9 failed: the traced drift test and all 8 abolition cells
Remove budget: benefit from Scottish Child Payment	the package refuses to import with RegistryError: ... only ordered ['scottish_child_payment']
Restore main's two formula modules	33 passed: these are equivalence tests, so they pass on both sides. The first run of this check failed the fresh-interpreter package case, which showed the submodule was reachable only through the variable modules' imports; __init__.py:18 now imports it, and the rerun passed

Aggregate impact report

Aggregate impact on certified data: pass

Base 2da212abcf against head 498e9f74b0 [the PR merge commit], both on enhanced_frs_2024_25.h5 (policyengine-uk-data 1.58.1), fiscal years 2025-26 to 2031-32.
Checks: 0 broken by this change, 1 failing on both sides, 0 fixed. Moves flagged for review: 0
Standard reform costings: 0 in every unsuppressed year.
No 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.

The check failing on both sides is sign:vat (VAT excluding private school fees goes negative for 76 units), which this PR does not touch.

Skipped

• Step 3 is for other repositories, as the brief says.
  • policyengine.py: replace UK_PROGRAMS and the imports of the two lists (Update validation page for new UKMOD country report #724) with policyengine_uk.programs.budget_programs(). Use is_tax/budget for the sign and group for grouping, and keep dropped_when/uprating so rows add up under those reforms.
  • policyengine-sim-api: pass budget, group and kind through beside the is_tax flag that Uprating benefit levels for 2025/26 #972 adds.
  • policyengine-app-v2: drop BENEFIT_PROGRAMS and sign rows from the API field.
• No survey-data test ran locally. The 9 certified-dataset tests in test_aggregate_impact_certified.py skip without a token; CI's Aggregate impact job ran on the certified data.
• No programs entry was added for the 16 kind: programme terms (for example Statutory Sick Pay and Carer Support Payment), so the coverage page does not change. Each needs verified years that parameters and tests confirm before it moves into programs.

Judgment calls

1. Non-programmes. vat_change, private_school_vat_change, both employer NI incidence terms, both unspecified incidence changes, the wealth, land value and carbon taxes, basic income, disability basic income and the Two Child Limit Payment are kind: reform_measure. The three in-kind service terms are public_service. They sit in budget_terms, never in programs, and BudgetProgram.is_programme is False for them. test_budget_programs_have_labels_and_kinds checks that none appears in the coverage list. I put the Two Child Limit Payment here because its variable documents that the regulations were not laid and the baseline pays zero.
2. Components of listed programmes (esa_income, jsa_contrib, working_tax_credit, council_tax_benefit, expected_sdlt and so on) are programme_measure with a program link, not new coverage entries. New entries would show ESA, JSA, tax credits and SDLT twice on the coverage page.
3. Real programmes the coverage list lacks are kind: programme in budget_terms, with no verified-year fields. The schema requires verified_start_year on programs entries, and I did not verify those years.
4. budget_order. Membership comes from the flags, but float32 sums depend on order, and the entries cannot be reordered without breaking the open PRs. So a separate list fixes the order, and the loader requires it to match the flags exactly. The cost: adding a term means flagging it and appending it there.
5. Behaviour in the registry. Council tax abolition is budget_dropped_when on three terms, and uprating is budget_uprating with a default of both rates.
6. Groups are deliberately narrow: property_transaction_taxes, council_tax, vat, employer_ni_incidence, esa, jsa, tax_credits, statutory_pay, winter_heating_payments, childcare (Tax-Free Childcare, the three English entitlements, the Wales offer), student_support, carer_benefits and public_services_in_kind. Taxes and benefits never share a group.
7. DATA_BUILD_SURFACE now includes programs.py, because it now determines the totals. This changes the build fingerprint value once.

Pre-Budget lists

• PRE_BUDGET_CHANGE_HOUSEHOLD_TAX_VARIABLES (17 names) is a subset of the registry's tax terms in the same relative order. It lacks 12: the ISA charge, the High Value Council Tax Surcharge, the four newer duties, private school VAT, the four incidence terms and student loans. Its formula drops only council_tax on abolition.
• PRE_BUDGET_CHANGE_HOUSEHOLD_BENEFIT_VARIABLES (36 names) is a subset of the benefit terms but not in the same relative order: esa_contrib and jsa_contrib sit later. It lacks 21 terms, including Statutory Paternity Pay, both winter payments, the childcare and student support terms, the in-kind services, Healthy Start, the Two Child Limit Payment and Scottish Child Payment.
• Both are frozen pre-Budget definitions that the registry neither feeds nor describes. Both are unchanged.

---

Generated by Claude Code

[MaxGhenis]

Picking this up for the UK hub. Reply here if you're mid-change and I'll stop.

---

Generated by Claude Code

[MaxGhenis]

PR_TITLE.txt

Build the budget totals from the programme registry

---

Generated by Claude Code

[MaxGhenis]

PR_BODY.md

Addresses #2702
Reproducer: n/a: this changes no result
axiom: n/a: this changes how the package lists programmes and changes no rule or parameter value

## What the code does now

`policyengine_uk/programs.yaml` defines the budget totals as well as model coverage.

- Each `programs` entry whose `variable` a total adds carries `budget: tax` or `budget: benefit`, and an optional `group` for terms that move together (for example the five childcare entitlements, or ESA's two parts).
- Terms that no programme entry carries sit in a new `budget_terms` section. Each has a `kind`: `programme_measure` (the amount counted for a listed programme, such as `esa_income` for ESA, with `program` naming it), `programme` (a programme the coverage list does not carry yet), `reform_measure` (not a programme: a hypothetical tax or benefit, or a change from baseline such as `vat_change` and the employer NI incidence terms) or `public_service` (public services in kind). These entries stay out of `programs`, so the coverage page never lists them.
- `budget_dropped_when: gov.contrib.abolish_council_tax` sits on council tax, the High Value Council Tax Surcharge and Council Tax Reduction. `budget_uprating` sits on the State Pension (`[all]`) and basic income (`[]`); every other benefit term takes both `gov.contrib.benefit_uprating` rates.
- `budget_order` gives each total's summation order. The loader refuses a registry whose flags and order disagree, a misspelt field whose name starts with `budget`, and any `budget_terms` field outside the documented ones.

