The Withdrawal Rate That Survived Every Historical Crash

Four percent is safe because it always has been. That is the belief, and it is the one that gets people hurt, because the word always is carrying a great deal of weight that most of it cannot hold.

a $1,000,000 portfolio, $40,000 pulled out in year one, and a rule everyone says survived every crash. Four percent is not a rule. It is a historical result from US stock and bond studies, and it travels with conditions attached: a 60/40-ish mix, a 30-year horizon, withdrawals rising 2.5% a year for inflation, and fees small enough not to matter. The Trinity study and the Bengen updates carry the survival claim, or a calculator like FIRECalc or cFIREsim. The headline carries none of it.

A defensible rate is the one that still works when you test it against 1929, 1966, 1973 and 2000 on your own mix, your own 30 years, and Social Security at 67, not the one that survived somebody else's retirement.

That is the whole job here: a starting rate you can defend, the $1,000,000 example run through the worst start years, and one spending guardrail written down before the first crash rather than during it.

The Benchmark Rate Is a Starting Point

Four percent survived the historical record, but only for the particular plan those studies tested.

Set the plan on the table: $1,000,000 you can actually invest, split 60% stocks and 40% bonds, fund fees running about 0.30% a year, and a retirement you expect to last 30 years. Take $40,000 out in year one and that is 4.0% on the nose. No pension in this picture. Social Security pays $2,400 a month starting at 67, and taxes come out of the $40,000 separately rather than sitting inside it.

Where that number came from matters more than the number. Researchers ran US market history through a stock-heavy portfolio held at low cost over 30-year windows, and 4% cleared almost all of them. Anyone repeating the figure should check the papers and calculators behind it directly.

The inflation step was baked into the test, not bolted on afterward.

Withdrawals were assumed to climb 2.5% each year, so year two is $41,000, year three a little more than that. Assume a steeper raise, say 3.5%, and you have swapped in a harder exam with a lower passing score.

Treat $40,000 as the opening bid. Every section ahead puts that dollar figure under pressure and reports back what survives.

Your Horizon Changes the Safe Rate

A retirement that runs 40 years instead of 30 needs a lower starting rate, because the last decade of withdrawals has no fresh money behind it.

Run the same $40,000 out over three decades and you are testing the plan those studies actually graded. Stretch the finish line to 40 years and you add ten more years of withdrawals, plus ten more rolls of the dice on a crash landing in the opening years, when the damage is hardest to recover from.

The calculator tells you the difference in failures, not averages. A rate that came through every 30-year start on record can still run dry in several 40-year ones, and that count is the answer you are looking for.

Bringing Social Security forward changes the arithmetic too. At 67, $2,400 a month covers $28,800 a year, so the portfolio is only funding the gap plus the taxes on the full $40,000. Retire at 60 and the portfolio carries the whole $40,000 for seven years with no benefit check arriving until 67.

Asset Mix Moves the Survival Odds

Your mix changes the odds because the 4% figure quietly assumes 60/40, and an 80/20 or a 50/50 holder is running a different historical test with a different survival percentage for the same $40,000. Put 60/40 next to 80/20 and the 80/20 carries a higher expected return; it lasts longer across average history. But its early crashes are deeper, so the spread of outcomes around that $40,000 widens in both directions, better on the good runs and worse on the bad ones.

In a bad first stretch, the withdrawal gets paid from whatever is up.

That is the mechanism people miss. The bond side of a 60/40 is what keeps shares from being sold at the bottom of a crash, and the mix, not just the average return, decides survival. An 80/20 owner in 2000 spent down stocks to fund spending while the market fell, and those shares never came back for the recovery. Two years of $40,000 held in Treasury bills or a money market fund, $80,000 in all, cushions that sequence risk and removes any forced stock sale in a crash, though $80,000 parked at T-bill yields instead of market returns drags on growth if it grows past two years of spending.

Test the rate on the mix you actually hold. The generic 60/40 is somebody else's test.

Stress-Test Against the Worst Start Years

A withdrawal rate gets tested against a crash by running the plan through the specific start years where early damage broke retirees, not by looking at average returns. You do not take the mix and horizon you have now and ask whether the average year works. You drop that plan into 1929, 1966, 1973 and 2000 and see whether it is still standing thirty years later.

Those four years are not famous by accident. A person retiring in 1929 put their first withdrawals into a market that was collapsing underneath them, and 2000 did the same thing to a different generation. 1966 and 1973 broke people a different way: the portfolio went sideways in real terms while rising prices pushed the withdrawal up every single year.

Pass or fail is binary, and that is the point. If the $1,000,000 holds above zero through year 30 in all four runs, the rate survives. If even one run hits zero, the starting number comes down and the entire set gets rerun at the lower figure. There is no partial credit and no averaging away the bad one.

