Key takeaways
- On the rules in force for 2026/27, putting £5,000 into a pension costs a basic-rate payer £4,000, a 40% payer £3,000 and a 45% payer £2,750.
- A flat 30% rate would price that same £5,000 contribution at £3,500 for everyone: £500 cheaper for a basic-rate payer, £500 dearer at 40% and £750 dearer at 45%.
- HMRC estimates 57% of Income Tax relief on pension contributions was given at the higher rate in 2024 to 2025 and 14% at the additional rate, against 29% at the basic rate.
- Gross pension Income Tax and NICs relief came to £83.9 billion in 2024 to 2025. Net of tax on pensions in payment, HMRC puts the cost at £53.8 billion.
- One pension change is already announced: from 6 April 2029, salary sacrifice above £2,000 a year attracts Class 1 NICs, scored at £4,845 million in 2029 to 2030.
The same £5,000 contribution costs three people three different amounts
You want to know whether the Budget is about to make your pension contributions more expensive, and by how much. Here's the honest state of it. Nothing about the rate of relief has changed, and no official document proposes changing it. A pension tax relief change at a Budget arrives with a commencement date attached, as the salary sacrifice measure did. Nothing of that kind exists for this one. What has a date on it is the Budget itself: the Chancellor's letter to the Treasury Select Committee, published on 31 July 2026, says "The Budget will be held on 28 October 2026".
So the useful thing isn't a prediction. It's knowing what the options on the table would actually cost you, and that can be worked out from rules that already exist.
Start with today. GOV.UK puts it plainly: "You can get tax relief on private pension contributions worth up to 100% of your annual earnings". Relief comes at your marginal rate, so the same gross contribution has a different price depending on which band it comes out of. For 2026/27 the Personal Allowance is £12,570, the basic rate of 20% runs to £50,270, the higher rate of 40% runs to £125,140, and 45% applies above that.
Put £5,000 into a pension and here is what leaves your pocket:
- Basic-rate payer, 20% relief: £4,000.
- Higher-rate payer, 40% relief: £3,000.
- Additional-rate payer, 45% relief: £2,750.
That £1,250 gap between the top and bottom of the table is the whole argument. It isn't a loophole. It's the direct consequence of relieving a contribution at whatever rate the money would otherwise have been taxed at, which is how pension tax relief has worked for decades.
Pension tax relief and the Budget: what is confirmed and what is not
Two things are confirmed. The Budget date, and one pension measure that is already announced and already costed.
That measure is salary sacrifice. HMRC's policy paper, published 4 December 2025, "removes the OpRA excluded exemption for employer pension contributions for Class 1 National Insurance contributions only, where salary sacrifice arrangements exceed the £2,000 contribution limit". It takes effect from 6 April 2029. HMRC scores it at £4,845 million in 2029 to 2030 and £2,585 million the year after. On HMRC's own count, 7.7 million employees use salary sacrifice for pension contributions, 3.3 million of them sacrifice more than £2,000, and 4.3 million are fully protected by the threshold. What the salary sacrifice cap 2029 actually costs the 3.3 million above the line turns on their salary as much as on the size of the sacrifice.
Everything else is commentary on pension tax relief and the Autumn Budget, and it should be read as commentary. BDO's Autumn Budget 2026 predictions page, updated 8 September 2026, describes pension tax relief as "the most-discussed lever that has never quite been pulled". Pensions Age reported on 9 September 2026 that AJ Bell's analysis of FCA data found speculation ahead of the 2024 Budget prompted "around £10bn of excess withdrawals from retirement accounts". Neither is a policy document, and neither is a figure you can plan around. The same line between policy and commentary runs through the pension tax free lump sum, where the £268,275 allowance is unchanged and only pots above four times any cap would be reached.
Government has looked at this before. In July 2015, HM Treasury's consultation Strengthening the incentive to save (Cm 9102) asked whether relief should be restructured, and floated a system where "pension contributions are taxed upfront (a 'Taxed-Exempt-Exempt' system like ISAs), and then topped up by the government". The summary of responses, published in March 2016, records that submissions included "models featuring a 'flat rate of relief', where rates of relief currently based on marginal rates of income tax are replaced by a consistent flat rate incentive for all". Nothing came of it. On the question of sustainability, respondents agreed consensus mattered but "did recognise that total consensus was unlikely to be achieved".
