IoM Pension Explorer — Public Sector & State Pension

Two separate pension obligations, two separate pots: the Civil Service/GUS pension (PSPA, paid from General Revenue) and the Manx State Pension (paid from the National Insurance Fund — a reserve that runs dry on current policy). Explore the combined scenario, or drill into either scheme's own numbers.
● IoM population scenario Drives the Public Sector tab and the Combined tab's PSPA side — same model as observatory.coalfinch.com's Demographics Explorer
ACTUARIAL ASSUMPTIONS (300/yr): The Government Actuary's (GAD) own principal-projection net migration assumption from the 2023-GD-0005 quinquennial review -- the same report the State Pension tab's NI Fund figures are built from. Lands at 87,163 by 2041 -- between Historical trend and Recent trend.
Combined (Gap)
Public Sector (PSPA)
State Pension
Assumed inflation, FY2026-27 onward 2.00%/yr
Constant prices — FY2024-25 pounds. History to FY2025-26 is deflated on the Isle of Man's own measured CPI. From FY2026-27 future years are chained off that last actual at the rate set above. Growth you can still see here is real growth.
Active members, growth per year flat
2024
Year
20182028203820482058
Historical data
25-year demand — what goes out, what comes in £m, FY2024-25 prices Tick items to build up each side — anything ticked in the same column adds together
Paid out
real money from someonePaid in
your levers, £0 until movedModelled fix
the balancing itemLeft to find
Not applied to this scheme — both levers are modelled against the PSPA gap only.
"Left to find" is a balancing item, not a funding source — both entries are whatever remains after everything else, so the totals meet by construction rather than by evidence. Full explanation under Data Governance & Methodology below.
Balances behind those flows same years, same scale of money — stocks, not annual flows
The National Insurance Fund is a real reserve, exhausted during 2047-48. The Public Sector Pension scheme has no fund at all — the flat line is the whole position, not an omission.
Total paid out (all items)
Total Income (selected)
Public Sector Pension paid out
State Pension paid out
State Pension levers
State Pension age+0 yrs
Delays when people start drawing the State Pension — softens the near-term bulge as the 1990s finance-sector cohort crosses 65.
State Pension uprating3.75%/yr
How fast the Manx State Pension rises each year. 3.75% is the modelled default — roughly what a triple-lock-style uprating has delivered, and well above assumed inflation. GAD's own review assumes about 4.2%/yr long-run.
Stress-test: NI Fund already exhausted (backstop active from FY2024-25, instead of the actual 2047-48)
Public Sector Pension (PSPA) levers
Public Sector Pension age+0 yrs
Same idea, for PSPA/Civil Service pensioners specifically — delays when they start drawing, softening near-term pensioner-count growth.
PSPA contribution rate+0 pts
How much more (or less) gets paid into the PSPA pension pot each year, on top of normal wage growth — e.g. a future pay deal that includes a higher pension contribution.
Benefit indexation (PSPA)2.0%/yr
How much PSPA pensions rise each year to keep pace with the cost of living. Currently assumed at 2%/yr (Hymans Robertson's long-term assumption) — drag higher to see what a more generous upgrade would cost.
Salary growth (PSPA)4.0%/yr
How fast the public-sector paybill grows, which is what contributions are levied on. Assumed at 4%/yr (Hymans Robertson). This is the single most powerful number in the model — contributions compound on it for sixty years, and the two percentage points between it and benefit indexation above are what eventually closes the gap.
Modelled fix — paying for it
Dept. cuts diverted to pensions0%
Cut non-protected departments' employee-cost budgets by this much, and redirect the saving to close the PSPA funding gap — shown against the department budgets on the right.
Tax increase diverted to pensions0%
Alternative to cuts — raise this much extra General Revenue (as a % of the whole department budget envelope) and redirect it to close the PSPA funding gap.
Protect Manx Care from cuts (42% of budget — health)
Protect Education from cuts (21% of budget)
Both levers act on the PSPA gap only — neither is modelled against the State Pension side.
Department budgets vs the gap
Year by year
Data Governance & Methodology
The two reference lines, and what "Left to find" really is. Both "Left to find" entries are calculated as whatever remains after everything else — payments out, minus contributions in, minus the NI Fund's own investment return, minus anything the levers divert. With every income item ticked, the dashed Total Income line meets the solid Total paid out line exactly. That equality is true by construction rather than by evidence, and it means these two residuals absorb every error in the model. The four paid-out items are drawn unstacked, each showing its own scheme at face value, which is why an explicit total line is needed: stacking them would make the upper lines cumulative and no value on the chart could then be read directly. Both residuals are floored at zero, so when contributions outrun payments the income line rises above the total-paid-out line rather than a negative bar appearing. On the PSPA side the residual genuinely falls on General Revenue in the year it arises. On the State Pension side it is drawn down from the National Insurance Fund — until that Fund is exhausted during 2047-48, which now happens inside this chart rather than beyond its edge. From FY2048-49 the same shortfall is still there, but there is no longer a fund to meet it. Nothing in the flow chart changes shape at that moment: the money still has to be found, only the source runs out.

