You have probably heard the 4% rule quoted as if it were a law of nature: multiply your annual spending by 25, and that is your number. The rule shows up in call-room conversations, in physician online forums, and in the back of your mind during a brutal stretch of nights when you find yourself calculating how far you are from done. But the 4% rule is not a law. It is a summary of two research papers from the 1990s, tested against one country's market history, under assumptions that may or may not describe your retirement. Before you build a 30-year plan on top of it, you should know exactly what the evidence says — and what it never claimed.
Bengen 1994: one planner, one question, one dataset
William Bengen was a financial planner, not an academic, when he published "Determining Withdrawal Rates Using Historical Data" in the October 1994 issue of the Journal of Financial Planning. His question was narrow and practical: if a retiree withdrew a fixed percentage of the starting portfolio in year one, then adjusted that dollar amount for inflation every year afterward, what initial percentage would have survived every 30-year retirement in United States market history?
The test design matters more than the headline. Bengen used annual United States stock and intermediate-term government bond returns and examined rolling retirement start dates — including the retiree who quit in 1929 into the Depression, and the one who quit in 1966 into fifteen years of inflation and flat markets. He tested portfolios between 50% and 75% stocks and recommended holding as close to 75% stocks as the retiree could tolerate. His finding: an initial withdrawal rate in the neighborhood of 4% survived even the worst historical start dates over 30 years, while 5% and 6% failed in the bad cohorts.
Note what is inside that result and what is not. Inside: a 30-year horizon, United States returns from the twentieth century, a rigid inflation-adjusted withdrawal that never responds to markets, and a stock-heavy portfolio. Not inside: a 45-year early retirement, international diversification, advisory fees, taxes, or a retiree who adjusts spending when markets fall. The 4% figure was the answer to one specific historical question, not a forward-looking guarantee — and Bengen himself later revised his views as he extended the dataset and asset mix.
Trinity 1998: the confirmation that became a slogan
Four years later, three finance professors at Trinity University — Philip Cooley, Carl Hubbard, and Daniel Walz — published "Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable" in the February 1998 issue of the AAII Journal. The venue is worth noticing: this was a practitioner-facing publication for individual investors, not a peer-reviewed finance journal, which is part of why the result traveled so fast.
The Trinity study extended Bengen's method rather than replacing it. Using stock and corporate bond data from 1926 through 1995, the authors computed historical "success rates" — the fraction of rolling 15-, 20-, 25-, and 30-year periods in which a given withdrawal rate and asset allocation did not exhaust the portfolio. Portfolios with 50% to 75% stocks withdrawing 4% with inflation adjustments succeeded in the overwhelming majority of rolling 30-year periods. That table of success percentages is the artifact that hardened into "the 4% rule."
Important
A historical success rate is not a probability. When the Trinity study reports that a strategy succeeded in a given share of rolling periods, it is counting overlapping windows drawn from a single 70-year sample of one country's markets — a country that won two world wars, dominated the global economy, and delivered some of the best equity returns on record. Treating that count as the probability your specific retirement succeeds imports an optimism the data cannot support.
What the next 25 years of research added: the order of returns is the whole game
The most important refinement after Trinity was naming the mechanism of failure. Portfolios that died in the historical simulations did not die from low average returns. They died from bad returns early, while withdrawals were being taken. A retiree who experiences a 40% drawdown in years one through three, while simultaneously selling shares to eat, does permanent damage that the recovery never repairs — the shares sold at the bottom are gone. A retiree who experiences the identical crash in year twenty barely notices. This is sequence-of-returns risk, and it is why two retirees with the same average return over 30 years can end with wildly different outcomes. The mechanics are covered in depth in the drawdown sequencing module.
Sequence risk reframed the research agenda. If the danger is concentrated in the first five to ten years, then a withdrawal rule that responds to markets during those years should outperform a rigid one — and that is what the dynamic-withdrawal literature found. The most cited framework is the guardrails approach developed by Jonathan Guyton and William Klinger in "Decision Rules and Maximum Initial Withdrawal Rates," published in the March 2006 issue of the Journal of Financial Planning. Their decision rules cut withdrawals by 10% when the current withdrawal rate drifts more than 20% above its initial level, and raise withdrawals by 10% when it drifts more than 20% below. Under those rules, their modeling supported initial withdrawal rates meaningfully above 4% for portfolios holding at least 65% stocks — because the retiree who agrees in advance to take small, rule-based pay cuts in bad markets buys the right to start with a bigger paycheck.
