Topic

The State Pension — the new State Pension, the triple lock, and State Pension age

The State Pension is the foundation of UK retirement income and the backdrop to every private pension. Three things define it today: the single-tier new State Pension, the triple lock that uprates it, and a State Pension age that is still rising.

The new State Pension

For people reaching State Pension age on or after 6 April 2016, the new State Pension replaced the old basic-plus-additional system with a single-tier benefit — “the biggest reform of the State Pension since it was created more than a century ago” ^HMT. It requires 35 qualifying years of National Insurance for the full rate and a minimum of 10 years to receive anything; the full rate launched at £155.65 a week ^DWP and is £241.30 a week in 2026–27, with each post-2016 qualifying year worth about £6.89 a week ^DWP. Existing contributors received a “starting amount” — the higher of their entitlement under the old or new rules — and any excess over the full rate is held as a price-indexed protected payment ^DWP. The reform also ended contracting-out of the additional State Pension from 6 April 2016, removing the reduced-NI rebate — the legacy that drives GMP equalisation ^DWP.

A decade on, the old system is still the majority system. Answering in the Lords on 23 September 2026, DWP put the caseload for the quarter ending March 2026 at approximately 13.3 million State Pension recipients, of whom around 5.4 million (41%) were on the new State Pension and around 7.9 million (59%) on the basic State Pension, drawn from DWP Stat-Xplore ^WQ. That split is the practical reason so much of this article is about two parallel regimes rather than one: for a clear majority of today’s pensioners the relevant rules are still the pre-2016 basic pension plus additional pension — and for them the triple lock reaches only the basic component, while the earnings-related additional pensions, Graduated Retirement Benefit, protected payments and deferral increments are uprated by CPI instead (raw/gad/2026-01-13-report-to-parliament-on-the-2026-re-rating-and-up-rating-orders-tables.md; see The National Insurance Fund — the pay-as-you-go plumbing behind the State Pension and Pension indexation and increases — statutory minimums, RPI vs CPI, and the 2030 RPI reform). The cohort is turning over slowly — new State Pension recipients are only those reaching State Pension age from 6 April 2016 — so the administration, communication and correction workload on the old rules (below) has years left to run.

How slowly: the forecast trajectory. DWP put numbers on that turnover in a further Lords answer on 7 October 2026 (UIN HL3530). Of people in Great Britain over State Pension age, 57% are forecast to be in receipt of the new State Pension in 2030, rising to 74% in 2035 and 85% in 2040, on “DWP’s internal modelling consistent with Spring Statement 2026 forecasts and the ONS’ 2024-based population projections” ^WQ. Three things follow for anyone planning around the two regimes:

• The crossover is imminent but the tail is long. On this path the new rules become the majority system in the late 2020s — and even in 2040, nearly a quarter-century after the reform, roughly one in seven of the over-SPa population is still on the pre-2016 rules. The basic-pension uprating mechanics, additional-pension CPI treatment and GMP legacy above therefore remain live administration for the whole of the 2030s.
• It is modelled, not measured. This is a forecast resting on a fiscal forecast and a population projection, not a statistical outturn, so it should not be set against the Stat-Xplore caseload figures as a trend series.
• The denominators differ. The 41% above is a share of the ~13.3 million people receiving a State Pension; the 57/74/85% series is a share of people in Great Britain over State Pension age ^WQWQ. Those are not the same population — the receipt base includes recipients abroad (below) and the GB base includes people over SPa with no entitlement — so the 41% is not the 2026 point on this curve.

Contains Parliamentary information licensed under the Open Parliament Licence v3.0.

Knowing your entitlement: the forecast-checking gap

Because the new State Pension turns entirely on the individual’s National Insurance record — 35 qualifying years for the full rate, 10 for anything, plus a transitional starting amount and any protected payment ^DWP — nobody can know their own entitlement without checking it. HMRC’s own research suggests a substantial minority never do. A survey of 5,206 consumers, fielded 27 February to 12 March 2026, found 12.5% (one in eight) of UK adults have never checked their State Pension forecast — around 6.9 million people on a UK adult population of 55,022,253 — while almost a third have checked via the HMRC app or online ^HMRC. The gap is worst where it matters most for planning: 45 to 54 year olds were likelier than any other age group never to have checked, despite being close enough to retirement to still act on what they find (same source).

The stated reasons matter more than the headline, because each maps to a different remedy. Respondents cited retirement feeling too far away (26%), worry about losing track of pots from previous jobs (24%), uncertainty about how career breaks affect entitlement (20%), only thinking about pensions at financial moments such as a new tax year (17%), not knowing how to check (9.5%) and expecting it to be too complicated (5%) ^HMRC. Two of those are not attitudinal at all: the lost-pots worry is the problem Pensions Dashboards is being built to solve, and the career-break concern is precisely the defect the Home Responsibilities Protection correction exercise had to fix retrospectively — 12,379 underpayments worth about £104m arising from missing HRP on the NI records of pre-2000 Child Benefit claimants ^DWP. A forecast check is the individual-level control on exactly that class of error.

HMRC’s channel is the HMRC app, which shows the forecast, identifies potential gaps in the NI record and lets eligible users top up contributions, alongside its “Tax Confident” guidance site; the push was timed to Pension Awareness Week, 15–18 September 2026 ^HMRC. This is also a measurable baseline for a commitment made elsewhere: DWP’s action plan following the PHSO finding undertakes to run an ongoing awareness campaign for the ‘Check your State Pension age’ tool, with paid elements running since Autumn 2025 ^DWP. The 2026 figures are the state of play part-way through that campaign — and a reminder that the department’s answer to a communications failure is a tool people must choose to use.

The pre-campaign baseline, and who the tool does not reach

DWP’s Planning and Preparing for Later Life survey measured the same thing five years earlier, on a probability sample of 2,655 people aged 40 to 75 in Great Britain interviewed by telephone between 14 November 2020 and 7 February 2021 at a 29% response rate ^DWP. Among those below State Pension age, 51% had checked their State Pension age and 40% the amount — rising with proximity, from 37% and 26% at 40 to 49 to 73% and 60% at 60 to 65 (same source). So checking is not rare; it is late.

Accuracy is the more useful measure, and it is poor. 56% identified their own State Pension age correctly, 24% thought it later than it is and 20% thought it earlier, with those who underestimated wrong by an average of 3.17 years (same source). The error is concentrated exactly where the law has moved: 75% were correct where their State Pension age is 66, 54% where it is 67, and 19% where it is 68 — and underestimation rises from 5% to 23% to 55% across those three groups (same source). The cohorts whose State Pension age has been legislated furthest into the future are the cohorts who do not know it, which is the forward-looking form of the communications problem the PHSO found in the 1950s cohort, and a standing caution for the State Pension age reviews.

On the amount, knowledge is weaker still: against a full new State Pension of £175.20 a week in 2020/21, only 3% named it exactly and 22% came within 10%, while 65% underestimated it by more than 10%; the median estimate was £150 and the mean £143 — closer to the old basic State Pension of about £138 than to the new rate, which the report reads as people not yet having absorbed the 2016 reform (same source).

The tool works, and reaches the wrong people. Those who had used the “Check your State Pension” service identified their State Pension age correctly at 69% against 39% of those who had not (same source) — strong evidence that the remedy is effective. But the groups with most at stake check least: of people whose only expected retirement income is the State Pension, just 36% had checked their State Pension age and 19% the amount, against 51% and 40% overall; those not in paid work were correct 47% of the time against 54% of employees; and 31% of people with low financial literacy underestimated their State Pension age against 13% with high literacy (same source). DWP’s own conclusion is that because low financial literacy tracks lower income and savings, “people who are more dependent on the State Pension for their retirement income are generally less well informed about the State Pension than those who are less reliant” (same source).

That is the structural objection to a self-service answer, stated in DWP’s own evidence five years before the campaign the action plan commits to: an opt-in tool raises accuracy by 30 points among people who choose to use it, and is least used by the people for whom the State Pension is everything. Self-reported knowledge shows the same skew — 17% rated their State Pension knowledge “good” and 14% “little or nothing”, with men at 22% “good” on pensions generally against 11% of women (same source), a gap also recorded in The Gender Pensions Gap.

What a forecast check actually converts to (12 October 2026)

The campaign’s purpose is not checking for its own sake: a check is useful if it leads someone to act, and the action available is paying voluntary National Insurance contributions to fill a gap. Two Treasury written answers answered on 12 October 2026 (UINs 29831 and 29835, both with identical text and both asked by reference to the 14 September press release) are the first source in this corpus to put numbers on that funnel — and on what HMRC does not know ^WQ.

What HMRC does not hold is the measure that would value the campaign. The answers state that “[t]he HMRC-commissioned consumer survey did not provide information on how many people with an incomplete National Insurance (NI) record have never checked their State Pension forecast, nor how many could increase their State Pension entitlement by making voluntary NI contributions. HMRC has also not produced an estimate of these figures” (same source). So the 6.9 million never-checkers cannot be intersected with the population who would actually gain from checking. And the limit is structural, not merely a gap in the survey: “HMRC cannot forecast an individual’s future income or NI contributions to determine whether they could increase their State Pension by making voluntary contributions” (same source) — because entitlement turns on “contribution history, credited years and any future qualifying years they may accrue before reaching State Pension age” (same source). A gap in the record today may close itself through later employment, which is why neither HMRC nor the member can treat a payable gap as a shortfall needing payment without a view of the rest of the working life. That is a real caution for anyone advising on voluntary contributions, and it is the department’s own.

What is given is the conversion funnel. HMRC published operational management information for customers who paid voluntary NI online after checking their forecast online (figures “subject to change”):

^WQ

Three readings, in descending confidence.

The constraint is identification, not willingness. Of those identified with a payable gap, roughly two in three paid it — 67% in 2024/25, 69% in 2025/26 and 67% in the first four months of 2026/27. That is a strikingly high and stable conversion at the final step. The attrition is all earlier: payable gaps were found for 1.4% of online checkers in 2024/25, 0.7% in 2025/26 and 0.3% so far in 2026/27. Whatever the campaign’s reach, the filter that decides the outcome is how many checkers turn out to have a gap that can be paid at all — so “get more people to check” and “get more people to top up” are not the same policy, and the second is overwhelmingly determined by the first’s yield, not by persuasion.

The volumes of checking are large, and not comparable with the survey. More than 8 million unique customers checked online in 2024/25 alone. That cannot be read against the survey’s one-in-eight never-checked finding: the table counts unique customers per financial year through one online service, while the survey measures whether an adult has ever checked by any route ^HMRC. The two are different bases and no captured source bridges them.

The series falls steeply, and nothing captured explains why. Both the checker count and the payable-gap count decline sharply across the three periods, the payable-gap count by more than an order of magnitude on an annualised basis. (Corpus gap: no captured source explains the fall in checkers or in payable gaps across 2024/25 to 2026/27 — the answers give the table without commentary, and the obvious candidate, the closing of an extended window to pay voluntary contributions for earlier years, is not evidenced anywhere in this corpus. An HMRC statistical commentary or a capture of the voluntary-contribution deadline guidance would close it.) Until it is explained, the figures should not be read as a trend in member engagement.

Two caveats the answers state themselves. The figures “relate only to customers who paid voluntary NI contributions online after checking their forecast through the online Check your State Pension forecast service”, and exclude those who checked and then paid outside the service, and those who obtained a forecast through DWP and paid outside it (same source) — so they are a lower bound on voluntary contributions, not a total. And on the self-service objection set out above, the answers do name a non-digital route: HMRC “recognises that digital services are not suitable for everyone, and customers who cannot use the online service can contact the DWP Future Pension Centre to request a forecast” (same source). That is a fallback channel for the forecast; the answers say nothing about a non-digital route to paying, and nothing about how either channel reaches the low-financial-literacy and State-Pension-only groups DWP’s own 2020/21 evidence identifies as least likely to check. On the related gap for HMRC’s member-facing tax tools, see HMRC tax administration for scheme administrators — Managing Pension Schemes, relief at source and the returns calendar.

Finally, the questions asked what further action was required. No assessment and no new action is stated — the answer is the table and the caveats, followed by encouragement to use the app ^WQ.

The triple lock

Since 2011 the basic and new State Pension have been uprated each year by the triple lock — the highest of the growth in average earnings, price inflation (CPI), or 2.5%. Introduced by the Coalition government, it has applied every year since with one temporary suspension for 2022/23, when post-pandemic distortions pushed the earnings measure to around 8%. The 2.5% floor binds when earnings and inflation are both lower (the April 2021 rise was 2.5%), while in other years earnings or inflation drive it (April 2026 rose 4.8% in line with average weekly earnings). Its long-run cost and sustainability are a recurring policy debate ^HoC.