`policyengine_uk/programs.py` parses the YAML (with PyYAML's C parser when it is available) with no import from the package's top level. `budget_programs()` returns one `BudgetProgram` per term (variable, label, `tax` or `benefit`, group, kind, programme id, dropped-when parameter, uprating rates), taxes first, each in summation order. `household_tax` and `household_benefits` build their lists from it, and `HOUSEHOLD_TAX_VARIABLES` and `HOUSEHOLD_BENEFIT_VARIABLES` stay in their modules with the same contents in the same order. `build_metadata.DATA_BUILD_SURFACE` now includes `programs.py`.

policyengine-core loads the whole of `programs.yaml` as the system's `modelled_policies`, so that attribute, and any metadata built from it, now carries `budget_terms`, `budget_order` and the new fields on each `programs` entry.

## Invariants

- `HOUSEHOLD_TAX_VARIABLES` and `HOUSEHOLD_BENEFIT_VARIABLES` equal `main`'s lists, element by element.
- Each total's formula reads the flagged terms and no others, less those whose `budget_dropped_when` parameter is true. `household_tax` reads them in `budget_order`; `household_benefits` first reads them in the order its three `add` calls imply (the terms the `all` rate scales, then those the `non_sp` rate scales, then the rest), each in `budget_order`.
- `budget_variables_in_effect` and `uprated_benefit_variables` return subsequences of `budget_order`, in order.
- For each total and each switch, `fiscal_partition.yaml`'s `dropped_when` names the same terms as the registry's `budget_dropped_when`.
- Every `budget_programs()` variable exists in the system and appears once, and its sign agrees with the list it came from.
- For any registry, the loader returns the terms in `budget_order` and refuses one whose order drops or repeats a flagged term.
- `household_tax`, `household_benefits` and `household_net_income` equal `main`'s values exactly: the totals add the same float32 terms in the same order.

## Tests

New tests:

- `tests/test_budget_programs.py` (122 tests): drift between the registry and what each total's formula reads (traced, at current law and with council tax abolished, including the benefit total's first-read order); the partition's `dropped_when` against the registry's; the differential test against `main`'s two lists, frozen as literals in order, and its council tax and uprating rules; `budget_programs()` from a fresh interpreter, importing `policyengine_uk.programs`, the `household_tax` module or the package first; the existence check parametrized over all 86 terms; a Hypothesis property that the in-effect and uprated selections keep `budget_order`; Hypothesis properties on generated registries; and one negative case per loader check.
- `tests/test_budget_totals_pinned.py` (17 tests): `household_tax`, `household_benefits` and `household_net_income` for 13 hand-built households, one of them a London owner who pays the High Value Council Tax Surcharge from 2028, in 2023, 2025, 2026 and 2028 under four reforms (current law; `abolish_council_tax`; both uprating rates; both together with a flat basic income), 208 cells, with exact equality. The values come from `main` at 2da212ab, and the file passes on `main`'s code too.

| Run | Result |
| --- | --- |
| `test_budget_programs.py` | 122 passed |
| `test_budget_totals_pinned.py` on this branch, and on `main`'s code | 17 passed each |
| The new files with the related suites: `test_program_registry.py`, `test_fiscal_partition.py`, `test_household_benefits_properties.py`, `test_gov_spending_properties.py`, `test_uc_statutory_pay_properties.py`, `tests/aggregate_impact`, `test_build_metadata.py` | 671 passed, 10 skipped (9 need the certified dataset, 1 needs `policyengine_bundles`) |
| `tests/code_health` | 3088 passed |
| ruff 0.17.2 `format --check`, `check`, `check --select F811,F821`; `git diff --check` | clean |
| Wheel from `uv build` | holds `policyengine_uk/programs.py` and `policyengine_uk/programs.yaml` |

Mutation checks on the two new test files. I made each change in a scratch copy:

| Change | Result |
| --- | --- |
| Swap `income_tax` and `national_insurance` in `budget_order.tax` | 1 failed: the differential test (the pinned sums happen not to move) |
| Drop the State Pension's `budget_uprating: [all]` | 9 failed: the rules test and every uprating cell of the pinned test |
| Drop `budget_dropped_when` from the High Value Council Tax Surcharge | 5 failed: the rules test, the partition test for tax, the traced drift test with council tax abolished, and the 2028 `abolish_council_tax` and `abolish_uprating_basic_income` cells, where the surcharge payer's `household_tax` gives 76,493.6015625 against 72,993.6015625 |
| Make `household_tax` ignore `budget_dropped_when` | 9 failed: the traced drift test and every council tax abolition cell |
| Sort the output of `uprated_benefit_variables` (`test_budget_programs.py` only) | 3 failed: the subsequence property and both benefit drift cases |
| Reverse the benefit list in `household_benefits` (`test_budget_programs.py` only) | 2 failed: both benefit drift cases |
| Disable each of the loader's 18 checks in turn (the loader tests only) | each fails at least one test; disabling the check on flagged `programs` entries stops the package loading |
| Remove `budget: benefit` from Scottish Child Payment, order unchanged | the package refuses to load: `budget_order.benefit and the entries flagged budget: benefit differ: only ordered ['scottish_child_payment'], only flagged []` |
| Restore `main`'s two formula modules | 139 passed, as an unchanged result requires |

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---

Generated by Claude Code

[MaxGhenis]

Ready for hub review

Head be07af7dadb5f9b6d193a46de9426c287fb6fb0c on branch claude/apply-review-nits-cinyah, one commit on #2809's head e3308b11e6d78c9b2097bbcceef123d3652d45f1. That branch has no pull request, so no CI has run on it. #2809's branch is untouched.

