Key takeaways
- In the third quarter of the 2025 to 2026 financial year, 11.8% of people who had just been enrolled opted out within a month. In the first quarter of 2014 to 2015 it was 6.2%.
- Participation among eligible employees went from 55% in 2012 to 90% in 2025, or 22.6 million savers. In the private sector it moved 46 percentage points, from 42% to 88%.
- Opting out rises with age rather than falling: 15.1% of those aged 60 to State Pension age left within a month, against 11.5% of 22 to 29 year olds.
- Around 3 in 10 private sector defined contribution savers, 4.3 million people, contribute exactly the 8% legal minimum. Around 7.5 million save 8% of total pay or less.
- Because the 8% applies only to earnings between £6,240 and £50,270 in the 2026 to 2027 tax year, a £20,000 earner's minimum is £1,101 a year, which is 5.5% of their actual pay.
Almost nobody leaves once they have been put in
If your employer enrolled you in a workplace pension without asking, you can get out. Almost nobody does. In the third quarter of the 2025 to 2026 financial year, 11.8% of people who had just started saving opted out within a month of being enrolled. That figure comes from HMRC payroll records, published by the Department for Work and Pensions in July 2026.
It has roughly doubled since the 6.2% recorded in the first quarter of 2014 to 2015, and it's still a small minority. Here's why that one number carries so much weight. In 2012, before the duty started, 55% of eligible employees in Great Britain saved into a workplace pension. By 2025 it was 90%, or 22.6 million people. Nothing about the economics of saving changed in between. What changed was which way the paperwork runs.
The second Pensions Commission's evidence pack, published in May 2026, puts it in exactly those terms: automatic enrolment "has hugely increased pension saving, through the power of defaults". So the interesting question isn't whether defaults work on participation. They plainly do. It's what they do to the size of the contribution, which is the rest of this piece.
What the auto enrolment opt out rate actually measures
"Opting out" has a narrow meaning in these statistics, and it's worth knowing before you compare the figure to anything. The Pensions Commission glossary defines the opt-out rate as "the proportion of new enrolments in workplace pensions who choose to stop saving within a month of being enrolled, with the employees receiving a full refund for any contributions made". One month, and your money back.
Anyone who drifts out later is counted separately as "stopping saving", and that runs at under 2% of active savers against roughly 10% of new savers opting out. So the auto enrolment opt out rate is a measure of immediate rejection, not of long-run commitment. It tells you how many people read the letter and acted on it.
The chart above splits the latest quarter by age band, and the result is the opposite of the usual assumption. Younger workers are supposed to be the ones who cannot afford a pension and bail out. In fact 11.5% of 22 to 29 year olds opted out, below the 12.7% of 30 to 39 year olds and well below the 15.1% of people aged 60 to State Pension age. Men opted out at 13.3% and women at 10.3%.
There's an obvious candidate explanation for the over-60s, though these statistics don't test it: they're close enough to the age at which a pension can be taken that a small new pot competes directly with the pay. The 30 to 39 peak is harder to explain from the data alone, and the statistics don't attempt to either.
Participation rose 46 points in the private sector and 5 in the public sector
The single most useful comparison in the whole dataset is between the two sectors, because one of them already had the default the policy was trying to create. Workplace pension participation among eligible private sector employees went from 42% in 2012 to 88% in 2025. Among public sector employees it went from 88% to 93%.
That's as close to a control group as this policy has. Where joining a decent occupational scheme was already the thing that happened unless you objected, there was very little for automatic enrolment to do. Where the default was to do nothing, it moved 46 percentage points.
The effect is sharpest on the people least likely to have filled in a form. Among private sector eligible employees aged 22 to 25, participation went from 18% in 2012 to 84% in 2025, a rise of 66 percentage points. The Pensions Regulator's compliance data gives the scale from the employer side: 2,734,305 employers had completed a declaration of compliance by May 2026, covering 11,452,000 eligible jobholders automatically enrolled since July 2012.
