Key takeaways
- A full year of Class 3 voluntary National Insurance costs £956.80 in the 2026 to 2027 tax year, and adds about £358 a year to the state pension.
- That is a payback period of 2.7 years before tax, and 3.3 years for a pensioner who pays 20% Income Tax on the extra income.
- Paid at state pension age, £956.80 returns roughly £6,700 over the 18.7 years ONS gives as male life expectancy at 65, an implied 37% a year.
- Paid 17 years before state pension age, the same cash flows imply about 8% a year. The return is set by the wait, not by longevity.
- Gaps can be filled only 6 years back, and GOV.UK states that voluntary contributions do not always increase your State Pension.
The whole sum in one line: £956.80 buys £358 a year
You've found a gap in your National Insurance record. The government offers to sell you the missing year. Is the price fair?
Here's the arithmetic, and it's short. In the 2026 to 2027 tax year a Class 3 voluntary contribution costs £18.40 a week. A full year runs to 52 weeks, so the year costs £956.80. HMRC's own direct debit schedule for 2026 to 2027 confirms the total: eight payments of £73.60 and four of £92.00.
What you get is one more qualifying year. GOV.UK values one at about £6.89 a week. Across 52 weeks that's £358.28 a year, paid for as long as you live, and rising each April.
Divide the second number into the first and the year repays its own cost in 2.7 years, or 32 months. Put another way, the first year's payments alone hand back 37% of what you paid. Then they keep coming.
That is the headline, and it's genuinely unusual. Three things complicate it, and any of them can turn the sum on its head: when you pay, how long you have to wait, and whether the year counts for you at all.
Where £6.89 comes from, and why GOV.UK's own pages disagree about it
The new state pension has a deliberately simple design. The full rate is £241.30 a week in the 2026 to 2027 tax year. You need 35 qualifying years to reach it if your record started after April 2016, and 10 to get any new state pension at all.
So one qualifying year is worth the full rate divided by 35. £241.30 ÷ 35 is £6.894, which is the £6.89 that GOV.UK quotes on its contracting-out page.
Then look at DWP's guidance note, Your State Pension explained. It says each qualifying year adds "about £6.89 a week, this is £230.25 divided by 35". £230.25 was the full rate in the 2025 to 2026 tax year. Divided by 35 it gives £6.58, not £6.89. The answer has been updated and the working has not.
It's a small inconsistency with a large implication. The £6.89 is not a constant. It's a quotient, and the numerator moves every April. The full rate rose 4.8% on 6 April 2026, from £230.25 to £241.30 a week, under the triple lock, which raises the pension by the highest of average earnings growth, CPI inflation, or 2.5%.
The cost side moves too, and not in step. Class 3 was £17.45 a week for both 2023 to 2024 and 2024 to 2025, £17.75 for 2025 to 2026, and £18.40 now. That's a rise of 3.7% this April against 4.8% on the pension. The ratio drifted in the buyer's favour by about a percentage point in a single year, for no reason other than that two different formulas were applied to two different numbers.
The return depends far more on when you pay than on how long you live
Treat the contribution as buying an income stream and you can compute an internal rate of return, meaning the discount rate that makes £956.80 today equal to £358.28 a year afterwards.
ONS national life tables for 2022 to 2024 put life expectancy at age 65 in the UK at 18.7 years for males and 21.2 years for females. Cohort tables, which assume mortality keeps improving, are more generous: someone aged 65 in 2023 could expect a further 19.8 years if male and 22.5 if female.
Run those horizons through the calculation, assuming the contribution is paid in the year you reach state pension age and the increment stays flat in cash terms:
- Over 18.7 years of payments, an implied 37.3% a year. Total received about £6,700, roughly seven times the outlay.
- Over 21.2 years, 37.4%. Total about £7,600.
- Over only 10 years, if things go badly, still 35.7%.
Notice how little those move. Once you're past the 2.7-year payback, extra years add to the total but barely touch the rate of return. Longevity is not the variable that matters here.
This is. Pay 10 years before you reach state pension age and the same cash flows imply about 11.2% a year. Pay 17 years early, which is a 50-year-old with a state pension age of 67, and it drops to roughly 7.8%. The money does nothing in the gap. It's the waiting, not the living, that sets the return, and that inverts the usual instinct to deal with these things early.
Two things pull the headline down. The state pension is taxable income. The Personal Allowance is £12,570 and the 20% band runs to £50,270. A pensioner whose income already clears the allowance keeps £286.62 of the £358.28, so payback stretches from 2.7 years to 3.3. At 40% it's 4.5 years. The chart above puts those three payback periods side by side. Second, the life tables are population averages. They describe a cohort, not a person, and your own health history is not in them.
Deferring the whole pension is the same trade at six times the price
There's a second way to buy state pension income, and the comparison is the most useful benchmark available. Delay claiming and your weekly payments rise by "just under 5.8%" for every year deferred.
Take 5.8% of £241.30 and you get £14.00 a week, or £728.00 a year. The price is a full year of pension foregone, which at the full rate is £12,547.60. That's a payback of 17.2 years.
Same income, same payer, roughly six times the cost per pound of annual pension. A Class 3 year is not cheap because state pension income is cheap. It's cheap because the price was set as a weekly National Insurance rate and the value was set as a share of the full pension, and nobody tied the two together. Deferral, by contrast, is priced to be roughly neutral.
Contracting out is why your forecast doesn't match the headline rate
If your record started before April 2016, the 35-year rule probably doesn't describe you.