The government has repeatedly confirmed in Parliament that the triple lock is a manifesto commitment for the whole of this Parliament, with no means-testing and no cap ^HansHans. The recent upratings were 8.5% (April 2024, ~£900 a year), 4.1% (April 2025, ~£470) and 4.8% (April 2026, up to ~£575) ^HansHans. On OBR projections the full new State Pension rises by around £2,100 a year over the Parliament, and State Pension spending ends it more than £30bn a year higher than in 2024–25 ^Hans.

The wedge is quantified, but the reform options are not costed. A Lords written answer of 23 September 2026 asked the Government to assess the ten-year Treasury savings from two specific dilutions — dropping the 2.5% element, and applying the triple lock only to the value equivalent to the basic State Pension while uprating the remainder by prices. DWP declined, saying such an assessment “would incur disproportionate cost” ^WQ. What the answer did disclose is the arithmetic underneath the debate: the OBR’s long-term annual growth assumptions in its March 2026 Economic and Fiscal Outlook are CPI 2.0%, average earnings 3.75% and the triple lock 4.31% (same source). That is the ratchet in one line — because the lock takes the highest of three measures each year, it is assumed to outrun earnings by about 0.56 percentage points and prices by about 2.3 percentage points every year, indefinitely, which is why the Government Actuary names it as one of three structural drivers of National Insurance Fund expenditure growth (raw/gad/2026-01-13-up-rating-report-2026-report-on-the-national-insurance-fund.md; see The National Insurance Fund — the pay-as-you-go plumbing behind the State Pension).

For practitioners the point is the asymmetry: the long-run cost assumption is published and precise, while the savings from any particular reform are not modelled on request. A written question is therefore not a route to a costed triple-lock alternative, and the live analytical work sits instead with the third State Pension age review and the revived Pensions Commission.

The adjusted triple lock from April 2030

Six days after that answer, on 29 September 2026, DWP published an ad hoc analysis that does model a specific alternative — because the alternative is now the government’s own. The publication restates the commitment to the triple lock “throughout this parliament”, under which the basic and new State Pension are expected to “reach a record high relative to earnings over the coming three years”, and then states that “[f]rom April 2030 the government will adjust the Triple Lock” ^DWP. DWP updated the same publication on 6 October 2026 to add a “Triple Lock reform factsheet”; the analysis text is word-for-word unchanged, so the two captures differ only by that addition (raw/dwp/2026-10-06-state-pension-uprating-analysis-2026-tables.md — the current version; the factsheet is taken up below). The two statements are consistent rather than contradictory: GAD expects the April 2030 up-rating to “fall in the next parliament” ^GAD, so the adjustment is timed to begin after the manifesto commitment recorded above has run its course.

The adjusted mechanism is described in a single sentence: the pension will go up “by at least inflation or 2.5% – and anything more that is needed to retain that value” — that value being the record high relative to earnings reached under the current lock (same source). Read against the existing rule, this is a change of kind rather than of generosity in any one year:

• the current lock runs an annual three-way race — the highest of earnings growth, CPI or 2.5% — so what happens in one year does not depend on where the pension already sits relative to earnings;
• the adjusted lock keeps a floor of the higher of CPI or 2.5%, but turns the earnings comparator into an earnings target: a top-up only to the extent needed to hold the pension at its established ratio to average earnings.

DWP states the long-run consequence itself — “over time the State Pension is set to rise in line with average earnings” (same source). That is the ratchet closed. On the OBR determinants disclosed the week before (CPI 2.0%, average earnings 3.75%, triple lock 4.31%), the current lock is assumed to outrun earnings by about 0.56 percentage points a year indefinitely ^WQ; an instrument that tracks earnings once a target ratio is held is, by construction, the removal of that compounding excess rather than a cut in the level of the pension.

Scope is narrower than the headline suggests. The mechanism reaches the basic State Pension and the full rate of the new State Pension only, and “[c]omponents of the pre-2016 State Pension that are not covered by the Triple Lock are assumed to continue to be uprated by CPI” ^DWP — which restates, and does not change, the split described above, under which the lock already reaches only the basic component for the 7.9 million recipients (59%) still on the pre-2016 rules.

The modelling is Pensim3, DWP’s dynamic microsimulation model, run over 2018 to 2100 for Great Britain on administrative and survey data including the Family Resources Survey, and on the ONS 2024-based national population projections (same source; see Longevity and mortality assumptions — base tables, the CMI model and the data).

How far the modelling can be checked was put to the government directly, and the answer matters for anyone relying on the figures below. Asked on 5 October 2026 whether Pensim3 is open access, “including information on the underlying assumptions and formulae”, DWP answered that “Pensim3’s baseline data contains information from real individuals and is subject to data protection controls”, that its “economic and demographic assumptions align with OBR’s Fiscal Risks and Sustainability Report and ONS’s population projections”, and that “[o]ther assumptions are outlined in the relevant published results”, pointing to the Analysis of future pension incomes statistics ^WQ. Read precisely, that is a “no” on the baseline data, with a stated and unobjectionable reason, and no answer at all on the formulae — the question asked about both and the reply addresses only the data and the assumptions. So the position is that the demographic and economic inputs are externally verifiable because they are someone else’s published numbers, the remaining assumptions are discoverable only to the extent a given publication chooses to describe them, and the model’s internal structure is not published. A reader cannot reproduce the savings table below; they can only check that its inputs match the OBR’s and ONS’s. The same answer is the authority for the cross-model assumption consistency noted in Longevity and mortality assumptions — base tables, the CMI model and the data, and the publication it points to is the one underpinning Retirement Adequacy and the Revived Pensions Commission.

Savings are the difference between long-run projected State Pension expenditure under the adjusted and the current lock, presented in both nominal and real terms, as direct Annually Managed Expenditure impacts that “do not account for tax or debt interest effects” (same source). Three caveats travel with the figures and should travel with any use of them: the savings “should not be used to then derive further point estimates for savings in an individual year and should instead be relied on for long-term estimates”; the analysis “isolates the modelled effect of uprating policy, rather than behavioural responses or wider policy and macroeconomic changes”; and “[t]he analysis is illustrative and are not forecasts of future government policy” [sic] (same source). On a fair reading that last caveat attaches to the illustrative uprating path modelled rather than to the decision itself, which the same document states as settled — but the document does not say so, and the two sentences sit awkwardly together.

The savings figures are now available. The third table re-capture batch (10 October 2026) restored the two tables that had been captioned-but-empty through four earlier passes, and they give the scale of the reform in two years only:

Both tables are rounded to the nearest billion and a negative value denotes a saving ^DWP. The 29 September and 6 October captures carry identical figures, so the 6 October update changed the publication’s framing and not its arithmetic ^DWP.

Three features of the presentation constrain how the numbers can be used. They are in-year savings at two widely spaced snapshots, not an annual path and not a cumulative total, so no figure for any intervening year and no running total can be derived from them — which is the same point DWP’s own caveat above makes when it says the estimates should not be used “to then derive further point estimates for savings in an individual year”. They are direct AME impacts only, excluding tax and debt-interest effects. And the real-terms series shows the compounding plainly: by 2049-50 the nominal saving is more than three times the 2039-40 figure but the real saving less than three times it, so roughly two-fifths of the headline £50 billion is price inflation rather than reduced purchasing power transferred away from pensioners.

(Corpus gap: no captured source quantifies the “record high relative to earnings” that the adjusted mechanism is designed to hold — the phrase appears in prose in both captures and in the factsheet, but no table attaches a ratio or percentage to it, so the target the mechanism locks in cannot be stated here. Re-checked 10 October 2026 against the restored tables: the publication contains only the two AME tables above, so this half of the gap is a genuine publisher omission rather than a capture defect and will not close through re-capture. Re-characterised 13 October 2026 against the Commons statement below, which explains why it will not close by any capture: the target is “the level of average earnings that it reaches in April 2030” ^Hans, so it is defined prospectively by the outturn of four further upratings and is not yet a knowable figure. The statement does supply the first magnitude — “over 30% of average earnings by the end of the Parliament” — but as a forecast of the destination, not the locked-in ratio. Narrowed on the measure limb on 15 October 2026: asked in the Lords how the record-high conclusion was arrived at, DWP answered that “[h]istorical average earnings growth and the Consumer Prices Index used in the analysis are consistent with the Office for Budget Responsibility’s economic assumptions published at the Spring Statement 2026” ^WQ. That pins the provenance of both series — including the historical earnings series, not only the forecast — to a single dated OBR vintage, which is more than the publication itself discloses and is enough to date the analysis. It does not name the underlying earnings measure, the averaging period or the lag, and it attaches no ratio. Note also the shape of the exchange: the question asked how the conclusion was reached, and the answer describes only where its inputs came from — consistent with the re-characterisation above, under which the ratio is not yet a knowable figure and so cannot be given however the question is put. Resolved as to why, on 15 October 2026: in the Lords repeat the minister said the earnings measure “will be determined before any legislation is made on the way earnings are measured” ^Hans, so the measure limb is not an omission from the published analysis but a decision not yet taken — it will not close by capture until the government takes it, and the sequence is measure first, legislation after. The pensioner-poverty result is similarly unavailable in tabular form: it is carried by Chart 1, an image, which the re-capture does not restore, and the surrounding prose gives relative pensioner poverty after housing costs falling from about 14% in 2024-25 to about 8% in 2049-50 without any comparison against the unadjusted lock — so the reform’s distributional effect, as opposed to its cost, remains unevidenced. The minister restated those two percentages at the despatch box on 15 October, adding that 8% “would be a record low”, but offered no counterfactual either (same Hansard source).)

One consequence is already visible in the Fund projections. GAD’s report to Parliament on the 2026 orders states that its principal projection assumes the lock continues “for all future years … including the April 2030 up-rating which is expected to fall in the next parliament” ^GAD — the very uprating now to be adjusted. The National Insurance Fund projections to 2030-31 therefore rest on a policy assumption the sponsoring department has since changed; see The National Insurance Fund — the pay-as-you-go plumbing behind the State Pension. The same GAD report also finds that in April 2028, April 2029 and April 2030 the highest of the three components is the 2.5% minimum (same source). If that holds, the adjusted lock’s first uprating would deliver 2.5% under either rule, because the floor binds — which would locate the savings in later years, when earnings growth would otherwise have driven the increase. Neither source states that: it is what the two together imply on GAD’s assumptions.

The analysis’s other half projects pensioner poverty. Relative low income after housing costs among those over State Pension age is projected to fall from “around 14% in 2024 to 2025 to around 8% in 2049 to 2050” under the adjusted lock ^DWP. That figure needs reading with care — the chart is framed against 2024–25, not against the current triple lock, so it does not show the effect of the adjustment on poverty; see WPC pensioner-poverty inquiry 2025 — challenges, mitigations and the Pension Credit gap.

The 6 October 2026 reform factsheet

A week after publishing the analysis, DWP added a “Triple Lock reform factsheet” to the same page (6 October 2026). It carries no new modelling — the analysis text is unchanged — but it does two things the analysis did not.

First, it states the new rule as a three-limb test in the government’s own voice, in the same drafting form as the existing lock. From April 2030 the basic State Pension and the full rate of the new State Pension “will rise each year by the highest of”: “inflation (the rise in the consumer price index)”; “2.5%”; and “the amount required to return or maintain the value of the State Pension relative to earnings” (raw/dwp/2026-10-06-state-pension-uprating-analysis-2026-tables.md, emphasis added). That third limb is a sharper formulation than the analysis’s “anything more that is needed to retain that value”: “return or” makes the earnings limb restorative as well as protective. On the face of the words, if a year in which the CPI-or-2.5% floor binds leaves the pension below its target ratio to earnings, the following year’s increase can exceed earnings growth to bring it back — so the adjusted lock is a two-way tracker around a ratio, not a one-way cap. Neither document says that in terms, and no captured source defines the target ratio, the measure of earnings used, or the lag at which the comparison is made.

Second, it addresses the transition: “[p]ensioners who currently receive a State Pension will continue to receive their Triple Lock pension until April 2030”, and from April 2030 “they will receive the adjusted State Pension, which will continue to rise every year and pensioners will continue to be protected against inflation” (same source). So the change is date-based, not cohort-based — it applies to existing pensioners from the same uprating as to new ones, unlike the single-tier reform, which took effect only for those reaching State Pension age on or after 6 April 2016 (raw/hmt/2016-04-05-new-state-pension-for-millions-of-new-pensioners.md; see The 2016 abolition of contracting-out — and the legacy SIs still in force). The factsheet also supplies the government’s framing — “[t]hese reforms ensure that the State Pension is sustainable, protecting pensioners of today and today’s workers when they’re pensioners” (same source) — which is the same sustainability argument the third State Pension age review is weighing on the other lever.