The default contribution rate became the national contribution rate
Auto enrolment minimum contributions have been 8% of qualifying earnings in total since April 2019, of which at least 3% comes from the employer and 5% from the employee, with basic rate tax relief counted inside the employee's share. That's a number chosen in regulations, and it's now where most private sector savers sit.
The DWP's own analysis of saving levels found that around 3 in 10 private sector defined contribution savers, 4.3 million out of 14.2 million, contribute at the minimum. Around 1 in 2, or 7.5 million, contribute 8% of total pay or less. The Pensions Commission puts the same point as an average: the median earner contributes only 1.7% of pay above the minimum, and more than 3 in 4 private sector employees save less than 12% of total pay.
None of this should be surprising, because it was described before the UK data existed. Madrian and Shea studied one large US employer's 401(k) records from June 1997 to June 1999, across a switch from opt-in to automatic enrolment. Participation at comparable tenure went from 37% to 86%. They also found this: "while three-quarters of non-automatically enrolled 401(k) participants have contribution rates of 6% or greater, three-quarters of automatically enrolled participants contribute at the default rate of 3%." Same inertia, both directions.
Where a worker lands depends heavily on who employs them. Around 53% of private sector defined contribution savers at firms with 1 to 4 employees sit at the minimum, against 20% at firms of 250 to 4,999. By industry the spread is wider still: 55% in hospitality against 5% in finance and insurance. Some of that is pay, and some of it is how generous the employer chose to be above the floor. The route the money takes matters too: gov.uk notes that paying by salary sacrifice can mean both you and your employer pay less tax and National Insurance on the same contribution.
8% of qualifying earnings is never 8% of your pay
This is the part that catches people, and it's pure arithmetic. The 8% isn't charged on your salary. It's charged on your earnings between £6,240 and £50,270 a year for the 2026 to 2027 tax year, the band the rules call qualifying earnings. The first £6,240 is excluded, so the headline percentage always overstates what reaches your pension.
| Salary | Qualifying earnings | Minimum total, 8% | Of which employer, 3% | Total as share of salary |
|---|---|---|---|---|
| £12,000 | £5,760 | £461 | £173 | 3.8% |
| £20,000 | £13,760 | £1,101 | £413 | 5.5% |
| £30,000 | £23,760 | £1,901 | £713 | 6.3% |
| £35,000 | £28,760 | £2,301 | £863 | 6.6% |
| £50,000 | £43,760 | £3,501 | £1,313 | 7.0% |
| £70,000 | £44,030 | £3,522 | £1,321 | 5.0% |
The DWP reaches the same place from its own data: "the AE minimum saving level for an individual earning £35,000 per year is £2,300 per year." The £70,000 row shows the other end of the band working. Above £50,270 nothing more is charged, so the minimum stops growing and the share of pay starts falling again.
Two consequences follow. A £12,000 earner on the minimum is saving 3.8% of pay, not 8%, which is why around 48% of those earning £10,000 to £20,000 sit at the minimum and why the minimum does so little for them. And the employee's 5% share includes tax relief, so what actually leaves your payslip depends on whether your scheme uses relief at source or net pay, which your employer picked and probably never mentioned.
The strongest objection is that joining and saving enough are different problems
Critics of automatic enrolment rarely dispute the participation numbers. They dispute what those numbers are taken to prove. The Pensions Commission, looking at the same data, estimates that around 15 million working-age individuals are set to miss their Target Replacement Rates, the share of pre-retirement income their pension would need to replace.
The timing arithmetic in the same evidence pack is blunter. For an average earner, "someone saving from age 40 needs to contribute at 13% of pay to hit their TRR but this falls to 7% for those starting from age 22". The statutory minimum is 8% of a band that is less than pay. So the default is adequate for a 22 year old who never stops, and well short for someone who starts late.
There's a second objection, pointed the other way, and it's the strongest evidence in favour of defaults rather than against them. Chetty and four co-authors studied 41 million Danish savings observations and found that "85% of individuals are 'passive savers' who are unresponsive to subsidies but are instead heavily influenced by automatic contributions made on their behalf". Pass-through from employer contributions to total pension saving exceeded 90%, meaning people didn't offset the money by saving less elsewhere. Tax subsidies, by contrast, mostly moved money between accounts.
