Key takeaways
- Flooring the moderate living standard for a single 65-year-old cost £448,289 on the best RPI-linked annuity rates of 29 October 2026, once income tax is counted in.
- The full new State Pension of £12,547.60 a year covers 88.2% of the minimum standard by itself, so the floor underneath that standard cost about £29,991.
- A lifetime RPI annuity bought at 65 undercut an index-linked gilt ladder priced the same day beyond 22.58 years of income, which is age 87.6 for that buyer.
- After the moderate floor, a £500,000 pot had £51,711 left for upside, and a 5% draw on that remainder is £2,586 a year.
- On UK real returns from 1872 to 2020, a 5% real draw on an all-equity upside sleeve lasted 25 years in 77.3% of 20,000 bootstrapped paths.
What the floor and upside strategy costs before you reach the upside
Buy secure income for the essentials, invest whatever is left for growth. That's the floor and upside strategy, and it's almost always described without a price on it. Here's the price, for one person, at one age, on one week's quotes.
A single person who wants the moderate Retirement Living Standard needs £32,700 a year after tax, on Pensions UK's 2026 modelling. The full new State Pension pays £241.30 a week in the 2026 to 2027 tax year, which is £12,547.60 a year. Gross that £32,700 up for income tax, using the £12,570 personal allowance and the 20% basic rate, and the taxable income required is £37,732.50. The State Pension covers part of it. The other £25,184.90 a year has to be bought.
On 29 October 2026 the best single-life RPI-linked annuity in Hargreaves Lansdown's weekly survey paid £5,618 a year on a £100,000 purchase at age 65, with a five-year guarantee. That's £17,800 of capital for every £1,000 a year of inflation-linked income. Buying £25,184.90 of income cost £448,289.
That figure is 89.7% of a £500,000 pension pot. At that pot size the upside isn't a design choice. It's a remainder.
The State Pension does the cheap work and income tax does the expensive part
A floor is only the income you don't already have. The minimum Retirement Living Standard is £13,900 a year after tax for one person, which grosses up to £14,232.50. The State Pension covers 88.2% of that, leaving £1,684.90 to buy. At the same 29 October 2026 rate, that cost £29,991.
The comfortable standard is £45,400 after tax. Grossing it up pushes income over the £50,270 top of the basic rate band, so part of it is taxed at 40%, and the gross requirement becomes £54,720. The gap after the State Pension is £42,172.40, and flooring it cost £750,666.
| Standard (single, outside London) | After tax | Gross needed | Income to buy | RPI annuity cost at 65 |
|---|---|---|---|---|
| Minimum | £13,900 | £14,232.50 | £1,684.90 | £29,991 |
| Moderate | £32,700 | £37,732.50 | £25,184.90 | £448,289 |
| Comfortable | £45,400 | £54,720 | £42,172.40 | £750,666 |
Spending rises by a factor of about three across those rows. The cost of flooring it rises 25-fold. Two things drive that. The State Pension is a flat amount, so it covers a shrinking share of a bigger target, and the tax grossing-up compounds on top.
Giving up the inflation link makes the bill much smaller. A level annuity at 65 paid £8,146 per £100,000, so £12,276 bought £1,000 a year that never rises again. The moderate floor on that basis cost £309,169 rather than £448,289. A 3% escalating annuity sat between them, at £16,410 per £1,000 a year and £413,274 for the moderate floor. The cheap floor is cheap because it shrinks in real terms for as long as you hold it.
Age moves the price more than anything else on the page. The chart plots what £1,000 a year of RPI-linked income cost at each age on those quotes, from £22,292 at 55 to £13,075 at 75. Retiring before State Pension age costs twice over, because the State Pension isn't there to do its share. State Pension age reached 66 in October 2020 and rises to 67 between 2026 and 2028, so nobody retiring at 60 has it yet. A 60-year-old flooring the same moderate standard has to buy the whole £37,732.50, and at the age-60 rate of £4,922 per £100,000 that cost £766,609. That's 1.71 times the age-65 bill for the same standard of living.
Buying a guaranteed income floor or building it from index-linked gilts
An annuity isn't the only way to floor spending. You can buy index-linked gilts maturing in each year you need the money, which is a ladder: a row of bonds that pay out one after another. The Bank of England describes that market as two types of inflation-linked bond, old-style and new-style, "both linked to the UK Retail Price Index (RPI)", and it publishes the implied real yield curve they price off every business day. On 30 October 2026 the real spot rate was 0.8168% at three years, 1.9537% at ten, 2.4297% at twenty and 2.3692% at thirty.
