Why the 4% rule breaks outside the US

The 4% rule is the most repeated piece of retirement advice in the English-speaking world. A retiree with a balanced portfolio can withdraw 4% of their starting balance, adjust for inflation each year, and expect the money to last thirty years. It is the foundation of most FIRE calculations, most retirement planning software, and most financial journalism on the subject.

It is also a rule derived from a single country's historical returns, published in 1994, updated occasionally since then, and treated as universal despite being nothing of the sort.

When researchers extend the underlying methodology to other developed markets, the 4% rule fails. Not narrowly, not in edge cases, but catastrophically and repeatedly. A UK retiree using the 4% rule historically had a 3.05% SAFEMAX. An Italian retiree had a 4% rule that failed in 80.5% of historical periods. A Japanese retiree's portfolio, if invested in domestic equities over the thirty years from 1990 to 2019, lost 21% in real terms — not a bad sequence around an average, an actual negative return across the entire retirement window.

This post explains why the 4% rule is a US-market artefact, what the best international research actually shows, and what a retiree outside the US should take from it.

Where the 4% rule came from

William Bengen published the original SAFEMAX research in 1994 in the Journal of Financial Planning. His method was straightforward: take every historical 30-year period in US equity and bond market data, apply a constant-dollar withdrawal starting at 4% of initial balance with annual inflation adjustments, and find the highest withdrawal rate at which no historical period produced portfolio depletion. His answer was 4.15%, which the financial press immediately rounded to 4%, and the 4% rule was born.

The Trinity Study in 1998 extended the research to multiple asset allocations and success-rate thresholds, generally confirming Bengen's findings. Bengen himself updated the figure to 4.5% in 2006 and 4.7% in a 2025 book, reflecting longer data series and refinements to the methodology. Morningstar's 2026 State of Retirement Income report puts the current base-case starting rate at 3.9% for a 30-year horizon, reflecting forward-looking capital markets assumptions rather than pure historical back-testing.

All of these figures share one feature: they are derived from US market data. US equities from 1926 to 2019 produced a real equity premium of roughly 6% a year, one of the strongest records in any developed market over any comparable period. US bonds offered meaningful real yields during most of the back-testing window. US inflation, with the exception of the 1970s, was relatively contained. These conditions produced the particular sequence of returns that made 4% survivable in the worst historical US cases.

None of those conditions are guaranteed to hold outside the US, and in many cases they demonstrably did not.

Pfau's international study

Wade Pfau's 2010 paper in the Journal of Financial Planning was the first systematic attempt to apply Bengen's methodology internationally. Using Dimson-Marsh-Staunton (DMS) data for 17 developed countries from 1900 to 2008, Pfau calculated each country's SAFEMAX — the highest withdrawal rate that would have survived every historical 30-year period in that market.

The results were striking. For a 50/50 stocks/bonds portfolio, only five countries exceeded 4%: Sweden, Canada, New Zealand, Denmark, and the US. Every other country in the sample — France, Germany, Italy, Japan, Spain, Belgium, Switzerland, Ireland, the UK, Finland, Norway, Australia, South Africa — had historical SAFEMAX values below 4%.

The UK came in at 3.05% for 50/50 stocks and bonds, or 3.43% for 50/50 stocks and bills. Italy's 4% rule failed in 80.5% of rolling 30-year periods — meaning an Italian retiree following standard advice had worse than a four-in-five chance of running out of money. Japan and Germany both came in below 3%.

Pfau's paper was titled, with academic restraint, "An International Perspective on Safe Withdrawal Rates: The Demise of the 4 Percent Rule?" The question mark was generous.

The Anarkulova study and the 2.31% number

The most important recent work in this field is Anarkulova, Cederburg, O'Doherty, and Sias (2025), published in the Journal of Pension Economics and Finance. The authors assembled a dataset of approximately 2,500 years of asset-class returns across 38 developed countries from 1890 to 2019, making it by some distance the largest dataset ever used for withdrawal-rate research.

