This guide is for households retiring with money in both the US and UK systems who want to know how much they can spend each year without running out. The short answer is that no single percentage is safe for everyone. The well-known 4% rule comes from US market history, and research across other countries points lower for a fixed, inflation-linked withdrawal. The most reliable answer comes from testing your own plan against bad sequences of returns, currency swings and a long life, and deciding in advance how you would adjust.
How much can I safely spend in retirement?
There is no universally safe number. The research gives a range. At the cautious end, about 2.3% of the starting portfolio holds for a fixed, inflation-linked withdrawal judged across many countries’ market histories[5]. Around 4% held on US history alone[1]. Starting rates above 5% are possible if you accept cuts in bad years[6]. Where you sit in that range depends on how long the money must last, how flexible your spending is, what else you have coming in, and how much risk of having to cut back you can live with.
A “withdrawal rate” is the first year’s withdrawal as a percentage of the portfolio at retirement. In the classic studies, that cash amount then rises each year with inflation, whatever markets do. On a £1,000,000 portfolio, 4% means £40,000 in year one, 3% means £30,000 and 2.31% means £23,100 (our arithmetic). These are amounts before tax, before charges, and on top of any state or defined benefit pensions.
| Study | Data | Headline finding |
|---|---|---|
| Bengen, Journal of Financial Planning, October 1994[1] | US shares and intermediate Treasuries, from 1926 | 4% inflation-adjusted never exhausted a portfolio in under 33 years; 50–75% in shares |
| Cooley, Hubbard & Walz (“Trinity”), AAII Journal, February 1998[2] | US, 1926–1995 | 4% inflation-adjusted over 30 years: 95% success (50/50), 98% (75/25) |
| Guyton & Klinger, Journal of Financial Planning, March 2006[6] | US, with decision rules, 40-year horizon | 5.2–5.6% starting rates at 99% confidence with at least 65% in shares, if spending is cut after poor years |
| Pfau, Journal of Financial Planning, December 2010[3] | 17 developed countries, 109 years | 4% “safe” in only 4 of 17 countries |
| Anarkulova, Cederburg, O’Doherty & Sias, Journal of Pension Economics & Finance, 2025[5] | 38 developed countries | 2.31% for a 65-year-old couple accepting a 5% chance of ruin |
Where does the 4% rule come from?
From a 1994 paper by William Bengen, a US financial planner. He tested every rolling retirement period in US data from 1926 onwards and asked what first-year withdrawal, raised each year with inflation, would have survived the worst of them. With between 50% and 75% in US large-company shares and the rest in intermediate-term Treasuries, 4% “in no past case” exhausted a portfolio in under 33 years[1]. The periods that did most damage were those starting in the late 1960s, when retirees met the 1973–74 bear market and high inflation early on[1]. Bengen later raised his figure to about 4.5% after adding smaller-company shares[8].
The 1998 “Trinity study” by Philip Cooley, Carl Hubbard and Daniel Walz, of Trinity University in Texas, took a similar approach using 1926–1995 data and reported success rates rather than a single safe maximum. A 4% inflation-adjusted withdrawal lasted 30 years in 95% of periods for a 50/50 portfolio and 98% for 75/25. The authors stressed that the answer depends on asset mix, time horizon and the retiree’s tolerance for risk[2].
Three features of these studies matter more than the headline figure. They assume a fixed real withdrawal: spending never adjusts to markets. They use 30-year horizons: someone retiring at 60 may need 35 or 40 years. And they ignore charges and taxes, both of which reduce what the portfolio can support. Estimates are updated regularly. For example, Morningstar’s research for 2026 put a safe starting rate for fixed, inflation-adjusted spending at 3.9% over 30 years at a 90% success rate[9].
Does the 4% rule work for UK and cross-border retirees?
With caution. The 4% rule describes the experience of a US-dollar investor holding US assets through the twentieth century, one of the most successful markets on record. When Wade Pfau applied the same method to 17 countries using 109 years of data, 4% would have been sustainable in only four of them: the United States, Canada, Sweden and Denmark. The UK was not among them[3][4]. A 2025 study using 38 developed countries, and treating each retiree’s outcome as uncertain rather than tied to one country’s history, put the rate for a 65-year-old couple accepting a 5% chance of ruin at 2.31%[5].
