Loading…
Learn
In 1994, financial planner William Bengen sat down with 70 years of US stock and bond market data and asked a simple question: what is the highest withdrawal rate a retiree could have used in any historical period and never run out of money over a 30-year retirement? The answer was 4.15%. Bengen called it SAFEMAX — the safe maximum withdrawal rate.
That finding, rounded to 4%, became the single most influential number in personal finance. It is the foundation of the FIRE number. It is why the "25x rule" exists. And it is simultaneously the most used and most misunderstood concept in retirement planning.
The 4% rule does not mean you are guaranteed to have money for 30 years. It means that in every historical 30-year window Bengen studied — including the 1929 crash, the Great Depression, and the 1970s stagflation — a retiree who started at 4% and inflation-adjusted upward each year afterward never hit zero. That is a powerful finding. It is also a historically specific one, with important caveats that every FIRE planner should understand.
Bengen published "Determining Withdrawal Rates Using Historical Data" in the Journal of Financial Planning in October 1994. He studied rolling 30-year periods from 1926 to 1992, using a 50/50 stock-bond portfolio based on S&P 500 and intermediate US Treasury data. The worst case — the 1966 retiree, who faced both the 1970s bear market and rampant inflation — still survived 30 years at 4%. Bengen later revised the figure upward to 4.5% when adding small-cap equities to the analysis, but 4% remained the canonical round number.
In 1998, three professors at Trinity University — Philip Cooley, Carl Hubbard, and Daniel Walz — published what became known as the Trinity Study. They expanded Bengen's framework across five withdrawal rates (3% to 7%) and four time horizons (15 to 30 years), producing a table of portfolio success rates that planners still reference today.
Here is an approximation of that table for a 100% equity portfolio, which most FIRE planners use:
| Withdrawal rate | 15 years | 20 years | 25 years | 30 years |
|---|---|---|---|---|
| 3% | 100% | 100% | 100% | 100% |
| 3.5% | 100% | 100% | 98% | 98% |
| 4%← classic rule | 100% | 100% | 95% | 95% |
| 4.5% | 100% | 98% | 87% | 87% |
| 5% | 98% | 93% | 74% | 68% |
Approximate historical success rates from the Trinity Study (Cooley, Hubbard, Walz 1998), 100% equity portfolio. Green = 95%+, black = 80–94%, red = below 80%.
Data vintage
The table above reflects the original 1998 Trinity Study. Since then, multiple authors — including Cooley, Hubbard, and Walz themselves (2011), Wade Pfau, and the Early Retirement Now "Safe Withdrawal Series" — have re-run the same analysis with longer datasets and found broadly similar results, with modest downward revisions at higher withdrawal rates. For your own scenario, use the Spend page — it backtests against Shiller's full 1871-present dataset rather than Trinity's 1926-1995 window.
30-year success rates by withdrawal rate
The 4% rule gives us the simplest formula in FIRE planning. If you can safely withdraw 4% of your portfolio each year, then you need a portfolio worth 25 times your annual expenses to be financially independent. This is the 25x rule — it is just 4% expressed as a multiplier.
The formula
FIRE Number = Annual Expenses ÷ Withdrawal Rate
At 4%: Annual Expenses × 25
At 3.5%: Annual Expenses × 28.6
At 3%: Annual Expenses × 33.3
The formula works in both directions. Increasing your withdrawal rate from 4% to 5% cuts your required portfolio by 20% — from 25x to 20x. But it also materially increases the chance of running out of money over a long retirement. Conversely, dropping to 3.5% increases your required savings by 14% but purchases significantly more historical safety margin.
For early retirees, the practical question is not just "what is my FIRE number" but "at what withdrawal rate am I comfortable with the historical odds." That depends on your retirement horizon, your spending flexibility, and how much buffer you want.
The 4% rule is a research finding, not a guarantee. Understanding where it breaks down is as important as understanding where it holds.
It was tested on 30-year retirements
Bengen studied 30-year windows. A FIRE retiree at 35 may need their portfolio to last 55 or 60 years. The historical record does not contain enough 50-year windows to be statistically robust, and longer horizons dramatically increase the chance of encountering a catastrophic sequence of returns. Big ERN's analysis suggests that truly safe withdrawal rates for 50-year retirements may be closer to 3.25–3.5%.
It used US stock data only
Both Bengen and the Trinity Study relied primarily on US market returns. The 20th century was extraordinarily good for US investors by global standards. Research by Dimson, Marsh, and Staunton covering 21 countries found that US returns were exceptional in a global context. A 4% rule calibrated on US data may be optimistic for portfolios with significant international exposure, and it says nothing about what would have happened to a German or Japanese retiree in the same period.