The balance panel. Two schemes, two very different positions. The National Insurance Fund is a real reserve: still accumulating today, turning over around the middle of the next decade, exhausted during 2047-48 — which is why the window runs to FY2049-50 rather than stopping short of it. The Public Sector Pension scheme has no fund at all; its PSEPR buffer was drawn down through FY2023-24 and has been at zero since, so the flat line is the whole position rather than an omission. Every pound of PSPA pension is paid from that year's contributions plus a top-up from General Revenue. Read the panels together: the Fund's decline is what pays the State Pension shortfall above it, and when it reaches zero that shortfall does not disappear — it moves onto General Revenue, alongside the PSPA top-up that was always there.

On covering pensions by employing more people. Because contributions are (active members) × (salary), headcount growth raises contributions immediately and raises liabilities only decades later, so it always improves the near-term picture. It should be read with three things in mind. About two-thirds of each contribution is the employer share, which is General Revenue — so most of the apparent new income is the government paying itself. Members' own contributions fund roughly 22% of the PSPA bill; the rest is public money either way. And every additional active member accrues a pension this model prices only through the 65+ cohort, so the later cost is understated here relative to the earlier benefit. Headcount growth is therefore a deferral, not a funding source.
"State Pension payments" = Basic + Age Addition + Premium + new single-tier MSP + Pension Supplement, all combined (estimated split from FY2018-21 per-benefit actuals, applied to FY2021-25 aggregate NI-funded totals). Both expenditure items are gross liabilities (total benefits payable), not netted against contributions. On the income side: "Employee Contributions" and "Employer (Dept) Contributions" split PSPA's combined contribution total using the same real Unified Scheme 2011 employer-share ratio used on the Public Sector tab. "NI Contributions" is the real, GAD-sourced contributions figure from the same 2023-GD-0005 report the State Pension tab's own chart uses (interpolated between its published anchor years) — not modelled by this tab, borrowed from the real series. "Dept. cuts diverted to pensions" and "Tax increase diverted to pensions" are the revenue raised by those two levers, redirected here to close the PSPA gap (both are £0 unless their sliders are moved). "Funding the Gap (PSPA)" = PSPA gross minus PSPA's own contributions minus any diverted cuts/tax revenue (can be negative, i.e. a surplus, if diverted revenue plus contribution growth outpaces benefit growth). "Funding the Gap (State Pension)" = State Pension payments minus the real NI Contributions figure — this is the closest this tab comes to reproducing the real NI Fund's year-by-year draw-down, using this tab's own demographic-driven demand against GAD's own real income, so treat it as indicative rather than a substitute for the GAD-sourced chart on the State Pension tab. Two changes on 15 August 2026. The NI Fund's own investment return is now shown as income on the State Pension side, taken from the annual reconstruction built for the State Pension tab — this tab previously ignored it entirely, which overstated what had to be found from elsewhere. And both residuals are now clamped at zero: left unclamped they went negative whenever contributions and diverted revenue exceeded payments, drawing a downward red band labelled as a top-up when the truth was a surplus. A surplus now shows as the dashed Total Income line rising above the paid-out lines, which is what a surplus looks like. The consequence is worth stating: ticking every income item makes the totals coincide exactly only while there is a shortfall. In surplus years income legitimately exceeds outgo and the chart now says so rather than hiding it in a negative bar.