Read together, those two findings are uncomfortable. The mechanism that got 22.6 million people saving works precisely because they aren't paying attention, and the amount they end up with is therefore set by a figure in regulations rather than by anything they decided.
Where the default does not reach at all
Automatic enrolment applies to employees aged 22 to State Pension age earning over £10,000. Three groups fall outside it, and the contrast tells you how much of the 90% is the default rather than a change in attitudes.
The self-employed get no default. Just 4% of people whose income is entirely self-employed save into a pension, and the Pensions Commission estimates around two-thirds of them would be eligible if the employee criteria were applied. The two groups aren't otherwise identical, because the self-employed also have no employer contribution and less predictable income. But the gap between 4% and 90% is the largest single number in this argument.
Low earners get a partial one. Around 2.5 million private sector employees earned under £10,000 in 2023, and only around 1 in 4 of them saved into a workplace pension; 69% of that group are women. And the very smallest employers remain behind: only around 55% of eligible private sector employees at firms with fewer than 5 employees are saving at all.
Age cuts the other way too, since eligibility ends at State Pension age, which is where that 15.1% opt-out rate for the over-60s starts to make sense. One thing that doesn't break the default: statutory maternity pay, statutory paternity pay, shared parental pay and statutory sick pay all count as qualifying earnings, so contributions continue through those periods on the pay you receive.
What this evidence cannot tell you
Start with the measurement. Participation and contribution figures come from the ONS Annual Survey of Hours and Earnings, a survey of employer payroll returns, not a census. The DWP applies a cleaning algorithm to it that infers missing tax relief and rounds contributions "slightly below AE minimums" up to the minimum, and states the effect is "less than 1 percentage-point on average contribution levels". That is a judgement about a figure the whole argument rests on.
The two vintages also differ. Participation runs to 2025; the contribution analysis runs to 2023. And the sources disagree about how many people sit on the default: the DWP notes that "Nest Insight's Employer Survey found around 4-in-10 employees were saving at default AE levels (this analysis finds around 3-in-10)". A 10 point gap on the central claim of this piece is not a rounding error, and neither estimate is obviously right.
The before-and-after comparison isn't a clean experiment either. The £10,000 earnings trigger has been frozen for years, which pulls more people into eligibility as wages rise: coverage went from 79% of employees in 2012 to 86% by 2026 partly for that reason alone. Some of the fall in average contribution rates is also composition rather than behaviour, because savers moved from defined benefit schemes, where the median total contribution was 25% of pay in 2023, into defined contribution schemes, where it was 8%.
Finally, the inertia research is mostly not British. Madrian and Shea studied one US firm over the two years to June 1999. Chetty and co-authors studied Denmark, which has a different tax system and different mandatory schemes. Both are NBER working papers, which are circulated before peer review. The UK statistics measure what happened here; the US and Danish papers explain why, and borrowed explanations are weaker than measured ones.
What would change the conclusion
If the thresholds move, the arithmetic above moves with them. The Pensions (Extension of Automatic Enrolment) Act 2023 gives the Secretary of State power, by regulations and after consultation, to cut the age criterion and to reduce or repeal the lower earnings limit. Section 1 isn't in force; it commences on a day appointed by statutory instrument. The 2017 review costed both: lowering the age to 18 would bring in "a further 900,000 young people", and removing the lower earnings limit "would bring an extra £2.6 billion into pension saving". Remove the £6,240 floor and the £20,000 earner's 5.5% becomes a genuine 8%.
If opting out keeps climbing, the premise changes. Going from 6.2% to 11.8% is a doubling in 11 years, concentrated in the last two. The official commentary calls the levels low, which they are. But a rate that doubles again stops being a rounding error, and it's the single series worth watching, because it's the only one that measures whether the default still has consent.
And if the minimum rose, the evidence says most people would save more without ever deciding to. That's the part worth sitting with. The same inertia that moved private sector participation from 42% to 88% is why 4.3 million people are saving exactly what the regulations say and not a pound more. The default isn't a floor that savers build on. For most of them, on this evidence, it is the whole decision, taken by someone else.