Findings

1. Benefit order untested. Resolved.
   • test_budget_programs.py:207 (test_totals_read_exactly_the_registry_terms): for the benefit total, the test now asserts the trace's first-read order, list(dict.fromkeys(read)), against the order the formula's three add calls imply (terms the all rate scales, then non_sp, then all), computed from the raw YAML by raw_uprated (line 233).
   • test_budget_programs.py:399 (test_selections_are_ordered_subsequences_of_budget_order, Hypothesis, 100 examples): budget_variables_in_effect and uprated_benefit_variables return ordered subsequences of budget_order for every council tax switch state, both rates and any subset of benefits.
   • mA now fails 3 tests and mB 2 (table below).
2. Loader accepts misspelt fields. Resolved.
   • programs.py:92 _check_budget_fields refuses any field starting with budget outside budget, budget_dropped_when, budget_uprating, and refuses those fields on an entry without a budget flag. It runs on programs and budget_terms entries.
   • programs.py:174 refuses a budget_terms field outside TERM_FIELDS (line 38), so a dropped_when without the prefix also fails. programs.py:177 refuses a budget_terms entry without variable or budget. programs.py:128 refuses a non-string budget_dropped_when.
   • The reviewer's loader_typo_repro.py, run against this head, now raises RegistryError: council_tax: unknown budget fields ['budget_droped_when'].
   • test_budget_programs.py:514 loads a valid example registry, and test_loader_refuses_malformed_registry (line 606, cases from INVALID at line 539) has 17 negative cases, one per check, each matching the error message.
3. No surcharge payer in the grid. Resolved. test_budget_totals_pinned.py adds london_high_value_owner (exactly the household the review gave) and main_residence_value to HOUSEHOLD_INPUTS, with its 16 pins. I computed them by running main at 2da212a on the grid with the new household added. With the household present, main reproduced all 192 existing pins (0 mismatches), so adding it moves no other household's pin. The completeness check now asks for at least 13 households. mG now fails the 2028 abolish_council_tax and abolish_uprating_basic_income cells: 76,493.6015625 against 72,993.6015625.
4. modelled_policies claim. Resolved in PR_BODY.md. I read policyengine-core's tax_benefit_system.py:832-836: it loads the whole file with yaml.FullLoader and stores it as modelled_policies. The body now says that attribute and any metadata built from it carry budget_terms, budget_order and the new fields. I didn't read policyengine-api or the coverage page, so the body makes no claim about how they render.
5. Attribution and process text. The body now ends with 🤖 Generated with [Claude Code](https://claude.com/claude-code) and no session link. I didn't edit the earlier hand-off comment on Build the budget totals from the programme registry #2809, which still carries process text. Whether to edit or hide it is the hub's call.
6. CONTRIBUTING recipe. Resolved. .github/CONTRIBUTING.md:84-99 gives four numbered steps: flag the entry; append to budget_order; add a fiscal_partition.yaml head listing the total, plus its dropped_when entry if the term carries budget_dropped_when; and for a benefit, GOV_SPENDING_VARIABLES or NOT_PAID_BY_GOVERNMENT with the reason. It also says that misspelt budget fields stop the package loading.
7. Cross-specification invariant. Resolved. test_budget_programs.py:244 (test_partition_drops_the_registry_terms, for tax and benefit): for every switch in either file, partition.dropped(total, [switch]) equals the registry terms of that total carrying that budget_dropped_when.
8. Existence property covers 50 of 86. Resolved. test_each_budget_program_exists_once_with_its_list_sign (line 357) is parametrized over budget_programs(), giving 86 cases.
9. Registry consistency. Resolved. private_school_vat_change gains program: vat (programs.yaml:885). expected_ltt is now "Land Transaction Tax on household purchases" (programs.yaml:814).
10. YAML parse cost. Resolved in this repo. load_registry uses yaml.CSafeLoader when PyYAML has it, and falls back to SafeLoader (programs.py:40, :89). Measured here on one call: 50.4 ms with SafeLoader, 5.7 ms with CSafeLoader, identical parsed objects. Core's per-build parse (FullLoader, tax_benefit_system.py:835) lives in policyengine-core, so this job leaves it alone.
11. End-user documentation exposes YAML field names. Resolved. household_tax.py:12-16 and household_benefits.py:16-21 now describe the totals in plain words: no field names, and the voice-table openers.
12. Changelog type. Applied. changelog.d/2702.changed.md becomes changelog.d/2702.added.md, with the text unchanged.

Voice rows

Where	Done
Body, invariants	"Each total's formula reads the flagged terms and no others, less …"
Body, "No law changes, so this cites none."	cut
Body, mutation intro	"I made each change in a scratch copy"
programs.py:6	"PolicyEngine's simulation API reports the change in household_tax less the change in household_benefits as the headline budget impact." (wording from the review, which checked the sim-api; I did not read that code)
programs.py:10-12	"imports only the standard library and yaml … while the package builds the tax-benefit system"
programs.py:147	"Every call returns the same tuple."
programs.yaml:21, CONTRIBUTING.md:71	"A benefit term without it takes both."
programs.yaml:31-34	"; it stays zero until a reform sets it"
programs.yaml:799-802	"some measure part of a listed programme, some a programme the list doesn't carry yet, some no programme"
corporate_sdlt, corporate_incident_tax_revenue_change, consumer_incident_tax_revenue_change notes	"; the model shares it across households by shareholding" / "by consumption"
Header and the three public_service notes	"at their cost to government"
two_child_limit_payment note	"whose regulations the Scottish Government decided on 3 December 2025 not to lay" (date from the review)
household_tax.py:13	"Total taxes the household owes: …" in plain words
household_benefits.py:17	"the benefits the household receives"
CONTRIBUTING.md:61	"Four fields set a term's place in the totals:"
model-structure.md:71	"in programs.yaml"
test_budget_programs.py:1-7	"add the budget terms … and no others"; "The drift tests parse the YAML themselves rather than through policyengine_uk.programs"; "reads the flagged terms and no others"
test_budget_programs.py:217-218	"the trace lists its reads in summation order"
test_budget_totals_pinned.py:9	"The test compares with exact equality, because …"

Tests run at this head

Run	Result
test_budget_programs.py	122 passed
test_budget_totals_pinned.py, on this branch and on main's code (2da212a worktree, this test file copied in)	17 passed each
New files plus test_program_registry.py, test_fiscal_partition.py, test_household_benefits_properties.py, test_gov_spending_properties.py, test_uc_statutory_pay_properties.py, tests/aggregate_impact, test_build_metadata.py	671 passed, 10 skipped (9 test_aggregate_impact_certified.py need the certified dataset; 1 test_build_metadata.py:83 needs policyengine_bundles)
tests/code_health	3088 passed
ruff 0.17.2 format --check, check, check --select F811,F821 policyengine_uk; git diff --check	clean
uv build --wheel from git archive HEAD	wheel holds policyengine_uk/programs.py and policyengine_uk/programs.yaml

Mutation and revert checks (each applied to the working tree, then restored with git checkout)