Discount £1,000 a year back along that curve and you get the cost of a self-built floor. Twenty years of inflation-linked income cost £16,277 per £1,000 a year. Twenty-five years cost £19,146. Thirty years cost £21,718.
Set that beside the £17,800 the RPI annuity charged for the same £1,000 a year, for life. The ladder is cheaper over 20 years and dearer over 25. The crossover sits at 22.58 years of income, which is age 87.6 for a 65-year-old buyer. Past that point the annuity was the cheaper way to hold the same real income, and it was 18.04% cheaper than funding 30 years yourself.
The same comparison in nominal terms crosses sooner. A level annuity at 65 cost £12,276 per £1,000 a year. A conventional gilt ladder paying £1,000 for 20 years cost £12,066 on the nominal curve of 30 October 2026, where the thirty-year spot rate was 5.7233%. The break-even is 20.68 years, or age 85.7.
What you're buying above those crossovers is pooling. An annuity provider sets one price across many buyers who don't all live for the same length of time, so it can pay a survivor more than that survivor's own capital would have funded. A ladder has no pool, so it has to cover every year you might live out of your money alone. The Office for National Statistics puts cohort life expectancy at 65 in 2023 at a further 19.8 years for men and 22.5 years for women. Both crossovers sit just beyond the male figure and either side of the female one. Life expectancy is an average rather than a median, so this doesn't split buyers neatly in half. What it does say is that the crossover lands inside the range of ordinary outcomes rather than out in the tail. Buyers who die earlier would have done better with the ladder. Buyers who live longer would otherwise have run out of it.
One caveat on the ladder arithmetic. The Bank's real curve starts at 2.5 years, so the first two years of income can't be priced from it, and they're counted here at face value as cash. That slightly overstates the ladder's cost, by whatever those two years would have earned. It doesn't move the crossover by much, because the discount factors near the front of the curve are close to 1 anyway at a 0.8168% real yield.
What is left for the upside, by pot size
Once the floor's price is fixed, the upside is simple subtraction. These rows use the £448,289 moderate floor at 65.
| Pot | Floor cost | Upside left | Share of pot in upside | 5% draw on the upside |
|---|---|---|---|---|
| £300,000 | £448,289 | not affordable | none | none |
| £500,000 | £448,289 | £51,711 | 10.3% | £2,586 |
| £750,000 | £448,289 | £301,711 | 40.2% | £15,086 |
| £1,000,000 | £448,289 | £551,711 | 55.2% | £27,586 |
The first row is the one that matters. At £300,000, and at £400,000, the moderate floor can't be bought at all on these rates. The strategy doesn't degrade gracefully there; it just isn't available at that standard of living, and what's left is a choice between a smaller floor and no floor.
Couples get a better deal than the single-person rows suggest, because two State Pensions and two personal allowances both count. Splitting the £45,400 couple standard evenly, each person needs £25,232.50 gross, so each buys £12,684.90 of annuity income. The pair cost £451,581, which is close to what one person paid for the moderate single standard.
What the upside sleeve has done, and what failing there means
The point of flooring essentials is that the rest of the money can take risk. The Jordà, Schularick and Taylor macrohistory database carries UK total returns and consumer prices back to the nineteenth century. Over 1872 to 2020, UK equities returned 5.198% a year after inflation, with an annual standard deviation of 17.619%. Bonds returned 1.742% real. The worst year in the series was 1974, at minus 56.97% real; the next year was plus 103.43%.
Run a constant real withdrawal against 20,000 bootstrapped paths drawn from those annual equity returns, and you get the survival rates below. Survival here means the sleeve still had money at the end of the horizon.
| Real draw on the upside sleeve | 20 years | 25 years | 30 years |
|---|---|---|---|
| 3% | 97.9% | 95.9% | 93.3% |
| 4% | 94.0% | 88.7% | 83.4% |
| 5% | 86.5% | 77.3% | 69.7% |
| 6% | 75.4% | 62.9% | 53.4% |
| 7% | 60.9% | 47.0% | 37.5% |
The actual record is kinder than the simulation at low draws and harsher at high ones. Rolling 25-year windows of the real series gave a 5% draw an 81.6% survival rate, against 77.3% from the bootstrap. At 7% it goes the other way: the real windows lasted 30 years only 30.8% of the time, against 37.5% in the bootstrap. The difference is sequence, because real history has runs of good and bad years and drawing years at random throws those runs away. Our piece on sequence risk shows what reordering one fixed set of returns does to a retirement.