It assumes rigid, inflation-adjusted spending
The rule takes 4% in year one and adjusts upward with inflation every year after — no matter what markets do. In practice, virtually no retiree spends this way. Most people cut spending in bad market years and increase it in good ones, whether consciously or not. Flexible withdrawal strategies — like Guyton-Klinger guardrails or CAPE-based dynamic rules — can support higher starting rates precisely because they allow adaptive response to market conditions.
Sequence of returns still matters
The 4% rule survived every historical window, but some windows were close calls — the 1966 cohort in particular. A retiree who started at 4% in 1966 was not in comfortable territory for most of their retirement. If future returns are lower than historical US averages — due to high starting valuations, lower structural growth, or other factors — the 4% floor from the past may not hold going forward.
The 4% rule has been updated, challenged, and refined many times since 1994. Here is where the research stands today.
Big ERN: 3.25–3.5% for early retirees
Karsten Jeske (earlyretirementnow.com) has published the most rigorous modern analysis of safe withdrawal rates across different horizons and starting valuations. His Safe Withdrawal Rate series runs thousands of historical simulations and concludes that for 40–60 year retirements, the safe floor under rigid rules is roughly 3.25–3.5%, not 4%. The gap widens when markets are expensive at retirement (high CAPE), shrinks when they are cheap. His methodology is transparent and widely respected in the FIRE community.
Flexible rules can support higher rates
A rigid 4% rule that never adapts is inherently conservative because it has no mechanism to respond to bad markets. Strategies like Guyton-Klinger guardrails, CAPE-based dynamic withdrawals, and variable percentage withdrawal (VPW) can support higher starting rates because they incorporate spending cuts when conditions deteriorate. Morningstar's annual retirement research has consistently found that flexible strategies meaningfully extend portfolio survival compared to rigid inflation-adjusted rules at the same starting rate.
The CAPE ratio as a valuation signal
Research by Wade Pfau and others has shown that the CAPE ratio (Shiller P/E) at the time of retirement is a meaningful predictor of subsequent safe withdrawal rates. Retiring when CAPE is high (above 25–30) is historically associated with lower safe rates. This suggests that the static 4% rule may be too generous in high-valuation environments and too conservative in low-valuation environments. CAPE-adjusted dynamic strategies attempt to incorporate this signal directly into the withdrawal calculation.
Bengen's own update: 4.7%
In 2021, Bengen updated his original analysis using a broader dataset that extended back to 1926 and incorporated small-cap stocks. With this expanded dataset, SAFEMAX rose to 4.7%. He also clarified that his original 4% was always a conservative floor — he expected most retirees to do better. The distinction matters: 4% is not the "expected" withdrawal rate, it is the worst-case floor from historical data.
The 4% rule is a useful starting point, not a final answer. Here is how to think about it in the context of your own plan.
Use 3.5% if your horizon is 40+ years
The additional buffer gives you meaningful protection against the combination of long horizon, potential sequence-of-returns risk, and lower future returns. The cost is a larger required FIRE number — about 14% more savings.
Add Social Security and other income sources
Social Security benefits, pension income, or part-time work all reduce what you actually need to withdraw from your portfolio. A $20K/yr Social Security benefit effectively lowers a $60K expense plan to a $40K withdrawal need — reducing your effective withdrawal rate considerably.
Build in flexibility on spending
A plan that can cut 10% spending in a bad market year is far more robust than a rigid plan at the same withdrawal rate. The mechanical ability to adapt — even modestly — dramatically improves historical survival rates and lets you start retirement at a higher rate.
Run the historical scenarios, not just the average
Your success rate should be tested against specific historical cohorts — the 1929 cohort, the 1966 cohort, the 2000 cohort. Calcifer's historical backtest does this automatically, running your plan against every available window in Shiller's 150-year dataset.
The 4% rule is one of the most robust empirical findings in personal finance. It emerged from genuine historical analysis, it has been replicated and extended many times, and it remains a sensible planning anchor for most retirees with 30-year horizons. The FIRE community has made it more complicated — rightly so, given longer retirement horizons and higher starting valuations.
But the most important insight is not the specific number. It is the framework: your required portfolio is a multiple of your spending, that multiple depends on your withdrawal rate, and your withdrawal rate depends on your horizon, your flexibility, and your tolerance for historical uncertainty. Calcifer models all of this — and shows you the full historical distribution, not just the headline.