One figure is new, and is weaker than it looks: “[t]he full nSP is set to rise by over £2,000 and reach a record high relative to earnings over this parliament” (same source). It is a cash rise over the parliament under the existing lock, not a figure about the adjusted one, and the factsheet gives no start year, no weekly or annual rate to measure it against, and no statement of whether it is nominal or real — so it cannot be converted into a level, a percentage or a comparison with the adjusted mechanism. It does not quantify the “record high relative to earnings” that the new third limb is designed to hold.

The 13 October 2026 Commons statement — announced at last, and tied to social care

On 13 October 2026 the Secretary of State for Work and Pensions made an oral statement to the Commons “on the Government’s plan for the future uprating of the state pension” ^Hans. That is the thing the two preceding passes recorded as missing: the reform is no longer only an annexe to an ad hoc analysis but a policy announced to Parliament in its own right, answerable at the despatch box. Three things in it are new, and one conspicuously is not.

First, the benchmark is identified — as a date, not a number. The gap recorded above was that nothing quantified the “record high relative to earnings” the new third limb is designed to hold. The statement does not quantify it either, but it does say what it is: the pension will be “linked to the level of average earnings that it reaches in April 2030”, “instead of outpacing average earnings as the current system does” (same source). So the target ratio is not a figure chosen in advance but whatever ratio the existing triple lock happens to deliver by the April 2030 uprating — which means it cannot be known until then, and that it depends on the earnings and price path over the intervening four upratings. The statement does attach a magnitude to the expected destination, which no earlier capture did: this year’s uprating “will take the new state pension to over 30% of average earnings by the end of the Parliament and it is forecast to be a record high” (same source). That is a forecast of where the ratchet ends, not a statement of the locked-in target, and the statement still names no earnings measure, no reference period and no lag.

The operative wording also differs from the factsheet’s in one respect worth noting. At the despatch box the third limb is “what is required to maintain its record high level against average earnings” (same source), where the factsheet had “the amount required to return or maintain the value of the State Pension relative to earnings” ^DWP. The restorative limb discussed above is therefore in the written factsheet but not in the spoken statement. Nothing captured says which formulation will be legislated, and on a matter this consequential — whether the adjusted lock is a two-way tracker or a one-way cap — the discrepancy should be treated as unresolved rather than as a drafting slip in either direction.

Second, the savings are now stated by a minister and hypothecated. The statement puts the figures in the government’s own voice — “[t]he savings from the changes will reduce spending by an estimated £15 billion a year by the end of the next decade, rising substantially to around £50 billion a year by 2050” — which matches the nominal AME figures in the restored tables above for 2039-40 and 2049-50 ^DWP, though the statement’s looser year labels should not be read as a re-dating of them. What is genuinely new is what the money is for: the savings “will make a significant contribution to a new settlement for older people: a rising, more sustainable state pension alongside free personal care through a new national care service” ^Hans. Two limits on that are in the statement itself — it is a “significant contribution”, not full funding, and the service’s “pace and scale” are left to a separate review reporting “next summer” (same source) — and the statement gives no cost, eligibility rule, delivery body or start date for the care service. The timing asymmetry is plain on the face of the plan and was put to the Secretary of State in the exchange: the savings begin only in April 2030, while care need is current. The reply was that “[i]t is right that we do not wait until April 2030 before trying to improve the system”, that “[e]xtra money is going into the social care system as a result of decisions made by this Government”, and that more is needed for the care workforce “through the fair pay agreement that this Government have promised” — no amount, no date and no source of the interim money being given (same source). For a pensions reader the hypothecation matters less as care policy than as politics of durability — a saving earmarked to a named benefit is harder to reverse than an unallocated one, which is the implicit answer to the comparison the Opposition front bench drew with the Winter Fuel Payment reversal (same source).

Third, the alternatives are expressly rejected. The Secretary of State said that “[e]conomists and commentators have proposed a range of options — as has the Conservative party, over the years — from removing the earnings link to scrapping the triple lock altogether”, and that “[s]ome have called for the state pension to be means-tested. We have rejected those proposals” (same source). That closes off, on the government’s current position, the three options most often canvassed against the lock — and it confirms that the April 2030 change is intended as an alternative to them rather than a step towards them. The statement also repeats the transition rule the factsheet gave: the current lock holds “for the duration of this Parliament”, this year’s full new State Pension rise is £575, and the pension is set to be “over £2,000 a year more by the end of this Parliament compared with when we took office in 2024” (same source) — both cash figures measured under the existing lock, as above.

What the statement does not do is name a legislative vehicle, and that is now the whole of the remaining gap. Nothing in the record names a Bill, an uprating order, a statutory instrument, a consultation or a Budget measure for the adjusted lock, and no legislative date or stage is given; “[f]rom April 2030” is the effective date of the new rule, not a parliamentary step. The one Budget reference in the exchange — “[t]he Chancellor will announce further details at the Budget” — was given in answer to a question about income tax on those reliant on the State Pension, not about the uprating mechanism (same source), and should not be read across. The Pensions Commission has been asked “to include suggestions on further supporting lower-income pensioners in its final report, to be delivered next year” (same source), which is a remit on mitigation rather than on the mechanism; see Retirement Adequacy and the Revived Pensions Commission.

One objection from the exchange is worth recording because it is the correct way to state the reform’s effect and was not disputed. A back-bench member put it that “setting future indexation to a proportion of average earnings means that, over time, the state pension will be lower than under the current system”; the reply was that the gains to April 2030 are locked in and the pension is thereafter linked to the April 2030 earnings level (same source). Both propositions are true at once — the adjusted lock is lower than the unadjusted lock would have been, while still rising every year in cash and holding its earnings ratio. Nothing in the statement claims otherwise, and the £15bn and £50bn savings are the arithmetic of exactly that difference.

The 15 October 2026 Lords repeat — the mechanics, and the cohort the benchmark misses

The statement was repeated in the Lords on 15 October 2026, and the exchange that followed it is more useful than the repeat, because three of the things the Commons statement left open were put directly to the minister ^Hans. The repeated statement text itself is the Commons text verbatim, including the third limb as “what is required to maintain its record high level against average earnings” — so the despatch-box “maintain” formulation has now been given twice, in both Houses, against the factsheet’s “return or maintain” once.

The overshoot mechanics are described for the first time. Answering the Opposition front bench, the minister set out what happens when the pension runs ahead of the benchmark: under the present lock, “[w]hen the state pension increases faster than earnings, that becomes permanent. There is no mechanism through which wages can ever catch back up”, whereas “[o]ur adjustment means a rise in the state pension above the level of earnings in one or more years will mean that future increases will track inflation or 2.5% until wages catch up” (same source). That is the first captured account of how the earnings link is held rather than merely stated, and it tells a reader two things. An overshoot is never corrected in cash — the pension does not fall, and does not rise by less than the CPI/2.5% floor; the correction happens by letting earnings grow past it. And the third limb is therefore additive only: it can top an uprating up above the floor, never pull it below. On the unresolved “maintain or return” question above this is a half-answer. It describes the above-benchmark case, which on this account needs no restorative rule because the floor does the work. It says nothing about the below-benchmark case — whether a pension that has slipped under the April 2030 ratio is uprated above earnings to climb back, which is what the factsheet’s “return” limb would require. So the discrepancy is narrowed on the downside and untouched on the upside.

The earnings measure is unstated because it is undecided. Asked to explain the 30%-of-earnings figure, the minister said the modelling “is based on Pensim3 … a dynamic model that the department uses”, using “the projections that are out there and projections forward on CPI”, and then — on the measure itself — “that will be determined before any legislation is made on the way earnings are measured” (same source). Two points follow, and both matter more than the sentence’s syntax suggests. The absence of an earnings measure recorded as a gap through four passes is now explained: no capture names it because the government has not chosen it. And this is the first captured statement, from the government rather than from a questioner, that legislation is expected at all for the pension change — the Commons statement had named no parliamentary step of any kind. It still names no Bill, no instrument, no session and no date, and the only other legislative commitment given in the exchange is “[w]e will legislate, obviously, for changes to social care, and any legislation we publish will have detailed impact assessments at that point” (same source), which is the care service and not the uprating mechanism. What it does establish is a sequence: measure first, legislation after.

The benchmark covers the new State Pension, not the basic one. The sharpest point in the exchange came from a former Secretary of State for Work and Pensions, and it is arithmetic rather than politics: the “30% of average earnings” destination is a figure about the new State Pension, while “[r]ight now, 59% of pensioners are on the lower basic state pension”, falling to “only 43%” in four years, and the basic “operates only at about 75% of the new state pension”, so “43% of pensioners on the old basic state pension will not be getting the 30% of earnings” (same source; the 59%, 43% and 75% are the questioner’s figures, not the government’s, and no captured source corroborates them). This is a real feature of the reform rather than a debating point. The adjusted lock applies the same uprating rule to both rates — the basic State Pension’s triple-lock treatment is unchanged — but the benchmark being locked in is expressed as a ratio for the new rate, and a pre-2016 pensioner on the basic rate holds a materially lower proportion of average earnings and will continue to. See The 2016 abolition of contracting-out — and the legacy SIs still in force for why the two cohorts exist.

The government’s answer to that was not a change to the basic rate. It was that people on the basic State Pension “have the great advantage … of potential access to pension credit”, that Pension Credit “is there to lift up those who are either on the basic state pension or have an income below the level of pension credit”, and that “[t]hat and pension credit will stay in place” — adding, significantly for anyone modelling the floor, that Pension Credit “has never been the subject of the triple lock, either in previous Governments or in the current Government”, with decisions “made on that in the light of those decisions” each year (same source). So the mitigation for the cohort outside the benchmark is a means-tested benefit carrying no statutory uprating guarantee at all — which is the same structural point Pension Credit and pension income — how the means test treats pensions records about Pension Credit’s take-up, viewed from the other end. The questioner also asked whether the change, not having been in the manifesto, “may not be subject to the Salisbury convention”; that was not answered (same source).

Two further answers are worth recording. On the contributory principle, a back-bencher asked twice “would the intention be to abolish the contributory principle of the state pension and to abolish the National Insurance Fund?”; the replies were “please do not set any hares running” and “nothing like that has been suggested” (same source). That is a denial rather than a policy statement, but it is the only captured government word on whether the adjusted lock disturbs the Fund — see The National Insurance Fund — the pay-as-you-go plumbing behind the State Pension. On Scotland, the minister confirmed there is “no question that the pension will be different in Scotland”, that Scottish pensioners will see it rise “by the higher of inflation or 2.5%, and higher if necessary to maintain the links with earnings”, and that there will be “more funding for social care in Scotland as a result of the changes” (same source); see Pensions across the four nations — NI parity legislation, devolved schemes and tax divergence.

Against the hypothecation argument, the Liberal Democrat front bench put the Institute for Fiscal Studies’ view that “[w]e should not expect this reform to save enough that it could fund universal social care in the next parliament”, adding that the saving is itself very uncertain (same source) — a characterisation by a peer of a third party’s analysis, not a captured IFS publication, but the natural counterweight to the care link recorded above. The minister’s figures in reply were the published ones: the pension “around £2,000 higher” by 2030, around 30% of average earnings by April 2030, and relative pensioner poverty falling “from 14% in 2024-25 to 8% by 2050, which would be a record low” (same source) — the last restating in Hansard the projection that the 29 September analysis carries only as a chart image, though still without the comparison against the unadjusted lock that would show the reform’s own distributional effect.

(Corpus gap: no legislative vehicle for the adjusted triple lock is captured. The announcement half of this gap closed on 13 October 2026 with the oral statement above — the reform is now announced to Parliament as policy in its own right, so the earlier marker’s “no Budget statement, written ministerial statement, consultation or legislative vehicle” is reduced to its last limb. What is still missing is how and when the new mechanism acquires legal footing: each year’s uprating is delivered by a draft Up-rating Order laid with the Government Actuary’s report to Parliament ^GAD, and the current lock’s earnings limb sits in the Social Security Administration Act 1992, so an adjusted mechanism needs either primary amendment or a different basis for the April 2030 Order — and no capture says which, or whether a consultation will run. Narrowed on 15 October 2026: in the Lords repeat the minister said the earnings measure “will be determined before any legislation is made on the way earnings are measured” ^Hans, which is the first government statement that legislation is expected for the pension change and fixes the order of events — measure, then legislation. It still names no Bill, instrument, session or date, and the only express legislative commitment in the exchange was for social care. Also unresolved: which of the two competing formulations of the third limb (“maintain” or “return or maintain”) is the operative one — the Lords exchange supplies the above-benchmark mechanics (overshoots decay under the CPI/2.5% floor until wages catch up) but says nothing about the below-benchmark case the “return” limb would govern. See The National Insurance Fund — the pay-as-you-go plumbing behind the State Pension and Pension indexation and increases — statutory minimums, RPI vs CPI, and the 2030 RPI reform.)