Key insight
Every dynamic strategy is the same trade wearing different clothes: you exchange certainty of income for a higher starting income. A rigid 4% withdrawal makes your spending predictable and your final portfolio value wildly unpredictable. Guardrails make your final outcome more predictable and your year-to-year spending less so. Neither is free. The question is which uncertainty you would rather hold — and for a physician who can flex clinical hours, spending flexibility is far cheaper than it is for most retirees.
The honest critiques: other countries, other valuations
Two lines of criticism deserve to be stated without softening.
First, the international evidence. Researchers who reran Bengen-style historical tests on other developed markets — using long-run return datasets covering countries such as Japan, Germany, France, and Italy across the twentieth century — found that the withdrawal rate that survived every historical period was frequently well below 4%. Investors in countries that lost wars, suffered hyperinflation, or experienced multi-decade equity stagnation would have needed materially lower withdrawal rates. The United States sample that produced the 4% result is, by global standards, a lucky draw. You cannot know in advance whether the next 30 years of United States returns resemble its own past or the broader international experience.
Second, the valuation critique. Historical safe withdrawal rates were lowest for retirees who started when stock valuations were high and bond yields were low, and highest for those who started after crashes. This is intuitive — expensive markets have historically delivered lower subsequent returns — and it means the "safe" rate is not one number but a range that depends on conditions at your retirement date. Researchers disagree about how much to adjust and whether valuation signals are reliable enough to act on. What no serious reading of the evidence supports is treating 4% as equally safe at every starting point, for every horizon, in every market.
| What was tested | Bengen 1994 | Trinity 1998 | Guyton-Klinger 2006 |
|---|---|---|---|
| Horizon | 30 years | 15–30 years | 40 years |
| Data | US stocks and government bonds, rolling history | US stocks and corporate bonds, 1926–1995 | Modeled returns, two historical calibration periods |
| Withdrawal rule | Fixed, inflation-adjusted | Fixed, inflation-adjusted | Dynamic, rule-based cuts and raises |
| Stock allocation | 50–75%, favoring 75% | 0–100% tested; 50–75% highlighted | 50%, 65%, 80% tested |
| Headline finding | ~4% survived worst US history | 4% succeeded in most rolling 30-year periods | Higher initial rates sustainable with decision rules |
Why the 4% rule is a planning heuristic, not a product you can buy
Here is the correct use of all this evidence. The 4% rule gives you a defensible first-pass estimate of the portfolio that makes work optional: annual spending times 25. If your household spends $160,000 per year after taxes, the heuristic points at roughly $4,000,000 — a number to aim at, stress-test, and refine, which is exactly how the freedom number module uses it. What the rule cannot do is operate your actual retirement. No one retires by setting a withdrawal on autopilot for 30 years while ignoring a 45% crash in year two; real retirees adjust, which is precisely why the rigid-rule failure scenarios overstate real-world failure and why the guardrails literature exists.
Example calculation
Assumptions, stated explicitly: household spending of $160,000 per year in retirement, a 30-year horizon, and the Bengen-style 4% initial withdrawal heuristic. Taxes on withdrawals are ignored here and must be layered on separately.
Target portfolio at 4%: $160,000 ÷ 0.04 = $4,000,000 Same spending at a more conservative 3.5%: $160,000 ÷ 0.035 = $4,571,000 (rounded) Same spending at a guardrails-style 5% initial rate: $160,000 ÷ 0.05 = $3,200,000
The spread between the cautious and flexible readings of the same evidence is roughly $1,370,000 — about three to four years of a typical attending's savings. Which end of that range you plan toward should depend on your horizon, your flexibility, and your willingness to cut spending in bad markets, not on which blog post you read last.
The physician overlay: part-time medicine is the guardrail other retirees cannot buy
Now apply all of this to your specific situation, because physicians hold an asset the withdrawal literature never models: a durable, high-priced license to generate income on demand.
Consider what sequence risk actually demands. The danger zone is the first five to ten years of retirement, and the fix — in every dynamic strategy — is reducing portfolio withdrawals when markets fall. A 68-year-old former sales manager can only achieve that by cutting spending. A 58-year-old internist can achieve it by picking up six locum tenens weeks or one clinic day per week. At plausible attending rates, even 400 hours per year of clinical work generates $80,000 to $150,000 depending on specialty — enough to cut a $160,000 withdrawal need by half or more during exactly the years when selling depressed shares does the most damage. In guardrails terms, you carry a prosperity rule and a capital preservation rule in your pocket, and it renews with your medical license.
