UK local, bus fare, and CGT source-fact gaps

[juaristi22]

Summary

This issue records the UK source-fact gaps found while implementing the local facts in #202 and auditing the bus fare and Capital Gains Tax inputs used by policyengine-uk-data and Microcosm.

Chronicle owns preserved publisher artifacts, source cells, and emitted facts. It does not own target profiles, target selectors, measurement bindings, period or geography transformations, activation rules, or calibration logic. Those belong in the Microcosm consumer. This issue must not add or change Chronicle target profiles.

Status of the Chronicle work

#202 implements the local, bus, and CGT source asks with local validation passing:

• PIPR average private rents from January 2023 through June 2026: 14,656 monthly facts, including 316 England and Wales local authorities in every month plus the publisher's coded region, country, UK/GB, and Scottish BRMA rows.
• DfC Northern Ireland Universal Credit claimants from January 2023 through May 2025: 841 monthly person-level facts for 18 PCON24 constituencies and 11 local government districts. The source does not publish household counts or child-presence splits at these grains.
• DfT bus revenue, support, usage, and fare-index facts: 152 facts from BUS05i for 2023-2025, NTS0705 for 2023-2024, and BUS0415 for all published quarter ends from March 2023 through March 2026.
• HMRC CGT Tables 3, 5, and 6 for tax year 2023-24: 214 numeric facts covering gain by taxable-income band, UK country and English region, and age.

These local facts remain source facts only. Microcosm owns their consumer target definitions and support-aware activation.

The audit inventory below records what #202 closes and the source or consumer gaps that remain.

Bus fare and bus subsidy source gaps

#202 adds source packages for the official bus revenue, support, usage, and fare-index tables behind the bus inputs in policyengine-uk-data.

The following official ODS files were checked against the current uk-data constants:

Publisher table	Verified publisher facts	What it closes
DfT BUS05i, sheet BUS05ai	Passenger fare receipts for the year ending March 2025: England £3,417.389m, London £1,347.435m, England outside London £2,069.954m	Replaces the hardcoded England fare total and London/outside-London split
DfT BUS05i, sheet BUS05bi	Total estimated net support for the year ending March 2025: England £3,024.904m, London £1,130.214m, England outside London £1,894.690m	Replaces the hardcoded England bus-subsidy anchor
DfT NTS0705, sheet NTS0705a	2024 local-bus trips per person by real household-income quintile: 47.610, 37.451, 24.842, 18.452, and 12.978	Replaces the five hardcoded income-quintile distribution weights
DfT BUS0415	Quarterly local-bus fare indices by London, England outside London, England, Scotland, Wales, and Great Britain	Supplies publisher facts that Microcosm can use for explicit period alignment

The BUS05i and NTS0705 facts close the reproducibility gaps in uk-data's existing England anchors. They do not close an exact UK bus-fare or bus-subsidy total:

• BUS05ai and BUS05bi publish the required fare-receipt and net-support measures for England, London, and England outside London, not for every UK nation.
• BUS04i publishes operating revenue for England, Scotland, and Wales, but operating revenue is not the same measure as passenger fare receipts or net support and must not be substituted silently.
• DfT's bus tables cover Great Britain, so Northern Ireland requires a separate publisher.
• Any England-to-UK uplift, period conversion, or use of an operating-revenue proxy is a Microcosm transformation, not a Chronicle fact.

Chronicle implementation for bus facts

#202 adds facts-only source packages for:

1. dft-bus05i-revenue-support-2025: 24 facts preserving BUS05ai passenger fare receipts and BUS05bi net support for all numeric published geographic series and fiscal years 2023-2025.
2. dft-nts0705-local-bus-trips-2024: 24 facts preserving both London-bus and other-local-bus trips per person for five real household-income quintiles plus the all-income row in 2023 and 2024.
3. dft-bus0415-fares-index-2026: 104 facts preserving eight geographic fare-index series at all 13 published quarter ends from March 2023 through March 2026.

BUS04i should only be ported if a consumer needs operating-revenue facts in their own right. It does not close the fare-receipt or subsidy-total gap.

Separate Scotland, Wales, and Northern Ireland sources are still needed before a consumer can construct an exact UK total without an explicit estimation method.