Change	Result
Swap income_tax and national_insurance in budget_order.tax	1 failed: the differential test (the pinned sums happen not to move)
Drop the State Pension's budget_uprating: [all]	9 failed: the rules test and every uprating cell of the pinned test
Drop budget_dropped_when from the High Value Council Tax Surcharge	5 failed: the rules test, the partition test for tax, the traced drift test with council tax abolished, and the 2028 abolish_council_tax and abolish_uprating_basic_income cells, where the surcharge payer's household_tax gives 76,493.6015625 against 72,993.6015625
Make household_tax ignore budget_dropped_when	9 failed: the traced drift test and every council tax abolition cell
Sort the output of uprated_benefit_variables (test_budget_programs.py only)	3 failed: the subsequence property and both benefit drift cases
Reverse the benefit list in household_benefits (test_budget_programs.py only)	2 failed: both benefit drift cases
Disable each of the loader's 18 checks in turn (the loader tests only)	each fails at least one test; disabling the check on flagged programs entries stops the package loading
Remove budget: benefit from Scottish Child Payment, order unchanged	the package refuses to load: budget_order.benefit and the entries flagged budget: benefit differ: only ordered ['scottish_child_payment'], only flagged []
Restore main's two formula modules	139 passed, as an unchanged result requires

I ran the first six rows and the last row against both new test files unless the row says otherwise.

Skipped

• The 9 certified-dataset tests and the policyengine_bundles test (above). I used no survey data.
• CI: this branch has no pull request. The hub moves Build the budget totals from the programme registry #2809's branch.
• The review's seeded 3,150-value differential: this change touches no formula arithmetic. The pinned grid on both trees covers the totals.

Judgment calls

• The loader allow-lists every budget_terms field, but for programs entries it checks only fields starting with budget, because those entries carry the coverage schema's own fields.
• The benefit drift test asserts first-read order rather than the full trace, because the formula reads each term up to three times.
• The household docstrings name council tax abolition as the example of a removed term, because it is the only budget_dropped_when switch today.
• I kept the changelog text as it was. The review flagged nothing in it, and only the type changes.
• I left the earlier hand-off comment unedited (finding 5).
• PR_TITLE.txt stays unchanged.

---

Generated by Claude Code

[MaxGhenis]

Picking this up for the UK hub. Reply here if you're mid-change and I'll stop.

---

Generated by Claude Code

[MaxGhenis]

Picking this up for the UK hub. Reply here if you're mid-change and I'll stop.

---

Generated by Claude Code

[MaxGhenis]

PR_TITLE.txt

Build the budget totals from the programme registry

---

Generated by Claude Code

[MaxGhenis]

PR_BODY.md

Addresses #2702
Reproducer: n/a: this changes no result
axiom: n/a: this changes how the package lists programmes and changes no rule or parameter value

## What the code does now

`policyengine_uk/programs.yaml` defines the budget totals as well as model coverage.

- Each `programs` entry whose `variable` a total adds carries `budget: tax` or `budget: benefit`, and an optional `group` for terms that move together (for example the five childcare entitlements, or ESA's two parts).
- Terms that no programme entry carries sit in a new `budget_terms` section. Each has a `kind`: `programme_measure` (the amount counted for a listed programme, such as `esa_income` for ESA, with `program` naming it), `programme` (a programme the coverage list does not carry yet), `reform_measure` (not a programme: a hypothetical tax or benefit, or a change from baseline such as `vat_change` and the employer NI incidence terms) or `public_service` (public services in kind). These entries stay out of `programs`, so the coverage page never lists them.
- `budget_dropped_when: gov.contrib.abolish_council_tax` sits on council tax, the High Value Council Tax Surcharge and Council Tax Reduction. `budget_uprating` sits on the State Pension (`[all]`) and basic income (`[]`); every other benefit term takes both `gov.contrib.benefit_uprating` rates.
- `budget_order` gives each total's summation order. The loader refuses a registry whose flags and order disagree, a misspelt field whose name starts with `budget`, and any `budget_terms` field outside the documented ones.

`policyengine_uk/programs.py` parses the YAML (with PyYAML's C parser when it is available) with no import from the package's top level. `budget_programs()` returns one `BudgetProgram` per term (variable, label, `tax` or `benefit`, group, kind, programme id, dropped-when parameter, uprating rates), taxes first, each in summation order. `household_tax` and `household_benefits` build their lists from it, and `HOUSEHOLD_TAX_VARIABLES` and `HOUSEHOLD_BENEFIT_VARIABLES` stay in their modules with the same contents in the same order. `build_metadata.DATA_BUILD_SURFACE` now includes `programs.py`.

policyengine-core loads the whole of `programs.yaml` as the system's `modelled_policies`, so that attribute, and any metadata built from it, now carries `budget_terms`, `budget_order` and the new fields on each `programs` entry.

The branch merges `main` at 46a60678ff52d829253572393abdcc6a03ceb6a7, which brings #2757's corrections to programme agencies and national coverage in `programs.yaml` and `test_program_registry.py`, and #2751's rounding of projected State Pension weekly rates to the nearest 5p. `programs.yaml` keeps `main`'s corrections alongside the budget fields; the merge needed no conflict resolution.

## Invariants

- `HOUSEHOLD_TAX_VARIABLES` and `HOUSEHOLD_BENEFIT_VARIABLES` equal `main`'s lists, element by element.
- Each total's formula reads the flagged terms and no others, less those whose `budget_dropped_when` parameter is true. `household_tax` reads them in `budget_order`; `household_benefits` first reads them in the order its three `add` calls imply (the terms the `all` rate scales, then those the `non_sp` rate scales, then the rest), each in `budget_order`.
- `budget_variables_in_effect` and `uprated_benefit_variables` return subsequences of `budget_order`, in order.
- For each total and each switch, `fiscal_partition.yaml`'s `dropped_when` names the same terms as the registry's `budget_dropped_when`.
- Every `budget_programs()` variable exists in the system and appears once, and its sign agrees with the list it came from.
- For any registry, the loader returns the terms in `budget_order` and refuses one whose order drops or repeats a flagged term.
- `household_tax`, `household_benefits` and `household_net_income` equal `main`'s values exactly: the totals add the same float32 terms in the same order.

