Bengen's 4.7% Safe Withdrawal Rate Faces the 50-Year Test

Bill Bengen, the advisor whose 1994 research gave us the 4% rule, spent late 2024 telling anyone with a microphone that the safe withdrawal rate is now 4.7%. Podcast clips did the rest. If you are 38, sitting on a low-cost index portfolio and a savings-rate spreadsheet, that headline reads like a promotion letter: at $80,000 of annual spending, moving from 4.0% to 4.7% cuts the portfolio you need from about $2.0 million to about $1.7 million. Roughly $300,000 of additional permission.

The catch is that 4.7% is a 30-year number wearing a universal label. Bengen measured it on retirements designed to end at year 30, on US data, with a portfolio whose edge comes mostly from small-cap stocks. An early retiree planning a 40 to 60 year horizon is importing a number from a test that never examined years 31 through 50. Most of the bump is a horizon artifact, not new safety, and the rest of this article prices exactly how much of it you can defensibly bank.

The Claim and the $300,000 Math

Start with arithmetic, because it is the only part of this debate without a debate. Divide $80,000 of planned annual spending by each candidate withdrawal rate:

• 4.0%: a $2,000,000 target
• 4.7%: a $1,702,128 target

The gap is $297,872, call it $300,000, or a required portfolio about 15% smaller. For a mid-career saver contributing $5,000 a month, that is 60 months of contributions, five full years, before counting a dollar of investment growth. The decimal places in a withdrawal rate decide whether you resign at 43 or at 48.

The real question is not whether Bengen is wrong. It is whether his 4.7% transfers from the 30-year retirements he tested to the 50-year retirement you are planning. The rest of this piece answers that in dollars and months.

What Bengen Actually Changed

His original 1994 study tested 30-year retirements on US market data back to the mid-1920s, withdrawing an inflation-adjusted amount each year. Safety had a precise meaning: the portfolio survives the single worst historical starting month, not the average one. That definition produced roughly 4% (4.15% in his later retellings), and the Trinity study later confirmed the neighborhood.

The revision keeps the definition, the data window, and the 30-year clock, and changes the portfolio. In the interview transcript circulating widely, Bengen describes spreading equities well beyond the S&P 500 into US small-caps and international small-caps, alongside a large intermediate-bond allocation and a small cash sleeve. Those extra sleeves, he reports, lift his worst-case 30-year rate from about 4.15% to 4.7%, and to roughly 5.5% if you accept occasional spending cuts.

One fairness point before the criticism: Bengen himself frames 4.7% as a 30-year result. The FIRE internet performed the extrapolation to 50-year plans, and that extrapolation is where the trouble starts.

Where the Extra 0.7% Comes From

Any backtested withdrawal rate is the product of three inputs: the data window, the definition of safety, and the portfolio. Bengen changed only the third. So the bump is not evidence that retirement got safer. It is evidence that a particular mix of assets backtested better over one particular century.

The engine inside that mix is the small-cap premium. In the standard Fama-French factor dataset, small-cap stocks beat large-caps by about 3.6% a year from 1926 to 1981, and by essentially nothing in the four decades since. A backtest that leans on 1926-1981 small-cap dominance is leaning on returns that stopped arriving before most current retirees were born, and some researchers argue that trading costs and stale-price biases meant even historical investors never captured the full premium.

Independent replications sharpen the doubt. Practitioner simulations using monthly retirement cohorts, real fund fees, and pre-1926 market history reproduce the new allocation with worst-case 30-year rates closer to 4.2% than 4.7%, and AAII's assessment of 4.7% captures a research community that is interested but unpersuaded. Part of the gap is methodological detail, cohort frequency, data series, fee assumptions. Part may be optimism. None of it is reassuring at 50 years.

A 30-Year Safety Test Applied to 50-Year Retirements

Here is the structural problem, and it survives any data dispute. Bengen's safety criterion is binary: the portfolio must last 30 years from the worst start. A cohort that lands at exactly zero in month 360 passes. The same portfolio on a 50-year clock fails in month 361.

Sequence of returns risk explains why the clock matters this much. A retiree who hits a deep drawdown in the first decade sells shares to fund withdrawals and never recovers the compounding, even when average returns across the whole period look fine. The historically dangerous start dates (1929, 1968, the 1973 oil shock) were identified under a 30-year lens. Extend the horizon and those same cohorts must survive two additional decades of withdrawals with no salary to refill the tank. The test that generated 4.7% literally cannot see the years where an early retirement withdrawal rate lives or dies.

That is why horizon extension is brutal in every long-horizon study: the safe withdrawal rate for 50 years is not 4.7% minus a rounding error. In most practitioner research it sits near 3.25% to 3.5%.

How Many 50-Year Retirements the Data Can Test

There is also a sample-size problem no backtest can engineer away. The standard US dataset, popularized by Ibbotson and running in Ibbotson return data since 1926, covers about a century. A century contains roughly two non-overlapping 50-year retirement sequences, and only a few more 40-year ones. Every 50-year withdrawal figure you have ever seen, 3.25%, 3.5%, or 4.7%, rests on effectively two independent observations.