This is the single most underpriced fact in physician retirement planning: the ability to earn meaningful income part-time converts sequence risk from an existential threat into an inconvenience, and it justifies retiring on a thinner portfolio than the rigid 4% arithmetic suggests — provided you actually maintain the license, the certification, and the willingness to use them.
The honest caveats: clinical skills and credentials decay, malpractice tails and licensing fees are real carrying costs, and the physician who swears they will "just do locums if things get bad" at 55 may feel very differently at 63. The option is valuable only while you keep it alive. How that option interacts with an early exit date is covered in the physician early retirement article, and the broader independence math lives in the physician FIRE guide.
Quick takeaway
Read the evidence as a stack, not a slogan. Bengen 1994 established that roughly 4% survived the worst 30-year windows in United States history. Trinity 1998 confirmed it with success-rate tables. The sequence-risk literature explained why failures happen early. Guyton-Klinger 2006 showed that agreeing to flex spending buys a higher starting withdrawal. The international and valuation critiques warn you the American past was generous. And your medical license is a spending-flexibility instrument none of those papers could model.
Common questions
Is the 4% rule still valid in 2026?
As a first-pass planning heuristic, yes; as a guarantee, it never was one. The rule summarizes United States historical worst cases over 30-year horizons. If your retirement is longer than 30 years, if future United States returns look more like the international historical record, or if you retire into high valuations, a rigid 4% carries more risk than the historical tables imply. If you retain spending flexibility — especially the physician option to earn part-time — you can defensibly plan at or above 4%.
Does the 4% rule include taxes?
No. Both Bengen and the Trinity study modeled pre-tax portfolio withdrawals and ignored taxes and fees entirely. The 4% you withdraw must cover the income tax due on it, which for a physician drawing from a large traditional or can be substantial. Your spending number and your withdrawal number are different figures, and the gap between them is a tax-planning problem, and the withdrawal-ordering side of it is covered in the drawdown sequencing module.
What withdrawal rate should I use for a retirement at 50?
The original studies stop at 30-year horizons, so a retirement at 50 extends past the data. Longer horizons pushed historical safe rates down — research extending the horizon toward 40 to 50 years generally lands in the low-to-mid 3% range for rigid withdrawals. In practice, most physicians retiring at 50 should plan around a 3.25% to 3.5% rigid-equivalent rate, then credit themselves back toward 4% or higher to the extent they genuinely retain part-time earning capacity.
Are guardrails strategies better than the fixed 4% rule?
They solve different problems. Fixed withdrawals maximize income predictability; guardrails-style rules such as Guyton-Klinger 2006 raise the starting withdrawal in exchange for pre-committed cuts when markets fall. The evidence favors dynamic rules for total lifetime income, but only for retirees who will actually execute the cuts. If a 10% spending reduction in a bear market would be intolerable for your household, price that honestly and start lower instead.
What to do next
- Compute your rigid baseline: multiply expected annual retirement spending by 25, and by 29 for a 3.5% rate, to see your planning range before touching any assumptions.
- Read your own horizon honestly — count years from your realistic exit age to age 95, and if it exceeds 30, shade your planning rate down.
- Estimate your taxes separately: your withdrawal must fund spending plus the tax on the withdrawal, so model your effective rate on traditional-account distributions.
- Price your flexibility: write down what one clinic day per week or six locum weeks would actually pay in your specialty today, and what license, DEA, certification, and tail costs keeping that option alive would run.
- Choose your rule in advance — rigid rate, guardrails with specified triggers, or a hybrid — and write it down before retirement, because the worst time to design a spending policy is inside a crash.
- Revisit the plan annually against actual portfolio value and spending, the same way you would re-titrate any long-running protocol.
The withdrawal-rate literature rewards physicians who read it the way they read a trial: check the population studied, the intervention actually tested, and the endpoints measured before applying the conclusion to the patient in front of you. The patient in front of you is a household with a long horizon, a large tax-deferred balance, and an earning option most retirees would pay dearly for — and the protocol above works with or without us. This is education, not individualized financial advice.