## Tests

New tests:

- `tests/test_budget_programs.py` (124 tests): drift between the registry and what each total's formula reads (traced, at current law and with council tax abolished, including the benefit total's first-read order); the partition's `dropped_when` against the registry's; the differential test against `main`'s two lists, frozen as literals in order, and its council tax and uprating rules; `budget_programs()` from a fresh interpreter, importing `policyengine_uk.programs`, the `household_tax` module or the package first; the existence check parametrized over all 86 terms; a Hypothesis property that the in-effect and uprated selections keep `budget_order`; Hypothesis properties on generated registries; and 19 negative loader cases, one for each condition the loader refuses (its 16 refusals, with the two field-name checks counted once for `programs` and once for `budget_terms`, and the `budget_terms` check for a missing `variable` or `budget` counted once per field).
- `tests/test_budget_totals_pinned.py` (17 tests): `household_tax`, `household_benefits` and `household_net_income` for 13 hand-built households, one of them a London owner who pays the High Value Council Tax Surcharge from 2028, in 2023, 2025, 2026 and 2028 under four reforms (current law; `abolish_council_tax`; both uprating rates; both together with a flat basic income), 208 cells, with exact equality. Every value comes from a run of `main` at 46a60678ff52d829253572393abdcc6a03ceb6a7 with this file copied in. Six of the 208 moved from the earlier pins taken at 2da212ab, all in 2028: the pensioner couple under current law and `abolish_council_tax`, and the East of England household under all four reforms. #2751 rounds the projected 2028 new State Pension weekly rate from £259.2357 to £259.20, which moves each 2028 new State Pension by one or two float32 units in the last place: £11,000.0010 becomes £10,999.9990, £9,000 becomes £8,999.9990 and £10,500.0010 becomes £10,500. The branch, `main` at 46a60678 and `main` at 6c6e229a5 all reproduce the 208 pins exactly.

| Run | Result |
| --- | --- |
| `test_budget_programs.py` | 124 passed |
| `test_budget_totals_pinned.py` on this branch, on `main`'s code at 46a60678 and on `main`'s code at 6c6e229a5 | 17 passed each |
| `test_program_registry.py` | 119 passed |
| The new files with the related suites: `test_program_registry.py`, `test_fiscal_partition.py`, `test_state_pension_rate_rounding.py`, `test_household_benefits_properties.py`, `test_gov_spending_properties.py`, `test_uc_statutory_pay_properties.py`, `tests/aggregate_impact`, `test_build_metadata.py` | 716 passed, 10 skipped (9 need the certified dataset, 1 needs `policyengine_bundles`) |
| The new files, `test_program_registry.py`, `test_fiscal_partition.py` and `test_state_pension_rate_rounding.py` on this branch merged with `main` at 6c6e229a5 | 337 passed |
| `tests/code_health` | 3090 passed |
| ruff 0.17.2 `format --check`, `check`, `check --select F811,F821`; `git diff --check` | clean |
| Wheel from `uv build` of `git archive HEAD` | holds `policyengine_uk/programs.py` and `policyengine_uk/programs.yaml` |

Mutation checks on the two new test files. I made each change in a scratch copy:

| Change | Result |
| --- | --- |
| Swap `income_tax` and `national_insurance` in `budget_order.tax` | 1 failed: the differential test (the pinned sums happen not to move) |
| Drop the State Pension's `budget_uprating: [all]` | 9 failed: the rules test and every uprating cell of the pinned test |
| Drop `budget_dropped_when` from the High Value Council Tax Surcharge | 5 failed: the rules test, the partition test for tax, the traced drift test with council tax abolished, and the 2028 `abolish_council_tax` and `abolish_uprating_basic_income` cells, where the surcharge payer's `household_tax` gives 76,493.6015625 against 72,993.6015625 |
| Make `household_tax` ignore `budget_dropped_when` | 9 failed: the traced drift test and every council tax abolition cell |
| Sort the output of `uprated_benefit_variables` (`test_budget_programs.py` only) | 3 failed: the subsequence property and both benefit drift cases |
| Reverse the benefit list in `household_benefits` (`test_budget_programs.py` only) | 2 failed: both benefit drift cases |
| Disable each of the loader's 19 refused conditions in turn (the loader tests only) | each fails its own negative case; disabling either order check also fails the generated-registry property |
| Remove `budget: benefit` from Scottish Child Payment, order unchanged | the package refuses to load: `budget_order.benefit and the entries flagged budget: benefit differ: only ordered ['scottish_child_payment'], only flagged []` |
| Restore `main`'s two formula modules | 141 passed, as an unchanged result requires |

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---

Generated by Claude Code

[MaxGhenis]

Ready for hub review

Head 3d234c759fbd48c1d7bd69bbe25cb682a7becc4e on branch claude/merge-main-pr-2809-nits-yrnrwq (the branch this session's harness pins). The branch starts from 46b788c77f9bba60bfba4bc5861e9ae2bc7f50de, the stopped session's head on claude/merge-main-pr-2809-nits-hqke0f, and adds two commits:

• f171f24a adds two loader cases;
• 3d234c75 corrects the date wording in the Two Child Limit Payment note.

No pull request has this branch as its head, so no CI has run on it. #2809's branch is still at e3308b11e.

Checking the stopped session's two commits

Merge 9f16d9c69 (item 1). I ran git merge-tree --write-tree be07af7 46a60678ff52d829253572393abdcc6a03ceb6a7. It merged cleanly to tree d97260a9dbe5, the same tree as 9f16d9c69. So the merge commit is exactly what git produces: programs.yaml keeps #2757's agency and coverage corrections next to the budget fields, and test_program_registry.py is main's. import policyengine_uk works. test_program_registry.py gives 119 passed and test_budget_programs.py gives 124 passed (122 before the two new cases).

