Solve Pension Credit take-up over entitled benefit units by MaxGhenis · Pull Request #510 · PolicyEngine/policyengine-uk-data

MaxGhenis · 2026-10-01T14:08:16Z

Stacked on #490: this PR's branch is #490's branch plus six commits. It targets main so that CI runs, since this repo's CI only runs on PRs into main. Until #490 merges, the diff here includes #490's commits; merge #490 first.

Results (real builds, FY2025-26, Great Britain)

Production builds (PE_UK_DATA_OA_CLONES=1, 512 epochs) with identical trained imputation models:

C2: Calibrate Housing Benefit to DWP's GB figures by age group #490, which this stacks on.
D′: this PR at 0598636, on the locked policyengine-uk 2.102.5. Its tree is identical to e386209 apart from one test and some wording. It predates d798: build G2 ("d798 implemented" above) describes the current rule.
E: this PR plus uk-data#513 (FRS benefit-unit capital for Pension Credit), on policyengine-uk main plus #2018, #2019, #1899, #1940 and #1925 (v2.104.2). It shows where the take-up solve lands once the model finds reporters entitled.

Earlier build D, at 381e398, solved over UK units and anchored SPI-synthetic reporters. Its figures differed from D′ by under 1%.

d798 implemented (10/5). Max ruled 37%: units not entitled in the calibration year claim Pension Credit at DWP's Savings Credit-only caseload take-up, with reporters and the solved fill for entitled non-reporters unchanged. Commit "Draw units not entitled in the calibration year at 37% Pension Credit take-up" builds it. The rate is a new take-up parameter, pension_credit_newly_entitled: 37% from FYE 2024 and 42% for FYE 2023, from DWP's table PC1.

Build G2 is this PR with that rule, at ba97fb1 on #490 9f18a3d. The head is now 6565630, on #490 f4d8f95; #490's later commits change only docs and comments. Both sides are measured on G2's own weights, with the old flags rebuilt from the same seeded draws. Draws and flags align, with no violations across 61,212 benefit units.

	G2 (37% rule)	same weights, old rule	G (old rule, own build)
GB Pension Credit 2025, saved file	£6.19bn, 1,414k	£6.19bn, 1,414k	£6.23bn, 1,428k
GB Pension Credit 2026	£6.836bn, 1,410k	£6.836bn, 1,405k
+10% guarantee: newly entitled units / share claiming	218k / 50%		213k / 20%
+10% guarantee: newly entitled non-reporters claiming	43%		9%
+10% guarantee: added spending (to newly entitled)	£1.91bn (£71m)		£1.89bn (£28m)

2025 is unchanged. No unit whose flag changed is entitled in the shipped 2025 file. The step and the saved file can disagree on 2025 entitlement (Rail, bus and fuel calibrations run after base-year materialisation, so the shipped file is not uprate-invertible #479), but here none of those units flipped.
2026 changes by 4.9k claims and £0.6m (£0.56m, from unrounded totals).
The take-up step is otherwise identical to G's. Fill 12.87%; 37.2% of GB units not entitled in 2025 are flagged.
Calibration log. Pension Credit at 1.04x for claims and 1.03x for spending; HB at 1.03x. 84.2% of national targets are within 10%, against 84.1% for G.
Other outputs, 2025.
- UK Pension Credit: £6.32bn.
- Relative poverty BHC/AHC: 20.18%/23.58%.
- Pensioner relative poverty AHC: 17.50%.
- Each is within 0.1pp or £0.04bn of G. G and G2 are separate unseeded builds, so that comparison includes calibration noise (Seed the calibration weight dropout so builds are reproducible #507). The same-weights columns above have none.
Caveat on the rate. Savings Credit-only units are a closed cohort (no Savings Credit for people reaching State Pension age from April 2016), and DWP no longer publishes breakdowns of them. People reaching State Pension age after 2025 also draw at 37%, though many will get Guarantee Credit-sized awards (DWP: 69%).

