Key takeaways
- A one-person "other retired" UK household spent £343.30 a week in the year to March 2025, which is £17,851.60 a year against a £676.60 all-household average.
- The full new State Pension of £241.30 a week is £12,547.60 a year, leaving £5,304.00 of that spending for a portfolio to fund.
- At a 3.5% withdrawal rate the gap converts to £151,500 of capital. At 4.0% it is £132,600, and at 3.0% it is £176,800 — a £44,200 range.
- Using the £32,700 moderate Retirement Living Standard instead of observed spending returns £575,800, moving the answer 9.6 times further than the rate does.
- A National Insurance record of 25 qualifying years rather than 35 adds £102,400, which is 2.3 times what the withdrawal-rate range moves it.
Start from what retired households spend, not from a multiple of your salary
You want to know how much capital you need before you can stop working. The usual answer is some multiple of your current income. That answer has a problem: your salary is a fact about your working life, not about your retirement.
There's a route that doesn't need it. Take what retired households actually spend, from published expenditure data. Subtract the income that arrives whether or not you own anything, starting with the State Pension. Divide what's left by a withdrawal rate, which is simply the conversion between an annual flow and the capital that has historically supported it.
Run that on UK figures and a one-person retired household comes out at roughly £151,500, at a 3.5% withdrawal rate. Now the important part. That number is derived, and it moves a very long way when you change the assumptions feeding it. The same household, the same method, needs anywhere from £132,600 to £575,800. The spread is what this piece is about. A single headline figure would hide it.
UK retired households spent £343.30 a week in the year to March 2025
The Office for National Statistics runs the Living Costs and Food Survey and publishes the results as Family Spending. The edition covering April 2024 to March 2025 was released on 11 June 2026. Average weekly household expenditure across all UK households was £676.60.
Table A23 splits that by household composition, and it separates retired households into two groups. One is households "mainly dependent on state pensions and not economically active". The other is everyone else who is retired — people with a private pension, a defined benefit entitlement, or savings alongside the State Pension. If you're reading a page about portfolio capital, the second group is your comparator.
- Retired, mainly dependent on the State Pension: £271.30 a week for one person, £498.10 for two adults.
- Other retired: £343.30 a week for one person, £691.40 for two adults.
On a 52-week basis, £343.30 is £17,851.60 a year. The two-adult equivalent is £35,952.80.
Two things about what those figures contain. Housing costs are in there, but the "housing, fuel and power" line is net: £72.40 a week for the one-person other-retired group, with mortgage interest, council tax and Northern Ireland rates moved into a separate "other expenditure items" category that is still inside the total. And it's expenditure, not income. Income tax hasn't been deducted from it, because it was never in it.
The State Pension comes off the spending before anything is divided
The full rate of the new State Pension for the 2026/27 tax year is £241.30 a week, up from £230.25. On a 52-week basis that's £12,547.60 a year of inflation-linked income for life, and it doesn't come out of your portfolio.
So the portfolio's job is the remainder:
- One person, other retired: £17,851.60 − £12,547.60 = £5,304.00.
- Two adults, other retired: £35,952.80 − £25,095.20 = £10,857.60.
For the households mainly dependent on the State Pension, the remainders are £1,560.00 and £806.00 a year. Those aren't rounding errors, but they're close to it. Two full entitlements, on this data, come within £806.00 of covering what a two-adult household in that group spends.
There's a tax detail sitting right on top of this. The standard Personal Allowance for the 2026/27 tax year is £12,570. The full new State Pension is £12,547.60. That leaves £22.40 of allowance before the basic rate of 20% starts applying to everything else, which for most people means the portfolio withdrawals. Every gap figure above is a net spending gap. The gross withdrawal needed to produce it is larger.