Taxation of the State Pension

The State Pension is liable to income tax — the long-standing view is that state pensions, like other “earnings-replacement benefits”, should be taxable ^HoC. But DWP pays it gross: no tax is deducted at source. Instead the State Pension is folded into the PAYE calculation applied by a pensioner’s private or occupational pension provider, so tax on total income is collected there. A pensioner whose only income is the State Pension generally pays no tax in practice, because that income has fallen below the personal allowance — the tax-free threshold ^HoC.

That last point is now under strain. Applying the triple lock to uprate the basic and new State Pension while the personal allowance is frozen at £12,570 (its 2021/22 level) means the tax-free headroom above a full State Pension steadily narrows — a fiscal-drag effect pulling a growing number of pensioners into paying some tax on their State Pension. Those affected may have to file a self-assessment return, or, for straightforward affairs, receive an HMRC bill under simple assessment ^HoC.

The 2025 Budget addressed both sides of this. It extended the personal allowance freeze to April 2031 — in the Financial Secretary’s own words to the Lords, the Budget “maintains all income tax and equivalent national insurance thresholds at their current level for a further three years from 2028”, a decision he expressly accepted “will affect working people” ^Hans — prolonging the drag; but it also announced that from 2027/28 pensioners whose sole income is the basic or new State Pension will not be charged small amounts of tax via simple assessment even once the State Pension exceeds the personal allowance — a targeted carve-out for the State-Pension-only cohort. The government has still not published how this easement will be implemented: the Commons Library re-issued this briefing on 22 September 2026 with that sentence unchanged from its 22 July edition, so two months on the mechanism — who is taken out of simple assessment, and by what legislative route — remained unspecified ^HoCHoC. For trustees and administrators that matters mainly in negative: until it is specified, nothing in the easement changes how a scheme operates PAYE on the private or occupational pension it pays. The wider allowance-and-thresholds interaction with private pensions is covered in pensions tax.

How narrow the headroom now is

The 2026 parliamentary record puts an exact figure on the gap. The uprating order debated on 10 February 2026 set the full new State Pension for 2026-27 at £241.30 a week ^Hans — £12,547.60 a year, against a personal allowance of £12,570. The headroom is £22.40 a year ^Hans. On any plausible uprating, a full new State Pension exceeds the frozen allowance at the next uprating. (The 15 June debate quotes the weekly figure as £240.30 alongside the same £12,547.60 annual figure; the two are not consistent with each other, and the £241.30 in the uprating order is the authoritative one.)

That collision was the subject of a Westminster Hall debate on 15 June 2026 on an e-petition signed by over 120,000 people seeking a doubled or protected personal allowance for State pensioners. The government ruled the proposal out: doubling the allowance for pensioners would cost “several billion pounds each year” and is inconsistent with its stated priorities of raising the State Pension and funding the NHS, and the minister said flatly that “no political party will deliver a doubling of the personal allowance for pensioners” ^Hans. The government’s answer is instead the simple assessment easement: ministers confirmed the Chancellor’s commitment that pensioners whose sole income is the State Pension will not have to deal with small tax bills from 2027-28, with the enabling legislation “included in the next Finance Bill” ^Hans. Written answers of 8 September 2026 repeat the same formulation — simple assessment “from 27/28,” with further detail “in due course” ^WQ — so as at the latest source held here the mechanism is still announced rather than legislated.

The freeze has a measured caseload. A Treasury written answer of 5 August 2026 (HL2380) put numbers to the drag. The count of individual income taxpayers over State Pension age was 8,160 in 2023-24 — the only outturn figure — then 8,780 (2024-25), 9,080 (2025-26) and 9,580 (2026-27), the last three being projections from the 2023-24 Survey of Personal Incomes on assumptions consistent with the OBR’s March 2026 Economic and Fiscal Outlook. That is a rise of about 17% in four years. The answer attributes it directly to the freeze: the previous Conservative government’s decision “to maintain income tax thresholds at their current levels from April 2021 until April 2028… is reflected in the numbers” ^WQ. Note the freeze end-date given there is April 2028; the 2025 Budget extension to April 2031 recorded elsewhere in this article post-dates that framing.

(Corpus gap: the capture of HL2380 does not state the units of that table, and the GOV.UK income-tax-liabilities statistics it cites are not themselves in the corpus. A count of pensioner taxpayers is conventionally in the millions, so the figures are only coherent read as thousands — roughly 8.16m rising to 9.58m — but that is inference, not something the capture states, so the raw values are given above as printed. Re-checked 5 October 2026 against the Treasury answer below: it gives a figure in millions but no table and no units statement, so it makes the thousands reading more plausible without establishing it.)

The same answer carries a government commitment in broader terms than the simple-assessment easement described above: individuals “whose only income is the basic or new State Pension, without increments, will not pay income tax over this Parliament”, with “further details… to follow” ^WQ. That is a commitment about liability, where the June and September sources describe a commitment about collection — not having to deal with small tax bills from 2027-28 ^HansWQ. No captured source reconciles the two formulations or names the mechanism for either, so which is the operative promise is not established here.

Two framing points from that debate are worth keeping, because they recur. First, the taxation of the State Pension is not new or anomalous: over 80% of pensioners were already paying income tax by the end of the previous government, and the system has treated state and private pensions as taxable income throughout ^Hans. Second, the age-related personal allowance — the more generous pensioner-specific allowance that once existed — was abolished under the coalition government, so the comparison is with a policy both main opposition parties have previously ended (same source).

Both framings were restated by the Treasury on 5 October 2026, answering a Lords question that asked bluntly “whether they intend to tax state pensions” (HL3707). The answer did not address the intention at all. It gave the legal position — “It has always been the case that pension income, from private pensions and the State Pension, is taxable” — and then the caseload: “Under the previous Government, 8.3 million pensioners, the vast majority, paid income tax” ^WQ. Three cautions on using that number. It carries no tax year — “under the previous Government” is a period, not a date — and no cited source, so it cannot be matched to any row of the HL2380 table above; a “pensioner” count and a count of “individual income taxpayers over State Pension age” need not share a definition in any event. And the answer says nothing about the frozen threshold, the triple lock, the personal allowance or the 2027-28 easement, so it neither advances nor retracts the liability commitment recorded above. Its practical use is as framing: the department’s standing line is that taxing the State Pension is the long-standing position rather than a change, which is the same argument as the 80% figure from the June debate, now given by Treasury in writing and in a later Parliament.

A separate administrative failure sits alongside the policy one: up to 8.7 million pensioners were reported to have been overcharged tax, with around £43.5 million collected in error and an HMRC fix expected in summer 2026 (raw/hansard/2026-06-15-state-pensioners-personal-allowance-commons.md — figures cited in debate from press reporting, not a government statistic).

What HMRC now expects the pensioner to do

While the policy question stays open, HMRC has been building out the self-service side of it. On 5 October 2026 it published a GOV.UK tool, Check if you have to pay tax on your pension, which returns an estimate of “whether you need to pay tax on your pension” and “how much tax you may need to pay this year” ^HMRC. Three features of it matter here.

First, its stated basis is the frozen threshold itself. The result assumes the user has “the standard Personal Allowance of £12,570” (same source) — the same 2021/22 figure the triple lock is converging on, so HMRC’s own public estimator is built directly on the collision described above, with no pensioner-specific allowance in it. That is the abolition of the age-related allowance showing up in the tooling.

Second, the inputs are the total-income model made operational. The tool asks for the types of pension expected (State Pension or private pension), the State Pension amount (“you can use your annual, weekly or 4-weekly amount”), the private pension amount, “the amount of any other taxable income you’ll get (for example, from employment or income from savings interest)”, and whether the user receives or transfers Marriage Allowance ^HMRC. That is the gross-payment-plus-aggregation structure set out at the top of this section: the State Pension is not taxed where it is paid, so liability can only be worked out by adding everything together. Note what is absent from the asked-for inputs — the tool is about pension income in payment, and nothing in the capture extends it to lump sums, drawdown withdrawals or trivial commutation, which are the over-taxation cases handled by the reclaim forms in HMRC tax administration.

Third, it has express limits that fall on the cohorts most likely to need it. The result assumes the user started “to get your State Pension after 5 April 2010”, and the page states “You cannot include any foreign income or Blind Person’s Allowance you get” ^HMRC. The post-2010 assumption excludes the oldest pensioners from a reliable answer; the foreign-income exclusion excludes anyone with an overseas pension or overseas-source income, which also limits its use for the cross-border cases in Brexit and cross-border pension schemes — the EU Exit SIs and what happened to EU-derived pensions law. The capture names no alternative route — no phone or postal fallback — for anyone the tool cannot serve.

Read against the policy timeline, the sequencing is the point. This tool and the companion postal-reclaim guidance of 1 October 2026 ^HMRC arrived within five days of each other, both addressed to the member rather than the scheme, and both about a pensioner working out their own position. Meanwhile the 2027/28 simple-assessment easement — the government’s actual answer to the headroom problem — remains announced and unlegislated, with implementation detail still unpublished as at the Commons Library’s 22 September 2026 re-issue ^HoC. So for 2026-27 the operative position is the self-service one: the pensioner checks, and HMRC assesses afterwards. For trustees and administrators that changes nothing about the PAYE they operate, but it is the answer to give a member who asks how to find out whether their State Pension is now taxable.

State Pension age

State Pension age (SPA) is currently 66. Under the Pensions Act 2014 it rises to 67 between 2026 and 2028, and under the Pensions Act 2007 to 68 between 2044 and 2046 ^DWP. The law requires a review at least every six years, each informed by a Government Actuary’s report and an independent report. Three independent reviews have run: John Cridland (2017), whose “Smoothing the Transition” recommended bringing the rise to 68 forward to 2037–2039 ^DWP; Baroness Neville-Rolfe (2023), after which the government confirmed the 67 rise but declined to legislate the 68 bring-forward, deferring to a further review ^DWP; and the third review (2025) under independent reviewer Dr Suzy Morrissey, examining the merits of linking SPA to life expectancy and international automatic-adjustment mechanisms ^DWPGAD. In his 13 July 2026 Mansion House speech the Minister for Pensions confirmed he had asked the revived Pensions Commission to use its final report (due early 2027) to feed into the Secretary of State’s State Pension age review, “adding to the input from Dr Suzy Morrissey and the Government Actuary” — tying the long-run adequacy work directly into the statutory SPA decision ^DWP. The pressure is demographic and fiscal: the over-SPA population is projected to grow from 12.6m (2025) to 19.5m by 2075, and State Pension spending from £146bn (2025–26) toward a projected 7.7% of GDP by the early 2070s ^DWP.

Watch which “68” a projection is using. The OBR’s July 2026 Fiscal Risks and Sustainability report works on its own assumption that SPa reaches 68 between 2037 and 2039 — the Cridland timetable that was never legislated — because the OBR is independent in setting its forecast assumptions ^WQ. HM Treasury has confirmed it has not assumed or quantified any savings to public service pension schemes from such a bring-forward, and that actuarial valuations under Treasury Directions use the currently legislated timetable, under which the rise to 68 takes place in 2044–2046 (same source). Two official bodies are therefore projecting on different SPa timetables, and neither is wrong: the OBR is forecasting likely policy, the Treasury is valuing enacted law. Any cost or savings figure quoted in this area should be read back to which of the two it assumes. See State Pension age reviews — the statutory machinery, Cridland, Neville-Rolfe and the third review.

On the fiscal path itself, the OBR’s sustainability analysis projects State Pension expenditure rising from around 5% of GDP to around 9% of GDP by FYE 2076, while net public service pension expenditure falls from around 1.2% to around 0.9% of GDP over the next 50 years — a divergence that is easily missed when the two are discussed together as “the pensions bill.” Government’s position is that these commitments are “fully funded within our fiscal rules” ^WQ.

How future rises are managed and communicated is itself under parliamentary scrutiny. The Work and Pensions Committee ran an inquiry on the Transition to State Pension age and published its 2nd Report of the session (HC 76) on 11 July 2026 ^WPC. The inquiry follows the notice-period and communications failures at the heart of the women’s SPA communications chapter below, on which the same committee took the DWP Permanent Secretary’s 1 July 2026 correspondence confirming the action plan’s publication.