On the State Pension residual. It is labelled "drawn down from the NI Fund" rather than General Revenue, and over this tab's twenty-year window that is right: the Fund is not exhausted until 2047-48, two years past the right-hand edge of the chart. Nothing on the State Pension side of this tab falls on General Revenue within the period shown — the General Revenue problem begins just after the window closes, which is itself worth noticing. Ticking all income items makes the total match the two expenditure lines added together while a shortfall remains. The two Paid out items are drawn as unstacked lines, each showing that scheme's own payments, so they can be read against each other directly. "Total Income (selected)" is the dashed line and sums only whichever income items are currently ticked, so it moves as you tick boxes and drag levers.
Bars = FY2025-26 department employee-cost budgets (Pink Book 2025-26 Table 14B), grown at 3.0%/yr. Red segment = amount removed by the "Dept. cuts diverted to pensions" lever. Manx Care and Education are protected by default (63% of total department paybill between them) but their checkboxes in the levers panel can be unchecked to stress-test cutting them too.
"Top up from Reserves" (stat cards and table) = PSPA net deficit (gross minus contributions), plus the NI Fund backstop shortfall if that stress-test toggle is on — a narrower, backstop-gated figure than the chart's own "Funding the Gap" lines above, kept as-is for consistency with the levers and department-budget sections. "After levers" additionally nets off whatever the "Dept. cuts" and "Tax increase" levers are diverting. FY2024-25 figures are confirmed PSPA/NI Fund actuals; everything to the right compounds from sourced or clearly-flagged assumptions.
Price basis. The model is built and stored in cash (nominal) terms throughout, which is how PSPA, the Government Actuary and the Pink Book all publish. The "constant prices" toggle at the top of the page converts at display time only — no stored figure is altered, and switching back and forth is lossless. History to FY2025-26 uses the Isle of Man's own measured CPI (financial-year averages, April–March; Statistics Isle of Man Historic Data Sets December 2025, index January 2008 = 100, rebased here to FY2024-25 = 1.000); FY2025-26 is a part-year in that release, covering April–December 2025 only. Beyond that, no Manx inflation projection is published by anyone, so future years are chained off the last actual at a user-set rate defaulting to 2.0% — the figure Hymans Robertson's benefit-indexation assumption implicitly treats as long-run inflation. That is an assumption, not a measurement, which is why it is a control rather than a constant. Because deflation applies a single common factor to every item in a given year, the identity that income sums exactly to expenditure holds in both bases.

Why inflation does not rescue this. Inflation erodes debt only when the debt is nominal and fixed. Neither condition holds here. The Isle of Man is a net creditor, not a net debtor — the National Insurance Fund is an asset of about £1.03bn, so inflation destroys its real value rather than a liability's. GAD's own report records exactly this: the Fund balance falls in 2023-24 in constant price terms "due to high CPI from the previous year increasing benefits". And the liabilities are index-linked by statute and policy — the Triple Lock is assumed to continue indefinitely, at around 4.2%/yr in the long run — so inflation passes straight through to the benefit line while the asset side is merely nominal. Higher inflation makes this Fund's position worse, not better. What does shrink the liability in real terms is uprating by less than inflation, which is a political decision rather than an economic event, and it falls entirely on pensioners: GAD models it as a variant, and dropping the Triple Lock to earnings-only after April 2025 moves Fund exhaustion from 2047-48 to 2051-52. Four years, bought from pensioners. Note also that the deflator control on this page rescales but does not re-model. Raising it alone makes every figure look smaller while nothing real has changed — at 6% inflation the 20-year gap reads £6.6m instead of £13.8m purely as an artefact of the arithmetic. Raise the three uprating levers alongside it and the gap goes to £33.5m: worse than where it started, because adding the same number of percentage points to every rate compresses real differentials rather than preserving them. Inflation is not a way out of this. It is a way of making the way out harder to see.