Backtests paper over this with rolling cohorts: start a retirement in every month of the century and you can manufacture a thousand simulated retirements. But the 1951 cohort and the 1952 cohort share 49 of their 50 years of returns. Overlapping windows create the appearance of statistical power without the substance. They are not a thousand independent tests of a withdrawal rate; they are two long market histories sliced thin.

And those histories are American. One market, which happened to become the best-performing economy of the 20th century, supplies the entire evidence base for a number now being marketed as universal.

Building the Allocation With Low-Cost Index Funds

Suppose you accept Bill Bengen's new allocation anyway. Can a low-cost FIRE portfolio actually hold it?

The US small-cap allocation for retirement portfolios is a non-issue; index funds deliver it for pennies. International small-cap is the bottleneck. The fund menu is short, the category is less liquid, spreads run wider, and expense ratios sit above broad-market funds even at Vanguard. Absolute costs stay modest, but failsafe withdrawal math is a basis-point business: replications that include a realistic 15-basis-point drag for fees and tracking already shave meaningful points off the headline number.

The practical conclusion is uncomfortable. Most FIRE portfolios can cheaply capture the US sleeves and only partially capture the international small-cap sleeve, so they should expect only part of the backtested benefit. Partial replication of a marginal edge is not a plan. And note that Bengen's described mix has shifted between tellings, which is exactly what you do not want from a portfolio you must hold for five decades.

What a Safe Withdrawal Rate Looks Like at 50 Years

If 4.7% cannot survive the transfer, what can? Three bodies of evidence triangulate an answer for 40 to 60 year horizons.

International history is the first. Research compiling return histories across a range of developed markets finds that 4%-class fixed withdrawal rates failed in many of them over long horizons. The US record that produced the 4% rule is closer to a best case than a global law, and the 4.7% update, built on that same US-only data, inherits the survivorship rather than escaping it.

Valuation conditioning is the second. Kitces's valuation research shows that withdrawal outcomes depend heavily on the CAPE ratio at retirement: starting when equities are expensive historically required lower rates, and the 4% rule's clean record largely reflects cohorts that retired into cheaper markets. For early retirees, valuations at the starting line matter more, not less.

Current practitioner estimates are the third. Morningstar's annual retirement income research has suggested starting withdrawal rates in the mid-3% to 4% range in recent editions, and long-horizon simulation series focused on FIRE consistently land 50-year failsafes near 3.25% to 3.5%. Against that distribution, 4.7% is not the new consensus. It is the optimistic tail of 30-year estimates, one methodological disagreement away from 4.2%.

Portfolio Targets at 3.5%, 4.0%, and 4.7%

Convert the evidence into the units behind every "how much do I need to retire early" spreadsheet. At $80,000 of spending:

Contribution months ignore investment growth, so real timelines run shorter; the deltas are what matter, because growth affects all three rows similarly. The distance between planning at 3.5% and banking the full headline is about $584,000, nearly ten years of contributions. The distance between a defensible 4.0% and the viral 4.7% is five.

Notice the asymmetry of regret. Retire on the 4.7% row and, if the long-horizon research is right, you face a meaningful probability of forced spending cuts in your 70s with no employment fallback. Retire on the 3.5% row and the cost is concrete: extra working months you can count, rather than a tail risk you cannot.

A Decision Rule for Early Retirees

The verdict is conditional. The bump is not fake, but at 40 to 60 years you can defensibly bank only a fraction of it. A workable rule:

Bank part of the bump and plan around 3.75% to 4.0% if most of these hold:

• Your horizon is closer to 40 years than 60 (retiring in your late 40s, or a pension and Social Security covering late-life spending)
• You can genuinely cut spending 10-15% for years at a stretch without misery
• You hold a high equity allocation, 65% or more, since long horizons punish bond-heavy mixes
• You retire into mid-range rather than peak valuations

Ignore the bump entirely and plan at 3.25% to 3.5% if any of these are true:

• Spending is fixed by health costs, kids, or a mortgage you will not flex
• You want a bequest or a capital cushion at the end
• You are retiring at 40 with 50+ years on the clock and elevated valuations
• You would rather work four more years than gamble five in your 70s

Treat the Bengen 4.7% rule as what the evidence supports: a 30-year, US-only, small-cap-tilted backtest, one careful replication away from 4.2%. Use it as a ceiling to argue against, never as the floor you budget from.

At $80,000 of spending, the roughly $300,000 discount the headline promised was never free. It was borrowed from years 31 through 50 of your retirement, collateralized by two non-overlapping 50-year market histories and a small-cap premium that went quiet in 1981. If your plan survives being priced honestly at 3.5% to 4.0%, retire. If it only works at 4.7%, the gap is about five more years of saving, and knowing that now beats any headline.