Pins 46b788c77 (item 2). I checked out 46a60678 in a separate worktree, copied this branch's test_budget_totals_pinned.py in, and ran its whole grid with that worktree's policyengine_uk on the path. I confirmed that the import came from the worktree, then recorded all 208 cells. Every pin in the file equals that run bit for bit: 0 differences. I edited no pin. Six of the 208 pins moved from the earlier ones taken at 2da212a, all in 2028:

Reform	Household	Moved cell
baseline	pensioner_couple	benefits 22,573.3203125 → 22,573.318359375; net income 20,293.3203125 → 20,293.318359375
abolish_council_tax	pensioner_couple	benefits 20,473.3203125 → 20,473.318359375; net income 20,293.3203125 → 20,293.318359375
baseline	east_of_england_mixed	benefits 11,700.0009765625 → 11,700.0
abolish_council_tax	east_of_england_mixed	benefits 11,700.0009765625 → 11,700.0
uprating	east_of_england_mixed	benefits 12,405.0009765625 → 12,405.0
abolish_uprating_basic_income	east_of_england_mixed	benefits 15,593.0009765625 → 15,593.0

The input that moves them is #2751's rounding. In 2028, the projected new State Pension weekly rate falls from 259.2357292385871 to £259.20, and the basic rate rises from 198.6435405562153 to £198.65. state_pension scales each reported amount by that rate, in float32. At 2da212a and at 46a6067, it gives these values for the people with a reported new State Pension:

• pensioner couple, a: 11,000.0009765625 → 10,999.9990234375;
• pensioner couple, b: 9,000.0 → 8,999.9990234375;
• East of England household, b: 10,500.0009765625 → 10,500.0.

None of the other 2028 cells moved:

• In the pensioner couple's two uprating cells, the totals round to the same float32 values.
• The Scottish pensioner's basic_state_pension changes from 10,329.4638671875 to 10,329.7998046875, but their state_pension is the reported £12,500 at both commits.
• 2023 to 2026 aren't projected years, so nothing moved there.

Against main at 6c6e229a5. I ran the same grid from a 6c6e229a5 worktree: 0 differences from the pins, and pytest test_budget_totals_pinned.py passed 17 there. So I left the merge at 46a60678. The pins hold on both commits.

I also checked a merge of this head with 6c6e229a5: git merge-tree is clean (tree e4df5cf22abd). On that tree, test_budget_programs.py, test_program_registry.py, test_budget_totals_pinned.py, test_fiscal_partition.py and test_state_pension_rate_rounding.py give 337 passed.

Items 3 to 6

3. Mutation and revert table. I reran every row at this head. Each change went into the working tree, ran, and was restored with git checkout HEAD. Results:

Change	Result
Swap income_tax and national_insurance in budget_order.tax	1 failed: test_household_tax_variables_equal_main
Drop the State Pension's budget_uprating: [all]	9 failed: the rules test and all 8 uprating cells of the pinned test
Drop budget_dropped_when from the High Value Council Tax Surcharge	5 failed: the rules test, test_partition_drops_the_registry_terms[tax], test_totals_read_exactly_the_registry_terms[tax-abolished], and the 2028 abolish_council_tax and abolish_uprating_basic_income cells; the surcharge payer's household_tax gives 76,493.6015625 against 72,993.6015625
Make household_tax ignore budget_dropped_when	9 failed: the traced tax drift test with council tax abolished and all 8 abolition cells
Sort the output of uprated_benefit_variables (test_budget_programs.py only)	3 failed: the subsequence property and both benefit drift cases
Reverse the benefit list in household_benefits (test_budget_programs.py only)	2 failed: both benefit drift cases
Disable each of the loader's 19 refused conditions in turn (loader tests only)	each fails its own case, and only that case, except the two order checks, which also fail the generated-registry property
Remove budget: benefit from Scottish Child Payment, order unchanged	the package refuses to load: budget_order.benefit and the entries flagged budget: benefit differ: only ordered ['scottish_child_payment'], only flagged []
Restore main's two formula modules (from 46a60678)	141 passed (124 + 17)

Apart from the row counts noted, every count matches the earlier table. The revert row moves from 139 to 141 because of the two new loader cases.

4. Loader cases. programs.py has 16 raise statements in the loader. Counting each refused condition separately gives 19:

• _check_budget_fields runs on both programs and budget_terms entries, so its two checks count twice;
• programs.py:177 refuses a budget_terms entry that lacks variable or budget, and that counts once per field.

The 17 earlier cases covered 17 of these 19. I disabled each condition in turn against be07af7's test file, and two survived (19 passed each):

• a budget field on a budget_terms entry without a budget flag (programs.py:100-103 via :173);
• a budget_terms entry without variable (programs.py:177).

test_budget_programs.py adds a case for each: budget_term_field_without_flag and budget_term_without_variable (lines 601-608, with the helper _unflag_term_with_dropped_when at line 538). With these, all 19 mutants fail. I can't say which 18 checks the earlier hand-off counted, so the body now gives the count of 19 and how it's made up.

5a. programs.py:6-8, the sim-api sentence. I couldn't read it. This session's GitHub access covers only policyengine-uk, and gh api repos/PolicyEngine/policyengine-sim-api/contents/... returns 403 ("GitHub access to this repository is not enabled for this session"). I didn't fetch it by another route. The sentence is unchanged; it rests on the review's reading and the hub's reading of build_budgetary_impact.

5b. Two Child Limit Payment note. in_effect.yaml cites "Scottish Government confirmation that the regulations will not be laid, 3 December 2025". That dates a confirmation, not a decision. The note said "decided on 3 December 2025 not to lay"; programs.yaml:1099 now says "whose regulations the Scottish Government confirmed on 3 December 2025 it will not lay".

6. PR text. I posted PR_BODY.md above. It keeps the first three lines and adds a paragraph on the merge at 46a60678. It also:

• gives 124 tests in test_budget_programs.py and 19 loader cases;
• says the pins come from main at 46a60678, that 6 of the 208 moved, and why;
• says the pins pass on the branch and on main's code at both SHAs;
• updates every count in the test and mutation tables.

PR_TITLE.txt is unchanged.

Tests run at this head

Run	Result
test_budget_programs.py	124 passed
test_program_registry.py	119 passed
test_budget_totals_pinned.py: on this branch, on main's code at 46a60678, and on main's code at 6c6e229a5 (worktree, this file copied in)	17 passed each
Grid derivation (208 cells) on the branch, 46a60678 and 6c6e229a5	0 differences from the pins on each
The new files, test_program_registry.py, test_fiscal_partition.py, test_state_pension_rate_rounding.py, test_household_benefits_properties.py, test_gov_spending_properties.py, test_uc_statutory_pay_properties.py, tests/aggregate_impact, test_build_metadata.py	716 passed, 10 skipped (9 test_aggregate_impact_certified.py need the certified dataset; test_build_metadata.py:83 needs policyengine_bundles)
This head merged with 6c6e229a5: the two new files, test_program_registry.py, test_fiscal_partition.py and test_state_pension_rate_rounding.py	337 passed
tests/code_health	3090 passed
ruff 0.17.2 format --check ., check ., check --select F811,F821 policyengine_uk; git diff --check be07af7 HEAD	clean
uv build --wheel from git archive HEAD	the wheel holds policyengine_uk/programs.py and policyengine_uk/programs.yaml

The environment came from uv sync --frozen --extra dev. main's uv.lock and pyproject.toml at both SHAs differ from the branch only in the package version, so I ran the main worktrees with the branch's environment and the worktree first on PYTHONPATH.