These lower figures do not mean 4% will fail. They mean a single country’s past, however long, is a small sample, and the US sample was a favourable one. For a household that lives in the UK, spends in pounds and holds a mix of US and UK assets, four further differences apply:
- Currency. The studies measure a US investor’s returns in dollars against US inflation. A sterling spender with dollar assets has a third source of variability: the exchange rate.
- Two tax systems. The same gross withdrawal can leave very different net income depending on which account it comes from and which country taxes it.
- Longer horizons. Retiring at 60 and planning to 95 is 35 years, beyond the standard 30. ONS cohort estimates give a 65-year-old in 2023 an average of about 19.8 more years (men) or 22.5 (women)[11], and a plan needs to cover the longer-lived partner, not the average.
- Other income. State pensions on both sides can cover much of a household’s essential spending. The full new UK State Pension is £241.30 a week in 2026/27[13], about £25,100 a year for a couple who both receive it in full (our arithmetic). Social Security rose by 2.8% for 2026[17]. The portfolio then only has to fund the gap.
A withdrawal rate works as a starting point for a conversation, not as a plan. See our guide to Social Security and the UK State Pension for the income side.
What is sequence-of-returns risk?
It is the risk that poor returns arrive early in retirement, while you are withdrawing, so that you sell more units at low prices and the portfolio never recovers, even if average returns over the whole retirement turn out fine. The same average return can produce very different outcomes depending on its order. Before retirement the order barely matters for a lump sum left to grow. Once withdrawals start, it matters a great deal.
That is why the years just before and after retirement carry the most risk. The worst periods in Bengen’s data were those in which a retiree met a bear market and high inflation in the first decade[1]. The worked example below shows the effect for a fictional couple: the same 7% average return, in three different orders, makes their money last to 79, 85 or 90.
The usual responses are to keep one to several years of planned withdrawals in lower-volatility assets, to be ready to spend less after a fall, and to delay large discretionary spending until the early years have passed. Each has costs: cash earns less over time, and cutting spending may be unwelcome. Which, if any, suits you depends on your circumstances.
Monte Carlo or historical replay: which should I trust?
Both, for different reasons. Neither predicts the future. A Monte Carlo simulation runs a plan through hundreds or thousands of randomly generated market histories and reports how many end with money left, together with the spread of outcomes. It can produce sequences worse than anything in the historical record, but it depends on its assumptions about average returns, volatility and how assets move together. A historical replay puts the plan through real past sequences, such as retiring into 1973 or 2000. It is concrete and easy to explain, but it contains only the history that actually happened, which for any one country is a limited number of independent 30-year periods.
| Monte Carlo simulation | Historical replay | |
|---|---|---|
| What it shows | A range of outcomes and the share of simulations in which money lasts | What would have happened in specific real periods |
| Strengths | Many scenarios, including some worse than history; a range rather than one line | Real inflation, real crashes, real recoveries; easy to relate to |
| Limits | Depends on assumed returns and volatility; may miss how markets actually behave in crises | A small sample; the future need not resemble the past |
| Best used for | Judging how much margin a plan has | Testing a plan against named stress events and sequence risk |
A Monte Carlo “success rate” needs careful reading. A 67% result does not mean a one-in-three chance of destitution. It means that in about a third of simulated histories the plan as written, with no change in behaviour, falls short of its aims at some point. Real households can adjust along the way, and the result depends on the return and volatility assumptions used. The size and timing of any shortfall matter as much as the percentage, and a plan showing 100% may be one that is spending less than it could.
What are spending guardrails?