Rebased on 10/4 onto #490's working-age change (#490 at 33092bf; this PR at 7c4972b). #490 now also calibrates working-age Housing Benefit, to DWP's working-age line less supported and temporary accommodation. Build G is this PR at 7c4972b, compared with F, which is #490 at 304baad:

The take-up step is unchanged. Fill probability 12.87% over 1,368k GB entitled units, the same as D′.
Pension Credit after calibration. GB claims 1,428k and £6.23bn, with take-up among entitled units of 73.5%.
Calibration log (calibration year). DWP Pension Credit lands at 1.05x for claims and 1.04x for spending; simulating 2025 from the saved file gives 1,428k claims (1.03x), the round trip in Rail, bus and fuel calibrations run after base-year materialisation, so the shipped file is not uprate-invertible #479. Pension-age and net working-age Housing Benefit are at 1.03-1.04x. 84.1% of national targets are within 10%, against 84.4% for F.
+10% minimum guarantee reform. 213k GB units are newly entitled, 20% of them claim, and spending rises £1.89bn.
Against F, 2025.
- UK Pension Credit: £6.35bn, against £6.16bn.
- Relative poverty BHC/AHC: 20.26%/23.62%, against 20.32%/23.58%.
- Pensioner relative poverty AHC: 17.54%, against 17.66%.

The D′ tables below therefore describe this PR before d798, which changes only the draw for units not entitled in 2025 (see "d798 implemented" for G2). The commits were replayed unchanged, except that 2371630's message was reworded (now 334b327) for the round-3 review's note on capital-gains donors.

Acceptance gate: reporter baseline before any fill (PolicyEngine/uk-triple-lock#14)

The take-up step's aggregates for 2025, on survey weights, before calibration:

	D′ (this PR)	E (+#513, PE-UK fixes)	DWP 2025-26
GB benefit units reporting Pension Credit	1,196k	1,192k	1,382k claims
GB Pension Credit they report	£4.99bn	£4.97bn	£6.14bn
GB reporters the model finds entitled	771k (64%)	1,105k (93%)
GB entitled units	1,368k	2,129k
solved fill probability (over GB)	12.9%	20.9%
GB claims and spending after the fill	859k, £3.64bn	1,341k, £6.17bn	1,382k, £6.14bn

Calibration never has to remove reporters. The GB reporter baseline is below DWP's caseload and spending.
When the fill is zero. That depends on a different condition: the fill is 0 only if entitled GB reporters already make up 62% of entitled GB units. Here they make up 56% in D′ and 52% in E. Calibration against DWP's absolute targets is a separate, later step.
On 2.102.5, entitlement and the fill set the shortfall together. About a third of GB reporters get no modelled entitlement, and the entitled pool is too small for 62% take-up to reach DWP's caseload (62% of 1,368k is 848k). Calibration closes the gap by upweighting.
With Use the FRS benefit-unit capital for Pension Credit's capital test #513's FRS capital, the gap closes before calibration. 93% of reporters are entitled, and pre-calibration claims and spending land at 0.97x and 1.00x DWP. The PolicyEngine UK fixes also move entitlement, so E mixes both changes.

After calibration (dataset simulated for 2025)

	C2 (#490, old flags)	D′ (this PR)	E	DWP
GB Pension Credit claims	1,400k	1,428k	1,403k	1,382k
GB Pension Credit spending	£6.00bn	£6.22bn	£6.27bn	£6.14bn
GB units reporting Pension Credit (calibrated weights, incl. SPI-synthetic)	1,510k	1,736k	1,309k
take-up among entitled units	80.7%	73.4%	60.2%	62%
non-reporters with `would_claim_pc`	69%	12%	20%

Fit to other targets. Calibration log, final epoch: median |ratio-1| is 4.1% for C2, 4.0% for D′ and 4.0% for E. The share within 10% is 84.0%, 84.5% and 83.9%.
Calibration-year vs saved file. D′'s calibration-year Pension Credit is close to the figures above. Simulating 2025 from the saved base-year file shifts them slightly (Rail, bus and fuel calibrations run after base-year materialisation, so the shipped file is not uprate-invertible #479).
Reporter upweighting. Counting only survey reporters, the ones the take-up step anchors, calibration raises GB reporters by 41% on 2.102.5 (1,196k to 1,687k) to reach DWP's caseload. With Use the FRS benefit-unit capital for Pension Credit's capital test #513 and the policyengine-uk fixes, the rise is 5% (1,192k to 1,251k). The calibrated reporter totals in the table also count SPI-synthetic units (48.9k in D′, 58.6k in E). Their reports are imputed. None of D′'s 154 SPI-synthetic reporters is entitled; 24 of E's 152 are.