A 6% draw that lasts 30 years in 53.4% of paths is close to a coin toss. As a whole-portfolio plan that's a failure rate nobody would accept for essential spending, which is what a safe withdrawal rate is built to avoid. On a floored portfolio the same 53.4% describes something different, because what runs out is the discretionary money. Failure means fewer flights and a smaller car, not a cold house. That asymmetry is the entire argument for the floor and upside strategy, and it's why the acceptable odds on the upside sleeve are lower than they'd be anywhere else.
Left alone, the upside sleeve does something the floor can't. Across the 125 rolling 25-year windows in the UK series, an untouched equity holding grew by a median of 3.74 times in real terms. The worst of those windows ended at 0.98 times, so flat in real terms after 25 years. The best ended at 29.82 times.
The strongest case against paying for the floor
The critics of flooring have three good arguments, and the first is irreversibility. Once the £448,289 has bought an annuity, the capital has gone: there's nothing left to leave, to spend on a new roof, or to redirect if your health changes. A gilt ladder keeps the residual capital, which is a real advantage that the crossover arithmetic above doesn't price.
The second is that the price is a snapshot of one day's yields. The £17,800 per £1,000 a year reflects a thirty-year real spot rate of 2.3692%. Real yields move, and at lower real yields the same floor costs materially more. Buying the floor is a bet on the level of real rates on the day you buy, and the same timing problem runs through the annuity vs drawdown comparison.
The third is that the floor can be defined much more cheaply than Pensions UK defines it. The basket behind the standards covers an eating-out budget, a car "(if applicable)" and a holidays and leisure allowance, but Pensions UK is explicit that "rent and mortgage payments are not included". A renter's floor is higher than any row in these tables, and an outright owner's is lower. Plenty of people's genuine essentials sit between £13,900 and £32,700, and the cost of flooring them falls roughly in proportion.
There's also a cheaper version of the same product for anyone whose health is poor. The 29 October 2026 table shows a 65-year-old smoker getting £6,219 a year of RPI-linked income from £100,000, against £5,618 for standard rates. That's 10.7% more income, which brings the moderate floor down to £404,967, a saving of £43,322.
What these numbers can't tell you
Every figure here is a point estimate from a single day, and the annuity rows are one broker's best quotes rather than an average of the market. Annuity rates move weekly. The gilt curve moves daily.
The return evidence is a backtest of one country and it isn't a forecast. The macrohistory sample ends in 2020, so it carries nothing from the last few years, and UK equities are one market rather than the world. The bootstrap assumes each year is independent, which discards both momentum and mean reversion. It also charges no fund fees, no platform costs and no tax on the upside sleeve, all of which reduce the survival rates in the table.
The tax arithmetic assumes the only taxable income is the State Pension plus the annuity, with the 2026 to 2027 allowances. Anyone with other income faces a different gross-up, and the sequencing choices in pension withdrawal tax change the answer again. The Hargreaves Lansdown table also doesn't state whether the RPI increases are capped, and a cap would make the RPI annuity a weaker match for an index-linked gilt ladder than it looks here.
What would change this arithmetic
Two prices decide whether a floor is worth buying, and they move together. The first is the real yield curve, which sets what a self-built floor costs. The second is the annuity rate, which sets what a pooled floor costs. On 30 October 2026 the gap between them favoured the insurer past 22.58 years, and the gap is the whole decision. If annuity pricing lags a fall in real yields, the pooled floor gets relatively cheaper; if it lags a rise, the ladder wins over longer horizons than it does today.
Longevity is the other input, and it's drifting. The Office for National Statistics projects cohort life expectancy at 65 rising to 21.8 years for men and 24.4 years for women by 2047. Every year added to that projection pushes the crossover horizon further inside a normal retirement, which makes pooling cheaper relative to building. The number to watch isn't the annuity rate on its own. It's the annuity rate against the thirty-year real gilt yield, because that spread is what you're paying for mortality pooling, and it's the only part of this arithmetic that isn't in your hands.