Guardrails are pre-agreed rules that raise or cut spending when the portfolio moves outside a set range, so you adjust gradually rather than in a crisis. They were set out by Jonathan Guyton and William Klinger in 2006. Their “capital preservation” rule cuts that year’s withdrawal by 10% if the current withdrawal rate has risen more than 20% above the starting rate. Their “prosperity” rule raises it by 10% if the rate has fallen more than 20% below. Combined with a rule that skips the inflation increase after a year of negative returns, they found starting rates of 5.2–5.6% sustainable at 99% confidence over 40 years for portfolios with at least 65% in shares[6].
The trade-off is real. A higher starting income comes with the possibility of cuts, sometimes several in a row. Guardrails suit households whose essential spending is largely covered by secure income such as state pensions or defined benefit schemes, so that cuts fall on travel and treats rather than on bills. They suit households with little secure income much less. Morningstar’s 2026 research likewise found that flexible approaches supported higher starting withdrawals than fixed real spending[9].
Guardrails are a behavioural commitment as much as a calculation. Agree the triggers and the size of any cut in advance, while markets are calm, and review them each year. Whether they suit you depends on your circumstances and attitude to risk, so take advice.
Will my spending stay the same throughout retirement?
Probably not. The classic studies assume spending rises exactly with inflation for 30 years. Real households tend to spend more in the early, active years, less in the middle years, and sometimes more again late in life when care is needed. David Blanchett’s 2014 analysis of US retiree data found real spending fell by about 1% a year on average. He called the shape the “retirement spending smile”: younger retirees spend on travel and leisure, and older retirees face higher medical costs[7]. His conclusion was that assuming spending rises with inflation every year can overstate how much a retiree needs[7].
UK benchmarks help set the levels. The Retirement Living Standards, compiled by Loughborough University for the Pensions and Lifetime Savings Association and published in June 2025, put the annual cost of a “moderate” retirement at £31,700 for one person and £43,900 for two, and a “comfortable” one at £43,900 and £60,600. These figures are after tax and exclude housing costs[12]. Many cross-border families also budget for regular transatlantic travel, which none of these benchmarks includes.
Planning in phases (active, slower and later-life) with a separate allowance for possible care costs usually gives a more realistic picture than a single flat line. It also gives a more cautious one where care is a real possibility.
How do currency and tax change the answer?
Currency. For a household spending in sterling and holding dollar assets, the risk is sterling strengthening. Each dollar withdrawn then buys fewer pounds, and the effect is largest if it happens early in retirement, in the same way as sequence risk. A fall in sterling helps such a household. Keeping enough of the portfolio in the currency you spend, while recognising that UK and US share markets and the exchange rate all move together in complex ways, is part of setting a sustainable spending level. Our guide to GBP/USD currency risk in retirement tests this against historical episodes.
Tax. Withdrawal rates are gross, but what you can spend is net. For a UK resident, 25% of a UK pension can usually be taken tax-free, up to a £268,275 lump sum allowance, and the rest is taxed as income[14]. The personal allowance is £12,570, with basic-rate tax at 20% to £50,270 and higher rate at 40% above that. The allowance is withdrawn between £100,000 and £125,140[15]. ISA income and gains are free of UK tax but remain taxable in the US for a US citizen. Traditional IRA and 401(k) withdrawals become compulsory from 73 under current rules (75 for those born in 1960 or later)[16], and US citizens also face US tax, reduced by credits for UK tax. The order in which accounts are drawn can change a household’s lifetime tax bill considerably. See our guides on how both tax systems apply and ISAs and PFIC rules. Tax rules change and depend on individual circumstances.
Worked example: sustainable spending, Monte Carlo and sequence risk (computed in Edale’s planner)
Fictional household, for illustration only. Alex (50) and Sam (48), UK residents, earn about £94,000 and £38,000 and plan to retire at 60. They have £430,000 in pensions, an ISA, a UK general investment account, a US brokerage account and a £650,000 home. All values are in today’s money. The results come from Edale’s planner on stated assumptions and are not guaranteed. Past or simulated returns are not a guide to future returns.
