The 4% rule was never a rule, and it was never really about 4%

What the study actually did

Bengen’s 1994 work in the Journal of Financial Planning ran a fixed real withdrawal against historical United States returns, starting in every available year, and asked which starting rate never exhausted the portfolio within thirty years. The Trinity study four years later reached compatible conclusions with a different method and different data, and between them they produced the number everyone now repeats without the sentence that follows it.

The withdrawal is a percentage of the balance on day one, raised by inflation each year afterwards. It is not 4% of the current balance. Those are different policies with different failure modes, and half the arguments about the rule are two people describing different ones.

Four assumptions carry all the weight

A thirty-year horizon. Retire at 50 and you have asked a question the research did not answer, and the safe rate falls as the horizon lengthens. Second, United States market history as the sample: that is the most successful large equity market of the twentieth century, which makes the result a survivorship-flavoured estimate rather than a general law about stocks.

Third, rigid spending. The simulation keeps withdrawing the same real amount into the fifth year of a bear market, which no living person does. Fourth, no fees and no taxes. A 1% total expense charge does not reduce the safe rate by 1%, but it comes off the top of a number that was already the worst case, and it is left out of almost every retelling.

Sequence of returns is the actual mechanism

Two retirees with identical average returns over thirty years can finish in completely different places if the bad years fall in different places. Withdrawals during a decline sell more shares to raise the same money, and those shares are not there for the recovery. Early losses are therefore far more damaging than late ones, even when the average is unchanged.

That is why the safe rate is a worst-case figure rather than something derived from the average return. The binding constraint is not what markets did over thirty years, it is what they did in the first ten. Anyone quoting an expected return as evidence that a higher withdrawal is fine has not engaged with the mechanism.

• Decide the horizon before the rate, because the rate depends on it.
• Subtract fees and taxes from the portfolio return before applying any rule.
• Test the same average return in a different order and watch the outcome move.
• Write down in advance what you would cut, and by how much, after a 30% fall.

What to use it for instead

Our position: 4% is a decent sizing heuristic and a poor withdrawal policy. As a target it gives you 25 times annual spending, which is a reasonable place to aim. As a policy it commits you to ignoring the single largest lever you have, which is the ability to spend less for two years.

Flexibility is worth more than portfolio construction here. A retiree who will cut real spending by 10% after a bad year can support a materially higher starting rate than one who cannot, and the reverse is also true: rigid spending in a long retirement makes 4% look optimistic rather than conservative. Model the cut you would actually make, and stop treating a single decimal as though it were a law.