Large numbers of people were contracted out of the Additional State Pension. Anyone in a final-salary scheme was, and so were members of many workplace and personal schemes before 2012. While contracted out, you or your employer paid less into the state system and more into a workplace or personal pension. A deduction was then applied when DWP worked out your starting amount for the new state pension in 2016.
GOV.UK states the consequence without hedging: "If you were contracted out, you will usually need more than 35 qualifying years to get the full rate of new State Pension."
Its own worked example shows the mechanics, and also shows why you can't run this from a web page. Sam has a state pension age of 67 and a forecast cut by contracting out to £200 a week. GOV.UK says "the £30 deduction will be cancelled out by adding 5 more qualifying years".
Check that against the same page's own full rate. £241.30 minus £200 is £41.30 a week, not £30. Five qualifying years at £6.89 add £34.45, which doesn't close a £41.30 gap. The example was written when the full rate was lower and the deduction really was about £30. The uprating moved the target and the example stayed put.
That's the second stale calculation on GOV.UK in one topic. Neither is a scandal. Both point the same way: the arithmetic on the public pages is illustrative, and the only figures that describe you are the ones in your own forecast.
Six years back, and an old gap does not cost an old price
You can't fill any gap you like. GOV.UK: "You can only pay for gaps in your National Insurance record for the past 6 tax years. The deadline is 5 April each year." Its example is that you have until 5 April 2032 to fill the 2025 to 2026 year.
Pricing inside that window works in a way most people get wrong. For the previous 2 tax years you pay the rate that applied then, so £17.75 for 2025 to 2026 and £17.45 for 2024 to 2025. Beyond that, "if you're paying for any earlier years, you'll pay the rate for 2026 to 2027". An older gap costs £18.40 a week, exactly the same as a current one.
Two consequences follow. Waiting never makes a gap cheaper, and it eventually makes the gap unfillable. Filling all 6 available years at the current rate would cost £5,740.80 and buy £2,149.68 a year, assuming every one of them counts. Buying all 35 years outright would cost £33,488 for a pension of £12,547.60 a year, which is the same 2.7-year payback seen from a different angle.
The self-employed sit in a different position entirely. Class 2 voluntary contributions cost £3.65 a week in 2026 to 2027 and buy the same qualifying year, which is where the arithmetic stops being merely good.
The strongest case against running this arithmetic at all
The best objection isn't that the numbers are wrong. It's that they may not apply to you, and GOV.UK says so in one flat sentence: "Voluntary contributions do not always increase your State Pension."
There are several routes to that outcome. If your starting amount in 2016 already matched or exceeded the full rate, more years add nothing. If working alone would take you to 35 qualifying years, or past the higher number a contracted-out record needs, a purchased year is redundant. Some gaps are already covered by National Insurance credits granted free for illness, unemployment, caring or parenting, and paying for those buys something you have.
The means-tested layer matters as much. Pension Credit "tops up your weekly income to £238 if you're single". For a pensioner below that line, extra state pension can displace Pension Credit rather than add to it, and Pension Credit is the gateway to help with housing, council tax and NHS costs. The 37.4% payout rate assumes the increment reaches you. Sometimes it substitutes for money you'd have received anyway.
Then there's the political exposure. The full rate, the 35-year divisor and the triple lock all sit in statute, and statute moves. The 2016 reform rewrote the calculation for everyone reaching state pension age after it. State pension age itself is legislated to rise from 66 to 67 between 2026 and 2028. A contribution made today is a claim on a formula Parliament can amend before it pays out.
What these numbers cannot tell you
Every figure here is a 2026 to 2027 figure. Both sides of the ratio reset each April, so the 2.7-year payback is a snapshot and not a constant.
The life expectancy figures are population averages, and the two ONS tables disagree by more than a year at age 65 because they assume different futures. Period tables hold death rates at their 2022 to 2024 levels; cohort tables project improvement. Neither is a forecast for one person, and the gap between them is a fair measure of how much confidence any of this deserves.
The implied returns assume flat nominal payments, annual timing, no tax and no means-tested offset. Real payments have risen with the triple lock, which pushes the true figure higher; Income Tax pushes it lower. The calculation also assumes you reach state pension age. Someone paying at 50 carries 17 years of mortality risk on a contribution that pays out nothing at all if they die first, which is the one scenario where the entire outlay is lost.
And none of it substitutes for a forecast. The forecast service shows which years have gaps, whether each one can be filled, what each costs and, crucially, whether filling it would raise your pension. That last question is the one the arithmetic cannot answer, because it depends on a starting amount calculated in 2016 from a record you probably can't reconstruct.
What would change the conclusion
If the Class 3 rate rose faster than the full pension for a run of years, the ratio compresses and the payback lengthens. This April it went the other way, 3.7% against 4.8%, but that's a policy outcome rather than a mechanism. Nothing links the two.
If the 35-year requirement changed, the value of a single year changes with it, because 35 is a divisor rather than a law of nature. A move to 40 would cut the increment by about an eighth at any given full rate.
If your marginal rate in retirement is 40% rather than 20%, payback moves from 3.3 years to 4.5 and the deal starts to resemble an ordinary investment. That is a live question for anyone with a substantial private pension alongside the state one, and it runs on the same logic as the two mechanisms that deliver pension tax relief on the way in. Both are decided by which side of a threshold your income lands on. It also feeds directly into the arithmetic behind a retirement number, because a guaranteed indexed £358 a year removes a slice of the portfolio that would otherwise have to fund it, which is the same substitution the international evidence on safe withdrawal rates keeps running into.
The number worth watching isn't the payback. It's the divisor. £6.89 is £241.30 shared over 35 years, and Parliament sets both. Check the forecast before the sum, because the forecast decides whether the sum applies to you at all.