Use FIRECalc or cFIREsim, and check how the calculator is set before you trust a single result. The fee at 0.30%, the 2.5% raise in withdrawals, the 30-year length, the $80,000 in Treasury bills and Social Security starting at 67 all change the outcome. Enter Social Security as an offset of $2,400 a month from 67, and the calculator is testing what the portfolio really carries: $40,000 minus $28,800, plus the taxes you owe on the whole amount.

Run the lowest start year first.

Fees and Inflation Quietly Eat the Rate

A plan fails on flat market returns because fees and inflation raise the cost of the same lifestyle year after year, and the starting rate has to cover both. Market return is only one input. The other two never appear in a headline and never stop moving.

Start with the fee, which is easy to dismiss at 0.30%. On $1,000,000 that is $3,000 in year one, pulled off the balance before you withdraw a dollar, and it is taken again every year on the same portfolio. Each extra 0.10% costs another $1,000 annually. Pay 0.80% instead and you have handed the fund $8,000 in year one, which is a fifth of your entire $40,000 withdrawal gone before you have bought anything.

Inflation is the quieter one, because it works on your spending rather than your balance. The $40,000 of year one becomes $41,000 in year two, and by year 10 you are withdrawing about $50,000. That is nearly a quarter more money leaving the portfolio each year, and it happened without a single bad market year in between.

Now stack on the thing no backtest can predict: a health event, a roof, a family member who needs help. Any of those pushes the spending path above the one the calculator tested, and the test that passed at 0.30% and 2.5% is no longer the test you are running. The gap gets paid out of the rate.

This is why the defensible starting number under heavier costs sits below 4.0%. How far below is a question for the calculator with your real fee and your real inflation assumption entered. The headline cannot answer it.

Set a Guardrail Before the First Crash

The answer to a crash in your first year is a spending rule you wrote down before retirement, because the decision is impossible to make well while the balance is falling. Pick the rule now, while the numbers are calm and the portfolio is whole.

Two versions work. The first: if the portfolio drops 20%, cut the withdrawal 10%, which takes $40,000 to $36,000 the following year. The second, gentler: freeze the inflation raise for one year, so $41,000 stays at $41,000 instead of climbing to roughly $42,025.

Then lock in the asymmetry, which is the part people skip. You only raise spending again after the portfolio has recovered its inflation-adjusted high. A cut made this way is temporary, and you have not spent the recovery in advance before it has actually shown up.

The $80,000 in Treasury bills or a money market fund is what makes the rule survivable: two years of $40,000 sitting outside the market means a crash does not force you to sell stocks at the bottom. Refill that reserve in the good years, before you raise your spending, not after. The cash bucket rule shows how to size that reserve so a crash never forces a sale.

This is the piece you control. A rigid 4.0% has no answer to a bad first decade. A written rule for cutting and restoring spending is the difference between a backtest that looks fine on paper and a plan you can actually follow when the statements turn red. Work out the exact trigger with your own numbers, or run it past a fee-only fiduciary before you commit to it.

Frequently Asked Questions

Is 4% still safe if I retire at 60 instead of 65?

Probably not at 4.0% if your horizon stretches to 40 years. A rate that clears every 30-year historical start can fail several 40-year starts, because you have ten more years of withdrawals and ten more chances for a bad early decade. The fix is either a lower starting rate or a written spending guardrail, not a different stock mix.

Does my portfolio mix change the safe withdrawal rate?

Yes. A 60/40 mix and an 80/20 mix produce different survival percentages for the same $40,000. More stocks means a higher expected return but deeper early crashes, which widens the range of outcomes. Run the calculator with your real allocation, not a generic one.

How do I test a withdrawal rate against a crash?

Test by start year, not average return. Run your plan starting in 1929, 1966, 1973, and 2000, and check that the portfolio stays above zero through year 30 in every run. Use FIRECalc or cFIREsim with your actual fee, inflation rule, and Social Security start date.

Why does my plan fail if the market returns are fine?

Because fees and inflation quietly raise the cost of the same lifestyle. A 0.30% fee on $1,000,000 is $3,000 in year one and grows with the balance; a 2.5% inflation raise takes $40,000 to about $50,000 by year 10. Higher fees or higher inflation shift the whole path up, and the starting rate has to absorb them.

What do I do if the market crashes right after I retire?

Follow the guardrail you wrote before retirement. Cut the withdrawal by 10% after a 20% drop, or freeze the inflation raise for one year. Keep two years of withdrawals in Treasury bills or a money market fund so you don't sell stocks at the bottom, and raise spending again only after the portfolio recovers its inflation-adjusted high.

Pick Your Number, Run It, Write One Rule

Start with a rate. For a 30-year horizon on a 60/40 mix at 0.30% fees, $40,000 on $1,000,000 is the benchmark. Run it through FIRECalc or cFIREsim with your actual asset mix, your real Social Security start date, and your own fee and inflation assumptions. If 1929, 1966, 1973, or 2000 break the plan, lower the rate and rerun.

Then write one guardrail before you retire: the withdrawal cut you'll make after a 20% drop, and the rule for when spending goes back up. That's the part you control, and it's what turns a historical backtest into a plan you can actually follow.