Paired reform: minimum guarantee +10% from 2025, same seeded flags in both runs

	C2 (old flags)	D′	E
GB units newly entitled	239k	214k	529k
share of them that claim	70%	20%	25%
share of newly entitled non-reporters that claim	66%	9%	18%
added GB Pension Credit spending	£2.00bn	£1.89bn	£1.89bn
of which to newly entitled units	£0.109bn	£0.028bn	£0.081bn

Under the rule in these builds, newly entitled non-reporters claimed at about the solved fill probability, which is low because reporters supply most of the 62%. d798 has since changed that to 37% (see "d798 implemented" above).

Take-up before and after calibration (d798)

The fill probability is solved on survey weights after the imputations, before cloning and calibration. It is the residual share of entitled non-reporters that tops FRS reporters up to DWP's 62%. Take-up among GB units entitled in 2025:

	entitled units	caseload take-up	spending take-up
before calibration, this PR's flags	1,370k	62.7%	70.4%
before calibration, old flags (main)	1,370k	86.6%	89.9%
after calibration, D′ (this PR)	1,944k	73.4%	76.9%
after calibration, C2 (#490, old flags)	1,736k	80.7%	81.1%
DWP FYE 2024 (PC1/PC2)	2,140k	62%	71%

Before calibration, spending take-up lands on DWP's 71% although only caseload is solved.
Calibration raises take-up to 73%. It meets DWP's caseload and spending by upweighting reporters, whose share of entitled units rises from 56% to 70%. The entitled pool stays below DWP's.
d798, ruled 37% on 10/5 and built (see above). Units not entitled in the calibration year draw at DWP's Savings Credit-only take-up. Their mean award of £9 a week is the closest DWP group to a reform's marginal entrants; Guarantee Credit is 69% at £87 a week.

The before-calibration rows use the post-imputation dataset rebuilt on 10/4 from #490's code. D′'s own run solved 12.9% over 1,368k units.

Other outputs (2025)

	C2 (#490)	D′ (this PR)	E
relative poverty BHC / AHC	20.37% / 23.70%	20.29% / 23.63%	20.50% / 23.71%
pensioner relative poverty AHC	17.70%	17.49%	18.43%
GB pension-age HB	£7.86bn, 1,148k	£7.81bn, 1,146k	£7.76bn, 1,152k
UK Pension Credit	£6.18bn	£6.34bn	£6.50bn
ESS of all households	1,305	1,324	1,365

E's pensioner-poverty change mixes the data change with policyengine-uk changes (State Pension age by date of birth, DWP rates, mixed-age couples), so these two builds can't separate them.

The targeted suite passes on D′'s dataset at e386209: 70 tests, including the take-up, HB target, registry, aggregate and reform-impact checks. The full suite passed on build G at 7c4972b: 657 passed, 2 skipped, 1 xfailed. The full suite passed on build G2 at 68e740c, which is 6565630 apart from #490's comment-only f4d8f95: 660 passed, 2 skipped, 1 xfailed.

