Safe Withdrawal Rate for a 50-Year Retirement

Quick Answer

For a 50-year retirement, I use 3.25%. That's well below the famous 4% rule, and it's the rate I can defend with both historical backtests and Monte Carlo. The 4% rule was calibrated for a 30-year horizon. Stretch the horizon to 50 and the failure rate roughly doubles. The research lands somewhere between 3.0% and 3.5% for someone planning to live on portfolio income from their 30s into their 80s.

Anyone telling you a 50-year retirement can sustain a 4% rate is either selling something or hasn't read the papers. Usually both.

Why the 4% Rule Doesn't Stretch

Bill Bengen's 1994 paper tested withdrawal rates against rolling 30-year windows of U.S. market data. He found a 4.0% initial rate, adjusted for inflation, never caused a portfolio to fail in any 30-year window from 1926 forward. That's the foundation of the 4% rule.

The catch is in the words "30-year." Bengen was answering the question a typical retiree had in 1994: how much can I pull from a portfolio across a roughly 30-year window from age 65? The math does not extrapolate cleanly. Each additional year of retirement adds another withdrawal against a portfolio that may already be drawing down. Sequence-of-returns risk compounds. A 50-year retirement is not 1.67 thirty-year retirements stacked end to end. It's a fundamentally different problem that gets fundamentally wrong answers when you pretend it isn't.

Wade Pfau's extensions of the Trinity Study found that pushing the horizon from 30 to 50 years requires dropping the safe rate from 4.0% to roughly 3.0% to 3.5% to keep historical failure rates near zero. Big ERN at Early Retirement Now ran 1500+ historical paths and arrived at 3.25% as a practical floor for a 50-year horizon on a 75/25 portfolio. The numbers cluster. That's not coincidence. It's what the arithmetic says when nobody has an AUM fee riding on the answer.

What "Safe" Actually Means

The phrase "safe withdrawal rate" is used loosely. It almost always means: the highest initial inflation-adjusted withdrawal rate that, when run against historical or simulated returns, leaves the portfolio above zero at the end of the horizon at least 95% of the time. Some research uses 99%. Some uses 90%. The word "safe" is doing a lot of work in those cracks, and the glossy retirement-planning brochures never tell you which definition they're using.

For a 50-year retirement at 3.25%, historical backtests on U.S. data show a failure rate near 1% to 2%. Monte Carlo with reasonable forward-return assumptions shows higher failure rates, often 5% to 10%, because the simulations include scenarios that haven't happened yet but mathematically could. The honest reading: 3.25% is safe under the conditions the U.S. market has historically produced, and somewhat less safe under conditions that are plausible but unprecedented. Anyone who promises you certainty on a 50-year horizon is lying or selling.

A Worked Example

Pretend I retire at 35 with $2,500,000 and need the portfolio to get me to 85.

At 4.0%: Year-one spending is $100,000. Historical 50-year backtests on a 75/25 portfolio show a failure rate of roughly 18%. Nearly one in five historical paths runs me out of money before 85. I would not retire on those odds. Nobody should.

At 3.5%: Year-one spending is $87,500. Historical failure rate drops to roughly 8%. Better. Still meaningful tail risk.

At 3.25%: Year-one spending is $81,250. Historical failure rate falls to roughly 2%. This is the rate I actually use.

At 3.0%: Year-one spending is $75,000. Historical failure rate is essentially 0%, and the median ending balance after 50 years is over $10M in inflation-adjusted dollars. You died with most of your money. That's its own failure mode.

The trade is real. Going from 3.25% to 3.0% costs me $6,250 per year, or $312,500 of cumulative spending over 50 years. In exchange I buy a roughly 2-percentage-point reduction in failure probability. Whether that trade is worth it depends entirely on how much you'd hate going back to work at 65 because the math broke.

Assumptions That Quietly Drive the Number

The "safe rate" you read in any paper depends on assumptions that usually go unexamined. Every one of them moves the number.

Asset allocation. The 3.25% figure assumes 75/25 stocks/bonds. A 100% equity allocation produces higher long-run returns and higher safe rates in good paths, but also higher volatility and worse failures in bad paths. A 60/40 lowers the safe rate by 25 to 50 basis points. A 40/60 closer to 100. There is no allocation that simultaneously maximizes upside and minimizes downside. Pick your poison.

Return assumptions. Historical U.S. returns have averaged roughly 7% real for stocks and 2% real for intermediate bonds over the past century. Pfau, Bogle, and others have argued forward returns are likely lower because current valuations are higher than the historical average. If you use 5% real for stocks instead of 7%, the safe rate drops another 25 to 50 basis points. The calculator that runs 7% forever is showing you a promotional future, not a planning one.

Sequence of withdrawals. Most simulations assume you withdraw the full year's spending at the start of the year and don't rebalance opportunistically. McClung's Prime Harvesting and Kitces's bond-tent research show modest but real improvements when you pull from cash and bonds during equity drawdowns instead of selling stocks at depressed prices. Layered in, these tactics buy roughly 20 to 40 basis points of safe-rate headroom. This is the work a good plan does for free and an advisor charges you 1% of NAV to do, badly, once a year.

Spending shape. The 4% rule assumes constant inflation-adjusted spending. Most retirees spend more in their first decade (the "go-go years") and less in their 70s and 80s. Bake a declining real-spending pattern in and the safe rate rises 25 to 50 basis points. Bernicke's "Reality Retirement" paper documented this with actual consumer-spending data.

