The 4% Rule Was Built for a 30-Year Retirement. Yours Might Last 50.

If you haven't already, The 4% Rule, Explained covers where the "25x expenses" shortcut comes from. This post goes deeper on the assumption doing the most hidden work in that shortcut: the length of the retirement it was actually tested against.

The 4% rule is the closest thing the FIRE community has to scripture: save 25x your annual expenses, withdraw 4% in year one, adjust for inflation every year after, and you're financially independent. It's the engine behind every "your FIRE number" calculator on the internet, including ours.

But the 4% rule was never a universal law of retirement finance. It was the answer to a very specific question: what withdrawal rate would have survived a 30-year retirement in historical U.S. market data? If your retirement is going to last 40, 50, or 60 years, which is the whole point of retiring early, you're applying a 30-year answer to a much harder problem.

This post digs into where the rule came from, what the research actually shows at different withdrawal rates and horizons, and how sequence-of-returns risk and inflation quietly do most of the damage.

Where the 4% Rule Actually Came From

In 1994, financial planner William Bengen published "Determining Withdrawal Rates Using Historical Data" in the Journal of Financial Planning.¹ Bengen tested rolling historical retirement cohorts using S&P 500 stocks, intermediate-term Treasuries, and CPI inflation, and asked: what initial withdrawal rate, adjusted annually for inflation, would have survived every 30-year period on record?

His findings, by withdrawal rate:²

Notice what's hiding in that table: 4% wasn't chosen because it lasts forever. It was chosen because its worst historical outcome cleared a 30-year bar with a few years to spare. Bengen himself later called it the "SAFEMAX", the maximum withdrawal rate that survived the single worst starting year in U.S. history.

Four years later, three Trinity University professors (Cooley, Hubbard, and Walz) reframed the question in terms of "portfolio success rates" using 1926–1995 data, testing withdrawal rates from 3% to 12% across payout periods of 15 to 30 years.³ The famous result: a 4% inflation-adjusted withdrawal from a 50/50 or 75/25 stock/bond portfolio succeeded in ~95–98% of historical 30-year periods. That's the "Trinity Study" everyone cites.

Again: 30 years. The original Trinity Study didn't even test horizons longer than that.

What Happens When the Horizon Stretches

The attached-string on the 4% rule is the expected term. Every additional year of retirement is another year the portfolio must survive bad return sequences, inflation shocks, and compounding damage. The research is consistent on this point:

Cooley, Hubbard, and Walz's follow-up work. In their 1999 study extending the analysis to longer payout periods, a 4% inflation-adjusted withdrawal from a 50/50 portfolio failed in roughly 15% of historical 35-year retirements, and the withdrawal rates that produced zero historical failures beyond 25-year horizons were often below 4%.⁴ Their 2011 update (data through 2009) confirmed the pattern: success rates decline steadily as payout periods lengthen, at every withdrawal rate.⁵

Early Retirement Now's 60-year analysis. Karsten Jeske's Safe Withdrawal Rate Series, probably the most exhaustive FIRE-specific treatment of this topic, extended the historical simulations to 50- and 60-year horizons. The headline result: a 4% withdrawal rate on a 50/50 portfolio had a ~95% success rate over 30 years but only ~65% over 60 years. The failure probability is roughly seven times higher at the longer horizon.⁶ His fail-safe conclusion for early retirees: withdrawal rates in the 3.25%–3.50% range survived even the most catastrophic historical starting points.⁷

Kitces' horizon research. Michael Kitces summarizes the academic consensus similarly: extending the horizon from 30 to 45 years reduces the historically safe withdrawal rate from ~4.1% to ~3.5%, but interestingly, it doesn't keep falling much beyond that. Around 3.5% appears to form a floor in the U.S. data, because over very long horizons compounding growth eventually outruns even conservative withdrawals if the portfolio survives the early years.⁸

Putting it together on a $1,000,000 portfolio:

The relationship is also nonlinear: each additional year of horizon raises the odds that at least one adverse sequence hits before the portfolio has had time to recover. Which brings us to the mechanism.

Sequence-of-Returns Risk: Why When Matters More Than How Much

Average returns don't kill retirements. The order of returns does. Two retirees can experience identical average returns and end up in completely different places depending on whether the bad years come early or late.

Here's a simplified example. Both retirees start with $1,000,000, withdraw $40,000 at year-end (growing 2%/year with inflation), and experience the same three annual returns, just in opposite order:

Identical returns, identical withdrawals, but Retiree A is already ~$35,000 behind after three years, because withdrawals taken from a depressed portfolio lock in losses permanently. Stretch this dynamic over a decade and the gap becomes unbridgeable: Retiree A is selling more shares at low prices every year, so the eventual recovery applies to a smaller base.

