The 4 per cent rule says that withdrawing 4 per cent of a portfolio in the first year of retirement, then raising that amount with inflation each year afterwards, historically survived thirty years. It comes from work by the financial planner William Bengen published in 1994, tested against US market returns since 1926, and later reinforced by the Trinity study of 1998.
It is a historical finding rather than a law, and the conditions it was measured under are specific enough to be worth stating before anyone applies it.
What does 4 per cent actually produce?
On a pot of $918,260.68, four per cent is $36,730.43 a year, or $3,060.87 a month. That pot is what saving $500 a month from age 35 to 67 on top of an existing $50,000 reaches at a 6 per cent nominal return.
The important second figure is what that income is worth. Deflated at 2.5 per cent inflation over the same 32 years, $3,060.87 a month has the buying power of $1,388.93 in today’s money. The projection is honest; the headline number is simply denominated in the money of 2058.
| At retirement | In today’s money | |
|---|---|---|
| Pot | $918,260.68 | $416,679.65 |
| Annual income at 4% | $36,730.43 | $16,667.19 |
| Monthly income at 4% | $3,060.87 | $1,388.93 |
What was the rule actually testing?
A narrow question: given a portfolio of roughly half US stocks and half US bonds, what starting withdrawal rate would have survived every thirty-year window in the historical record, including retirements that began just before the worst crashes in it?
The answer was slightly above 4 per cent, and 4 became the round number. Note what that question contains: one country, one asset mix, one currency, a thirty-year horizon, and no fees or taxes. Change any of those and the answer moves.
Bengen tested every historical starting year he had data for, including 1929, 1937 and 1966 — the retirements that began immediately before a crash or a decade of inflation. That is what makes the finding useful: it is not the average outcome, it is close to the worst one in the record.
Which assumptions matter most?
Four, roughly in order of how much they move the result.
- The horizon. Thirty years. Retiring at 55 with a plausible forty-year horizon is a different question, and the sustainable rate is lower.
- Fees. The study modelled index returns without costs. A 1 per cent annual charge comes straight off the withdrawal capacity.
- The market history used. US returns over the twentieth century were unusually strong compared with most other developed markets. A rule fitted to that history is optimistic elsewhere.
- Behaviour. The rule assumes you raise the withdrawal with inflation regardless of what the portfolio did. Retirees who cut spending in bad years do considerably better than the rule predicts.
How sensitive is the projection to the return assumption?
Enough that quoting a single figure without the assumption is meaningless. The same plan — $50,000 saved, $500 a month, ages 35 to 67 — at three different returns:
| Assumed return | Pot at 67 | Monthly at 4% | In today’s money |
|---|---|---|---|
| 5% | $719,221.21 | $2,397.40 | $1,087.87 |
| 6% | $918,260.68 | $3,060.87 | $1,388.93 |
| 7% | $1,180,825.39 | $3,936.08 | $1,786.08 |
One percentage point of assumed return changes the monthly income by roughly a quarter. That spread is larger than most of the decisions people agonise over inside the plan.
What does starting later cost?
More than the contributions you skip. The same $500 a month starting at 45 instead of 35 reaches $459,669.40 by 67 rather than $918,260.68 — half the pot for $60,000 fewer contributions. The missing decade contributes $398,591.28 of growth that never happens.
In monthly terms the later start delivers $890.02 in today’s money against $1,388.93. The ten years cost about 36 per cent of the retirement income.
Questions people ask
Is 4 per cent still safe? It is still the most-tested starting point and it is not a guarantee. Lower bond yields and higher valuations at the start of a retirement both reduce the margin the historical test found.
Should the withdrawal be nominal or real? The rule is a real rule: 4 per cent of the starting pot, then indexed to inflation. Taking 4 per cent of the current balance each year is a different and more variable strategy.
Does it account for a state pension? No. Any guaranteed income reduces what the portfolio has to produce, which is why the rule is a portfolio question rather than a retirement-income one.
What about tax? Not modelled. Withdrawals from a taxable account and from a sheltered one leave very different amounts in your hand from the same 4 per cent.
Is there a version for a longer retirement? Later work has explored lower starting rates for horizons beyond thirty years, along with rules that flex the withdrawal with performance. All of them trade a lower starting income for a smaller chance of running out.
Why 4 and not 4.15? Because a rule people can remember gets used and a precise one does not. The rounding down also builds in a small margin, which is the right direction for a rule of this kind to be wrong in.
Treat it as a starting point with its assumptions attached rather than as a number. The retirement calculator lets you move the return, the inflation figure and the withdrawal rate to see how much of the answer they own; the compound interest calculator shows the accumulation on its own, and the inflation calculator converts any future figure back into today’s money.