Pension Credit

Pension Credit is the means-tested top-up for the poorest pensioners, with a Guarantee Credit floor of £238.00 a week (single) / £363.25 (couple) and an average award above £87 a week ^DWP. Take-up is the perennial problem: around a third of eligible households do not claim — roughly 910,000 households missing out, with an estimated £1.38bn a year unclaimed by those entitled to £50+ a week ^DWPDWP. Entitlement turns on how a claimant’s pension arrangements are assessed, and DWP’s Decision Makers’ Guide is explicit that an undrawn money-purchase pot past qualifying age is treated as producing the annuity income it could generate — a rule that catches savers who assume leaving a pot alone keeps it out of the means test ^DWP. The mechanics are set out in Pension Credit and pension income — how the means test treats pensions.

Take-up became acutely salient when the Winter Fuel Payment was restricted to Pension Credit recipients from 2024–25: claims surged 152% in the eight weeks after the announcement, and a back-dating deadline of 21 December 2024 drove roughly 150,000 claims and 42,000+ extra awards over the following months, processed by 500 redeployed staff against waits of up to nine weeks ^HansHansHans. Even so, the structural gap persists — 800,000+ eligible households still not claiming as of early 2025 ^Hans.

One constraint on every take-up campaign is measured rather than assumed, and it should temper any digital-first strategy. DWP’s Claimant Service and Experience Survey 2017/18 — 15,514 telephone interviews across ten benefits between August 2017 and May 2018 — found that “[p]ension age claimants still report relatively low levels of access to the internet”, with 54% of Pension Credit claimants and 63% of Attendance Allowance claimants having internet access, against figures in the 90s for working-age groups ^DWP. Satisfaction with DWP service was meanwhile high among both pension-age groups — 93% for State Pension claimants (base 482) and 92% for Pension Credit (base 455), against 84% across all claimants — though State Pension satisfaction had fallen from 97% in 2015/16, and the share finding it easy to get in touch fell from 87% to 82% between 2016/17 and 2017/18, which the survey flags as a decrease rather than noise (same source). Two cautions on citing it: the pension-age bases are around 450 against roughly 6,000 for Universal Credit, so margins of error are materially wider and the survey itself recommends reading within-benefit trends rather than comparing across benefits; and it is not an Official Statistic, though it applies the Code of Practice for Statistics (same source). (Corpus gap: the survey’s numbered data tables, which the report cites throughout, are not held in the capture — only the report text, charts and methodology tables — so most State Pension and Pension Credit measures described in the prose as “stable” carry no retrievable figure.)

From winter 2025/26 that means-test was largely reversed. In June 2025 the government announced that all people over State Pension age would again be eligible for the Winter Fuel Payment, but that individuals with taxable income over £35,000 a year — and not also on a means-tested benefit — would have it recovered through the tax system under a new Winter Fuel Payments Charge, applied UK-wide and mirrored by the Scottish and Northern Irish administrations for their equivalents ^HoC. The charge is assessed on an individual, not household, basis — so where only one partner in a couple is above £35,000, only that partner’s payment is clawed back — and is collected either by a PAYE tax-code change or a self-assessment return, with an opt-out that persists until the person opts back in ^HoC. Reach widens sharply: around 12.3 million people across the UK are expected to receive a payment in 2025/26 and roughly 2.2 million to have it withdrawn by the charge, leaving about 10.2 million net beneficiaries — against just 1.5 million who qualified under the 2024/25 means-test ^HoC. On cost, the restored policy is expected to run £1.3–1.4bn a year more than the means-tested-only approach, while the charge itself saves only about £450m a year against full universality — a saving the briefing notes may be largely offset by the extra Pension Credit awards the 2024/25 take-up drive generated ^HoC. The government frames the £35,000 threshold as “balancing support for lower income pensioners with fairness to the taxpayer”, while the Resolution Foundation and the IFS flag the £35,000 “cliff edge”, the extra self-assessment burden, and that the threshold is not inflation-indexed ^HoC. Unlike WFP in England, Wales and Northern Ireland — which is not uprated — Scotland’s equivalent Pension Age Winter Heating Payment is now uprated (£203.40 / £305.10 for 2025/26) ^HoC.

A recurring fairness point concerns veterans: only £10 a week of an Armed Forces Compensation Scheme award is disregarded for Pension Credit, so a service-injury pension can sharply cut a means-tested entitlement. The government has defended this as consistency across the means-tested system, while a private member’s “Pauline’s law” seeks a review ^Hans.

A narrow but relevant change from 16 July 2026: SI 2026/681 inserts a capital disregard so that the lump-sum payments DWP is making to carers under the Carer’s Allowance overpayment-reassessment exercise are ignored as capital when assessing Pension Credit (as well as Housing Benefit and Universal Credit) — protecting the entitlement of pension-age claimants who receive a retrospective carer’s payment ^leg.

Frozen State Pensions overseas

The UK State Pension is paid worldwide, but the annual increase is paid only where there is a legal requirement to do so — i.e. to pensioners living in the UK, the EEA, Gibraltar or Switzerland, or in a country whose social security (“reciprocal”) agreement with the UK provides for uprating ^DWPDWP. DWP’s published list of agreement countries where increases are paid is short: Barbados, Bermuda, Bosnia-Herzegovina, Gibraltar, Guernsey, the Isle of Man, Israel, Jamaica, Jersey, Kosovo, Mauritius, Montenegro, North Macedonia, the Philippines, Serbia, Turkey and the USA. Canada and New Zealand have agreements with the UK but they do not provide for uprating, and Australia has none — so pensioners there, like those everywhere else outside the list, have their State Pension “frozen” at the rate first paid ^DWP. The freeze is tied to residence, not nationality: a pensioner who returns to live in the UK is restored to the current rate ^DWP.

The EEA leg of that list rests on treaty provision, not on a reciprocal agreement in the ordinary sense, and it is structured by cohort rather than by country. For people already in a UK–EU cross-border situation at 31 December 2020, uprating flows from the Withdrawal Agreement: a person within Article 32(1)(a) — anyone with periods of insurance, employment, self-employment or residence completed under UK or EU legislation before 1 January 2021 — “will have their UK State Pension paid in the EU at the same rate as in the UK (uprating)”, and payment must be made “without any reduction, amendment, suspension, withdrawal or confiscation” by reason of residence outside the competent state ^HMRC. For people who moved afterwards, the Protocol on Social Security Coordination in the Trade and Cooperation Agreement carries it: on the old-age benefits row the government’s own comparison of the two agreements records “No change. State Pension continues to be aggregated, exported and uprated” (same source). The EEA EFTA Separation Agreement and the Swiss Citizens’ Rights Agreement extend the equivalent treatment to Norway, Iceland, Liechtenstein and Switzerland, the TCA Protocol itself applying to the EU only (same source).

Two practical consequences for advisers. First, the uprating entitlement of a pensioner in the EEA is not conditional on a bilateral agreement that could lapse the way the Canada and New Zealand agreements failed to provide for uprating in the first place — it is carried by the Withdrawal Agreement for the pre-2021 cohort and by the TCA Protocol for everyone since, which is why this is the one part of the overseas picture Brexit left intact. Second, the mechanics are cohort-tested at 31 December 2020, so a client’s position turns on where they and their contribution record were on that date rather than on where they live now; the full apparatus — full scope under Articles 10 and 30, partial scope under Article 32, aggregation to meet the UK’s 10-year minimum, and what the TCA removed around the pension (Winter Fuel Payment, S1 pensioner healthcare) — is in the Brexit article. Note the source is an operational staff guide whose own version statement is dated 11 November 2020, and whose capture is truncated at 50 of 96 pages, so it should be read as the position at the start of the regime (same source).

The scale has been stable for years. DWP’s 2019 ad-hoc costing put the frozen population at around 510,000, 84% of them in Australia, Canada and New Zealand, and modelled the cost of uprating them from a March 2018 data extract ^DWP. That costing’s figures are: £600m in 2019–20, £610m in each of the next two years, £630m in 2022–23 and £640m in 2023–24 — £3,090m over the five years, nominal and rounded to the nearest £10m (same source). Two features are worth noting. The near-flat profile — a 7% rise in nominal terms across five years — reflects two offsetting forces the analysis models: uprating compounds the cost of each surviving award, while mortality in an ageing frozen population reduces the caseload, and the two roughly cancel. And it was built from a 5% extract of State Pension administrative data at March 2018, uprated against frozen amounts using mortality and claims forecasts and OBR economic assumptions (same source) — so it is an administrative-data estimate, not a survey.

The ministerial figure has since been updated to about £0.9bn a year (March 2022 data) to bring all frozen pensions up to current UK levels ^Hans, which is roughly 40% above the 2019 analysis’s final year — a gap that the high inflation of 2022–23 and the triple lock’s effect on the UK rate the frozen pensions are measured against would largely explain, though no captured source reconciles the two figures. MPs still cite “nearly half a million” affected pensioners ^Hans.

The government’s position has been unchanged across Conservative, coalition and Labour administrations: uprating is prioritised for UK residents because the uprating levels reflect the path of UK earnings and prices, and no new uprating agreements are being negotiated ^HansHans. DWP restated it twice on 2 September 2026, in answers to two separately-framed questions — one on the financial impact on pensioners in non-agreement countries, one on the equity of applying different uprating rules to identical National Insurance records — with identical text and the flat conclusion that “there are no plans to change this policy”, adding that the policy has been supported by successive governments, that setting pensioner-benefit expenditure prioritises those living in the UK, and that it has been upheld in the courts ^WQ.

That second answer settles a point advisers are regularly asked about. A National Insurance record does not travel. DWP’s answer is that NI contributions and credits determine entitlement to the State Pension and the level at which it is first paid, but do not determine whether it is uprated once in payment abroad — that depends solely on the country of residence and whether the UK has a legal basis for uprating there; uprating tracks UK earnings growth and price inflation, “which has no direct relevance where the pensioner is resident overseas” ^WQ. So two people with identical 35-year records can receive permanently different amounts on the basis of residence alone, and the government’s answer to that being inequitable is that the comparison is not the relevant one. For advisers the practical points are the residence test (a pensioner cannot split the year between countries for payment purposes), the International Pension Centre as the claims and change-of-circumstances channel, and the interaction with double-taxation agreements on where the pension is taxed ^DWP.

Underpayment corrections

Two correction exercises have addressed historic State Pension errors. The LEAP exercise — covering married women on a husband’s record, the over-80 uplift and widowed pensioners — completed in 2025, identifying 130,948 underpayments worth £804.7m in arrears ^DWP. The three categories behave very differently, and the breakdown matters more than the total for anyone judging where administrative risk sits:

(raw/dwp/2025-04-23-state-pension-underpayments-progress-on-cases-reviewed-to-31-march-2025-tables.md; hit rates calculated from the published columns). The widowed category is the expensive one — a third of the cases but 60% of the money, at an average of £11,725, because a missed inherited entitlement compounds over the whole of a survivor’s retirement. The over-80 category is the opposite: by far the highest strike rate (more than one case in three was underpaid) but the smallest average, reflecting a flat Category D uplift rather than a recalculated accrual. DWP cautions that the categories are not mutually exclusive — a claim checked for more than one cause of error is counted in more than one row, so the rows cannot be summed to a population — and that the averages include cases where the arrears actually paid were £0 because an offsetting overpayment of another benefit (its example is Pension Credit) cancelled them out; some cases also remain unpaid because the customer has died and no estate has been identified (same source). Category BL and D figures continued to move after LEAP closed for them in December 2023, because customers have up to two years to supply further information (same source) — a point worth noting before treating any instalment of this series as final.

How much any instalment can mislead is now visible, because an early one is held. At 28 February 2022, thirteen months in, the exercise had identified 14,239 underpayments worth £94.3m — 5,122 married (Cat BL), 3,626 widowed and 5,491 over-80 (Cat D) — from 38,131, 11,540 and 9,138 cases reviewed respectively ^DWP. Read against the final figures above, the hit rates fell sharply as the exercise progressed: 31% of widowed cases reviewed yielded an underpayment at February 2022 against 11% at completion, and 60% of over-80 cases against 37% (rates calculated from the published columns of the two releases). The average arrears moved too, and not in one direction — widowed rose from £9,506 to £11,725, while over-80 halved from £4,104 to £2,203 and married fell from £7,399 to £5,553. So the early management information overstated the strike rate in every category and misstated the average in all three. DWP’s own quality statement anticipated exactly this, warning that correction rates and arrears amounts were “likely to change as [the] exercise progresses” because of varying individual characteristics and the recruitment of extra staff ^DWP. Four further instalments are now held, and they change that lesson in two ways. At 31 October 2022 the exercise had identified 31,817 underpayments worth £209.3m, at 28 February 2023 46,716 worth £300.1m, at 31 October 2023 82,323 underpayments owed a total of £497.0 million, at 29 February 2024 97,016 underpayments, owed a total of £571.6m, and at 30 September 2024 119,050 underpayments, owed a total of £736 million ^{DWPDWPDWPDWPDWP}. With February 2022 and the 2025 completion, that gives seven points on the same three categories:

(hit rates calculated from the published columns of the seven releases; the averages are as published; bold marks the anomalous readings discussed below — the married hit rate’s two-year rise, and each other series’ turning point. The October 2023 release gives 270,024 married cases reviewed against 37,488 underpayments, 240,786 widowed against 17,894 and 83,154 over-80 against 26,941, with £220.3m, £219.9m and £56.7m repaid; February 2024 gives 317,955 / 43,367 / £243.8m, 298,099 / 21,175 / £262.3m and 89,634 / 32,474 / £65.5m; September 2024 gives 321,142 / 45,907 / £250.6m, 445,188 / 39,706 / £417.2m and 90,720 / 33,437 / £68.2m.)