Invariants and tests

Fill probability. For any weights, eligibility and reporting: either weighted expected take-up among eligible units equals the rate exactly, or the probability sits at a bound that the reporters alone (0) or all eligible units (1) force. Property test.
Draws. Reporters always claim. Entitled non-reporters claim when draw < p, and units not entitled when draw < 0.37. The newly entitled rate changes neither p nor any entitled unit's flag. Property test.
Rates. Weighted take-up among entitled units still meets 62% over 200,000 units, and the rest claim at 37%.
Statistical check. Over 200,000 weighted units, take-up among the eligible is within 0.5pp of 62%.
Transcription. Spending equals guarantee plus savings credit, and caseload equals its three claim types, using DWP's separate rows.
Registry. Only the DWP targets target pension_credit.
Northern Ireland. NI's entitlement, reporting and weights cannot change the GB probability or any GB flag. Entitled NI non-reporters are drawn at the GB probability, and NI units not entitled at the newly entitled rate (37%). Property test.
SPI-synthetic units are found through their household, in any benefit-unit order.
Built dataset. Every survey reporter has would_claim_pc, and GB claims and spending land within 0.5-2x of DWP.

axiom: n/a: data/calibration-only change

🤖 Generated with Claude Code

Replace the OBR table 4.9 Housing Benefit target, which is GB DWP-funded spending but was compared with UK-wide modelled Housing Benefit, with DWP Spring 2026 targets for Housing Benefit spending and claims over Pension Credit qualifying age in Great Britain. Targets can now name the countries they cover. The figures under Pension Credit qualifying age are kept but not calibrated: the model pays working-age Housing Benefit to too few records for a target to be met without concentrating it on a handful of them. Stop drawing would_claim_uc for benefit units whose adults have all reached State Pension age, keeping every other unit's draw unchanged. Require policyengine-uk 2.102.5, which lets pension-age families make new Housing Benefit claims (PolicyEngine/policyengine-uk#1901). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

State in the release note that modelled GB Housing Benefit falls to about £8.0bn. Mark the mixed-age classification as an assumption, date the benefit-group equality from 2024-25, and quote the calibrated build's working-age claims. Add a test that the calibrated columns build on the dataset, since the loss matrix skips a target whose column raises, and note that would_claim_uc uses survey-year State Pension age. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…flag The create_frs smoke test's fake simulation did not answer the variables the would_claim_uc rule now reads (person_benunit_id, is_adult, is_SP_age), which failed CI with KeyError: 'person_benunit_id'. Pick the claimant-and-partner flag the way the locked policyengine-uk does for Housing Benefit's pension-age route: is_claimant_or_partner where the release defines it, otherwise is_adult as in 2.102.5. Both the would_claim_uc rule and the age-group targets use it. Under 2.102.5 the base FRS is identical to the previous commit's in every column. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…513) - TOTCAPB4 (DWP's below-average-resources measure since 2019/20) in place of TOTCAPB3, which stays as the fallback for earlier survey years. On #510 build D plus the column: the same 87.7k non-entitled reporters as with TOTCAPB3, GB entitled units 2,080.7k vs 2,083.5k, gate claims 1,290.0k vs 1,291.8k. The docstring now calls it an approximation (financial assets only; no Sch V disregard or reg 19 valuation). - uprating_factors.csv and uprating_growth_factors.csv rows recomputed from policyengine-uk 2.107.0's per-capita GDP index (2024 1.204, 2025 1.256), not copied from the stale savings row, so calibration and runtime grow the capital alike; the test now checks the rows against the model. - policyengine-core 3.32.13 in uv.lock (3.32.12 leaks memory per simulation). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…accommodation Max ruled on d703 that working-age Housing Benefit should be calibrated too. policyengine-uk pays no Housing Benefit for specified (supported) or temporary accommodation, so the working-age awards it can pay are general-needs ones. DWP's working-age line includes both kinds, and DWP splits accommodation type only across all ages. The calibrated working-age target is therefore DWP's working-age line less all of its supported and temporary accommodation: £584.8m and 107k claims in 2025-26, a lower bound on working-age general-needs Housing Benefit. It exists for 2024-25 and 2025-26 only; from 2026-27 the all-age accommodation lines exceed the whole working-age line. Seeded calibrations of one saved 2025 input compared the options. Against the full working-age line, uncapped calibration loaded it onto about three effective records. Capping household weights at 20-40 times their prior reached 54-60% of DWP's spending and raised income-related ESA claimants to 1.6-2.0 times DWP's count. Against the net figure, the calibration meets it (1.04x) and no other national target crosses the 10% line compared with calibrating pension-age figures only. Tests: DWP's accommodation lines reconcile with its age split in every year; the net figure equals all-age general needs less pension-age Housing Benefit, the same quantity computed the other way; a property test of the netting; the calibrated working-age column equals the full line's. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

