Repository navigation
Project the pension-age population and new-cohort State Pension awards beyond the data year #1929
Description
Projected State Pension falls further below the OBR/DWP forecast every year after the data year, because two things are not modelled. Split from #1632, which covers the data year; found while fixing #1921 (#1922).
What the model does
extend_single_year_dataset(policyengine_uk/data/economic_assumptions.py) copies the data year's person table into each projected year, so ages are held fixed.household_weightis uprated bygov.economic_assumptions.yoy_growth.ons.population(policyengine_uk/data/uprating_indices.yaml), the same factor for every household. So the pension-age population keeps the data year's age structure, less the people the State Pension age rise moves below it.basic_state_pension,new_state_pensionandadditional_state_pensionpay the data year'sstate_pension_reported, uprated by the flat rate of the person's State Pension type in the period. A record whose cohort moves from basic to new State Pension keeps its reported amount.
Measured effect
Enhanced FRS 2024-25 (policyengine-uk-data-private 1.57.4), with #1899 and #1922 applied; full microsimulation runs. Comparators are from the DWP benefit expenditure and caseload tables, Spring Forecast 2026 (consistent with the OBR's March 2026 EFO), sheet "State Pension", rows 4, 13, 29 and 38.
| 2025-26 | 2030-31 | change | |
|---|---|---|---|
| Model State Pension recipients | 11.44m | 11.09m | -3.0% |
| DWP caseload less those paid outside the UK | 12.12m | 12.69m | +4.8% |
| Model State Pension per recipient | £10,912 | £12,702 | +16.4% |
| DWP spending per recipient, less paid outside the UK | £11,591 | £13,714 | +18.3% |
| Model State Pension | £124.8bn | £140.9bn | +12.9% |
| DWP State Pension less paid outside the UK | £140.4bn | £174.1bn | +23.9% |
DWP's new State Pension spending (excluding protected payments) rises from £56.1bn to £102.4bn over those years, and basic State Pension spending falls from £66.7bn to £57.8bn (rows 10 and 6).
What's needed
- The pension-age population should follow its cohorts: for example, weights uprated by age from ONS population projections, or records aged one year per year.
- Records that move to a new State Pension cohort should get that cohort's awards, for example from DWP's award distributions by cohort.
Each option should be measured with full before/after microsimulation runs against the table above, and given invariants and tests (for example, basic + new + additional still equals the uprated award for anyone over State Pension age, as #1922 tests). Changes to published aggregates need sign-off.
Activity
Compare State Pension with DWP's forecast and fix the gap explanation