The two 2024 instalments overturn two conclusions earlier passes drew from the five-point version of this table, and both corrections run the same way.

First, it is not only the married category whose hit rate is non-monotone — all three are, and the troughs are now dated. The over-80 rate was previously described here as declining smoothly from 60% towards its final 37%; it in fact runs 60% → 45% → 39% → 32% → 36% → 37% → 37%, bottoming at October 2023 and recovering through 2024. The widowed rate runs 31% → 31% → 18% → 7% → 7% → 9% → 11%, with its trough at February 2024 (7.10%, marginally below October 2023’s 7.43%) and the same recovery shape. So no category in this exercise supports a “worked the most likely cases first” reading, and the earlier statement that the over-80 series declines smoothly should not be relied on.

Second, the averages are not all monotone either, and with the 2024 points in only the married series is. Married falls at every step (£7,399 → £6,929 → £6,630 → £5,931 → £5,713 → £5,591 → £5,553). The other two both overshoot their final values and come back. Widowed peaks at £12,486 at February 2024 — not at October 2023, as this article previously recorded — before settling at £11,725, so the mid-exercise average for the largest category by value sat above its final level for at least a year. Over-80 undershoots, falling to £2,192 at February 2024 and then edging back up to £2,202 and £2,203; the movement is only £11, but it disposes of the “falls at every step” reading for that category too. The original conclusion — that all three moved monotonically towards their final values and so were directionally reliable from thirteen months in — now holds for one category of three.

What explains both corrections is visible in the marginal rates, and it sharpens the denominator problem rather than replacing it. Between February and October 2023 the reviewed-case counts explode — married from 82,225 to 270,024, widowed from 54,150 to 240,786, over-80 from 37,163 to 83,154 — while identified underpayments rise far more slowly, giving marginal strike rates of 8.1% (married), 4.3% (widowed) and 27.0% (over-80) on that tranche. The two 2024 instalments then show the pattern inverting category by category rather than all at once: October 2023 to February 2024 gives marginal rates of 12.3%, 5.7% and 85.4%, and February to September 2024 gives 79.7%, 12.6% and 88.7% (calculated from the published columns of the four releases). Each category in turn takes a bulk low-yield tranche and then a small high-yield one — widowed last, adding 147,089 reviewed cases between February and September 2024 — which is exactly the shape that produces an undershoot and recovery in every cumulative rate.

That is the behaviour the documented denominator change would predict. The widowed rate had already halved between October 2022 and February 2023 — reviewed cases more than doubling, 25,268 to 54,150, against identified underpayments rising only 26%, 7,876 to 9,928, an incremental strike rate of about 7% — and the February 2023 release is the first to carry the footnote that “[c]ases reviewed includes cases which have been deemed out of scope of the LEAP exercise through an automated process” ^DWP. The October 2022 release has no such note ^DWP, and the October 2023 release carries the “may be checked for more than one potential cause of error” and the £0-after-offset notes but states no out-of-scope volume either ^DWP. No release says how many cases are out of scope or whether earlier figures were on the same basis, so the attribution remains this article’s inference — but a denominator silently absorbing automatically-excluded cases, in bulk, through 2023, would generate precisely the pattern above, and it is the only mechanism any release offers. The practical rule stands and is now firmer: the denominators in this series are not like-for-like between instalments, so a hit rate should be quoted only from a single release, never as a trend, and the same caution carries to the HRP exercise below.

The final tranche then proves that rule outright, rather than leaving it as an inference from a footnote. Between September 2024 and completion, married reviewed cases rise by just 297 while married underpayments rise by 1,097, and over-80 reviewed cases rise by 33 against 246 further underpayments (calculated from the published columns of the September 2024 and March 2025 releases). As marginal strike rates those are 369% and 745% — arithmetically impossible, because a case cannot be found underpaid without being reviewed. So the two columns are demonstrably not on a common basis: “cases reviewed” is effectively frozen once a category closes, while “underpayments identified” keeps accruing. The releases supply the mechanism — Category BL and D figures continue to move after closure because customers have up to two years to provide further information, which the September 2024 release states as a note explaining why its own BL and D numbers differ from those in the 2023-24 Annual Report and Accounts ^DWP. So a hit rate from this series is not a rate at all once a category has closed.

A further defect, running through every instalment, matters to anyone scaling these figures. The published “average arrears payment” cannot be reconciled with the two columns beside it in any release: multiplying the average by the underpayments identified always exceeds the published “total amount repaid”, and the discrepancy grows as the exercise runs — for widowed, £1.7m (0.8%) at October 2023, £2.1m (0.8%) at February 2024, £55.5m (13.3%) at September 2024 and £105.9m (21.9%) at completion (calculated from the published columns of the four releases). The consistent direction points at the owed-versus-repaid confusion discussed next — an average computed over arrears owed, against a column reporting arrears actually paid, would behave exactly this way — but no release states which basis its average uses, so that remains inference. What does not depend on the explanation is the warning: the published average times the published case count overstates arrears repaid by up to a fifth, so the two should not be multiplied to reconstruct a category total.

DWP’s one forecast in this series also slipped, and the slip is quantifiable. The September 2024 release stated that DWP “has completed two of the three customer groups: the Cat BL and Cat D cases” and that “[p]rogress remains on track for the Widowed cases to be completed by the end of 2024” ^DWP. The exercise in fact ran to the release covering cases reviewed to 31 March 2025, and in the interval after that forecast completion date the widowed category added 20,128 reviewed cases and 10,555 underpayments — so 21% of all widowed underpayments, worth £66.2m of arrears, were identified after the date DWP said the category would be finished (calculated from the published columns of the September 2024 and March 2025 releases). For the category carrying 60% of the exercise’s money, that is a material miss, and it is the only forward-looking statement any instalment in the series makes.

Two further cautions on the October 2023 release itself. Its headline says the exercise “has identified 82,323 underpayments, owed a total of £497.0 million” while its table column is headed “Total amount repaid” and sums to £496.9m — the release does not reconcile owed with repaid, and attaches only “totals may not add due to rounding” (same source). And it names the three categories only as “Married (Cat BL)”, “Widowed” and “Over 80 (Cat D)” without defining the underlying legal error in each, gives no expected completion date, and — although it states that it updates the 23 March 2023 release — reproduces no figure from it, so the instalment-to-instalment comparison above has to be constructed from the releases rather than read off any one of them (same source). (The October 2022 page also carries DWP’s own “latest release … can be found in the collection” banner, so it is a superseded instalment by the publisher’s own marking and should only ever be cited as a point in this series.)

The chain of releases is itself broken, which is why the series above has to be assembled rather than followed. The February 2024 release names two different predecessors — its lead paragraph links “cases reviewed to 31 October 2023” while its body says it updates “the previous release of 30 November 2023” ^DWP. The September 2024 release then skips the February 2024 one altogether, giving its predecessor as “22 July 2024, within DWP’s Annual Report and Accounts” ^DWP, so part of this series was published inside the ARA rather than as a standalone release and is not held here as a separate capture. Neither 2024 release reproduces any figure from its predecessor, neither gives a caseload remaining, and the February 2024 one gives no expected completion date at all. The status label also moves between them: February 2024 says the publication is “not part of our regular Official and National Statistics Publications” while September 2024 says “not part of our regular Official and Accredited Official Statistics publications” (same sources) — the same management-information status under the post-2023 nomenclature, but worth noting for anyone quoting the caveat verbatim.

A separate Home Responsibilities Protection correction (for missing HRP on the NI records of people who claimed Child Benefit before May 2000) found 12,379 underpayments worth about £104m, at an average of £8,377 ^DWP. The pipeline behind that figure shows how much attrition sits between a mailshot and a payment. HMRC sent 370,018 letters; it processed 8,639 applications from people under State Pension age and 44,296 from people over it; only 22,781 cases reached DWP (HMRC passes on only those it identifies as missing qualifying years because of HRP); DWP processed 21,878; and 12,379 of those yielded arrears (same source). So fewer than one in twenty letters produced a correction — not because the others were wrong, but because many recipients already had enough qualifying years, already drew a higher inherited State Pension, or did not meet the conditions (same source). Two further details in the table deserve attention. As with LEAP, the £8,377 average includes cases paid £0 after benefit offset. And engagement with HMRC’s online eligibility checker collapsed: the exercise recorded 493,813 hits to end-September 2024, but from October 2024 to March 2025 only about 2,000 individuals a month used it (same source, which also notes the measurement method changed, so no consistent total spans the whole period). A self-service correction route that is reaching a couple of thousand people a month is not, on its own, going to find a cohort of this size — which is why the residual depends on the direct mailing rather than the tool.

Why the HRP mailshot did not convert — the non-applicant research

The attrition above is a set of volumes. DWP commissioned the explanation, and it is the most directly useful evidence in the corpus on why a correction exercise aimed at a very old cohort under-delivers. Verian interviewed 40 letter recipients over State Pension age who had not applied, in telephone interviews of about 30 minutes between 10 December 2024 and 27 January 2025, analysed against the COM-B behavioural framework ^DWP. Read the findings with the method in view: this is qualitative work on 40 people, all of them women, 34 in England, and it reports no proportions, no take-up rate and no arrears figures — the report’s own limits are that affected individuals cannot be precisely identified, that decades of elapsed time make recall and documentation unreliable, and that quantifying any intervention’s effect on take-up was out of scope (same source). Nothing below is a measured share of the cohort; the volumes stay with the progress release above.

The identification problem is the root cause, and it is irreparable. HRP should have attached automatically to Child Benefit claims, but HMRC did not always record a National Insurance number against claims made before 2000, and the Child Benefit records from the relevant period are no longer held — the research records them as deleted five years after a claim ends (same source). So HMRC cannot match its own records to find the affected people, and the exercise instead targets those with incomplete NI records who show no HRP, inviting them to self-identify. That design choice is what everything else follows from: a letter has to ask a woman in her seventies or eighties to reconstruct, unaided, which years she claimed Family Allowance or Child Benefit for a child under 16 between 1978 and 2010.

The report groups the barriers into trust and relevance/understanding, judging trust the most significant, with the digital-by-default route a third strand (same source). Within those:

• Scam suspicion was the most common reason for non-engagement — an unexpected letter from HMRC about money, which family members sometimes advised ignoring (same source). HMRC had anticipated this, running a “check a letter is genuine” webpage and media work, and the research found broadcast coverage did improve trust where people saw it (same source).
• The name defeats the letter. “Home Responsibilities Protection” was unfamiliar to almost all participants and read as insurance or a benefit, and the word “carer” was taken to mean caring for an adult rather than for one’s own children — so recipients concluded the letter was not about them (same source). Many also knew the payment as Family Allowance rather than Child Benefit, which is the term the letter uses.
• The unexplained ask aroused suspicion. The letter does not say why HMRC cannot determine eligibility itself, and participants found being asked to supply decades-old information bewildering and a reason to doubt the letter’s provenance (same source) — a direct consequence of the deleted-records problem above going unstated.
• The expected gain was unclear and the perceived risks were concrete. Most assumed they would gain little or nothing and wanted to know how much might be owed, while fearing the State Pension could go down, that other benefits (Pension Credit, Attendance Allowance, DLA) could be lost, that a tax bill would follow, or that a wrong detail would be treated as fraud (same source). Those fears are not all groundless: additional income and arrears do interact with Pension Credit and with income tax above the personal allowance (same source) — see Pension Credit and pension income — how the means test treats pensions. The research records a single participant who took advice from an advice agency, and the advice was not to apply, for fear of losing Pension Credit (same source).
• People self-assessed rather than using the checker, and the common wrong beliefs were that HRP is for carers of disabled adults, that it is means-tested, that divorce, widowhood or remarriage disqualifies them, or that an ex-partner’s data is needed (same source).
• The digital route lost people at the date questions. Some abandoned HMRC’s online eligibility checker because they could not recall the dates it asked for; the paper alternative was missed in the letter; and paper forms requesting Family Allowance numbers and dates were abandoned by participants who did not have them and did not know that level of detail might not be required (same source). Helpline experience — long waits, uncertain advisers, forms that did not arrive — compounded it. Most participants had little or no internet use and relied on family for digital tasks, and the research notes digital confidence did not track age, with some people in their eighties more confident than some in their sixties (same source).