The loss matrix fills a year a target doesn't list with the latest earlier value within three years (_resolve_value). That carried the 2025-26 net working-age Housing Benefit targets (£584.8m, 107k claims) into 2026-27 to 2028-29, where DWP's lines give a negative net figure. Targets gain a carry_forward flag (default True, so nothing else changes), and the two net targets set it to False. The build calibrates 2025 only, so the dataset is unchanged; any loss matrix built for 2026-28 now leaves the net targets out. Also reword the docstring: policyengine-uk has no supported or temporary accommodation rules, but the FRS can't identify those residents, so a reported award there is modelled under the general-needs rules and treating the model's working-age HB as general needs is an approximation. The unmodelled supported and temporary accommodation HB (£5.2bn, all ages) is recorded as a limitation (d946). "Within 12%" becomes "about 12% below". Tests: the net targets resolve in 2024-25 and 2025-26 only while the other HB targets still carry forward; a property test of _resolve_value for any values, year and flag. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…escribe the year resolver exactly The changelog and a test comment still said the model pays no supported or temporary accommodation Housing Benefit. It has no rules for it, and the FRS can't identify those residents, so treating its working-age HB as general needs is an approximation. The resolver docstring and the carry_forward comment now describe _resolve_value as it works: the nearest listed year, the earlier on a tie, used only if it is earlier and at most three years away. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

The FRS build drew would_claim_pc against an unweighted count of every benefit unit, so entitled non-reporters claimed at about the take-up rate on top of every reporter. PolicyEngine/uk-triple-lock's coverage run (data/results.json, Enhanced FRS 2024-25 on policyengine-uk 2.90.2) put 2026-27 Pension Credit claims at 1.82m UK-wide, against DWP's 1.31m for Great Britain. After the imputations, compute Pension Credit entitlement and solve the claim probability that makes weighted take-up among entitled units equal DWP's FYE 2024 caseload take-up (62%). Reporters claim. Every non-reporter, entitled or not, is drawn at that probability, so units a reform makes entitled claim at the same rate. Replace the OBR Pension Credit target with DWP Spring 2026 Great Britain spending and caseload targets. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…aseline The take-up step returns weighted aggregates for the calibration year: the solved probability, entitled units and reporters, and the Great Britain reporter baseline before any fill, which the build's summary prints. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

The Pension Credit reports on SPI-synthetic households are imputed by the second-stage QRF (imputations/frs_only.py) from their SPI incomes, not observed claims. In build D, 154 such benefit units reported Pension Credit and were all anchored as claimants. They are now drawn at the solved probability like any non-reporter. Most SPI-synthetic households have zero survey weight when the probability is solved. The capital-gains band donors (imputations/capital_gains.py) are copies of sampled households re-weighted to HMRC taxpayer counts, so a copy of an SPI-synthetic household keeps the flag and has positive weight. The solved probability still does not move in build D', because none of its SPI-synthetic reporters is entitled. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

DWP's 62% caseload take-up covers Great Britain, but the solve counted Northern Ireland's entitled units and reporters too, so NI's reporter mix moved GB take-up away from 62%. Solve over GB entitled units and draw every non-reporter, NI included, at that probability. A property test checks that NI composition cannot change the GB probability or any GB flag. Correct the docstrings: the fill is zero when entitled GB reporters already make up the rate of entitled GB units, which is separate from calibration against DWP's absolute targets. OBR's Pension Credit line is the same GB series as DWP's, not identical figures. The summary now reports GB entitled units. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

The built-dataset test asserted that every Pension Credit reporter claims, which no longer holds for SPI-synthetic units, whose reports are imputed and no longer anchor take-up. It now checks survey reporters. The module docstring and changelog now say the SPI reports are predicted by the second-stage imputation from SPI incomes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

