Key takeaways
- From 6 April 2029, salary sacrificed into a pension above £2,000 a year carries employee and employer National Insurance. At a £50,000 salary sacrificing 15%, that's £440 for you and £825 for your employer.
- The employee charge peaks at a salary of £52,270, where it hits £467 a year. At £120,000 sacrificing the same 15% it's £320, because the rate drops from 8% to 2% above £50,270.
- HMRC counts 7.7 million people sacrificing salary into a pension, of whom 3.3 million sacrifice more than £2,000. The average extra employee bill is put at £84 in 2029 to 2030.
- The employer pays more than the employee wherever a charge arises at all, in eleven of the twelve cases below. The secondary rate is a flat 15% with no ceiling, so at £120,000 and 15% it's £2,400.
- The OBR puts the static yield at £4.9 billion in 2029-30 but the final costing at £4.7 billion, with behaviour removing around 48% of the static yield by 2030-31.
What the salary sacrifice cap costs in 2029, at four salaries
You pay into your pension by giving up salary, and you want to know what the new limit does to that. Here's the short answer. The first £2,000 you sacrifice in a tax year is untouched. Everything above it is treated as earnings for National Insurance, yours and your employer's, from 6 April 2029.
HMRC's policy paper of 4 December 2025 states the operative date flatly: "The measure will take effect from 6 April 2029." The thresholds that decide the bill are already announced. HMRC's Budget 2025 note says the primary threshold stays aligned with a £12,570 personal allowance, and that the "UEL and UPL will remain aligned to the higher rate threshold at £50,270 until the 2030 to 2031 tax year". The rates are the ones GOV.UK publishes for 2026 to 2027: 8% employee National Insurance between those two figures, 2% above, and a flat 15% for the employer on everything above a £5,000 secondary threshold.
Run that across four salaries and three contribution rates and the grid looks like this.
| Salary | Sacrifice | Above £2,000 | Your NI | Employer NI | Combined |
|---|---|---|---|---|---|
| £30,000 | 5% (£1,500) | £0 | £0 | £0 | £0 |
| £30,000 | 10% (£3,000) | £1,000 | £80 | £150 | £230 |
| £30,000 | 15% (£4,500) | £2,500 | £200 | £375 | £575 |
| £50,000 | 5% (£2,500) | £500 | £40 | £75 | £115 |
| £50,000 | 10% (£5,000) | £3,000 | £240 | £450 | £690 |
| £50,000 | 15% (£7,500) | £5,500 | £440 | £825 | £1,265 |
| £80,000 | 5% (£4,000) | £2,000 | £40 | £300 | £340 |
| £80,000 | 10% (£8,000) | £6,000 | £120 | £900 | £1,020 |
| £80,000 | 15% (£12,000) | £10,000 | £200 | £1,500 | £1,700 |
| £120,000 | 5% (£6,000) | £4,000 | £80 | £600 | £680 |
| £120,000 | 10% (£12,000) | £10,000 | £200 | £1,500 | £1,700 |
| £120,000 | 15% (£18,000) | £16,000 | £320 | £2,400 | £2,720 |
Two things fall out of that table. Your own bill doesn't climb with your salary, and your employer's does. At £50,000 sacrificing 15% you're charged £440. At £120,000 sacrificing the same 15% you're charged £320, on more than twice the pay.
Why your own bill peaks at a salary of £52,270
The reason is the upper earnings limit. Employee National Insurance runs at 8% on earnings between £12,570 and £50,270, then falls to 2% on everything above. Sacrificing salary cuts your cash pay, and the sacrificed amount above £2,000 is then added back as notional earnings. So whenever your sacrifice runs past £2,000, your National Insurance earnings come to your salary minus £2,000.
So the slice that gets charged sits immediately below your salary. On a modest salary that slice is inside the 8% band. On a large one it's entirely above the upper earnings limit, where the rate is a quarter as high.
The turning point is exact. At a salary of £52,270, a 15% sacrifice is £7,840.50, of which £5,840.50 is above the limit, and cash pay of £44,429.50 plus that £5,840.50 lands precisely on £50,270. Every pound of it is charged at 8%, which is £467. That's the largest employee charge available anywhere on the 15% line.
Above it the charge falls, and keeps falling until cash pay alone clears the upper earnings limit. That happens at a salary of about £59,141, where the whole £6,871 excess is charged at 2% and the bill is £137. The chart plots the shape: up to £467 at £52,270, down to £140 at £60,000, then a slow climb back to £320 at £120,000. A reader on £60,000 pays less under the salary sacrifice £2,000 cap than a reader on £40,000 doing the same thing.