The empirical evidence on this is striking. Kitces found that the real return of equities in the first decade of retirement alone has a 0.79 correlation with the 30-year safe withdrawal rate, far more predictive than year-one returns or even full 30-year averages.⁹ Cohorts with strong first decades (12%+ real returns) could have safely withdrawn 7% or more; cohorts with negative first-decade real returns are where every historical 4% failure lives. Related Kitces Research found that no historical cohort with a first-decade average real return above 3.9% has ever failed at a 4% withdrawal rate, all failures cluster in the bottom third of first-decade outcomes.¹⁰

This is also why sequence risk concentrates early: the portfolio is largest relative to withdrawals in the first years. A 20% decline on $1,000,000 costs $200,000; the same decline on a partially-spent $400,000 portfolio costs $80,000.¹¹ For FIRE retirees, a longer horizon effectively means more chances to draw a bad opening decade, and less room to absorb it.

Inflation: The Silent Second Punch

Market crashes get the headlines, but Bengen's own data pointed elsewhere. He found the 1973–74 period was the "most devastating" event for retirees, not because of the stock decline, but because of the inflation that followed. The 1929 crash, remarkably, was less damaging to withdrawal plans because it was accompanied by deflation, which reduced the real withdrawal burden.²

The mechanism is brutal in a fixed-real-withdrawal plan: inflation forces your dollar withdrawals up at precisely the moment returns are weak. Consider a retiree taking $40,000 from $1,000,000 who hits a 1970s-style stretch of 8% annual inflation:

Within a decade, the withdrawal has doubled in nominal terms, and if the portfolio simultaneously suffered a bear market, the effective withdrawal rate on the depleted balance can climb from 4% to 8–10%, a level from which few portfolios recover. This combination, early bear market plus sustained inflation, is exactly why historical failures cluster in the mid-1960s retirement cohorts rather than appearing evenly across time. A 1966 retiree faced the 1969–70 bear, the 1973–74 crash, and the great inflation, back to back to back.

Three Real Retirements: 1966, 1982, and 2000

Everything above is abstract until you watch actual history play out. The interactive chart below simulates three real retirement cohorts using historical S&P 500 total returns, 10-year Treasury returns, and CPI inflation, each starting with a $1M portfolio, a 4% inflation-adjusted withdrawal ($40,000/year), a 60/40 annually rebalanced allocation, retirement at 55, and Social Security ($24,000/year with COLA) beginning at age 67. Click the cohort buttons to compare paths, toggle Social Security on and off, and switch between real and nominal dollars.

What the three lines show:

Retired 1966, the worst case. This retiree walked straight into the 1969–70 bear market, the 1973–74 crash, and the Great Inflation. In nominal dollars the portfolio looks like it's treading water for 15 years; in real dollars it loses roughly 70% of its purchasing power by 1981. Without Social Security, the portfolio is depleted at age 87, 32 years in, almost exactly matching Bengen's finding that 4% survives "no less than ~33 years" in the worst historical case.² Toggle Social Security on, though, and the story flips: from age 67, the $24K benefit covers 60% of the $40K real withdrawal, dropping the effective net withdrawal rate to ~1.6%, and the portfolio not only survives, it recovers into the 1980s bull market. This single toggle is the clearest illustration of why outside income is worth more than any withdrawal-rate refinement.

Retired 1982, the best case. Same rule, same portfolio, radically different fate. This retiree caught the greatest bull market in U.S. history right out of the gate: the portfolio doubles in real terms within a decade despite withdrawals, and finishes with over $12M in today's dollars by age 97, even without Social Security. This is the flip side of sequence risk that Kitces documented: with a strong first decade, the historically safe withdrawal rate wasn't 4%, it was 7%+.⁹ The 1982 retiree massively underspent, the failure mode nobody talks about.

Retired 2000, the recent stress test. This retiree absorbed the dot-com crash immediately and the Global Financial Crisis eight years later, two major bear markets inside the critical first decade. The portfolio grinds sideways in real terms for 15 years but never comes close to failing: by 2024 (age 79), the balance sits around $755K real without Social Security, or ~$1.18M real with it, actually above the starting value once benefits kick in. As Kitces noted in his retrospective on this cohort, even starting with two crashes, the 2000 and 2008 retirees are tracking well ahead of the historical worst cases.¹²

The spread between these three outcomes, depletion at 87, versus $12M+ surplus, versus a nervous-but-fine middle path, is the entire argument of this post in one picture. Same rule. Same discipline. The only variable was the year the paycheck stopped.