… take-up Until now every non-reporter claimed at the solved fill probability (about 13%), including units a reform newly entitles. That probability is a residual: the share of entitled non-reporters that tops FRS reporters up to DWP's 62%. Units with no entitlement in the calibration year are now drawn at DWP's caseload take-up for Savings Credit-only awards, 37% in FYE 2024 (table PC1; 42% in FYE 2023), the group with the smallest awards (£9 a week, against £87 and 69% take-up for Guarantee Credit). Reporters and the solved fill for entitled non-reporters are unchanged, so is the solved probability, and every entitled unit's flag. The rate is a new take-up parameter, pension_credit_newly_entitled. The generic assign_takeup_over_eligible helper, which applied one probability to every non-reporter, is removed; pension_credit_takeup_flags applies the two rates. Tests: reporters claim, entitled non-reporters draw at the solved probability and the rest at the new rate, and the new rate changes neither the probability nor any entitled unit's flag (property test); weighted take-up among entitled units still meets 62% and the rest claim at 37% over 200,000 units; Northern Ireland follows the same split; the parameter reads 37% for 2025 and 42% for 2022. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

MaxGhenis · 2026-10-05T19:29:10Z

Hand-off to the UK hub (owner session e913bdc1, 2026-10-05).

Head: 6565630, stacked on Calibrate Housing Benefit to DWP's GB figures by age group #490 at f4d8f95.
Base and CI: now main and ready for review, so CI runs. It was triggered by reopening and was pending at hand-off. Until Calibrate Housing Benefit to DWP's GB figures by age group #490 merges, the diff includes Calibrate Housing Benefit to DWP's GB figures by age group #490's commits; merge Calibrate Housing Benefit to DWP's GB figures by age group #490 first.
Reviews:
- r4 at 7c4972b: CHANGES_REQUESTED.
- r5 (GPT-6.1 Sol) at 6565630: APPROVE WITH NITS, no P0-P2. Its three P3 nits (a rounded delta, the NI invariant wording, D′ labelled as pre-d798) are fixed in the PR body and the d798 write-up; no code changed.
- Reviews: ~/reviews/uk-hb-dwp-targets-2026-09-30/review_510_r4/, review_510_r5/.
Max's ruling: d798 = 37% for units not entitled in the calibration year, built in "Draw units not entitled in the calibration year at 37% Pension Credit take-up".
Build: G2 at ba97fb1, which is code-identical to this head apart from Calibrate Housing Benefit to DWP's GB figures by age group #490's comment-only commits.
- Full suite on G2's dataset: 660 passed, 2 skipped, 1 xfailed.
- 2025 Pension Credit is unchanged on the same weights; 2026 adds 4.9k claims and £0.6m (d798_effect_G2.json).
Waiting on: CI, merge order (after Calibrate Housing Benefit to DWP's GB figures by age group #490), and the batch go (d833).
Lock: Use the FRS benefit-unit capital for Pension Credit's capital test #513 also changes the lock; whichever lands second relocks.
Follow-ups owed: none.

Conflicts resolved per runbook A4 step 10: datasets/frs.py keeps #514's assign_reported_takeup(...) calls for would_claim_pc and would_claim_uc, keeps #490/#510's comments, and ANDs #490's ~derive_all_claimants_over_state_pension_age(...) term onto the would_claim_uc draw (draw count unchanged); .gitignore takes the union of both sides; uv.lock reset to main's (b45c373), relocked in R. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

MaxGhenis · 2026-10-07T04:16:31Z

Queued in release PR #544 for the 10/8 uk-data batch. It lands only on Max's go (d833).

