when to retire

Updated for 2026

What is a safe withdrawal rate?

The 4% rule is the most cited number in retirement planning and the most misunderstood. Current research puts the answer anywhere from 3.9% to 5.7% depending on what question you're actually asking.

Last reviewed August 2026 · Sources cited throughout · Full reference list

Where the 4% rule came from

In October 1994, financial planner William Bengen published "Determining Withdrawal Rates Using Historical Data" in the Journal of Financial Planning. He ran every rolling 30-year period in the historical record and asked a specific question: what is the highest starting withdrawal rate that would have survived every single one, including the worst? His answer — roughly 4% — became known as SAFEMAX.

The mechanic matters and is often described wrong. The rule says to withdraw 4% of your portfolio in year one, then adjust that dollar amount for inflation each year afterward. You never recalculate against the current balance. Someone starting with $1,000,000 withdraws $40,000 in year one and then that same $40,000 of purchasing power every year after, regardless of whether the portfolio has doubled or halved.

The later Trinity Study reinforced the finding, reporting success rates above 95% for a 50/50 stock and bond mix over 30-year periods. That pair of results is why 4% became shorthand for retirement safety.

Why 2026 research disagrees

Three credible numbers are currently in circulation, and they conflict because they answer different questions.

Bengen himself now says roughly 4.7%. In his 2025 book A Richer Retirement, he revisited the original work with a more diversified portfolio — adding small-cap and international equities rather than the original large-cap-plus-intermediate-bond mix. Broader diversification improves the historical worst case, pushing SAFEMAX upward.

Morningstar says around 3.9%. Their annual State of Retirement Income research, published December 2025, arrives at a more conservative figure for a fixed-spending retiree over 30 years targeting a 90% success probability. The key methodological difference: Morningstar uses forward-looking return assumptions based on current bond yields and equity valuations, rather than historical averages. That figure has moved over time — 3.3% in 2021, 3.7% the following year, 3.9% for 2026 as bond yields improved.

Flexible strategies support meaningfully more. Morningstar's own research suggests retirees willing to adjust spending in response to markets may support rates as high as 5.7%. That's not a contradiction of the lower figure; it's the price of rigidity made visible.

None of these is wrong. Historical worst-case analysis and forward-looking simulation are different exercises, and a plan that can flex is genuinely safer than one that can't.

What actually moves your number

Time horizon. This is the single biggest factor. A 20-year retirement supports a materially higher rate than a 30-year one, and early retirees planning for 40 to 50 years face a stricter constraint — much FIRE-oriented research lands nearer 3.25% to 3.5% for those horizons. More years means more opportunities to meet a bad sequence and less time to recover.

Portfolio composition. Bengen's own revision demonstrates the point: the same withdrawal strategy against a broader mix of assets produced a different safe maximum. Allocation isn't a detail layered on top of the withdrawal rate, it's part of what determines it.

Spending flexibility. The gap between roughly 3.9% fixed and 5.7% flexible is the largest single lever available to most retirees. A plan that can trim discretionary spending after a bad year tolerates a much higher starting rate than one that cannot.

Other income. Social Security, a pension, or annuity income reduces how much the portfolio itself has to carry, which changes the required withdrawal rate rather than the safe one.

Sequence-of-returns risk

Two retirees can experience identical average returns over 30 years and end up in completely different places. What separates them is when the bad years arrived.

A severe decline in the first few years of retirement is far more damaging than the same decline two decades in, because you're selling assets into the loss to fund spending. Those shares are gone and can't participate in the recovery. The same crash late in retirement hits a portfolio that has already had years of growth behind it and fewer remaining withdrawals ahead.

This is precisely why a simple average-return projection is misleading, and why the simulator runs your scenario against every historical starting year rather than a single assumed return. The spread between best and worst outcomes is sequence risk made visible.

Sequence of returns risk illustration Two portfolios with identical average returns end at very different values depending on whether the losing years occur early or late in retirement. $1.4M $700k $0 Years in retirement → Good years first, bad years later Bad years first, good years later
Both portfolios experience the same set of annual returns and the same average — only the order differs. Withdrawing during the early losses means selling assets that never participate in the recovery.

Guardrails: a middle path

Guyton and Klinger's 2006 paper "Decision Rules and Maximum Initial Withdrawal Rates," also in the Journal of Financial Planning, proposed a structured compromise between rigid and fully discretionary spending. Withdrawals stay fixed in real terms unless the withdrawal rate drifts beyond preset boundaries — commonly 20% above or below the starting rate — at which point spending adjusts by around 10% to bring it back in line.

The appeal is that the flexibility is defined in advance rather than improvised during a downturn, which is when improvising is hardest. Those same parameters are implemented in the simulator's guardrails mode, so you can compare it directly against fixed and percentage-of-portfolio strategies using your own numbers.

How to use any of this

Treat published rates as reference points rather than answers. Run your actual horizon, your actual allocation, and your actual spending, then look at where the failures cluster rather than fixating on a single success percentage. A plan that fails only in the worst historical sequences, and that you could rescue by trimming spending, is in a different category from one that fails routinely.

None of this constitutes financial advice, and none of it accounts for taxes, healthcare costs, or your specific circumstances. It's a framework for asking better questions of a professional, or of your own plan.

Advertisement

Common questions

Is the 4% rule dead?

No, but it was never meant to be a fixed law. Its own author has revised his figure upward, while forward-looking research has landed lower. It remains a reasonable planning anchor for a 30-year horizon and a poor fit for a 45-year one.

What rate should an early retiree use?

Longer horizons demand lower rates. Research aimed at 40-plus year retirements commonly suggests something in the 3.25% to 3.5% range with dynamic adjustments, rather than the 4% figure derived from 30-year periods.

Does a 90% success rate mean I'll probably be fine?

It means 90% of the simulated or historical paths ended with money remaining. It says nothing about which world you'll land in, and it assumes you never adjust spending — an assumption most real retirees would violate the moment things looked bad. That's why flexible strategies test better.

Should I withdraw a fixed dollar amount or a percentage?

A fixed inflation-adjusted amount gives predictable income but can deplete a portfolio in a bad sequence. A percentage of the current balance can never mathematically hit zero but produces income that swings with the market. Guardrails sit between the two. The simulator runs all three so you can compare them on your own numbers.

Sources cited on this page

Market data underlying the simulator comes from Robert Shiller (Yale) and Aswath Damodaran (NYU Stern). Full details on the methodology and references section.

Important: This page is educational and does not constitute financial advice. Safe withdrawal rate research describes historical and modelled outcomes, not guarantees, and none of it accounts for your individual tax situation, healthcare costs, or circumstances. Consider consulting a qualified fee-only fiduciary advisor before making retirement decisions.

Advertisement