IOMG headcount is an assumption, and it is now set separately. PSPA contributions are (active members) × (salary), so the number of public servants assumed to be paying in is as consequential as the rate they are paid. Until August 2026 active membership was bolted to the population scenario selected at the top of the page: choosing a higher-migration path also hired civil servants, and their contributions were then counted as income against the pension bill. On the Island's own "Our Island, Our Future" target path that amounted to assuming roughly 4,350 additional active members by FY2044-45 — an implicit staffing policy arriving through a demographic control. Headcount now defaults to flat at the FY2024-25 actual of 10,854, with population-tracking available as an explicit tick-box for comparison. Pensioner headcount continues to follow the 65+ cohort, which is a genuine demographic driver rather than a policy choice. Two cautions on any setting above flat: the employer share of the extra contributions (about 66%) is General Revenue, so it is a transfer within government rather than new money; and every additional active member is a future pensioner whose accrual this model captures only through the 65+ cohort, so headcount growth flatters the near term more than the long run.

On the real-terms wedge. The three price levers — benefit indexation 2.0%, salary growth 4.0%, State Pension uprating 3.75% — are the published assumptions, but they are also, taken together, the model's most consequential input. At 2% assumed inflation they imply public-sector salaries rising 2% a year in real terms indefinitely while PSPA pensions hold flat in real terms, and the State Pension rises 1.75% a year in real terms. Those spreads, compounding, are what drive the PSPA funding gap towards closure in the long run — not the ageing population, whose contribution to the trajectory is a fraction of a percentage point a year. Any reading of this tool that treats the long-run PSPA result as demographic is reading it wrong. All three are now sliders; set salary growth equal to benefit indexation and the result changes character entirely.
Sources & method: PSPA base (£143.3m gross / £95.2m contributions, FY2024-25) and demographic decomposition (2%/yr benefit indexation, 4%/yr salary growth — Hymans Robertson long-term assumptions per PSPA Board minutes 25 May 2020). Pensioner and active-member headcount growth comes directly from the population scenario selected at the top of the page (the same pyramid-driven GUS model used on the Public Sector tab) rather than a fixed historical CAGR — switch the scenario pill and this tab's PSPA figures move with it. State Pension core and Pension Supplement bases from LBB Table 6.8g/6.8h actuals FY2018-19 to FY2020-21 (retirement-family = 86.3% of total NI-funded benefits ex-COVID, applied to FY2024-25 aggregate £277.7m). The State Pension side's volume growth is locked to the "Actuarial Assumptions (300/yr)" population path regardless of which scenario pill is selected for PSPA — it deliberately does not follow Historical/Recent/Target, since GAD's own 300/yr migration assumption is what actually underlies the real State Pension tab's NI Fund figures, and letting this side swing with the PSPA scenario would mean silently second-guessing GAD's own number. Pension Supplement statutory phase-out: no awards to anyone reaching pension age after 5 January 2039 (gov.im); modelled here on its observed −3.6%/yr decline, a conservative proxy for the true post-2039 acceleration, independent of population scenario. Department budgets: Pink Book 2025-26 Table 14B. This is a scenario tool for exploring trade-offs, not a forecast — it uses a different, bottom-up parametrized model of the state pension than the Government Actuary's own published baseline shown on the State Pension tab; the two will not match line-for-line, by design.
Population by age — IoM Actuarial Assumptions
Under 45
45–64
65+
~2,300
Aged 55–64, approaching retirement
1.59:1
GUS active members per pensioner
£48m
Net PSPA cost to General Revenue
£143m
Total gross PSPA payments
PSPA pension cashflow, by funding source £m, FY2024-25 prices Driven by the population pyramid — active & pensioner headcounts scale with the IoM age profile, not a flat rate
Member contributions
Funded by Dept (employer)
PSEPR buffer drawdown
Funded by GRA (historic) / Funding gap (projected)
Gross payments (demand)
Data Governance & Methodology
Methodology — demographic model: gross payments and contributions are no longer two independently-fitted growth rates. Each is now (headcount from the population pyramid) × (price escalator). Pensioner headcount and active-member headcount for FY2025-26 onward scale directly off the same population-by-age model shown on the left — pensioners scale with the 65+ cohort, active members scale with the 20–64 working-age cohort — so if the pyramid shows the working-age pool flattening as the 45-64 wave ages through, active-member growth (and so contribution growth) flattens with it, rather than being assumed to escalate indefinitely. Price escalators are Hymans Robertson's long-term actuarial assumptions endorsed by the PSPA Board (25 May 2020): benefit indexation 2.0%/yr, salary growth 4.0%/yr. Base year FY2024-25 is confirmed from the PSPA Annual Report (gross £143.3m, contributions £95.2m, net £48.1m; active 10,854 / pensioners 6,835). Population held flat beyond 2051 (the model's last demographic snapshot) — years beyond that are a flatter, more conservative extension than a true bulge-driven scenario would show.