| Measure | Result |
|---|---|
| Monte Carlo: share of 300 simulated market histories in which the plan stays whole | 66.7% |
| Sustainable spending solver: extra annual spending the plan could support and still be met | about £10,900 a year |
| Usable money at age 85 (central projection) | about £1.14m |
The two headline results can look contradictory, but they answer different questions. The sustainable spending solver finds the most the couple could spend each year and still meet their plan on its stated growth, inflation and tax assumptions, and shows the working. Here that is about £10,900 a year more than they intend. The Monte Carlo result tests the plan at the spending they intend, not the higher figure, and shows how often it stays whole when market returns vary: in about two out of three of 300 simulated histories. Spending more would lower that share. Read together, the results describe a plan with headroom on its central assumptions that is also sensitive to poor markets. Neither result is a recommendation to spend more or less; how to treat the headroom, and whether measures such as agreed guardrails suit them, is a decision for the household to make with advice.
Show the figures
| Measure | Result |
|---|---|
| Simulated histories | 300 |
| Share in which the plan stays whole | 66.7% |
| Extra sustainable spending on central assumptions | £10,900 a year |
The range chart shows total wealth across the 300 histories: the median, and the 10th and 90th percentiles. The band widens with every year after retirement. This is why a single projected line gives false precision.
Sequence risk. To isolate the effect of order, the planner reran the same couple spending 40% more. It applied ten-year return sequences averaging 7% a year, in a good-years-first order and a bad-years-first order, starting at retirement, with 7% a year otherwise. The average is identical in every case. Only the order changes.
| Return sequence (same 7% average) | Money lasts to age |
|---|---|
| Good years first | 90 |
| Steady 7% every year | 85 |
| Bad years first | 79 |
Show the figures
| Return order in the first ten years | Money lasts to age |
|---|---|
| Good years first | 90 |
| Steady 7% a year | 85 |
| Bad years first | 79 |
An eleven-year gap, from 79 to 90, comes purely from the order of returns in the first decade of retirement. At this higher level of spending, poor early returns force larger sales of assets at low prices, and the portfolio cannot catch up later. Tax adds to the effect, because selling more also means realising more taxable income or gains.
How the Edale Wealth Hub helps
A spending level is only as good as the plan behind it. Edale’s planner lets an adviser and client test one plan in several independent ways, across both countries.
- Sustainable spending solver. It finds the most a household can spend each year and still meet its plan, and shows the proof. Spending can be set in phases, with care costs modelled separately.
- Monte Carlo and real history. The plan runs through hundreds of simulated market histories, reported as a range and as the probability that money lasts. It is also replayed through real market history, including the same average return in best and worst order to show sequence risk.
- Currency episodes. Twelve named GBP/USD and GBP/EUR episodes back to 1980 can be moved along the timeline, for example placing a period of sterling strength at the start of retirement.
- Longevity. Plans can be tested against ONS life and healthy-life expectancy, not a single assumed age. ONS period figures for 2022–2024 give 18.7 more years at 65 for men and 21.2 for women, as averages[10].
- Tax in both countries. UK income tax, National Insurance and CGT, US federal tax, NIIT and the interaction between them for US persons, so spending is judged on what the household keeps. Alternative withdrawal orders can be compared.
- Guided what-ifs. Retiring earlier or later, a market crash, first death, downsizing, gifts, or moving between the UK and the US, compared side by side with the base plan.
The planner models; it does not predict. Results depend on the assumptions entered, and they change as markets, tax rules and circumstances change, so plans should be reviewed regularly. US estate tax is not modelled. Tax outputs are summaries to share with your accountant or tax preparer, not returns. Clients can read how the method works in plain English in the in-app Philosophy & Technical guide. To discuss your own plan, arrange a conversation or read more about the cashflow planner.
Frequently asked questions
Is the 4% rule safe for UK retirees?
It is less reliable than its reputation suggests. It comes from US data. In Wade Pfau’s study of 17 countries, 4% would have held in only four, and the UK was not one of them. A 2025 study of 38 developed countries put a comparable rate at 2.31%. For a UK or cross-border household, 4% is a starting point to test, not a safe figure.
What is a sustainable withdrawal rate for a 35-year retirement?