Capital Gains Tax source gaps

Chronicle already contains the HMRC facts used for the current aggregate and size-of-gain calibration anchors:

• HMRC Table 1 aggregate taxpayers, gains, and liabilities in hmrc-cgt-statistics-2025.
• HMRC Table 2.1a size-of-gain bands in hmrc-cgt-size-of-gain-2025: nine retained bands at or above the £12,300 annual exempt amount, with taxpayer-count and gains facts for each band.

There is no additional Chronicle source gap for the current uk-data aggregate CGT targets or the 18 size-band facts. The remaining gap is in Microcosm: the size-band facts are present in its fixture but are not yet bound to consumer targets.

The following HMRC tables would add useful source facts beyond the current anchors:

HMRC table	Verified content	Consumer use and limitation
Table 3: size of gain by taxable income	Joint gain-size and taxable-income bands for individual taxpayers, with counts and gains	Closes a provenance gap for Microcosm's existing source-stage redraw. It is not automatically a calibration target: Microcosm currently fences these rows because taxable income is a proxy-conditioned dimension and the rounded/suppressed margins require reconciliation.
Table 5: country and region	2023-24 taxpayers, gains, and tax by UK country and English region	Directly supports consumer candidates for capital_gains by region, subject to Microcosm measurement and support rules.
Table 6: age	2023-24 individual taxpayers, gains, and tax across nine age bands	Directly supports consumer candidates for capital_gains by person age, subject to Microcosm measurement and support rules.

Chronicle implementation for CGT facts

#202 adds facts-only source packages for HMRC Tables 3, 5, and 6, preserving the publisher's tax-year periods, units, populations, dimensions, rounding, and suppression markers:

• hmrc-cgt-gain-by-income-2025: 142 numeric facts across ten gain bands plus the all-gains margin and six taxable-income bands plus the all-incomes margin. The package retains [Fewer than 1] count cells as guarded source cells and does not invent numeric values for them.
• hmrc-cgt-country-region-2025: 42 taxpayer-count, gain, and liability facts across the UK, its four countries, and the nine English regions.
• hmrc-cgt-age-2025: 30 taxpayer-count, gain, and liability facts across nine age bands plus the all-ages margin.

HMRC Tables 4, 7, 8, and 9 do not close current calibration gaps for capital_gains. They describe relief-specific gains, asset type and holding period, residential-property gains, and carried interest, while the current UK model exposes only aggregate capital_gains. They can be added later for validation or after corresponding model/data variables exist.

Chronicle acceptance boundary

The Chronicle source packages named above:

• preserve the official artifact with a pinned URL and checksum;
• retain the relevant source cells or rows and emit publisher-stated facts with explicit periods, units, geographies, and dimensions;
• preserve publisher rounding, suppression, and population notes;
• have deterministic package tests and pass the Chronicle validation and build suites;
• avoid computed UK totals, uprating, aging, reconciliation, target IDs, target profiles, selectors, measurement bindings, or calibration rules.

Microcosm consumer follow-up

Microcosm, not Chronicle, should:

• bind the existing 18 CGT size-band facts to consumer targets and reconcile them with aggregate HMRC margins;
• decide whether Table 3 remains source-stage evidence or becomes eligible for targets after a documented rounding/suppression reconciliation;
• define any region and age CGT targets from Tables 5 and 6 with explicit population and support rules;
• map BUS05ai to bus_fare_spending, BUS05bi to bus_subsidy_spending, and NTS0705a to the income-quintile distribution;
• declare all fare-index period alignment and any England-to-UK estimation method explicitly;
• leave exact UK bus totals inactive until matching Scotland, Wales, and Northern Ireland source measures are available, unless the consumer documents an estimation target separately from publisher facts.

Existing non-source gaps

The following remain out of scope because the publisher does not expose the required facts at the needed grain or because they are consumer modeling choices:

• Scotland and Northern Ireland local-authority rent facts matching the English/Welsh PIPR measure and grain.
• City of London and Isles of Scilly, which do not appear in the PIPR table.
• Northern Ireland Universal Credit household counts and child-presence splits at these area grains; the DfC workbook publishes claimant totals only.
• MSOA income transformations, derived rent targets, council-tax transformations, and all other consumer-side target construction.

[juaristi22]

Closed by #202