What a flat rate of 20%, 25% or 30% would do to that £5,000
A flat rate replaces your marginal rate with one number for everybody. The arithmetic is simple enough to do on the back of an envelope, and the chart above plots it: the net cost of a £5,000 gross contribution at each rate of relief, in pounds.
- Flat 20%: £4,000 for everyone. No change for a basic-rate payer, £1,000 more for a 40% payer, £1,250 more at 45%.
- Flat 25%: £3,750 for everyone. £250 cheaper at the basic rate, £750 dearer at 40%, £1,000 dearer at 45%.
- Flat 30%: £3,500 for everyone. £500 cheaper at the basic rate, £500 dearer at 40%, £750 dearer at 45%.
Higher rate tax relief on pension contributions is the only thing a flat rate below 40% takes away. Three things fall out of that table. The first is that a basic-rate payer is untouched by a flat rate of 20% and gains modestly above it. The second is that the loss to a higher-rate payer grows the further the flat rate sits below 40%. The third is the one that gets least attention: at any flat rate above 20%, a non-taxpayer gains something where today they may get nothing at all under a net pay scheme.
These are one year's figures on one contribution. Repeat the contribution for thirty years and the gap compounds, which is where the modelling below comes in.
Where pension tax relief goes now, in HMRC's own distribution
The distributional case for a flat rate rests on a single statistic, and it's worth getting it exactly right. HMRC's Private pension statistics, updated 30 July 2026, state that "in 2024 to 2025 it is estimated that 14% of Income Tax relief on total pension contributions was relieved at the additional rate of Income Tax, 57% at the higher rate, and 29% at the basic rate". Add the top two and 71% of Income Tax relief went to people paying above the basic rate.
That concentration isn't new. The 2015 Treasury consultation said "Over two thirds of pensions tax relief currently goes to higher and additional rate taxpayers," and put the gross cost of income tax and National Insurance relief at "nearly £50 billion in 2013-14".
The totals have grown a long way since. HMRC now estimates gross pension Income Tax and NICs relief at £83.9 billion in 2024 to 2025, against £79 billion the year before. The net figure, after subtracting Income Tax paid on pensions in payment, is £53.8 billion. Looked at relief by relief, HMRC's tax relief statistics of 22 January 2026 put Income Tax relief on registered pension schemes at £32,300 million in 2024 to 2025 and NICs relief at £23,100 million. For scale, HMRC counts 103 non-structural reliefs costing £218 billion in that year, so pensions are a large share of a very large number.
Two more numbers put the revenue question in proportion. HMRC's ready reckoner, updated 24 June 2025, estimates that changing the basic rate of Income Tax by 1p moves £6,900 million in 2026 to 2027, and changing the higher rate by 1p moves £1,600 million. The OBR, in its income tax page last updated 5 February 2026, says "In 2025-26 we estimate that income tax will raise £329 billion".
What none of those figures is, is the yield from abolishing anything. The Treasury's 2015 footnote on those relief costs is explicit that the estimates "do not represent the yield from withdrawing tax relief as there would be significant changes in taxpayers' behaviour". A headline relief cost is not a pot of money sitting on a shelf.
What flat-rate relief did to modelled savers
The most detailed public modelling of this is the Pensions Policy Institute's July 2021 write-up, Exploring the financial impact of flat rate pensions tax relief, sponsored by the PLSA. It ran four flat rates, 20%, 25%, 30% and 33%, across defined contribution and defined benefit vignettes. Its findings follow the band arithmetic closely.
For a non-taxpayer in a net pay scheme, who gets no relief today because there's no tax to relieve, "A flat rate of relief of 30% would add 40% to their pension saving". For a typical earner on the basic rate, "The increase in tax relief to DC pension savings of a flat rate of 30% increases DC savings by 14%". That improvement shrinks on the way out, because a bigger pot means a bigger tax bill in retirement and the State Pension dilutes the percentage.
For a higher-rate payer in a defined benefit scheme, the PPI found the member "could find their lifetime income reduced by around 4.6%. That is equivalent to a cost of around £2,000 a year". On its modelled scheme for a member earning £60,000, the extra annual tax liability is "around £1,800", which the PPI notes is "around 9% of the total contribution to the pension scheme".
The PPI also flagged the plumbing problem, and it's the part most summaries skip. Flat-rate relief on defined contribution schemes "would necessitate making employer pension contributions a taxable benefit otherwise higher rate taxpayers would be able to effectively claim a higher rate of tax relief through salary sacrifice". Defined benefit schemes are harder again: a flat rate above a member's marginal rate generates a rebate they may have to claim, and members who don't claim don't get it.