Three withdrawal rates on one gap: £132,600 to £176,800
Converting an income gap into capital needs a withdrawal rate, and the published research doesn't agree on one. The Pensions Policy Institute's June 2025 technical report for the Retirement Living Standards models income from capital "at an initial amount of 3.5% of the starting capital", rising with inflation, and puts the chance of exhausting it before death at "approximately 5%". Morningstar's 2025 research lands at 3.9% for a 30-year horizon at a 90% probability of funds remaining.
Rather than pick one, here's the £5,304.00 gap at three rates, rounded to the nearest £100:
- 4.0% → £132,600
- 3.5% → £151,500
- 3.0% → £176,800
That's a range of £44,200 across a full percentage point of withdrawal rate. The two-adult version at 3.5% is £310,200. The chart above plots the one-person figures alongside the two assumptions that follow.
Which rate is defensible is a genuinely contested question, and it's been argued elsewhere on this site — the cross-country evidence on the 4% rule is set out in the piece on safe withdrawal rates and the international evidence on the 4% rule, and the reason a rate that works on average can still fail for you is the subject of the analysis of sequence risk and identical returns in opposite orders. The reciprocal relationship between a rate and a spending multiple is worked through in the piece on whether the multiple is 25 times or 30 times. None of that gets restated here. What matters here is the size of the rate's effect relative to everything else in the sum.
Change the spending input and the answer moves 9.6 times further
The Retirement Living Standards, produced by Pensions UK with Loughborough University's Centre for Research in Social Policy, take the opposite approach to the ONS. Instead of measuring what households did spend, groups of people were asked what a given standard of living requires. The site puts a one-person household at £13,900 a year at the minimum standard, £32,700 at moderate and £45,400 at comfortable. Two-person figures are £22,500, £45,400 and £62,700.
Feed the moderate figure into the same three-step sum. £32,700 − £12,547.60 leaves £20,152.40, and at 3.5% that's £575,800 of capital.
| Assumption changed | Capital at stake | Versus the rate range |
|---|---|---|
| Withdrawal rate, 4.0% to 3.0% | £132,600 to £176,800 | £44,200 |
| National Insurance record, 35 years to 25 | £151,500 to £254,000 | £102,400, or 2.3 times |
| Spending input, observed to moderate standard | £151,500 to £575,800 | £424,200, or 9.6 times |
The spending input moves the answer £424,200, which is 9.6 times what a full point of withdrawal rate moves it. Neither figure is wrong. They measure different things. The ONS number is behaviour; the moderate standard is a specification. A reader who spends most of an argument on 3.5% versus 4.0% is polishing the smallest term.
The two standards also cover different ground. The Retirement Living Standards say plainly that "rent and mortgage payments are not included", and that "to meet these costs, you'll need enough income after tax". Social care, dependants and pets are excluded too. The ONS averages include renters and mortgage-holders, because they're averages of real households.
An incomplete National Insurance record costs more than a full point of rate
Everything above assumed the full State Pension. That assumption is a rule, not a certainty. GOV.UK is explicit: if your National Insurance record started after April 2016, "you will need 35 qualifying years to get the full rate of new State Pension", and "you'll need 10 qualifying years on your National Insurance record to get any new State Pension". GOV.UK describes each qualifying year added after 5 April 2016 as adding a fixed slice of the full rate, which is that rate divided by 35, so the amount builds in equal steps between the two thresholds.
Suppose the record reaches 25 years rather than 35. That's £241.30 × 25 ÷ 35, or £172.36 a week — £8,962.72 a year. The one-person gap widens from £5,304.00 to £8,888.88, and the capital at 3.5% goes from £151,500 to £254,000.
That's £102,400 of extra capital, or 2.3 times what the whole withdrawal-rate range is worth. It's also the one assumption of the three that can sometimes be changed directly rather than merely argued about, which is what the piece on what a voluntary Class 3 year buys examines.
In the same ONS data, spending falls 18.8% between the 65-74 and 75-plus groups
Every figure so far treats spending as flat. The ONS data itself says otherwise. Tables A15 and A16 average three years, from the financial year ending 2023 to the financial year ending 2025, and split households by the age of the household reference person. Households headed by someone aged 65 to 74 spent £547.10 a week. Households headed by someone aged 75 or over spent £444.30.
