Set the Personal Savings Allowance on rest-of-UK bands for Scottish taxpayers

[MaxGhenis]

Fixes #2347

axiom: ukpga/2007/3/12B encoded-correct (rulespec-uk c65fd7df, uk/statutes/ukpga/2007/3/12B.yaml, which carries the s. 12B(8)(a)(iv) and (b)(iv) Scottish and Welsh limbs at lines 85 and 110, as inputs, from 2024-04-06)

What changes

ITA 2007 s. 12B gives a Personal Savings Allowance of £1,000 with no higher-rate income, £500 with higher-rate but no additional-rate income, and nil with any additional-rate income. For a Scottish or Welsh taxpayer, s. 12B(8)(a)(iv) and (b)(iv) count as higher- or additional-rate income what "would, if the individual were not a Scottish taxpayer or Welsh taxpayer", be charged at those rates. So the allowance follows the rest-of-UK bands for everyone.

savings_allowance read tax_band, which follows the Scottish bands for a Scottish taxpayer. In 2026-27:

• taxable income from £62,430 (the Scottish advanced rate) to £125,140 got no allowance (law: £500);
• taxable income from £31,092 (the Scottish higher rate) to £37,700 got £500 (law: £1,000).

savings_allowance now works out the band itself on the rest-of-UK thresholds (rates.uk.thresholds[1] and [2]), with taxable income taken as adjusted net income less allowances, as tax_band does. Income exactly at a limit has none in the band above (ITA 2007 s. 10(2)-(3)). So £37,700 of taxable income gives £1,000 and £125,140 gives £500; the old >= gave £500 and nil. tax_band is unchanged: the Marriage Allowance conditions read it and do look at the Scottish rates.

This also fixes the periods 2016 and 2017, which tax_band's earlier formula covers. It put the higher band at thresholds[-2] (£150,000), so a higher-rate taxpayer on £60,000 got £1,000 (law: £500) and an additional-rate taxpayer on £200,000 got £500 (law: nil). The same formula still feeds Marriage Allowance in those years: #2356.

Impact (real run)

Measured on enhanced_frs_2023_24.h5 (local policyengine-uk-data storage, 6 October build) for 2026. This branch and main's code (c6d15c8) each ran a full simulation of the same data (impact_psa.py).

	main	This PR	Change
Income Tax	£349,674.5m	£349,666.1m	−£8.4m
of which savings income tax	£356.7m	£348.4m	−£8.4m

• Who changes: 957 records (393k people weighted), all Scottish taxpayers. Each gets £500 more allowance, either nil → £500 or £500 → £1,000. No one's allowance falls.
• Tax effect: 377 of those records (119k people) pay less tax, by up to £200 (£500 of interest at 40%). No one pays more.
• At the new head (49fad84) the totals are unchanged, at −£8.37m.

Tests

• New YAML (gov/hmrc/income_tax/allowances/savings_allowance.yaml):

  • a Scottish taxpayer on £80,000 with £1,000 of interest: £500 allowance and £200 savings tax (main: £0 and £400);

  • a Scottish taxpayer on £45,000: £1,000 and £0 (main: £500 and £100);

  • an English taxpayer on £80,000, unchanged;

  • a Scottish taxpayer on £130,000: nil;

  • 2016-17 at £60,000 (£500) and £200,000 (nil).

  • period 2017 (Case 7);

  • exactly at, and £1 over, each limit (Cases 8-11).

  On main, Cases 1, 2, 5, 6 and 7 fail and Cases 3 and 4 pass. At the previous head, Cases 8 and 10 failed.

• Changed YAML (allowances.yaml): the three allowance cases set tax_band as an input, which no longer drives the allowance. They now set adjusted net income and allowances for the same three bands.

• Property tests (test_savings_allowance_properties.py, Hypothesis, 12 single adults per example):

  • a Scottish and an English taxpayer with the same pay, interest and dividends get the same allowance;
  • the allowance equals the statute computed independently;
  • a higher pay never raises it.

  The first fails on main.

• The property tests always run four boundary rows: taxable income at, and £1 over, each limit.

• Run locally at 49fad84:

  • the full policy YAML suite, 3,019 passed;
  • the property tests, 2 passed. On main, the first fails.

• Review: a GPT-6.1 Sol review agreed with the s. 12B(8)(iv) reading and asked for the boundary fix, which is made here. Its follow-ups are handled as follows:

  • Marriage Allowance before 2017-18 is filed as tax_band before 2017-18 puts the higher band at £150,000, so 2016-17 higher-rate taxpayers get the Marriage Allowance #2356;
  • Gift Aid is round 1's Gift Aid is deducted from taxable income instead of raising the rate limits (ITA 2007 s. 414) #2253;
  • the allowance before 2016-17 is moot, because the model has no personal allowance before 2016.

🤖 Generated with Claude Code

[MaxGhenis]

CI run cancelled to free runners (UK hub, 2026-10-11 23:30Z). The PolicyEngine org is at GitHub's limit of 60 concurrent jobs, with about 360 queued. Fleet ops asked the hub to let the Budget-critical #2305 take runners first. This PR's queued run had not started, so I cancelled it. Nothing is wrong with the PR.

Please don't re-run it or push just to retrigger CI. The hub re-runs this run itself once #2305 has runners. The record is ~/reviews/uk-hub/ci-yield-2026-10-11.json; see notice 17 in ~/reviews/uk-hub/notices.md.

[MaxGhenis]

Hand-off to the UK hub (session local_66e4b1fd, PolicyBench UK round 2).

• Head: 60c6054a4ebaf2f1c7a1cdf05238c3719fceeb70 (includes main at ed7b42f95).
• Review: GPT-6.1 Sol, two rounds, then a delta review of the main merge; APPROVE at this head: ~/reviews/policybench-uk-round2/reviews/pe-uk-merge-delta-2327-2348.md (round 2: pe-uk-2348-review-r2.md).
• CI: not yet run at this head. The run was queued, and the hub cancelled it to free runners (notice 17, ~/reviews/uk-hub/ci-yield-2026-10-11.json); the hub re-runs it. No CI run has completed on this PR at any head: each was superseded by a push or cancelled while queued. Lint and Test are therefore the open gate.
• Local tests (~/reviews/policybench-uk-round2/fixes/evidence/), at the approved head before the main merge, which changed none of this PR's files (pr2348_approved_net.diff and pr2348_merged_net.diff are identical): 3,019 YAML cases and the 2 boundary property tests (pr2348_r2_full_yaml.log, pr2348_r2_py.log).
• PolicyBench check at this head: all 2,800 round 2 references recomputed on this head (audit/attribution/refs_psa_final.csv); 1 move, the same cells and values as at the reviewed head. With all five fix heads merged onto main ed7b42f95, 96 move, also unchanged.
• Impact (enhanced FRS, 2026, in the PR body): Income Tax −£8.37m; 393k Scottish taxpayers get £500 more savings allowance.
• Depends on: nothing. It and Band savings and dividends above property income #2352 both touch savings tax, in different files; they merge cleanly in either order.
• Decision: none. A bug fix to ITA 2007 s. 12B.
• Follow-ups filed, not owed here: tax_band before 2017-18 puts the higher band at £150,000, so 2016-17 higher-rate taxpayers get the Marriage Allowance #2356 (before 2017-18, tax_band puts the higher band at £150,000, which reaches the Marriage Allowance). Axiom parity: encoded correctly already (uk/statutes/ukpga/2007/3/12B.yaml).