Population pyramid — no assumed mortality curve, no invented migration figure: the 2021 baseline (17 age bands, 0-4 through an open-ended 80+) is the real IoM Cabinet Office Census 2021 Table 2.1. Every 5-year step after that ages the pyramid forward using cohort transition rates derived empirically from the Island's own real census history — comparing the actual population in each age band at one census against the actual population one band older at the next census, across all four available censuses (1996, 2001, 2011, 2021; source: observatory.coalfinch.com's Demographics Explorer pipeline, itself sourced to IoM Cabinet Office Census Tables 2.1-2.3, 1996 Table 3, 2001 Table 6). Because a census snapshot can't separate "died" from "emigrated" from "never arrived," these rates are net of mortality and migration exactly as the Island experienced it over 1996-2021 — which is why no separate migration figure is layered on top (that would double-count it). New births per step are the average of the three most recent confirmed years (2022-2024: 580, 609, 574 — Population Report 2025), replacing what was previously a flat, undocumented guess. One data-quality note: the Demographics Explorer's percentage-based pyramid series and the Cabinet Office's absolute 2021 headcount table don't agree perfectly in level (a discrepancy of 15-20%+ in some bands, despite both citing the same ultimate census). This model resolves that by anchoring absolute headcounts to the Cabinet Office table (independently verified) while taking only the ratios between census years from the Demographics Explorer series — ratios are far more robust to a systematic scaling difference between two sources than absolute levels would be. Caveat: net migration in 2022-2024 ran above the 1996-2021 long-run average baked into these rates (548 → 600 → 651/yr, per Population Report 2025), so this model may understate near-term working-age growth if that recent trend continues. No official Isle of Man population projection exists to check this against — Statistics IoM's own 2025 report contains no projection or mortality table, and this is confirmed independently in the Demographics Explorer's own sourcing notes.

Switch scenarios above to see how migration changes this chart: the pills at the top of the page toggle between four migration paths run off this real cohort-transition model — the same one behind observatory.coalfinch.com's Demographics Explorer, plus one added specifically for this page. Historical trend (1996-2021 average rates, no migration adjustment) is the plainest: IoM's total population actually declines slightly, 84,069 → 80,764 by 2041, so active-member growth — and contribution growth — is fairly muted. Actuarial Assumptions (300/yr) reproduces the migration assumption baked into the State Pension tab's own GAD-sourced NI Fund projection (2023-GD-0005), so the two tabs can be compared on a common footing — GAD doesn't publish its own age breakdown, so the age distribution is this model's, not GAD's; lands at 87,163 by 2041. Recent trend (adding the confirmed 2022-2024 average net migration, 600/yr) lands at 93,558 by 2041 and independently matches the Government's own 2024 OAHN housing-need evidence base almost exactly. Target — "Our Island, Our Future" (2022) back-solves the migration rate behind the 100,000-by-2037 ambition and reaches 103,963 by 2041. Moving from Historical up through Target, active-member headcount and contributions grow progressively faster, and the funding gap on this chart closes faster too — but pensioner headcount grows faster as well, so the net effect on the gap is smaller than the contribution growth alone would suggest. None of the four is presented as "the" forecast; Actuarial Assumptions (300/yr) is the default shown on page load because it's the same migration assumption the real State Pension tab is built on, not because it's the most likely outcome.