The OBR's own costing says the same thing from the other end. Its modelling applies an "increase in average employee NICs rate" of 2.7% to salary-sacrificed contributions and 2.0% to bonus-sacrificed ones, on the stated assumption "that all individuals making pension contributions using bonus sacrifice earn more than the upper earnings limit". A blended 2.7% is much closer to 2% than to 8%. Most of the money this measure touches sits above the limit, where the employee rate is small.
The employer's share is the larger number wherever a charge arises
Employer National Insurance has no upper limit. It's 15% above a secondary threshold of £5,000 a year, all the way up. So the employer side of the table rises in a straight line with the amount sacrificed: £150 on a £1,000 excess, £825 on £5,500, £2,400 on £16,000.
That matters because employer National Insurance is the larger half of the charge in all eleven rows where anything is charged. On a £120,000 salary sacrificing 15%, the combined cost is £2,720 and £2,400 of it is the employer's. HMRC's own guidance confirms the split survives the change: "All employer pension contributions will continue to be free of NICs", so only the sacrificed slice is caught. Whether an employer passes its 15% through to the pension, keeps it, or recovers it some other way is a scheme-by-scheme matter, and it's the same question that already decides who benefits from the National Insurance saving today.
HMRC expects this to land on 290,000 employers, at a one-off cost of £20 million and a continuing £30 million a year, with £1.9 million of HMRC IT work behind it.
£2,000 covers a minimum workplace pension until about £46,240 of salary
The salary sacrifice pension cap sounds small, and on auto-enrolment minimums it isn't. The statutory minimum has been 8% of qualifying earnings in total since April 2019, of which 5% is the employee's share. It's charged on earnings between £6,240 and £50,270 a year. Five percent of qualifying earnings reaches £2,000 when those earnings reach £40,000, which is a salary of £46,240. Those are the figures most people end up on, because auto enrolment minimum contributions are where around 3 in 10 private sector savers stay.
On a flat 5% of full salary, the crossing point is £40,000. Either way, somebody contributing the statutory minimum and nothing more stays under the limit until their salary passes £46,240. HMRC's own split points the same way: of 7.7 million people using salary sacrifice for pensions, 3.3 million sacrifice more than £2,000, so "44% of employees using salary sacrifice for pensions would be impacted by this measure, while 56% around 4.3 million people are fully protected". For those above the line, "the average additional employee National Insurance contributions liability is estimated to be £84 in the first year of impact".
£84 is a long way below the £440 in the table, and the gap is the point. The average is dragged down by people a few hundred pounds over the limit and by the 2% band at the top. The table shows what a deliberate 10% or 15% sacrifice costs, not what the median affected person pays.
Income tax relief doesn't move, and neither does adjusted net income
Nothing here touches income tax. HM Treasury's guidance of 26 November 2025 is explicit that contributions "will still be exempt from Income Tax (subject to the usual limits)". A 40% taxpayer sacrificing £12,000 still gets £12,000 into the pension out of pre-tax pay. The relief at the margin is unchanged, and so is the difference between relief at source and net pay for anyone contributing outside a sacrifice arrangement.
The second-order effect is the one people miss. Salary sacrifice reduces adjusted net income, which is what the personal allowance taper and Tax-Free Childcare are tested against. HMRC addresses it directly: "This measure will not change the impact of salary sacrifice on adjusted net income." The personal allowance still falls by £1 for every £2 of adjusted net income above £100,000. So the £120,000 earner sacrificing £18,000 still brings adjusted net income to £102,000, and still keeps whatever that is worth. The £320 of National Insurance is bolted on beside it, not instead of it.
The law fixes £2,000 once, then hands the rest to regulations
The 2029 salary sacrifice cap is not a proposal. The National Insurance Contributions (Employer Pensions Contributions) Act 2026 received Royal Assent on 29 April 2026. Section 1(3) says "The amendments made by this section have effect for the tax year 2029-30 and subsequent tax years", and section 1(4) requires that the first regulations "must specify the contributions limit as £2,000 for a tax year".
Read that twice, because the number is in the regulations rather than in the Act. The Act says the limit "is the amount specified by regulations", and it only pins £2,000 to the first set. There's no uprating duty. Parliament gave itself a hard look at one direction of travel and not the other. The Northern Ireland section spells out the test: an instrument containing regulations "making provision for a reduction in the amount of the contributions limit for a tax year shall not be made unless a draft of the instrument has been laid before and approved by resolution of each House of Parliament". Section 1(2) adds the same class of regulations to the Parliamentary control list in section 176 of the 1992 Act. Raising the limit, or leaving it at £2,000 while wages grow, is not on that list.
The mechanics are still blank. HMRC says "Details on the design and operation of the £2,000 contribution limit will be set out in secondary legislation in due course, following stakeholder engagement." The Act anticipates a per-pay-period version of the limit and allows the prescribed figure to exceed "by not more than £1" the arithmetical equivalent, which for a month is £166.67. Nobody has published the monthly number, how a mid-year job change splits the limit, or where a sacrificed bonus lands in the pay period.