Add the variance: optimistic returns plus declining spending plus a smart withdrawal sequence might justify 3.75%. Pessimistic returns plus flat spending plus naive withdrawals push you down to 2.75%. The published 3.25% is somewhere in the middle, and that's why I use it. It survives most of the assumption combinations. It does not survive all of them, and anyone claiming it does is moving a line on a chart instead of reading what the line says.

The Variable-Spending Alternative

If 3.25% feels too austere, the honest move is not to lift the static rate. It's to give up the static rate entirely.

Dynamic withdrawal strategies (Guyton-Klinger guardrails, Vanguard dynamic spending, Kitces ratcheting, the IRS RMD method applied to your own portfolio) let you start at a higher rate (often 4.0% to 4.5% for a 50-year horizon) in exchange for accepting that annual spending will move with the portfolio. Good markets, spend more. Bad markets, cut. The arithmetic that makes static 3.25% safe is the same arithmetic that makes dynamic 4.0% safe. You're just moving the safety margin from "lower spending" to "willingness to flinch."

This is the trade I push every long-horizon retiree to consider. Static 3.25% gives you a paycheck you can plan around. Dynamic 4.5% gives you more average lifetime spending at the cost of variability some people find hard to live with. Neither is a free lunch. Anyone who tells you there is a free lunch in a 50-year retirement plan is trying to get you to sign something.

What to Do With This

Three things, in order of importance.

First, use a real horizon. If you're 35 and plan to retire to 85, that's 50 years. Don't shave it to 40 to get a more comfortable safe rate. The model only works if the inputs are honest.

Second, run your own Monte Carlo with at least 1,000 paths and check the 5th percentile, not just the median. The median is the outcome that happens half the time. The 5th percentile is the one that determines whether you can sleep. The SWR Analyzer outputs both. The gap between them is the actual risk you're taking.

Third, decide in advance what you'll do if the portfolio is in trouble at year five or ten. The most expensive failure mode is a retiree who built a 50-year plan, hit a bad first decade, and refused to either cut spending or take part-time work. Whatever your plan B is, write it down before you need it. Plans written under stress are worse than plans written at a kitchen table with coffee.

Frequently Asked Questions

Why is the safe rate lower for a 50-year retirement than a 30-year one?

Each additional year adds a withdrawal that compounds against the portfolio. Sequence-of-returns risk dominates: a bad first decade is roughly five times more damaging in a 50-year retirement than in a 30-year one, because there are more remaining years for the deficit to compound. Dropping the rate is the cleanest way to absorb that risk.

Can I use the 4% rule if I'm willing to be flexible with spending?

Yes, but call it what it is. "4% with flexibility" is not the 4% rule. It's a dynamic withdrawal strategy with 4% as the starting rate. Look at Guyton-Klinger or Kitces ratcheting for published, testable versions. The unstated "I'll just cut if needed" plan rarely survives contact with an actual 30% drawdown.

How does international diversification affect the safe rate?

Modestly. Pfau's cross-country research found that retirees in countries with weaker historical equity returns (Japan, parts of Europe) needed materially lower safe rates. A globally diversified portfolio that doesn't put 100% on the U.S. market produces somewhat lower expected returns but also lower path-dependent failure risk. The effect is small (10 to 25 basis points) but mostly favorable.

What about Social Security and pensions?

Both reduce the rate you need from the portfolio, which is different from raising the safe rate. If Social Security will replace 30% of your spending starting at 67, your portfolio only needs to cover the other 70% from age 35 to 67, then 70% from 67 onward. Most modern Monte Carlo tools handle this with an "income floor" input. If yours doesn't, model the portfolio at the lower required spending after SS kicks in.

Is gold or crypto a good hedge for long-horizon retirement?

Gold has produced roughly 0% to 1% real return over the past century, which is lower than bonds, with much higher volatility. It's been a crisis hedge in some periods (1970s, 2008) and a drag in most others. Crypto has too short a history to draw any serious withdrawal-rate conclusion. Neither belongs in a core retirement portfolio at material weight if your goal is sustainable long-horizon withdrawals.

What's the difference between historical backtesting and Monte Carlo for figuring out the safe rate?

Historical backtesting uses actual sequences of returns from real market data (typically U.S. since 1926). Monte Carlo generates synthetic sequences from assumed return distributions. Both have weaknesses. Backtesting overfits to a single country's history. Monte Carlo bakes in any errors in the assumed distributions, especially for tail events. The safer practice is to run both and use the more conservative result.

How often should I revisit my safe withdrawal rate?

Annually, same date you check your portfolio. The rate itself shouldn't move much year to year, but your ability to live on it might. If a year of actual spending diverges significantly from the modeled spending, the gap between your plan and your reality is the thing to fix, not the published rate.

Closing

A 50-year retirement is not the 4% rule with extra confidence. It's a different problem with a different answer, and the answer is uncomfortable: roughly 3.25% if you want a static rate that survives most reasonable assumption combinations, or 4.0% to 4.5% if you're willing to accept the variability of a dynamic strategy.

The arithmetic is honest about the trade. The marketing material around early retirement often isn't. Bengen's original research, Pfau's extensions, and the early-retirement community's collective work all converge on the same point: the longer the horizon, the less the past 30 years of "safe rate" lore actually applies. Plan from the math, not the headline. Use the SWR Analyzer to see what your own numbers say. Costs nothing. Tells the truth.