Cash terms, not constant prices: every £m figure on this chart is in cash (nominal) terms — the year each pound is actually paid in — not deflated to a constant base year, same convention as the GAD's own State Pension figures on the other tab. So the upward slope you see is two effects stacked together: price (2%/yr benefit indexation, 4%/yr salary growth — both nominal, not real-terms) and volume (headcount, from the population pyramid).

Employer/member split — real or historically-derived only: years FY2018-19, FY2019-20, FY2021-22, FY2022-23 and FY2023-24 use the Isle of Man Government Unified Scheme 2011's own real employer/member contribution figures from its Annual Reports and Accounts (66.0-72.5% employer share; FY2018-19's 72.5% is shown as reported even though it's excluded from the forward-looking average as an outlier). Years without a sourced figure (FY2016-17, FY2017-18, FY2020-21, FY2024-25, and every projected year) use 66.2% employer share — the average of the three most recent real years (FY2021-22 to FY2023-24) — flagged as a historically-derived estimate, not an assumption invented for this chart. Unified is the largest of PSPA's five schemes but not the only one, and this ratio is applied to the whole PSPA-wide total since scheme-by-scheme splits for Teachers/Police/Judicial/Manual Workers aren't available. Red is "Funded by GRA" only for real historic years (FY2022-23 onward, when PSEPR ran dry) — for every projected year the same red segment is relabelled "Funding gap" because it hasn't actually been drawn from General Revenue yet, it's simply the calculated shortfall if nothing changes.
Population by age — IoM Actuarial Assumptions
Under 45
45–64
65+
£1,032m
NI Fund balance (start of yr)
£195m
Contributions received
£183m
Benefits paid (net of UK settlement)
2047-48
Projected exhaustion year
National Insurance Fund — contributions, benefits & fund balance £m, FY2024-25 prices GAD principal projection, 2022 quinquennial review
Contributions received
Investment return
Shortfall (historic / projected)
Benefits + admin (demand)
Fund balance (start of yr)
Data Governance & Methodology
Source: Report by the UK Government Actuary on the operation of the Social Security Acts in the Isle of Man, 1 April 2017 – 31 March 2022 (GD 2023/0005, published April 2023) — the actual quinquennial actuarial review, not a policy options paper. Historical figures (2017-18 to 2021-22) are real outturn from the Isle of Man detailed Government accounts. Projected figures are the report's own "principal assumptions" projection (current benefit/contribution policy continuing, net inward migration 300/yr, 1.8% long-term real earnings growth), published at 2022-23, 2032-33, 2042-43 and 10-year intervals thereafter to 2082-83 — years between those anchor points are reconstructed, not independently published, by the method set out below. The Fund is shown reaching zero during 2047-48 because the report states this explicitly in its narrative (paragraph 1.6), even though it isn't one of the tabulated anchor years. This is the "do nothing" baseline only — it assumes no changes to the Triple Lock, State Pension age, or contribution rates, and makes no allowance for any additional financing Treasury might introduce.

Price basis — checked, not assumed. The projections used here are Appendix I, Table I1 (p.71), which is stated to be in cash terms. This matters: GAD's main body — Section 2, Figures 2.1 and 2.2, the Section 3 variants — is stated in constant 2022-23 prices (para 2.4). Only Appendices H and I are cash. Had these anchors been taken from the main body, the "constant prices" toggle at the top of this page would have deflated them a second time and understated the entire State Pension side. They were not, and this was verified against the source PDF on 15 August 2026. As a read-check, the cashflow identity (contributions + investment return − benefits − administration) reproduces GAD's own published "Excess of income over outgo" row to within £0.001m at all seven anchor years.

Annual path between anchors — a disclosed reconstruction. GAD publishes seven columns spanning sixty years (2022-23, 2032-33, 2042-43, then decades to 2082-83). Any annual line drawn through them is a reconstruction. Until 15 August 2026 this tool drew straight lines between the anchors — also a reconstruction, but undisclosed and materially worse, because a straight line cannot represent curvature and so missed GAD's own stated turning points by several years. The method now used: contributions, benefits and administration are interpolated with a monotone cubic (Fritsch-Carlson), which passes through every published value exactly and cannot overshoot between them; investment return is applied as a rate on the fund balance, that rate interpolated between the three published anchors where the Fund is live (4.261%, 4.157%, 3.853%); and the Fund balance is then recursed annually as an output of the cashflow rather than drawn as a smoothed line of its own. One small mid-segment correction is applied to the benefit path so the recursion lands exactly on the next published Fund anchor — shaped so that it is zero at both published endpoints, meaning every figure GAD actually printed is reproduced untouched and only the path between them bends. The corrections required are +3.27%, +0.24% and −3.75%.