That's a fall of £102.80 a week, or 18.8%. It's the UK counterpart to the declining-spending research usually cited from American data, and it points the same way.
Read it carefully, though. These are different households, compared at one moment, not the same households followed over time. Some of the difference is cohort: today's 78-year-olds had different working lives and different pension entitlements from today's 68-year-olds. Some of it may be constraint rather than choice. A cross-sectional gap can't separate those.
Horizon matters for the same reason. ONS cohort life expectancy at exact age 65 in 2026, on the 2024-based principal projection, is 20.17 years for men and 22.86 years for women. Cohort figures allow for expected future improvements in mortality, which period life tables don't. Roughly half of people outlive their own life expectancy, which is what makes guaranteed income structurally different from capital — the comparison run in the piece on the break-even age between an annuity and drawdown.
The strongest objection: observed spending is what people could afford
Here's the serious case against starting from the ONS figures, and it deserves a fair hearing.
Family Spending measures what households did spend. It doesn't measure what they'd have spent with more money. If a retired household's income is low, its expenditure is low, and the survey records that as spending rather than as a shortfall. Building a capital target on it risks encoding other people's constraints as your plan. The Retirement Living Standards exist precisely because a group of people sat down and specified what a decent standard requires, independent of what anyone can currently afford.
The tenure evidence supports the objection. In Family Spending's Table A32, across all UK households rather than retired ones, net housing, fuel and power costs £70.70 a week for owners who own outright and £263.80 for private renters. The Retirement Living Standards assume the home is owned outright and exclude rent entirely. A renting household is looking at a different sum from either dataset.
So the honest reading is that the two figures bracket something. Observed spending is a floor that real households are living on. The moderate standard is a specification that a research panel endorsed. The distance between them, £424,200 of capital at 3.5%, is not a measurement error. It's the range within which your own answer sits, and the arithmetic can't narrow it for you.
What this arithmetic cannot tell you
Start with the sample. The one-person other-retired cell in Table A23 rests on 500 households in the survey. That's enough for a national average and nowhere near enough to describe any individual. Averages also hide dispersion: half of households in a group spend less than the middle figure, and half spend more.
The withdrawal-rate research carries its own caveat. The Pensions Policy Institute's 3.5% is a modelling convention with an approximately 5% chance of exhaustion, not a promise. Morningstar's 3.9% is built on forward-looking US asset-class assumptions and excludes non-portfolio income by design. Neither models UK income tax on the withdrawals, and this piece hasn't either — the gap figures are net of nothing because expenditure data comes with no tax attached.
The State Pension figures are the 2026/27 statutory rates. They're current, they aren't forecasts, and the arithmetic assumes they keep pace with the spending they're subtracted from. And the whole method assumes flat real spending, which the age tables above suggest is questionable and which the piece on withdrawal guardrails and the price of a flexible start treats as a variable rather than a constant.
What would change the answer
If the State Pension stopped keeping pace with prices, the largest single subtraction in the sum shrinks. £12,547.60 of inflation-linked income is doing more work here than any assumption about markets. A change in uprating policy would move every figure in the table.
If a UK longitudinal study confirmed the 18.8% fall is choice rather than constraint, the flat-spending assumption becomes indefensible and the capital figures come down. If it showed the opposite — that the fall reflects households running short — then planning around it would be circular, and the figures stand.
If you're renting in retirement, neither starting point applies without adjustment, and the £263.80-a-week private-rented housing line suggests the adjustment is large.
The thing worth watching isn't the rate. It's the spending figure you feed into the top of the sum, because that's where nine-tenths of the variation lives — and unlike the other two assumptions, it's the one you can measure directly from your own bank statements rather than infer from a survey of other people. LedgerTouch tracks the capital side of this; a year of your own outgoings is the other half, and no dataset can supply it for you.