Skip to main content
Investing ••13 min read•

International Stock Allocation for a 50-Year FIRE Portfolio

International stock allocation for FIRE: why the 4% rule is US survivorship bias, and what global 3 to 3.5% withdrawal rates do to your number.

International stock allocation spreads a FIRE portfolio's equity risk across global markets instead of one country's history.

The 4% rule has a country-of-origin problem. The Trinity research that produced it, the 25x shorthand you plan around, and the default portfolios in cFIREsim all backtest one market, the United States, which finished among the two or three best-performing equity markets of the twentieth century. Every canonical FIRE number is a reading taken near the top of the global distribution, and the international stock allocation question is really asking whether your plan survives contact with average markets instead of exceptional ones.

Stay in the loop.

Get the latest posts and exclusive content delivered to your inbox.

Join 16 readers. No spam. Unsubscribe in one click, anytime.

Run the trade in both directions and the stakes get concrete. Globalizing the portfolio has historically meant accepting a sustainable withdrawal rate closer to 3 to 3.5%, which reprices a $48,000-a-year lifestyle from $1.20M to as much as $1.60M. Staying 100% US dodges that haircut but makes an uncompensated single-country bet that a 50-year retirement cannot hedge. The resolution is a phase rule: at least 20% international through accumulation, gliding toward market-cap weight across the decade around your FIRE date. The rest of this piece builds that rule and prices it.

Why Every FIRE Number Is Calibrated on One Market

In the Trinity study, Cooley, Hubbard, and Walz tested withdrawal rates on US stock and bond returns from 1926 onward and found that 4% of the starting balance, inflation-adjusted, survived 30-year retirements in the large majority of stock-heavy cycles. The 25x rule is 1 divided by 0.04 with the citation stripped off. cFIREsim, the movement's default stress tester, runs on US market history because that is the deepest dataset available. None of these tools is wrong. All of them inherit the same sample, and the sample inherits the winner.

Survivorship bias, in one tight paragraph: we mine guidance from markets that survived and compounded, and the biggest winner writes the most confident backtests. An investor standing in 1900 had no way to know the US would out-compound nearly everyone. The sober consensus pick back then was Britain, the world's dominant financial market, and it materially lagged the US over the following century.

The size of the American win, per the Dimson-Marsh-Staunton cross-country dataset: US equities compounded at roughly 6.5% real a year since 1900 against roughly 4.5% for the world ex-US. Two points a year sounds like noise. Compounded across twelve decades it is an order of magnitude, on the order of $2,000 versus $200 in terminal real wealth per 1900 dollar. Your safe withdrawal rate was calibrated inside the $2,000 column. Call that the 25x rule's survivorship bias, quantified: the number works partly because it was estimated in the luckiest large market on record.

What Cross-Country Return Data Shows

International diversification lowers the odds that a 4% withdrawal plan depends on a single market's returns.

Start with the question a 1900 investor was actually entitled to ask, because it dictates the method. Nobody standing in that year could know which market would win the twentieth century, so the only honest backtest averages every national draw instead of crowning the one that already won. Reading the eventual winner's history as the base case is a victory lap, not a backtest. That is what a US-only sample is, structurally, every time you press run.

The Dimson-Marsh-Staunton dataset is the instrument that makes a fair sample computable: annual total returns, dividends and inflation included, for more than twenty markets back to 1900, first assembled in Triumph of the Optimists and maintained since in the annual Global Investment Returns Yearbook. Every country gets one draw, and the distribution across draws is the finding.

Wade Pfau's 2010 international analysis ran exactly that computation, applying Trinity-style withdrawal testing to seventeen developed markets from 1900 to 2009. The result is the payoff of the whole fair-sample argument: 4% real withdrawals failed in the majority of those markets over 30-year windows, and only a handful, Canada and Sweden alongside the US itself, reliably cleared the bar with stock-heavy mixes. If you have only ever backtested the United States, this is where you find out you have been reading one lucky draw as if it were the whole distribution. Globally diversified portfolios across the same sample sustained withdrawals closer to roughly 3 to 3.5%. So the direct answer to whether the 4% rule works outside the US is mostly no, and that is exactly why the global number prices lower.

Bessembinder's global study supplies the mechanism behind that dispersion: stock returns are lottery-shaped. A few percent of listed firms, concentrated in a handful of countries, accounts for effectively all net global stock market wealth creation above Treasury bills, while a large fraction of individual stocks underperformed bills over their lifetimes. The skew repeats one level up, across countries rather than firms. The US outcome since 1900 was one draw from that lottery, not a property markets have. A fair sample cannot tell you the US will revert, only that the variance was always there, and international diversification is the refusal to stake a 50-year retirement on a single ticket.