Lower than for 30 years, all else being equal, because the portfolio must last longer. Most classic studies used 30-year horizons, so someone retiring at 60 should not assume those rates carry over. Secure income from state pensions, flexibility to cut spending after poor markets, and falling real spending in later years can all support a higher starting level.
What does a Monte Carlo success rate of 67% mean?
It means that in about two-thirds of simulated market histories the plan as written, with no changes along the way, meets its aims. In the other third it falls short at some point. It is not a forecast of a one-in-three chance of running out: real households can adjust spending, and the figure depends on the assumptions used. How large and how late any shortfall is matters just as much.
What is the difference between Monte Carlo and historical back-testing?
Monte Carlo generates many random market histories from assumed returns and volatility, including some worse than anything seen. Historical back-testing replays actual past periods, such as retiring into the 1970s. Monte Carlo shows the range of outcomes; history shows specific real stresses. Both are useful, and neither predicts the future.
How do guardrails work in retirement?
You set a starting withdrawal and agree in advance to cut it, by 10% in the original rules, if poor markets push your withdrawal rate more than 20% above where it started, and to raise it if the reverse happens. Guyton and Klinger’s 2006 research found starting rates of 5.2–5.6% sustainable with such rules, at the cost of possible cuts.
Does exchange-rate risk affect how much I can spend?
Yes. If you spend in pounds and hold dollar assets, a strengthening pound reduces what each dollar withdrawal buys, and if that happens early in retirement it compounds like sequence risk. A weaker pound helps. Matching part of the portfolio to the currency you spend reduces, but does not remove, this risk.
Should I spend less as I get older?
Many retirees do so naturally. US research by David Blanchett found real spending fell by about 1% a year on average, with a “smile” shape as care and medical costs rise late in life. Planning spending in phases, with a separate allowance for care, is usually more realistic than a single inflation-linked figure. What suits you depends on your circumstances.
Sources
- William P. Bengen — “Determining Withdrawal Rates Using Historical Data”, Journal of Financial Planning, October 1994 (FPA reprint) (accessed September 2026)
- Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz — “Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable”, AAII Journal, February 1998 (accessed September 2026)
- Wade D. Pfau — “An International Perspective on Safe Withdrawal Rates: The Demise of the 4 Percent Rule?”, Journal of Financial Planning, December 2010 (accessed September 2026)
- Wade D. Pfau — “An International Perspective on Safe Withdrawal Rates”, Advisor Perspectives (accessed September 2026)
- Aizhan Anarkulova, Scott Cederburg, Michael S. O’Doherty and Richard Sias — “The safe withdrawal rate: evidence from a broad sample of developed markets”, Journal of Pension Economics & Finance, 2025 (accessed September 2026)
- Jonathan T. Guyton and William J. Klinger — “Decision Rules and Maximum Initial Withdrawal Rates”, Journal of Financial Planning, March 2006 (accessed September 2026)
- David Blanchett — “Exploring the Retirement Consumption Puzzle”, Journal of Financial Planning, May 2014 (accessed September 2026)
- Christopher M. Duquette — “Revisiting William Bengen’s SAFEMAX Portfolio Withdrawal Rate”, Journal of Financial Planning, November 2023 (accessed September 2026)
- Morningstar — What’s a safe retirement withdrawal rate for 2026? (accessed September 2026)
- ONS — National life tables: life expectancy in the UK, 2022 to 2024 (accessed September 2026)
- ONS — Past and projected period and cohort life tables: 2022-based, UK, 1981 to 2072 (accessed September 2026)
- Pensions UK (formerly PLSA) — Latest Retirement Living Standards, June 2025 (accessed September 2026)
- GOV.UK — The new State Pension: what you’ll get (accessed September 2026)
- GOV.UK — Tax when you get a pension: what’s tax-free (accessed September 2026)
- GOV.UK — Income Tax rates and Personal Allowances (accessed September 2026)
- IRS — Retirement plan and IRA required minimum distributions FAQs (accessed September 2026)
- Social Security Administration — 2026 Cost-of-Living Adjustment fact sheet (accessed September 2026)