How that reconstruction was checked. Three of GAD's published statements were deliberately held back from the fit and used as tests. GAD states expenditure first exceeds contribution income in 2033-34 (para 2.8) — the reconstruction crosses in FY2033-34. GAD states the Fund reaches a maximum in 2028-29 in constant price terms (para 1.6) — the reconstruction peaks in FY2027-28, within one year, the residual being the near-term CPI path. And the implied investment return stays inside the published 3.85–4.26% band across the whole live period, which is emergent rather than imposed. One thing that is not an independent check, and should not be read as one: the 2047-48 exhaustion date is used as the terminal condition of the final segment, so the tool reproducing it is arithmetic rather than corroboration.

Two basis errors corrected on 15 August 2026. Both were pre-existing, both were joins between the historic and projected series, and both flattered or distorted the picture. Fund valuation: Table D1 carries the Fund on two bases, "investments at cost" and "investments at market values". This tool took start-of-year at cost while joining it to GAD's projections at market value, so the line stepped up about £103m at the 2022/2023 join purely from the change of basis. Start-of-year is now the previous year's market value, joining continuously (£982.785m → £1,031.611m). FY2017-18 has no published market-value opening balance, so the Fund line now begins at FY2018-19 rather than showing a figure on the wrong basis. UK settlement: historic benefits in Table D2 are gross, before settlements with the United Kingdom, which Table D1 shows separately as income of £45–51m a year; GAD's projections are explicitly net of those settlements. Plotted together, the benefits line dropped £47.5m at the 2022/2023 join for no reason but the basis change. Historic benefits are now shown net — £230.242m − £47.703m = £182.539m for FY2021-22, against £182.775m projected for FY2022-23. That the two now join almost exactly is the evidence the diagnosis was right.

Investment return and administration are now shown. Table I1 carries both, and this tool previously discarded both. Their absence is why the chart could show a red funding-gap bar while the gold Fund line was still climbing — the £44–54m a year that reconciled the two simply was not drawn. The shortfall bar is now measured against all income (contributions plus investment return) versus all outgo (benefits plus administration), so the bars and the Fund line describe the same arithmetic. FY2020-21's shortfall is a real recorded outturn, mostly COVID-related supplementary payments; everything from FY2022-23 is projection. There is still no employer/employee split here — unlike the PSPA scheme, no year-by-year sourced breakdown of National Insurance receipts by payer was available, so none is invented.

One thing the constant-prices view reveals. On the Fund's own market-value basis, deflated to FY2024-25 money, the National Insurance Fund's highest real balance is at the start of the series — FY2021-22, about £1,159m — and it never regains that level. The rise to roughly £1,300m by 2033 that dominates the cash view is inflation. GAD's stated 2028-29 maximum is the maximum within the projection period, and the reconstruction reproduces it; but seen against the recent past, the real-terms picture is a Fund that is flat at best from 2022 and declining from the end of this decade.

Population panel: shown for demographic context only — it's switchable via the scenario pills at the top of the page (same real, census-derived model described on the Public Sector tab; defaults to Actuarial Assumptions (300/yr) on page load), but it does not feed into the figures on this chart. The GAD's own principal projection above already has its own built-in demographic assumption (net inward migration of 300/yr) — the "Actuarial Assumptions (300/yr)" pill reproduces that same 300/yr figure using this model's real 2024 migration shape, so you can see roughly what population path underlies the NI Fund numbers above (it lands at 87,163 by 2041, between "Historical trend" and "Recent trend"). GAD doesn't publish its own age breakdown, so this is this model's best reconstruction of GAD's assumption, not GAD's own age-banded data — the two aren't reconciled to each other and won't match exactly; see observatory.coalfinch.com's Demographics Explorer for the fuller scenario range.