Skipped

• The 9 certified-dataset tests and the policyengine_bundles test, as above. I used no survey data.
• Reading the sim-api source (5a): this session's GitHub access doesn't reach that repository.
• CI: this branch has no pull request. The hub moves Build the budget totals from the programme registry #2809's branch.

Judgment calls

• I kept the merge at 46a60678 because every pin passes on 6c6e229a5 and a merge with it is clean.
• I counted loader conditions per section and per field, giving 19, rather than guess at the earlier count of 18. The test comment above INVALID says that the field-name checks get one case per section.
• I reworded the Two Child Limit Payment note to "confirmed … it will not lay", matching the reference title, rather than state a decision date that the reference doesn't give.
• I left 46b788c77's pins and docstring as they were, because my run reproduced them exactly.
• I left the earlier hand-off comments on Build the budget totals from the programme registry #2809 as they were.

---

Generated by Claude Code

[MaxGhenis]

Ready for hub review

Verdict: APPROVE WITH NITS at 3d234c759fbd48c1d7bd69bbe25cb682a7becc4e.

I reviewed git show be07af7da, the merge 9f16d9c69 and git diff 9f16d9c69..3d234c759. I found no P1 or P2 problem. The merge matches what git produces, the 208 pins come from main at 46a60678 bit for bit, and the head gives the same 208 values. Both new loader cases fail when I disable the condition each covers. CI passes at the head: 20 checks succeeded, and one superseded duplicate run was cancelled. There are two P3 findings.

Findings

1. P3: two of the 19 "refused conditions" change only the error message, and the body's count rule isn't applied evenly

policyengine_uk/programs.py:100-103 (reached from :173) and :177-180; test_budget_programs.py:601-608; PR body, Tests.

• I disabled each condition in a scratch copy (mutrepo) and ran the loader tests (pytest test_budget_programs.py -k "loader or registr"). Unmutated: 30 passed.
  • m1: skip the "need a budget flag" check for budget_terms entries only. Result: 1 failed, 29 passed, and the failure is test_loader_refuses_malformed_registry[budget_term_field_without_flag] with AssertionError: Regex pattern did not match. The registry is still refused, because :177 refuses every budget_terms entry without budget. Only the message changes.
  • m2: change :177 to if "budget" not in entry:. Result: 1 failed, 29 passed, and the failure is [budget_term_without_variable] with KeyError: 'variable' at programs.py:107. The registry is still refused, but by a KeyError, not a RegistryError.
  • So both new cases kill their mutants, and the hand-off's claim holds: each fails its own case and no other. They kill them through the message or the exception type. The first condition can never refuse a registry that :177 would accept.
• The body counts _check_budget_fields's two checks twice, because two sections call the function. But _term's five checks (budget value, budget_uprating on a tax, unknown uprating, group type, budget_dropped_when type) also run on both sections, and the body counts each of them once. The count of 19 depends on that choice; it isn't a property of the code.
• Nothing is wrong at run time. Suggested body text, replacing the bracket after "19 negative loader cases, one for each condition the loader refuses":

  (its 16 refusals give 19 conditions when the two field-name checks count once for each section and the variable and budget test at line 177 counts once for each field; for budget_terms entries, line 177 already refuses an unflagged entry, so the flag check's case there tests its message)

• You could also delete the flag check's budget_terms branch as dead code. That's optional.

2. P3: previous finding 5 is only partly closed

• The body now ends with 🤖 Generated with [Claude Code](https://claude.com/claude-code) and no session link. That part is closed.
• The PR still has the earlier hand-off and status comments with process text: 6563708564, 6580155541, 6581377171 to 6581378034, 6581750666, 6585945640 and 6587481276 to 6587496667, each opening with "Ready for hub review", "Picking this up for the UK hub" or a file name. The author left them alone on purpose. Whether to hide them is the hub's call; I've listed it here so it isn't lost.

The twelve P3s from the previous round, at the head

#	Status	Evidence
1 Benefit order untested	Closed	I reran mA (sort the output of uprated_benefit_variables) and mB (reverse benefits in household_benefits) against test_budget_programs.py. mA: 3 failed, 121 passed (the subsequence property and test_totals_read_exactly_the_registry_terms[benefit-current_law] and [benefit-abolished]). mB: 2 failed, 122 passed (both benefit cases).
2 Loader accepts misspelt fields	Closed	_check_budget_fields (programs.py:92), the TERM_FIELDS allow-list (:174) and 19 INVALID cases. See finding 1 for the two conditions that change only the message.
3 No surcharge payer	Closed	london_high_value_owner is in the grid. mG (drop the surcharge's budget_dropped_when) gives 5 failed, 136 passed: the rules test, [tax-abolished], test_partition_drops_the_registry_terms[tax], and the 2028 abolish_council_tax and abolish_uprating_basic_income cells (expected (72993.6015625, 0.0, 107006.3984375), got (76493.6015625, 0.0, 103506.3984375)).
4 modelled_policies claim	Closed	The body now says core loads the whole file as modelled_policies. I didn't reread core.
5 Attribution and process text	Partly closed	See finding 2.
6 CONTRIBUTING recipe	Closed	.github/CONTRIBUTING.md:84-97 lists all four steps. GOV_SPENDING_VARIABLES (variables/gov/gov_spending.py:4) and NOT_PAID_BY_GOVERNMENT (tests/code_health/test_household_benefits_in_government_spending.py:28) are where it says.
7 Cross-specification invariant	Closed	test_partition_drops_the_registry_terms (test_budget_programs.py:244), and mG kills it.
8 Existence property 50 of 86	Closed	test_each_budget_program_exists_once_with_its_list_sign (:357) is parametrized over budget_programs().
9 Registry consistency	Closed	private_school_vat_change has program: vat (programs.yaml:885), and expected_ltt is "Land Transaction Tax on household purchases".
10 YAML parse cost	Closed in this repo	_YAML_LOADER = getattr(yaml, "CSafeLoader", yaml.SafeLoader). Core's own parse is outside this repo.
11 YAML field names in end-user docs	Closed	household_tax.py:12-16 and household_benefits.py:16-21 describe the totals in plain words.
12 Changelog type	Closed	changelog.d/2702.added.md.