Repricing Your FI Number Against Global Withdrawal Rates

A withdrawal rate is a target multiple turned upside down, so the global evidence moves your FI number before it touches your holdings. Stretching the horizon from Trinity's 30 years to a 50-year FIRE retirement pushes the defensible rate toward the bottom of any historical range, in US data and global data alike.

Assumed safe withdrawal rateTarget multipleFI number at $48,000/yrCost vs 25x
4.0%25.0x$1.20Mbaseline
3.5%28.6x$1.37M+$171k
3.25%30.8x$1.48M+$277k
3.0%33.3x$1.60M+$400k

The grid is a repricing of assumptions, not a forecast. Nothing above predicts what any market will deliver; it prices what history says survived.

Three honest ways to use it:

  1. Keep 25x and accept visibly higher failure odds once you grade outcomes against global rather than US-only history.
  2. Adopt 3.25% and save toward 30.8x, about $1.48M, as the robustness pick for a 50-year plan holding global equities.
  3. Keep 4.0% and treat the $400k delta as the explicit price of your conviction that the US repeats.

The middle row is the defensible default for most globally diversified 50-year plans. What the grid forbids is quoting 4% as a universal constant while your portfolio quietly carries one country's equity risk.

Why 100% US Stocks Is the Bigger Concentration Bet

Home bias is the one position in portfolio construction that carries no risk premium. Single-country concentration is idiosyncratic risk: it widens the range of outcomes without raising expected return, because the market does not pay you for risk you can diversify away. A 50-year retirement spans at least one regime change, monetary, geopolitical, or technological, and the cross-country dispersion above is what different regimes did to different home markets.

Japan is the canonical case, and it is the scenario a US-only backtest cannot see because the American sample contains nothing like it. The Nikkei peaked at 38,915 in December 1989 and took 34 years to reclaim that high, in February 2024. A Japanese investor who retired at the peak with a domestic portfolio spent three decades withdrawing from an equity engine with zero price recovery. Selling 4% a year into that drawdown compounds against principal, and the failure is slow, structural, and invisible to any tool whose dataset is the United States. Nothing about that outcome required a global catastrophe, only a portfolio that mistook one country's index for the whole market. A globally weighted portfolio, holding the US 1990s and 2010s alongside Japan's lost decades, behaves entirely differently.

So the question of whether 100% US stocks are risky for FIRE resolves to this: it is a bet that the next half century repeats the best case in a twenty-plus-country sample. Some bets pay. You just cannot call it the conservative option.

Sizing International Stock Allocation by FIRE Phase

A FIRE portfolio allocation phases international holdings from a 20 percent floor toward global market-cap weight.

How much international stock to hold for early retirement depends on when losses actually hurt, and that changes with phase. Kitces's research on sequence of returns risk shows outcomes are disproportionately determined by returns in roughly the first decade of withdrawals: strong early years carry most later paths, weak early years sink plans that average out fine on paper. In accumulation the asymmetry runs the other way. Underperformance mostly reprices future contributions, because you keep buying at lower prices. Losses in the first withdrawal decade have no such offset.

That asymmetry sets the rule:

  1. Floor in accumulation. Hold at least 20% in ex-US equities, the low end of common advisory ranges, no matter how dominant US returns look.
  2. Ramp through the final decade. Starting ten years before your FIRE date, add roughly two percentage points a year.
  3. Cap at market-cap weight. Stop at the global cap for a market-cap-weighted global portfolio, recently the mid-to-high 30s percent ex-US; a fund like Vanguard Total World Stock holds that weight in one ticker.
  4. Hold the cap through the first withdrawal decade, then rebalance normally thereafter.
international_weight(y) = max(20%, min(cap, 20% + 2% × max(0, 10 − y)))

where y is years to your FIRE date and cap is the current market-cap weight of ex-US equities.

Worked example: you are twelve years from FIRE with a $600,000 all-US portfolio. Today you move to 20% international, about $120,000. You hold that floor for two years, then the ramp begins: 24% at eight years out, 28% at six, 32% at four, 36% at two, and roughly cap weight, high 30s percent, on the day you quit. Your international allocation during retirement withdrawals then stays at cap through the sequence-risk window, precisely when global breadth is worth the most.