Their methodology differs from Bengen's in three important ways. First, they use block bootstrap resampling rather than rolling historical periods, which means their simulated retirements can draw from any country's market conditions in any combination, not just the specific sequence that happened in one country. Second, they use actual mortality tables rather than a fixed 30-year horizon, because real retirees die at varying ages and the 30-year convention understates longevity risk for healthy couples. Third, their sample mitigates two biases that inflate US-only studies: survivor bias (countries whose markets failed or were suppressed are included, not excluded) and what they call "easy data bias" (the tendency for high-quality data to come disproportionately from markets that performed well).

Their headline finding: a 65-year-old couple willing to accept a 5% chance of financial ruin over their remaining joint lifetime can withdraw only 2.31% per year from a diversified portfolio. Not 4%. Not 3%. Under 2.5%.

In the same paper, they note that the conventional 4% rule applied to a traditional 60% domestic stocks / 40% bonds portfolio produces a 17.4% chance of complete asset depletion before death. Roughly one in six retirees following the most-recommended retirement plan in the English-speaking world would run out of money, according to the broadest international dataset available.

The popular shorthand that has emerged from this research, particularly through Ben Felix's work at PWL Capital, is "the 2.7% rule" — a rounded figure that incorporates the benefits of global diversification rather than pure domestic exposure. The precise number depends on inputs, but the direction of the correction is unambiguous: the honest safe withdrawal rate is lower than 4%, and meaningfully so.

Why non-US markets perform worse

The question is not whether international SWRs are lower. That is settled. The question is why.

Four reasons matter.

Equity premium variance. The US delivered roughly 6% real equity premium over most of the 20th century. Many developed markets delivered considerably less. Japan's equity market produced a -21% real return over the thirty years from 1990 to 2019, meaning a Japanese retiree who bought a domestic equity index at the peak of the bubble and held for their entire retirement ended with less purchasing power than they started. Germany and Italy experienced decade-long periods of effectively zero real equity returns. The long-run US equity premium, treated as universal by most FIRE planning, is actually an upper-bound case.

Inflation shocks. Germany's hyperinflation in 1923, Italy's post-war inflation, the UK's inflation spike in the mid-1970s (peaking above 25% annually), and France's post-war devaluations all produced periods where bond holdings and conservative portfolios lost most of their real value in a few years. The US had its 1970s inflation period, but nothing comparable to the continental European experience. Bond-heavy portfolios that survive US inflation history do not necessarily survive European inflation history.

Wars and political disruption. The DMS dataset and Anarkulova's data both include country-periods that encompass the two world wars, the breakup of markets, capital controls, currency revaluations, and occasional market closures. These are not tail risks invented for academic stress-tests. They are documented historical events that destroyed retirement portfolios for people who had followed conservative advice for decades beforehand. The US, protected by geography and luck, experienced comparatively few such shocks. Most of the rest of the developed world experienced several each.

Equity premium survivorship at the dataset level. A subtle but important point: the indexes most commonly used in FIRE calculations are survivor-biased at the market level. The S&P 500 only contains companies that survived. The MSCI World Index only contains markets that continued to operate as investable entities. Countries whose markets collapsed or were nationalised get dropped from the data. The Anarkulova dataset was specifically designed to mitigate this by including country-periods throughout their developed history, not just the periods when data was easy to obtain.

Together, these factors mean that a retiree outside the US who plans to 4% based on US data is implicitly assuming that their country's future market returns will match the US past. This is not a forecast. It is an assumption imported from a foreign dataset without adjustment.

What this means practically

The obvious takeaway is to use a lower withdrawal rate. The less obvious takeaway is that the 4% rule's fragility outside the US is not fixed by simply knocking a percentage point off. Several structural responses matter more than a flat rate change.