The Go-Forward Problem: You Don't Get to Retire Into the Historical Average

Everything above uses U.S. historical data, and there's a serious argument that the U.S. sample from 1926 onward is a best-case scenario, not a baseline. The United States won two world wars, avoided invasion, and delivered arguably the best equity returns of any market in the world during that window.

Recent academic work takes this critique seriously:

• Anarkulova, Cederburg, O'Doherty, and Sias (2023) built a dataset spanning 38 developed countries and roughly 2,500 country-years of returns to strip out U.S. survivorship bias. Their result is sobering: a 65-year-old couple willing to accept only a 5% chance of financial ruin could withdraw just 2.26% per year, far below conventional advice.¹³ You don't have to accept that number as your plan (a 65-year-old couple's joint longevity is its own long horizon), but it demonstrates how much heavy lifting the exceptional U.S. sample does in the standard 4% result.
• Morningstar's annual "State of Retirement Income" research estimates a forward-looking safe starting withdrawal rate each year based on current valuations, yields, and inflation expectations, rather than raw history. Its base-case estimate for a 30-year retirement at a 90% success target has ranged from 3.3% (2021) to 4.0% (2023) depending on starting conditions, direct evidence that "safe" is regime-dependent, not fixed.¹⁴
• William Sharpe and coauthors critiqued the rule from a different angle in "The 4% Rule: At What Price?", arguing that fixed-real-spending rules are fundamentally inefficient: they pair a constant spending plan with a volatile portfolio, wasting money on surpluses in good scenarios while still failing in bad ones.¹⁵

The common thread: 4% is a product of a particular historical window. Forward-looking sustainability depends on your starting valuations, real yields, horizon, and, critically, your flexibility.

Flexibility Is Worth More Than a Lower Number

If fixed real withdrawals are brittle, the fix isn't necessarily withdrawing less forever, it's building rules that adapt. The research here is encouraging for FIRE planners:

Guardrails (Guyton-Klinger). Jonathan Guyton and William Klinger's 2006 decision-rules framework allows starting withdrawal rates of roughly 5.2%–5.6%, well above 4%, in exchange for pre-committed adjustments: skip the inflation raise after a losing year, cut spending 10% if your current withdrawal rate drifts 20% above the initial rate, and raise it 10% if it drifts 20% below.¹⁶ Simulations consistently show a ~5% guardrails start can match the failure rate of a static 4% rule while supporting higher lifetime spending, though critics (including Wade Pfau and ERN) note the original rules can force large, painful cuts in prolonged bear markets.¹⁷

Valuation-aware starting rates. Because first-decade returns drive outcomes, and because Shiller CAPE meaningfully predicts decade-ahead real returns, several researchers advocate setting the initial rate based on starting valuations, lower when CAPE is elevated, higher when markets are cheap. ERN's spreadsheet and series build this in directly.⁷

Structural mitigants. A rising-equity glidepath or "bond tent", holding more bonds through the vulnerable first decade and re-equitizing afterward, directly targets the sequence-risk window.¹⁸ Partial annuitization or future income (Social Security, pensions, part-time income) effectively shortens the horizon your portfolio must carry alone.

The practical hierarchy: flexible spending rules > lower fixed withdrawal rates > higher savings targets. If your budget has genuine slack, discretionary travel, a housing downsize option, willingness to earn some income in a bad decade, a 4% start with real cut discipline often outperforms a rigid 3.25% plan. If your spending is inflexible and there's no outside income coming, the long-horizon research says start at 3–3.5%.

The Bottom Line

The right question isn't "Is 4% safe?" It's "Safe for how long, under what spending rules, and with what flexibility?"

• 4% is a horizon-specific historical estimate, engineered to survive 30 years in U.S. data, not a universal constant.
• For a traditional 60-something retiree, 4% remains a reasonable, even conservative, anchor. For shorter horizons, it's arguably too conservative.
• For a 40–60 year FIRE horizon, the historical failure rate at 4% rises materially, from ~2–5% to ~15% at 35 years and ~35% at 60 years on a balanced portfolio. The long-horizon fail-safe in U.S. data sits around 3.25%–3.5%, and international evidence suggests even that may be optimistic.
• Sequence risk and inflation are the actual failure mechanisms, and both concentrate their damage in the first decade. Your first ten years of retirement largely decide your outcome.
• Flexibility is the highest-return lever. Guardrails, valuation-aware starting rates, bond tents, and adaptable spending do more for real-world sustainability than shaving a few basis points off a rigid number.

Run your own numbers at multiple withdrawal rates and horizons in our Retirement Calculator, and stress-test the plan against a bad first decade, not an average one. The average retirement was never the problem.

This article is for educational purposes only and is not financial advice. Historical returns do not guarantee future results.

Sources & Footnotes