How much do you actually need to retire?

Start from the income, not the pot

The question is almost always asked as "how big a number do I need", and that is the harder version. The tractable version is to start from the income you want to live on and work backwards, because income is something you can actually estimate from your current life.

The standard conversion is the 4% rule: a portfolio can support an initial annual withdrawal of about four percent of its value, rising with inflation each year, and historically survived thirty years. Inverted, that means a target of roughly twenty-five times the annual income you want.

So forty thousand a year implies about a million. Sixty thousand implies about one and a half million. The multiplier does the whole job, which is why arguing about the correct withdrawal rate matters so much — moving from 4% to 3.5% raises the target from twenty-five times to about twenty-nine, and moving to 3% raises it to thirty-three.

Two adjustments make the number more honest. Subtract any income that does not come from the portfolio — a state pension, a defined benefit scheme, rental income — because you only need to fund the gap. And use the income you expect to spend rather than your current salary, which is usually lower once commuting, saving for retirement itself, and often a mortgage have stopped.

What the 4% rule actually says

The figure comes from the Trinity study and the work around it, which tested historical US market data to see what withdrawal rate a portfolio could have sustained over thirty-year periods without running out. Four percent was the rate that survived essentially all of the historical windows tested.

It is worth being precise about the claim, because it is narrower than the way it gets quoted. It is a starting rate, not an ongoing percentage — you take four percent in year one and then increase that amount by inflation, regardless of what the portfolio does. It assumes a specific stock and bond mix. It was derived from a thirty-year horizon. And it comes from one country's market history during a period that included the strongest equity returns in recorded history.

Each of those is a reason for caution. Retiring at fifty-five rather than sixty-seven means planning for forty years rather than thirty, and the safe rate falls as the horizon lengthens. Applying US historical returns to a different market, or to a future that resembles the past less than we assume, is a genuine leap.

This is why later research often lands nearer three to three and a half percent, and why treating four percent as a floor rather than a ceiling is the wrong reading. It is a useful anchor for converting income into a target. It is not a guarantee, and it never claimed to be.

Why the projected number always looks wrong

A retirement projection produces a large nominal figure — a million, two million — and the size of it is misleading in both directions. It looks like an impossible amount to accumulate, and simultaneously like more money than you would ever need. Both impressions come from comparing a future number to present prices.

Inflation is what breaks the comparison. At two and a half percent, prices roughly double every twenty-eight years. A million dollars in forty years buys what somewhere around three hundred and seventy thousand buys today. That is the number to compare against a salary, and it is the one that tells you whether the plan works.

The same arithmetic runs the other way and is more encouraging: the contributions you make late in the plan are in inflated money too, so a fixed monthly amount becomes easier to afford over time as wages rise. A projection that assumes a constant contribution for forty years is usually conservative for that reason.

The practical rule is to do all planning in today's money. Use a real return — the return after inflation, historically around seven percent for broad equities over long periods, though with brutal variation across any single decade — and compare the result directly to today's prices. Nominal projections are for impressing yourself; real ones are for deciding.

Sequence risk, and why averages lie

Two retirees can experience the same average return over thirty years and get completely different outcomes, purely because of the order the returns arrived in. This is sequence-of-returns risk, and it is the most underappreciated danger in retirement planning.

The reason is that withdrawals interact with losses. While you are accumulating, a crash early on is close to harmless and arguably helpful — you keep buying at lower prices and the recovery lifts everything you own. Once you are withdrawing, a crash early on is severe, because you are selling assets at depressed prices to fund living costs, permanently removing shares that would have participated in the recovery.

A portfolio that falls thirty percent in the first two years of retirement while still paying out an inflation-adjusted income may never recover, even if the market subsequently delivers exactly its long-run average. The same crash in year twenty-five is usually survivable. The average return across the whole period is identical in both cases.

What mitigates it is having something other than equities to spend in a bad year — commonly a cash or bond buffer covering a couple of years of expenses — so that a downturn does not force selling. Flexibility helps too: retirees who reduce withdrawals modestly during bad years historically fare far better than the fixed-withdrawal model assumes, because the fixed model is what makes early losses permanent.

The variables worth arguing about

Retirement projections are unusually sensitive to their inputs, and it is worth knowing which ones move the answer and which are noise.

The retirement age is the strongest lever, because it works from both ends at once. Working three years longer means three more years of contributions, three more years of compounding on the whole balance, and three fewer years the portfolio must fund. That double effect is why a small change in age moves the required pot far more than a plausible change in return does.

Fees are the quiet one. A one percentage point difference in annual charges does not cost one percent of the final balance — it costs one percent of the balance every year, including the growth that money would have made, and over a full career it commonly consumes something like a quarter of the final amount. That is a larger effect than most people expect from a number that looks small on a statement.

The assumed return matters, but less than the confidence people place in it. Anything between five and seven percent real is defensible, and picking a single figure conveys false precision. Modelling a range and checking whether the plan survives the pessimistic end is more useful than optimising the central estimate — because the plan you want is one that works at six percent, not one that only works at eight.

Questions