The strongest objection: almost nobody pays the static number
Take the table as a static calculation and it overstates what this raises, which is also the best argument that it overstates what readers lose. The OBR's supplementary release of 5 February 2026 separates the two. The static yield is £4.9 billion in 2029-30 and £5.1 billion in 2030-31. The final costing is £4.7 billion and £2.6 billion, "with the behavioural impact reducing the static yield by around 48 per cent in 2030-31". Strip the temporary timing effects and the steady-state figure is "£3.2 billion in each of the final two years".
What closes that gap is people not paying it. The costing assumes "40 per cent of salary-sacrifice contributions above the cap and 20 per cent of bonus-sacrifice contributions above the cap remain in salary-sacrifice schemes, and the remaining 60 and 80 per cent move to net pay arrangements or RAS schemes". It assumes employers divert a share of future pay growth into ordinary employer contributions, which stay exempt, rising to 5% of pay growth by 2029-30. It assumes 80% of affected members in defined contribution schemes trim their own contributions, by 1.8% for higher earners and 2.75% for lower ones. It even prices a 5% reduction from employees smoothing contributions between years to make efficient use of the exemption.
So the honest reading of the table is as an upper bound on a static arrangement, not a forecast of what a given household ends up paying. The OBR rates the whole costing "medium-high" for uncertainty, with "a high level of uncertainty over the size of the behavioural response".
The counter-objection to the counter-objection is that most of those escape routes cost something too. Moving to a net pay arrangement loses the employer's 15% as well as your 8% or 2%. Moving to a relief at source scheme means higher-rate and additional-rate relief has to be reclaimed, and the OBR assumes 10% of it goes unclaimed even in this engaged population. The arrangement that avoids the charge is not the arrangement that was there before.
What these numbers can't tell you
The thresholds in the table are announced for 2029 to 2030. The rates are not announced at all, and even the thresholds aren't law yet: HMRC's note says plainly that "National Insurance contributions thresholds for future years have not yet been legislated" and are set in the annual exercise. The 8%, 2% and 15% used here are the 2026 to 2027 rates. A Budget on 28 October 2026 falls between now and April 2029, and it is the only one with a date on it.
The treatment of the excess is an assumption too, and a load-bearing one. The Act treats the amount forgone as "remuneration derived from the earner's employment", so this piece aggregates it with cash pay and charges it at the band it reaches. Regulations may instead "provide, in prescribed cases, for the amount treated as remuneration to be an amount other than the amount foregone". If the £2,000 is apportioned monthly, a single large sacrificed bonus could meet a different band mix than this annual calculation shows.
The two official figures for 2029 to 2030 also disagree. HMRC's policy paper scores the measure at £4,845 million; the OBR release dated 5 February 2026 gives £4.7 billion. That's a gap of roughly £145 million, and neither document explains it. For 2030 to 2031 they agree, at £2,585 million against £2.6 billion.
Finally, the behavioural numbers rest on survey data, not administrative records. The OBR says the main source is "the 2024 ONS Annual Survey of Hours and Earnings", that "HMRC does not hold administrative data on the use of salary-sacrifice schemes for pension contributions", and that bonus sacrifice is estimated from historic returns "where reporting is no longer required". A sample is not a census. The equalities note is a sample too: employees aged 31 to 50 are 52% of the affected group against 44% of employees generally, and males 59% against 50% of UK adults.
What would change the conclusion
If the upper earnings limit moved. The whole shape of the chart hangs on £50,270 and the 8% step down to 2%. That limit is announced as frozen through 2030 to 2031, which is also what makes the income tax threshold freeze bite. Unfreeze it and the £52,270 peak moves with it.
If the £2,000 were uprated, or weren't. Left in cash terms while wages grow, the limit catches more people every year, and the salary sacrifice contribution limit for 2029 becomes a smaller real allowance each April after that. Nothing in the Act requires it to move, and only a cut needs a vote.
If the start date changed. The salary sacrifice cap starts in April 2029 because section 1(3) says so, and a different date needs new primary legislation. The Chancellor's letter to the Treasury Committee of 31 July 2026 asked the Office for Budget Responsibility "to prepare an economic and fiscal forecast for publication on 28 October 2026", and added that "This will be accompanied by the Budget." That is the next scheduled moment anything could be reopened. Nothing published says it would be.
The figure worth watching isn't the £2,000. It's the gap between your salary and £52,270, because that single number decides whether your own share of this is charged at 8% or at 2%. LedgerTouch tracks contributions against the annual allowance; the band you're in is on your payslip. The cap was restated rather than changed on Budget day, and it sits alongside seven other Autumn Budget 2026 pensions measures with start dates running from Budget day to the next Parliament.