OG-UK follow-ups gated on Scorecard comparisons (oguk re-pin, one-process dynamic scoring deferred)

Re-scoping the OG-UK follow-ups per Max's steer (2026-08-16): PE model development centers on the Scorecard — compute counterparts to OBR's published scores first, add machinery only when a comparison needs it.

State

• PSLmodels/OG-UK#69 merged 8/3 (e0ff741): pin lifted to policyengine-uk>=2.89.2, and the pre-existing MTR bugs fixed (household-normalisation clipping had ~80% of mid-band earners at a 0.000 marginal rate; every OG-UK tax function to date — including the tax functions behind this repo's OG numbers and the #82 overlay factors — was fit on those). No oguk release yet (master still 0.3.2; not on PyPI). PSLmodels/OG-UK#70 lifts the now-stale ogcore<0.18 cap.
• integration/tests/test_og.py still describes the old world ("oguk 0.3.0 pins policyengine-uk==2.88.0") and skips the real solve whenever installed PE ≠ 2.88 — that skip is now stale.

Deferred (deliberately)

• Re-pinning oguk to e0ff741 and collapsing the two-environment og_payload pipeline into one-process dynamic scoring.
• Re-baselining quoted OG numbers.

These get built when PolicyEngine/policyengine-scorecard#55 has an OBR-published claim that an OG-UK counterpart is needed for (labour-supply/FTE effects of NICs cuts etc.), with PE-UK's static LSR answer sitting next to it. Until then the OG arm's correctness fix lives upstream and this repo carries no new OG machinery. Static OBR costings vs PE-UK is PolicyEngine/policyengine-scorecard#54.

Not deferred

• Any OG number currently quoted on the site or in papers that came from pre-#69 tax functions should be marked as such (fit on zeroed MTRs) or removed — a correctness note, not machinery.

Start with integration/tests/test_og.py and the State, Deferred, and Not deferred sections of the issue. Check the stale PolicyEngine version assumptions and the pre-#69 OG numbers; completion depends on the Scorecard comparison need, with affected numbers marked or removed and any required OG machinery defined afterward.