…ount (F1) test_full_target_set_available_offline (from #536) asserted at least 30 OBR targets offline. The release combines four member PRs that each replace OBR targets with better sources, which no single PR's CI saw together: #490 swaps obr/housing_benefit for DWP Housing Benefit targets, #510 swaps obr/pension_credit for DWP Pension Credit spend and caseload, #533 swaps the two OBR salary-sacrifice NI relief targets for HMRC's, and #530 merges the two OBR UC targets into obr/universal_credit. That leaves 29 OBR targets (34 on main), so release CI failed with 29 >= 30. The test now checks what it was for: offline, the committed workbooks give exactly the OBR targets that the same workbooks give when served, plus the receipts and NICs names. A broken fallback still fails it. Release-only fix for the 10/8 uk-data release (d833). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

MaxGhenis and others added 3 commits September 30, 2026 07:55

MaxGhenis force-pushed the pc-takeup-over-eligible branch from ae43f1f to 41a07c9 Compare October 1, 2026 14:09

juaristi22 mentioned this pull request Oct 2, 2026

UK: track the defects uk-data is fixing that microcosm still has PolicyEngine/microcosm#1095

Open

21 of 52 tasks

MaxGhenis force-pushed the pc-takeup-over-eligible branch from f09497f to e386209 Compare October 2, 2026 13:36

MaxGhenis mentioned this pull request Oct 2, 2026

Draw SPI incomes within earnings groups set by FRS employment status #529

Merged

MaxGhenis and others added 2 commits October 4, 2026 13:17

Update the release note for the working-age Housing Benefit target

d156176

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

MaxGhenis force-pushed the pc-takeup-over-eligible branch from e386209 to 9eaa2fb Compare October 4, 2026 18:39

Format the Housing Benefit target tests

33092bf

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

MaxGhenis force-pushed the pc-takeup-over-eligible branch from 9eaa2fb to 7c4972b Compare October 4, 2026 18:40

MaxGhenis mentioned this pull request Oct 5, 2026

Stop calibrating OBR total NICs on ni_employee #537

Open

MaxGhenis force-pushed the pc-takeup-over-eligible branch from 7c4972b to ba97fb1 Compare October 5, 2026 16:29

MaxGhenis and others added 8 commits October 5, 2026 13:36

Describe the year resolver exactly in the carry_forward comment

f4d8f95

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

MaxGhenis force-pushed the pc-takeup-over-eligible branch from ba97fb1 to 6565630 Compare October 5, 2026 18:15

MaxGhenis mentioned this pull request Oct 5, 2026

Calibrate Housing Benefit to DWP's GB figures by age group #490

Merged

MaxGhenis changed the base branch from hb-dwp-age-split-targets to main October 5, 2026 19:24

MaxGhenis marked this pull request as ready for review October 5, 2026 19:24

MaxGhenis closed this Oct 5, 2026

MaxGhenis reopened this Oct 5, 2026

MaxGhenis mentioned this pull request Oct 5, 2026

Calibrate DWP and OBR welfare targets on the countries their source covers #542

Open

MaxGhenis mentioned this pull request Oct 7, 2026

Release the 10/8 uk-data batch (15 PRs, relock to policyengine-uk 2.122.2) #544

Merged

MaxGhenis merged commit 152c3aa into main Oct 7, 2026
6 of 8 checks passed

MaxGhenis mentioned this pull request Oct 7, 2026

Regenerate the uprating factor table from policyengine-uk's load-time uprating #541

Open

Navigation Menu

Repository navigation

Solve Pension Credit take-up over entitled benefit units - #510

Solve Pension Credit take-up over entitled benefit units#510
MaxGhenis merged 15 commits into
mainfrom
pc-takeup-over-eligible

MaxGhenis commented Oct 1, 2026 •

edited

Loading

Uh oh!

MaxGhenis commented Oct 5, 2026

Uh oh!

MaxGhenis commented Oct 7, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

MaxGhenis commented Oct 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Results (real builds, FY2025-26, Great Britain)

Acceptance gate: reporter baseline before any fill (PolicyEngine/uk-triple-lock#14)

After calibration (dataset simulated for 2025)

Paired reform: minimum guarantee +10% from 2025, same seeded flags in both runs

Take-up before and after calibration (d798)

Other outputs (2025)

Invariants and tests

Uh oh!

MaxGhenis commented Oct 5, 2026

Uh oh!

MaxGhenis commented Oct 7, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

MaxGhenis commented Oct 1, 2026 •

edited

Loading