Global diversification. A UK retiree holding 100% FTSE 100 faces the UK's SAFEMAX of around 3%. The same retiree holding a global equity index faces the blended outcome of all developed markets, which is worse than the US alone but better than any single weak-performing country. Anarkulova's research explicitly shows that international diversification meaningfully improves outcomes, which is why the PWL Capital "2.7% rule" is framed as a globally-diversified number, not a single-country number.

Variable withdrawal rules. Guyton-Klinger guardrails and similar approaches that reduce withdrawals after bad years have an even larger effect outside the US than within it, because non-US sequences contain more severe bad periods that need to be responded to. A fixed 4% rule is brittle; a rule that adjusts to market conditions survives sequences that a fixed rule cannot.

Longer planning horizons. The 30-year convention in the original Bengen work reflected a traditional 65-year-old retiree. An early retiree at 50 has a 45-year horizon; a couple at 50 has potentially longer if one partner is in good health. Anarkulova's use of actual mortality tables, not a 30-year cap, makes their analysis particularly relevant to the early retirement case — where the longer the horizon, the more a bad international sequence has time to damage the plan.

Pension system offsets. The countries with the worst historical SWRs — France, Germany, Italy — also tend to have more generous public pension systems than the US or UK. A French retiree receiving a state pension plus employer pension may need the private portfolio to cover a smaller proportion of total retirement income, which means a lower SWR is tolerable because the stakes are lower. This does not invalidate the SWR research. It means the practical implications vary by country in ways that simple 4% rule applications miss.

The honest conclusion

The 4% rule is a useful piece of heuristic anchoring for someone in the US with a long US-equity-heavy investing history and a 30-year horizon. For anyone else, it is a starting point for more careful thinking, not an answer.

The Anarkulova research, which is the current state of the art, suggests that a globally-diversified retiree should think of 2.5% to 3% as the honest base rate, with variable spending rules and real-world flexibility providing the cushion between that rate and a workable retirement income. This is not a reason to abandon retirement planning. It is a reason to stop importing US-specific numbers into non-US plans without correction.

Where Endute fits

Endute's FIRE simulator supports country-specific capital markets assumptions rather than defaulting to US inputs. A UK user modelling a plan against UK equity and bond history produces UK-relevant SWRs. A user who wants to test their plan against the international distribution — as Anarkulova's research suggests everyone should — can do that too. The difference between a plan that works at 4% on US data and a plan that works at 2.7% on international data is usually the difference between "I can retire now" and "I need another five years of accumulation." Better to find out before you quit.

Sources

• Bengen, W.P. (1994). "Determining Withdrawal Rates Using Historical Data." Journal of Financial Planning 7(4):171-180.
• Cooley, Hubbard, Walz (1998). "Retirement Savings: Choosing a Withdrawal Rate That is Sustainable." AAII Journal. The Trinity Study.
• Pfau, W.D. (2010). "An International Perspective on Safe Withdrawal Rates: The Demise of the 4 Percent Rule?" Journal of Financial Planning, December 2010. 17 developed markets, DMS data 1900-2008.
• Anarkulova, A., Cederburg, S., O'Doherty, M.S., and Sias, R. (2025). "The safe withdrawal rate: evidence from a broad sample of developed markets." Journal of Pension Economics and Finance, Vol 24 Issue 3, pp. 464-500. 38 developed countries, 1890-2019, approximately 2,500 years of data.
• Morningstar (2025). The State of Retirement Income 2025. 3.9% base case for 30-year horizon, 3.3% for 40-year horizon, published December 2025.
• Dimson, Marsh, Staunton (various). Global Investment Returns Yearbook. Credit Suisse/UBS. Source data used in Pfau (2010) and related research.
• Felix, B. (PWL Capital). Popularisation of the Anarkulova findings as "the 2.7% rule." See Common Sense Investing YouTube channel and PWL Capital research notes.

This is the fifth post in a cluster on early retirement planning across UK, EU, and US markets. The preceding post explains sequence of returns risk in general terms; this one applies that framing specifically to international market data. The next post in the series examines different FIRE "flavours" — Coast, Barista, Lean, and full — and how the SWR discussion applies to each.