Checked and sound

• The merge. git merge-tree --write-tree be07af7da 46a60678f gives d97260a9dbe51a0c7b847b43307be1bc94b7b8b4, which is 9f16d9c69^{tree}. The merge's parents are be07af7da and 46a60678f. So programs.yaml holds Correct programme agencies and national coverage #2757's corrections next to the budget fields, with no hand resolution. test_program_registry.py (119) and test_budget_programs.py (124) give 243 passed at the head.
• The pins. I wrote a grid script that imports the head's test_budget_totals_pinned.py and evaluates all 208 cells. I ran it twice, once with the head on the path and once with main at 46a60678 first, and logged policyengine_uk.__file__ to confirm which tree each run imported. Results:
  • head against pins: 0 differences; main against pins: 0; head against main: 0.
  • Against be07af7's pins from 2da212ab: exactly 6 cells differ, the six listed in the body and hand-off, with the same old and new values (pensioner couple: baseline and abolish_council_tax; East of England: all four reforms; all in 2028).
  • pytest test_budget_totals_pinned.py gives 17 passed on main's code at 46a60678 and at today's main, 0080274d78e2b48e57682ac6a035fda6e11334e9. Each run had the file copied in and the worktree on PYTHONPATH.
• Why the pins moved. I printed the parameters and the 2028 values at 2da212ab and at 46a60678:
  • New State Pension weekly rate, 2027: 250.71102654036292 → 250.7. 2028: 259.2357292385871 → 259.2.
  • Basic State Pension, 2028: 198.6435405562153 → 198.65.
  • state_pension in 2028: pensioner couple a 11000.0009765625 → 10999.9990234375; b 9000.0 → 8999.9990234375; East of England b 10500.0009765625 → 10500.0. Scottish pensioner: 12500.0 at both. basic_state_pension: 10329.4638671875 → 10329.7998046875.
  • These shifts are 2, 1 and 1 float32 units in the last place, which agrees with the body's "one or two". create_rate_projection.py's docstring gives the same arithmetic as the body: £250.70 × 1.034 = £259.2238, rounded to £259.20, where the unrounded £259.2357 would round to £259.25.
• Merge with today's main. git merge-tree --write-tree 3d234c759 origin/main (0080274d) is clean, giving tree ba2d6e18. On that tree, test_budget_totals_pinned.py, test_budget_programs.py, test_program_registry.py, test_fiscal_partition.py and test_state_pension_rate_rounding.py give 337 passed.
• At the head. test_budget_totals_pinned.py, test_fiscal_partition.py and test_state_pension_rate_rounding.py give 94 passed. ruff 0.17.2 format --check . reports 2169 files already formatted, and check . reports all checks passed. git diff --check 9f16d9c69 HEAD is clean.
• The loader. programs.py has 16 raise statements in the loader. INVALID has 19 cases, and their messages match. Both new cases survive at be07af7's test file: under m1 and m2, the only failing case is the new one.
• The note. programs.yaml:1099 reads "whose regulations the Scottish Government confirmed on 3 December 2025 it will not lay". in_effect.yaml cites "Scottish Government confirmation that the regulations will not be laid, 3 December 2025". The two agree: the note dates a confirmation, as the reference does, and names the actor.
• The pinned file's docstring and comment now name 46a60678, and the values match that commit.
• CI at the head. 21 check runs: 20 succeeded, including Lint, YAML, all seven Python shards, the four install and smoke-import jobs, Aggregate impact and the documentation builds.
  • The one cancelled run is Fix evidence (job 125551418763, run 41525315732, pull_request_target, 21:06:40). pr_fix_evidence.yaml sets cancel-in-progress per PR, and a second Fix evidence run on the same SHA started one second later (run 41525318702) and succeeded.
  • The job's log is gone (BlobNotFound, HTTP 404), but the cancellation comes from concurrency, not from this delta.
  • The aggregate-impact comment, updated at 22:18:58 against base 0080274d78, reports 0 checks broken, 1 failing on both sides, and 0 moves flagged.
• The PR. Draft, mergeable: true, mergeable_state: blocked. The live body opens with Addresses #2702 and the two lines after it, and ends with the correct attribution line.

What I couldn't check

• programs.py:6-8 (the sim-api sentence). This session's GitHub access covers only policyengine-uk. gh api repos/PolicyEngine/policyengine-sim-api/contents/projects/policyengine-simulation-executor/src/policyengine_simulation_executor/simulation_output_budget.py returns HTTP 403 ("GitHub access to this repository is not enabled for this session"). The sentence rests on the hub's reading of build_budgetary_impact and the previous round's.
• I disabled only the two loader conditions the new cases cover, not all 19.
• I didn't rerun the other mutation rows (the income_tax/NI swap, the State Pension uprating drop, household_tax ignoring budget_dropped_when, removing Scottish Child Payment's flag, restoring main's formula modules), tests/code_health (the body says 3090), the 716-test related-suite batch, or the wheel build. CI's Python shards and install jobs pass at the head.
• I didn't run main at 6c6e229a5. I ran today's main (0080274d), which descends from it, instead.
• I used no survey data. The certified-data result comes from CI's comment.

---

Generated by Claude Code

[juaristi22]

Approved at exact head 3d234c759fbd48c1d7bd69bbe25cb682a7becc4e. Review COMPLETE, with no findings or material evidence gaps; applicable required checks pass at this head.

All 311 selected checks passed. The independent paired comparison substituted the exact merge-base formulas into the same combined-reform simulations: all 78 values across 13 households and six final household/government outputs were exactly equal. The source/contract review also confirmed the same ordered 29 tax and 57 benefit terms, council-tax abolition exclusions, benefit-uprating exceptions and coverage-programme content apart from additive budget metadata.

No new statutory rule or parameter value is introduced. The new budget descriptors keep programme measures, reform measures and public services distinct from the coverage list.