How to Backtest Both Portfolios in cFIREsim

You do not have to trust anyone's summary, including mine. The cFIREsim tutorial covers the mechanics; this is the comparison worth running:

  1. Set a 50-year horizon and your real annual spending, say $48,000.
  2. Run a baseline 100% US equity portfolio at that spending level.
  3. Duplicate it and split the equity sleeve into a US allocation and an international allocation, using the tool's international series to approximate ex-US exposure.
  4. Compare success rates and worst-cycle ending balances across cycles, never a single output.
  5. Rerun with fees loaded and different retirement start years to see how much the delta moves.

Expect a specific direction, stated without invented precision: on US-dominated historical data, adding international often trims the headline success rate modestly, because the sample crowns the one market you are diluting. You are measuring US versus international stock returns inside the tested window, and the window flatters the US. That is the sample talking, not diversification failing. A small success-rate cost means the robustness was cheap. A large one in your runs means your FI number, not the insurance, is the aggressive input.

Caveats to hold: the non-US series are shorter, so fewer overlapping cycles exist; currency effects arrive unhedged; international expense ratios run a touch higher. None of this breaks the test. It widens the error bars, which is the honest condition of the whole question.

The Honest Case Against International Exposure

The strongest objections deserve steelmanning, because the phase rule survives them only if you have actually heard them.

Bogle's argument against international funds

Jack Bogle contended that US investors can reasonably skip international funds because large US multinationals already earn a large share of revenue abroad; owning the S&P 500 hands you Apple's and Microsoft's global sales. That is genuine exposure to foreign earnings. It is not the same as owning foreign markets' valuations, currencies, and sector mix, but it is a real, low-cost partial hedge.

The orthodox range is wide

Major fund families have commonly recommended 20 to 40% international, while market-cap weight implies roughly 35 to 40%. Even the establishment brackets a two-fold range, which should temper anyone's certainty about a single right answer.

Valuations point the other way

Vanguard's capital markets model has recently projected similar or higher expected returns for ex-US equities over the next decade on starting-valuation grounds. Useful context, but long-horizon forecasts carry wide error bands, and the phase rule deliberately does not depend on them. It is robustness engineering, not return prediction.

Currency and behavior

Unhedged international adds currency volatility to year-to-year returns, and the 2010s asked US investors to rebalance into ex-US while it trailed by several points a year for a decade. Plenty of people quit at the trough. If you would capitulate at the cap weight, the rule fails you at the behavioral layer, and a lower weight you will actually hold beats a theoretical one you will not.

What would falsify the phase rule: credible evidence that country-level concentration is a compensated risk premium, structural evidence that US and global outcomes have converged so tightly that the distinction no longer matters, or your own revealed inability to hold the floor through a lost decade. Absent those, the asymmetry stands.

The Decision Rule in One Table

The full FIRE portfolio allocation decision compresses to four rows:

PhaseInternational weightPlanning withdrawal rateTarget multiple
Accumulation, 10+ years out20% floor3.5% to 4.0%25.0x to 28.6x
Final decade before FIREramp 20% toward cap (roughly 37 to 40%)3.25% to 3.5%28.6x to 30.8x
First withdrawal decadehold at cap3.0% to 3.5%28.6x to 33.3x
Later retirementrebalance to caprevisit as horizon shortensfalls as remaining years fall

Checkpoint list:

  • Recheck the cap weight yearly; market-cap weight drifts with relative returns.
  • Rerun cFIREsim whenever your FI number or spending changes, and record both portfolios' success rates.
  • Decide your withdrawal-rate band in writing before the bad decade, not during it.
  • Write the rebalancing rule down; a glide path you abandon in the first underperforming year is a random walk with extra steps.
  • If you hold all-in-one funds, verify the underlying ex-US weight matches the rule, not just the label.

The question that opened this piece was whether FIRE's canonical numbers survive average markets. On cross-country evidence 4% does not survive and neither does 25x, and the fix is a slightly bigger number, roughly 29x to 33x, plus an international stock allocation of 20% through accumulation rising toward cap weight as sequence risk approaches. The 1989 Japanese retiree is the scenario this rule exists to survive. What remains honestly uncertain: global datasets are shorter and messier than we would like, forward currency and fee effects vary, and no rule survives a saver who will not hold it. Price all of that in, and the phase rule still beats betting the best case repeats.

Stay in the loop.

Get the latest posts and exclusive content delivered to your inbox.

Join 16 readers. No spam. Unsubscribe in one click, anytime.

About the author

Dana Whitfield

Index-Fund Analyst

Dana spent a decade managing portfolios at a fiduciary firm before going independent to write for everyday investors. She breaks down asset allocation, fees, and long-term market strategy in plain English so readers can build wealth and actually sleep at night.

Related Posts