A small DWP-run companion study (six interviews drawn from 58 Fraud and Error cases with inaccurate NI records) found the same pattern compounded by health, care needs and cognitive conditions, plus assumptions that the applicant holds identity documents and can reach estranged family members for information (same source). Those participants had each had at least one DWP phone call, and the report attributes the difference in their findings — including on trust — to that difference in customer journey (same source), which is itself the most actionable thing in the research: human contact moved the trust barrier that no letter wording reached.

DWP’s stated response is to abandon the campaign model. The research concludes that a one-off campaign is poorly suited to an issue requiring historic information, and that changing the letter’s wording alone is unlikely to raise take-up, since even in interviews extensive personalised explanation was needed; and it records the government moving to an ongoing approach — year-round communications prompting people to check their NI record, revised GOV.UK wording, material added to DWP’s annual uprating letters, and closer work with local authorities, primary care, social housing providers and advice charities (same source). The uprating-letter channel is the significant one, because it reaches the cohort annually and by post without requiring them to recognise a one-off envelope. Set against the forecast-checking gap above, the same conclusion recurs from a different direction: information supplied on request does not reach people who do not know they have a question.

The redress machinery behind a correction — DWP’s special payments

Arrears restore the entitlement. Anything beyond arrears — for the delay, the distress, or a cost incurred — comes from a separate and much less visible scheme, and because this article discusses two maladministration findings and a correction exercise running to nine figures, it is worth setting out what that scheme can actually do. DWP’s staff guide describes payments made ex gratia under HM Treasury’s Managing Public Money rather than under any statute, with no right of appeal against a refusal ^DWP. The guide recognises three categories: loss of statutory entitlement; actual financial loss or costs, which includes professional fees and a “delay element” calculated at HMRC’s simple tax repayment rate and expressly not interest; and consolatory payments for the non-financial injustice of poor service (same source). The test is restorative — putting the claimant where they would have been but for the maladministration — with financial loss required to be its direct result, decided on the balance of probabilities and with the claimant’s own contribution taken into account (same source).

The number that matters is the consolatory band: payments “usually range between £50 and £500” (same source). That is the only monetary figure in the guide, and it puts the two adjudications discussed below in proportion. The Independent Case Examiner’s £35 award in the 2017–18 qualifying-years case sits at or below the bottom of DWP’s own range; the PHSO’s indicated Level 4 remedy of £1,000–£2,950 a person sits at two to six times its ceiling. So the Ombudsman was not asking DWP to pay more of what it routinely pays, but to pay on a scale its standing scheme does not contemplate — for a cohort numbering in the millions. That is a structural reason the recommendation was resistible quite apart from the aggregate cost argument DWP made, and it is this article’s reading rather than a point either body states.

Two further features of the scheme are worth knowing. The escalation route is internal re-look, then the Independent Case Examiner, then the PHSO via an MP (same source) — the same ladder described below, with no tribunal stage, since a special payment is discretionary and not an entitlement that could be appealed. And the guide is silent on pensions: it nowhere mentions the State Pension, Pension Credit, Winter Fuel Payment or HRP (same source). The application of this machinery to the underpayment and HRP exercises above is therefore an inference from its generality, not something the guide states. A dating caution for citers: although the GOV.UK page carries a 2025 date, the substantive guide within it is dated April 2012 (same source), so the £50–£500 band has not been restated for inflation in anything the corpus holds — on its own terms it has lost roughly half its real value over the period the LEAP and HRP exercises ran. The delegated authorisation limits sit in a separate “guide for special payment officers” that is not captured. (Corpus gap: no later edition of DWP’s special-payments guidance, and no capture of the separate special payment officers’ guide carrying the delegated limits, so neither the current consolatory band nor who may authorise what can be sourced here.)

The scheme-side analogue, for occupational schemes rather than DWP, is the Pensions Ombudsman’s award for non-financial injustice — see The Pensions Ombudsman (TPO) — role, jurisdiction and powers and Pension overpayment recovery — recoupment, set-off and the 'competent court' question.

When the underpaid pensioner has died — the route, and what it does not promise

The LEAP figures above note in passing that some cases remain unpaid “because the customer has died and no estate has been identified”. DWP publishes the mechanism by which such an estate is meant to find it, and the mechanism deserves to be read closely, because it is weaker in three respects than the correction statistics imply.

The proactive phase is over. DWP says that “where we knew their details, we wrote to the next of kin or executor of those who may have been affected”, and that it “did this between 11 January 2021 and 31 December 2024” ^DWP. The tense is past and no continuing tracing effort is described. A person may now contact DWP only where both limbs apply: they are the next of kin or executor, and DWP has not already contacted them (same source). So for every estate whose details DWP did not hold — which is, by construction, the population behind the “no estate identified” cases — the route since the start of 2025 has been entirely reactive: the bereaved family must know the exercise exists, work out that it might apply, and ring up. Set against this article’s forecast-checking evidence and the HRP non-applicant research, where self-service routes reached a couple of thousand people a month and DWP’s own conclusion was that the residual depended on direct mailing, a reactive route aimed at executors of people who died years ago is the weakest version of the same design.

The page promises information, not arrears. It is framed throughout as a way to contact DWP about someone who may have been underpaid. It states no service standard, no response time, no description of what DWP does after the call, and — the point that matters for an executor deciding whether it is worth the effort — no statement that arrears will be paid to the estate at all (same source). It is also silent on probate, on inheritance tax, and on how any arrears would interact with Pension Credit (same source), though the LEAP releases above record that arrears can be reduced to £0 by an offsetting benefit overpayment. Nothing here is a reason to think the underlying entitlement is weaker than for a living pensioner; it is a reason to note that the published route does not say.

The channel is a telephone line and nothing else. The only route given for Great Britain is 0800 731 0469, Monday to Friday 8am to 6pm, with Welsh-language, Relay UK and BSL video-relay alternatives, and separate pension centres for Northern Ireland and for a person who died abroad; no form, postal address or online route appears (same source). What the caller must supply is the deceased’s full name, date of birth, date of death, last known address and postcode, and — if they were married or in a civil partnership — the spouse’s or civil partner’s full name; the National Insurance number is explicitly optional, “it will help us to deal with your enquiry quicker” (same source). That is a deliberately low documentary bar, and no evidence of executorship, grant of probate or proof of relationship is asked for on the page. Read with the HRP research’s finding that a phone call was the one intervention that moved a trust barrier no letter wording reached, a phone-only channel is a defensible choice; read against a cohort of executors who may be elderly themselves, it is a single point of failure with no asynchronous fallback.

The eligibility restatement carries one asymmetry worth noting. For someone who reached State Pension age before 6 April 2016, the page lists all three LEAP categories: married, where they received less than the lower basic State Pension (Category BL) and did not get an increase through their spouse or civil partner; widowed, where they did not inherit some of their spouse’s State Pension; and aged 80 or over, where they received some basic State Pension but less than the over-80 (Category D) rate (same source). For someone who reached State Pension age on or after 6 April 2016, only one group is listed — widowed, where they “did not inherit State Pension they were entitled to” (same source). No married or over-80 route appears on the new-State-Pension side of the line. That is consistent with the structure described above — the Category BL and Category D mechanisms are features of the pre-2016 system and have no new-State-Pension analogue — but the page does not explain it, and an executor of someone who reached State Pension age after April 2016 should not read the single listed category as the limit of what can be asked about.

No deadline is stated for contacting DWP, and none for the arrears (same source). That cuts both ways: there is no cut-off to miss, but equally no point at which the department treats the exercise as closed for estates it never found. The page also reproduces the full weekly rate history — the new State Pension from £155.65 in April 2016 to £230.25 in April 2025, and the full and lower basic rates back to September 1971 (£6.00 and £3.70) — which makes it, incidentally, the longest State Pension rate series in the corpus (same source). Note for citers: its tables stop at April 2025, so the current-year figures come from the rates publication ^DWP, not from here.

Women’s SPA communications (the PHSO finding)

The most contested chapter is the communication of the 1995-Act equalisation of women’s SPA. The Parliamentary and Health Service Ombudsman found maladministration in DWP’s communications — a roughly 28-month delay in starting direct mailing — but no direct financial loss, and would have recommended level-4 compensation. In its decision of 29 January 2026 the government accepted the delay and apologised but rejected the injustice finding and ruled out any financial compensation scheme for 1950s-born women, promising instead an action plan on State Pension communications ^DWP.

The parliamentary handling was itself extraordinary. The PHSO laid its March 2024 report before Parliament under s.10(3) of the Parliamentary Commissioner Act 1967 — only the eighth time in its history, a step it takes when it anticipates a department will refuse to comply ^Hans. The decision was then made twice: the December 2024 decision (Kendall) ruled out compensation, and after a 2007 DWP evaluation — showing the personalised forecast letters had had only “negligible impact” — surfaced in legal proceedings, the government re-took the decision in January 2026 (McFadden), reaching the same conclusion ^Hans. The intention to re-take it was announced to the Commons on 11 November 2025, in a statement that set out both the trigger and the limits of the exercise: the 2007 evaluation had been “cited in legal proceedings” and had not been before the previous Secretary of State because its relevance “was not evident at the time”; the Court had been informed; DWP had been asked whether further survey or other evidence existed; and — the caveat that framed everything that followed — the government said re-taking the decision “should not be taken as an indication” that it would award financial redress. No deadline was given, only that work would begin immediately ^Hans. Read against the January 2026 outcome, that caveat was doing real work: the re-take was a procedural correction of the evidence base, not a reopening of the remedy question.

Where it stood on 12 October 2026: litigation, not policy. Pressed twice at Commons oral questions on what the department is doing for 1950s-born women, the Minister for Pensions confirmed he had “recently written to WASPI representatives to set out the Government’s position”, and then closed the subject on a ground that is the most operationally significant fact in this strand: “With legal proceedings ongoing, further comment is not appropriate” ^Hans. So nearly two years after the first decision and nine months after the re-take, the remedy question sits with the courts rather than the department, and ministers are declining to add to the published response on that basis. Asked specifically when a promised meeting between the Prime Minister and campaigners would take place, the Minister gave no date (same source). Challenged that the previous government had accepted the Ombudsman’s recommendations, he said this government “have considered the case set out by the ombudsman and provided a detailed response” — the 25 June 2026 document set out below — and disputed that the previous government had ever committed to a compensation scheme (same source). He also restated the political starting point that frames the government’s sympathy-without-redress position: he and the Prime Minister “opposed the 2011 acceleration of the state pension age by the coalition Government”, particularly because it gave some women “as little as five years’ notice”, and those women “have faced disadvantage in the workplace and very uneven burdens of caring responsibilities” (same source). On what the department is doing, the answer named only measures available to pensioners generally: raising the State Pension, the NHS, and “the largest ever pension credit take-up campaign”, which he noted matters particularly for low-income pensioners, “a disproportionate number of whom are women” (same source) — see Pension Credit and pension income — how the means test treats pensions. Nothing in the exchange is a new redress commitment, and nothing captured identifies the proceedings or their timetable. The government distinguishes throughout between the maladministration it accepts and the injustice/remedy approach it rejects, and confirmed that the 2024 Labour manifesto — unlike pre-election statements — contained no compensation commitment ^HansHans. The next statutory SPA review is due by March 2029, and the minister has indicated the government “would not be bringing forward such short-notice changes” again ^HansHans.

The government published its full, combined response on 25 June 2026, setting out in detail why it will not pay compensation ^DWP. It rejects the PHSO’s injustice analysis as logically flawed for not weighing whether an earlier letter would in fact have been read and acted on: DWP’s own evidence (the 2007 evaluation of Automatic Pension Forecasts, Research Report 447) suggests only around 45% of 1950s-born women read such mailings, while by 2006 — the year the PHSO said direct letters should have begun — about 90% of women aged 45–54 were already aware their SPA was rising, so “for a majority of 1950s-born women, an earlier letter would not have influenced the choices they could or would have made” ^DWP. On remedy, the PHSO had indicated Level 4 compensation (£1,000–£2,950 each), which on a flat-rate basis across all 1950s-born women the PHSO costed at £3.5bn–£10.5bn (a figure worth holding against the cost of the thing campaigners are often assumed to be asking for: DWP costed reversing the 1995 and 2011 Act State Pension age changes — women back to 60, men to 65, over 2010/11 to 2025/26 — at £215.2 billion in 2018/19 prices, of which £181.4bn relates to women ^DWP. The two numbers answer different questions and are two orders of magnitude apart; a maladministration remedy for defective communication is not a reversal of the timetable, and the £215bn figure is not an argument against the £3.5–10.5bn one. The construction of the reversal costing, and why the widely-quoted £77bn is not comparable with it, is set out in State Pension age reviews — the statutory machinery, Cridland, Neville-Rolfe and the third review); the government concludes that any individually-assessed or self-certified scheme would be unworkable (an estimated ~1.75m claims, ~1,850 staff and ~£275m to run an assessed scheme, ~20 years on, with high fraud-and-error risk) and not a fair use of taxpayer money ^DWP.

There is now a measurement of how well that general communication has actually landed, and it is sobering for both sides of the argument. DWP’s Planning and Preparing for Later Life 2024 survey tested adults below State Pension age (aged 40–65) on five true/false statements about the new State Pension. Most got the basics right — 73% knew the amount varies with a person’s National Insurance record and 79% that qualifying years can be built while self-employed — but only 55% correctly rejected the statement that qualifying years can only be built in paid employment (28% got it wrong), 67% knew that 35 qualifying years are needed for the maximum, and just 46% knew there is no minimum number of qualifying years before anything is payable, with 37% saying they did not know ^DWP. So in 2024, after three decades of the campaigns catalogued below, more than half of people approaching State Pension age did not know the entitlement floor, and the rules most relevant to people with interrupted work histories — carers, the self-employed, those relying on credits — are the ones least understood. Two readings follow, and they cut in opposite directions. The government’s case that an earlier letter would not have changed much draws some support: general awareness of entitlement mechanics remains patchy even now, which suggests information provision is a weak lever. But it equally undercuts the claim that DWP’s communications record is adequate, since these are the facts its own campaigns were meant to convey. (The survey is of 40–75-year-olds in Great Britain and tests knowledge of the new State Pension, not of State Pension age equalisation, so it is corroborating evidence on the difficulty of pension communication rather than a direct measure of the PHSO’s question.)

Underpinning that argument is Annex A of the response, a dated chronology of what DWP did communicate between 1991 and 2013 — the department’s evidential answer to the charge that women were not told. It runs from the 1991 green paper Options for Equality in State Pension Age (4,000 consultation responses) and the 1993 white paper, through leaflets that carried a date-of-birth-to-new-SPA table from 1995, a pensions order-line and State Pension forecasts on request from the same year, to multi-channel public campaigns: the 1998 campaign generated 1.5 million leaflet requests, the 2001 campaign was a £6.5m creative effort producing 1 million website visits, 2 million guides and 430,000 helpline calls, and a web SPA checker existed from 2001 ^DWP. The three entries that bear directly on the PHSO’s finding are the automatic pension forecasts — approximately 17.851 million sent to individuals between 2003 and 2006 — the 1.2 million letters to women affected by the 1995 Act sent across 2009–2011, and the 5 million letters and leaflets on the 2011 Act changes in 2012–2013 (same source). The chronology is what makes the government’s position internally coherent: it concedes the 28-month delay in individual letters while pointing to two decades of general communication either side of it, which is why its case turns on whether an earlier letter would have been read and acted on rather than on whether information was available at all. It also explains the PHSO’s counter-position, since the forecasts were sent after 2003 and the direct mailing only from 2009 — the gap the Ombudsman identified sits precisely between the general campaigns and the individual letters.

The response also records the public sector equality duty analysis under s.149(1) of the Equality Act 2010 that the compensation decision had to satisfy, and its reasoning is unusual enough to be worth noting. For every protected characteristic DWP concluded there was no discrimination, on the ground that a decision to compensate or not “would apply to and affect the entire cohort equally”, and that people outside the affected cohort were “not considered [an] appropriate comparator” ^DWP. On equality of opportunity it acknowledged that older women within the cohort had fewer years in which to add qualifying years and so “may feel they should be entitled to a greater amount of compensation”, and that the cohort as a whole is more likely than the general population to be male, white, disabled, married and Christian; but it treated the sub-group that actually “lost opportunities” — those who were in a contracted-out scheme and could not buy extra qualifying years — as too poorly evidenced to characterise, recording for most characteristics simply that its “makeup … is difficult to assess due to lack of relevant data” (same source). The grid therefore turns the data gap into a reason not to differentiate, and closes the circle with the observation that if DWP rejects maladministration-caused injustice at all, “the premise would be that any ‘loss of opportunity’ was not a matter for which the Department was responsible” (same source). Marriage and civil partnership is the one characteristic marked “not relevant”. For practitioners the transferable point is procedural: where a remedy decision is taken cohort-wide, the PSED analysis tends to be discharged by the uniformity of the decision rather than by evidence about who was actually harmed.

That action plan was published on 25 June 2026. It is entirely forward-looking — it makes no commitment to compensation or any remedy scheme, and the word “compensation” does not appear — and is scoped to improving communications and complaints handling so similar issues are less likely to recur ^DWP. It accepts that maladministration between August 2005 and December 2007 caused a 28-month delay in starting individual letters to 1950s-born women, and sets six actions — a State Pension age communications strategy for 2027–2030, an ongoing awareness campaign for the ‘Check your State Pension age’ tool (paid elements running since Autumn 2025), formalised feedback loops, a stakeholder working group, learning for high-volume complaint campaigns, and wider DWP learning — all due for completion by April 2027, overseen two-monthly by a Director-led Retirement Services Senior Oversight Group accountable to the Permanent Secretary and Ministers ^DWP.

The plan’s own status column, published alongside each action, is the part worth tracking, because it shows how little of the plan was under way when it was laid. As at 25 June 2026, three of the six actions had not started: the feedback-loop work (action 3, due April 2027), the stakeholder working group (action 4, due January 2027 — the one internal deadline earlier than April 2027), and the “is this scalable to other DWP services” review (action 6, due April 2027). The communications strategy (action 1) was in progress with a first iteration expected November 2026; the awareness campaign was split into a paid strand running to April 2027 and a low- and no-cost strand due by September 2026; and action 5 — applying the PHSO learning to future high-volume complaint campaigns — carries no date at all, being “triggered by future high-volume complaint campaigns” ^DWP. Each action names a Director as lead and the Oversight Group as the reporting route, so there is a traceable owner; but on the published face of it the plan was at that date a set of commitments with half its content not yet begun, and two of its internal milestones (the September 2026 campaign phase and the November 2026 first strategy iteration) have now fallen due without a captured progress statement. (Corpus gap: no DWP progress report against the action plan’s status column is held in raw/, so whether the September and November 2026 milestones were met cannot be sourced here; the plan’s own governance points to the Retirement Services Senior Oversight Group and to correspondence with the Work and Pensions Committee as the likely publication routes.) The plan also notes that, from June 2025, DWP began writing to the SPA-67 cohort ahead of their State Pension age, and cites the Pension Schemes Act 2026’s Guided Retirement duty and pensions dashboards as future supports for retirement planning ^DWP. The plan’s publication was formally confirmed to Parliament on 1 July 2026, when the DWP Permanent Secretary wrote to the Work and Pensions Committee to confirm publication of the action plan on State Pension age communications and complaints handling — a step consistent with the plan’s own governance, under which the Director-led Senior Oversight Group is accountable to the Permanent Secretary and Ministers ^WPC.

Before the Ombudsman: what the Independent Case Examiner found in 2017–18

The PHSO was not the first independent body to examine this complaint, and the earlier verdict went the other way. DWP’s internal complaints route ends with the Independent Case Examiner (ICE), a free independent review service that sits “at the end of the whole complaint process” for DWP and its contracted providers, and whose own decisions can be escalated further only by asking a Member of Parliament to refer the matter to the PHSO ^DWP. It is the State Pension counterpart to The Pensions Ombudsman (TPO) — role, jurisdiction and powers, which has no jurisdiction over DWP’s administration of the State Pension — a division of labour worth keeping in view when mapping Who regulates UK pensions — the bodies and how they fit together.

By 31 March 2018, 2,979 women — “most but not all part of the Women Against State Pension Inequality (WASPI) campaign” — had complained to the ICE that DWP “were maladministrative in failing to communicate changes in State Pension retirement age arising from the Pensions Acts of 1995, 2007 and 2011”, saying this had denied them the chance to make informed choices about their future pension provision (same source). Sixty-eight of those cases had been completed by year end, and the Independent Case Examiner’s statement of outcome was unequivocal: “I have not upheld any of the cases on that main complaint although some other matters such as issues with complaint handling and provision of information have been upheld” (same source).

The reasoning matters more than the count, because it is the proposition the PHSO later had to displace. In the published pensions case study on the point, the ICE did not uphold a complaint that the Pension Service had failed to give timely, appropriate information about SPa changes, on the ground that DWP “had no commitment to communicate changes to State Pension age to all those individuals affected by a change”, while accurate information — including a personalised State Pension forecast — was available on request (same source). The same case study shows where the ICE did find fault. On a parallel complaint about the rise in the qualifying years needed for a full State Pension, it again held there was “no automatic requirement to proactively communicate” the change, but upheld the element where DWP failed to give that information in answer to a direct enquiry, recommending an apology and a £35 consolatory payment (same source). The line the ICE drew was therefore between a general duty to inform, which it held did not exist, and accuracy when asked, which it enforced.

Read against the PHSO finding, the two conclusions are not strictly contradictory, and the distinction is worth stating precisely because it is easily elided: the ICE answered whether DWP was obliged to write to every affected woman, and said it was not; the PHSO answered whether DWP, having decided to write, took too long to start, and found a 28-month delay. (That reconciliation is this article’s reading, not a position either body states.) It does, however, explain why a grievance already running at scale in 2018 took a further six years to produce a maladministration finding, and it is the reason a practitioner should not treat “the Ombudsman found maladministration” as the whole adjudicative history — the first independent reviewer, on a narrower question, did not.

The report also records what a high-volume complaint campaign does to the body receiving it, which is the institutional problem action 5 of the action plan above is addressed to. The pensions strand accounted for 3,161 of the 5,845 referrals the ICE received across all seven DWP business areas that year — 54% of its entire intake — of which 1,810 were accepted, and the ICE attributed the increase directly to the WASPI campaign, noting that the volume and the complexity of the cases had lengthened both the time to bring complaints into investigation and the time to decide them (same source). Of the 115 pensions investigation reports it issued, 11 (10%) were fully upheld and 20 (17%) partly upheld, against 84 (73%) not upheld — the lowest uphold rate of the seven areas by a wide margin, where Child Support Agency cases ran at 30% fully upheld and Child Maintenance Service at 27% fully and 62% partly (same source). That gap is best read as a composition effect rather than a measure of DWP’s pensions administration: the strand was dominated that year by a single campaign complaint the ICE had decided it would not uphold.

Two cautions for anyone citing these figures. The report does not say how many of the 3,161 pensions cases were WASPI complaints, nor whether the 68 completed cases sit within the 134 pensions clearances, so the two counts must not be combined (same source). And the seven business-area tables sum to 2,782 accepted cases against a headline of 2,779, a three-case discrepancy the report does not explain (same source).

Related

The State Pension frames private provision — its contracting-out legacy underlies GMP equalisation, and its level and age shape retirement-income planning alongside pensions tax.

The single-tier reform also ended salary-related contracting-out on 6 April 2016 — see The 2016 abolition of contracting-out for the consequential SIs (CEP savings, scheme-modification power and GMP fixed-rate revaluation) that schemes still rely on for legacy contracted-out rights.

The Pension Credit take-up gap and Winter Fuel Payment linkage are examined in political and distributional context by the WPC pensioner-poverty inquiry 2025 — challenges, mitigations and the Pension Credit gap.

On divorce or dissolution the State Pension can itself be a shareable asset — but only the additional State Pension (pre-6 April 2016 proceedings) or the transitional Protected Payment (post-reform), never the core new State Pension — via the pension-sharing regime described in Pensions on divorce — pension sharing, pension credits/debits and the state-pension dimension.

The statutory review machinery behind the rising State Pension age — the Pensions Act 2014 six-yearly duty, the Government Actuary’s report and the appointed independent report that must inform each review, the Cridland, Neville-Rolfe and third-review recommendations (none of which has yet changed the legislated timetable), and the Work and Pensions Committee’s scrutiny of it — is set out in full in State Pension age reviews — the statutory machinery, Cridland, Neville-Rolfe and the third review.

The State Pension is not funded — it is paid out of the Great Britain National Insurance Fund, and the annual Government Actuary’s report to Parliament on each draft Up-rating Order is where the cost of the triple lock and the ageing population is quantified: see The National Insurance Fund — the pay-as-you-go plumbing behind the State Pension for the Fund’s mechanics, the projections to 2030-31 and the Treasury Grant test.