Scotland's bands make the same flat rate land differently
None of this is a single UK answer, because income tax rates aren't a single UK number. GOV.UK sets out the extra relief a Scottish taxpayer claims through Self Assessment: "1% up to the amount of any income you have paid 21% tax on; 22% up to the amount of any income you have paid 42% tax on; 25% up to the amount of any income you have paid 45% tax on; and 28% up to the amount of any income you have paid 48% tax on".
So a Scottish higher-rate payer on 42% pays £2,900 for that £5,000 contribution today, not £3,000. Against a flat 30%, they'd be £600 worse off rather than £500. At the 48% top rate the gap is wider still. Any flat rate set with the English and Welsh bands in mind produces different winners and losers north of the border, and the 21% band creates a group who gain from a rate that most basic-rate payers barely notice.
The strongest objection isn't about fairness, it's about coherence
The serious criticism of flat-rate relief doesn't dispute the distribution tables. It disputes the design. Relief on the way in is only half of the system; pension income on the way out is taxed at your marginal rate. Fix the first at 30% and leave the second alone, and a 40% payer who stays a 40% payer in retirement gets 30% going in and pays 40% coming out. A system meant to tax pension income once ends up taxing part of it twice.
The PPI describes the system in force as one where savings are “Taxed as income when money is taken from the pension savings”. That second half is what a flat rate leaves alone, and it's why the two halves are hard to separate: a flat rate on contributions implies a matching question about ISA vs pension treatment of withdrawals, including the tax-free lump sum, and about how the pension annual allowance of £60,000 interacts with a rate that no longer tracks anyone's tax band.
There's a practical counter to the counter-argument. The current system already breaks its own logic in a different place: a higher-rate payer in a relief-at-source scheme gets the extra 20% only by claiming it. A flat rate delivered at source would reach people the present one misses. Both objections are real, and the 2016 summary of responses records flat-rate models among the submissions with no agreement on any of them.
What these numbers cannot tell you
Quite a lot, and the gaps matter more than usual here because the subject invites guesswork.
The PPI modelling is a July 2021 exercise on a set of vignettes, not a forecast and not a population estimate. Its tax thresholds, contribution rates and assumptions are the ones that applied then, and its 4.6% lifetime-income figure belongs to one modelled defined benefit member, not to higher-rate payers generally. The PPI also excludes National Insurance relief from scope entirely, which on HMRC's numbers is a £23,100 million piece of the picture.
HMRC's ready reckoner is a static costing. It models the direct effect of a rate change and says so; it is not a behavioural forecast, and pension contributions are exactly the sort of thing people change in response to tax.
And the biggest limit is the obvious one. No official document sets out a flat rate of pension tax relief for this Budget. The £3,500, £3,750 and £4,000 figures above are the arithmetic of rates that have been discussed, applied to rules that exist today. They are a way of reading a measure quickly if one appears, not evidence that one is coming.
What would change the conclusion
A costing document. If Budget 2026 touches pension contributions, the policy paper and the scorecard carry the threshold, the operative date and the Exchequer impact, in the same shape as the salary sacrifice paper did in December 2025. Those three fields answer more than a month of speculation.
A change to the taxation of pension income. Flat-rate relief without a matching change on withdrawals leaves the double-taxation objection intact. One that moves both halves together is a different policy, and the £5,000 arithmetic above would need redoing from the other end.
A shift in where the relief goes. The 57% and 14% shares are the foundation of the distributional argument, and HMRC refreshes them with each annual Private pension statistics release. The salary sacrifice measure won't move them directly, because HMRC states that "The Income Tax relief on employee and employer pension contributions and National Insurance contributions relief on traditional (non-salary sacrifice) employer pension contributions remain unchanged". If the concentration at the top falls on its own, the case for forcing it falls with it.
Until then the position is stable, and the only number that changes anything for you is your own marginal rate. If you don't know which band your next contribution comes out of, that's the thing worth looking up, because every figure in this piece turns on it. LedgerTouch tracks pension contributions alongside the rest of a portfolio, which makes the band question easier to answer than a payslip does. The Budget has since been delivered, and the confirmed Budget 2026 pension changes left the rate of relief exactly where this piece found it.