Skip to content

Add the UK State Pension dynamics plan for the triple lock costing - #502

Merged
MaxGhenis merged 3 commits into
masterfrom
uk-state-pension-dynamics-plan
Oct 1, 2026
Merged

MaxGhenis merged 3 commits into
masterfrom
uk-state-pension-dynamics-plan

Conversation

@MaxGhenis

Copy link
Copy Markdown
Contributor

Adds docs/plans/uk-state-pension-dynamics.md: a plan for projecting the UK State Pension population and its entitlements to 2050-51 for PolicyEngine UK. uk-triple-lock's Burnham-plan costing is the driving use case and acceptance benchmark. Tracking issue: #501.

Documentation only. No code, gate, registration or committed run changes.

What the plan covers

  • Why. PolicyEngine UK's basic plus new State Pension bill is 12.8% below DWP's in 2026-27 and 23.3% below in 2030-31. The contributors are verified in code: fixed survey ages, £0 for anyone reaching State Pension age after the survey year, pinned types, and no pensioners abroad.
  • Phase 1.
    • Reweight fixed age slots to the ONS 2024-based projections.
    • Assign each slot its cohort's entitlement from a cohort entitlement model. The model covers mortality, State Pension age by date of birth, accrual to the transitional rate and protected payments, migration, and pensioners abroad with frozen and uprated pensions.
    • Deliver a path-independent sidecar and a synthetic abroad dataset that PolicyEngine UK runs.
  • Phase 2. Person-level trajectories as challengers on version 2 of the contract.
  • Engine contract. Seven policyengine-uk changes (E1-E7). The plan consumes Set State Pension age from date of birth, including the rise to 67 policyengine-uk#1899, #1904, #1922, #1939 and #1941.
  • Acceptance.
    • Four gates to be pre-registered: DWP Spring 2026 caseloads and spending by type in 2030-31, held-out Stat-Xplore full-rate bands, and an abroad backcast.
    • Registered comparisons with the OBR's July 2026 path, and with DWP's £15bn and £50bn as a single-path comparator against single-path runs only.
  • Cost and time (estimates). Phase 1 is about 26-41 agent-days and 10.5-14.5 human review days, 4-7 weeks elapsed. The expected-value build is about 456 path jobs, 3-6 hours on three workers.
  • Decisions needed. Nine, listed in the plan.

Boundaries

Review

  • Thirteen research dossiers were consolidated into one evidence base, with its load-bearing numbers re-checked against the saved primary sources.
  • Three independent reviews (claims and numbers, law and architecture, feasibility and hygiene) returned 99 findings, none blocking. All 36 major findings were applied.
  • An independent final audit confirmed those fixes and re-derived the totals, counts and legal citations. Its one further major finding and seven minor ones are applied.
  • 58 of the plan's 60 external links return 200. The OBR and House of Commons Library pages block automated fetches and were read from saved copies.

🤖 Generated with Claude Code

A scoping plan in docs/plans/: project the UK State Pension population and
its entitlements to 2050-51 for PolicyEngine UK, with uk-triple-lock's
Burnham-plan costing as the driving use case and acceptance benchmark.

Phase 1 reweights fixed age slots to the ONS 2024-based projections and
assigns cohort entitlements from a cohort entitlement model (mortality,
State Pension age by date of birth, accrual to the transitional rate and
protected payments, migration, pensioners abroad with frozen and uprated
pensions), delivered as a path-independent sidecar plus a synthetic abroad
dataset. Phase 2 is person-level trajectories as challengers.

Documentation only. Tracking issue: #501.

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

vercel Bot commented Oct 1, 2026 •

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated
social-security-model Ready Ready Preview Oct 1, 2026 3:26am UTC

Request Review

Applies the independent review of the pull request: the sentence on
uk-triple-lock#13 no longer sets the expected value against the £15bn as a
gap, the four unlabelled mentions carry the label, R3 no longer shows the
expected value beside it, and the note on policyengine-uk#1351 quotes the
closing comment without editorialising.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
R3 compares one path with one path, but not on DWP's own unpublished path.
This is the optional replacement from the approving re-review.

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

Copy link
Copy Markdown
Contributor Author

Merge record: every check passed on head 56d1070 and the PR is mergeable. An independent Opus 5.5 review on a Subfleet lane requested changes at 9e82012; those were applied in 7de8e09, which the re-review approved. 56d1070 applies that re-review's optional one-line wording change and nothing else. Before the PR, three independent reviews and a final audit covered the draft. This is a proposal-status plan requested by Max; its nine decisions stay open and are queued for him.

@MaxGhenis
MaxGhenis merged commit 75cd352 into master Oct 1, 2026
12 checks passed
@MaxGhenis
MaxGhenis deleted the uk-state-pension-dynamics-plan branch October 1, 2026 13:08

This branch was successfully deployed

1 active deployment
Preview — 56d10703 Deployed Oct 1, 2026 by vercel[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant