Global edition
Market data — delayed ~15 min · Last updated: November 16, 2026, 7:44 pm CST
S&P 5007,416.59-1.92%Nasdaq 10027,197.15-2.67%Dow Jones51,608.17-1.41%Shanghai Comp3,634.35-0.59%Nikkei 22559,078.52-3.38%FTSE 10010,902.60+0.36%DAX26,749.95-0.08%CAC 408,368.91+0.09%Hang Seng23,637.60-1.55%ASX 2008,906.20-1.94%Sensex81,592.45-0.39%
← Back to briefing Global edition · Explainers
Explainers

How Much Money Do You Need to Retire? How the Number Is Actually Built

The 25-times rule and the 4% rule are the same arithmetic read in opposite directions. Where they came from, what they assume, and their limits.

7 min read

Ask how much money you need to retire and you will be offered a number, usually a round and frightening one, as if it were a fact about the world rather than the output of a calculation. It is always a calculation. Every retirement target, from the crudest rule of thumb to a planner’s forty-page model, is built the same way: estimate what you will spend each year, subtract the income that arrives whether or not you saved anything, and multiply the gap by the number of years it has to last. The famous rules — save twenty-five times your spending, withdraw 4 percent a year — are just that arithmetic with the assumptions pre-filled. Understanding the assumptions is worth more than memorizing the number, because the assumptions are where your retirement differs from the average one baked into the rule.

The 4 percent rule is the ancestor, and its origin is more specific than its reputation suggests. In 1994 a financial planner named William Bengen published a paper in the Journal of Financial Planning, “Determining Withdrawal Rates Using Historical Data,” that asked a narrow question: starting with a portfolio split between stocks and bonds, what is the largest share of it a retiree could withdraw in the first year — raising the dollar amount with inflation every year after — and never run dry over thirty years, in any period of American market history? He ran the test against every retirement start date the data allowed, back to 1926. The binding case was the retiree of 1966, who walked straight into a bear market followed by a decade of high inflation, the worst sequence in the record. Even that retiree’s portfolio survived at a starting withdrawal rate of about 4.15 percent, which Bengen rounded down and which the profession shortened to the 4 percent rule. Four years later, in 1998, three Trinity University researchers — Philip Cooley, Carl Hubbard and Daniel Walz — published a broader version of the test in the AAII Journal, checking many mixes of stocks and bonds and reporting how often each withdrawal rate succeeded. Their work, the Trinity study, found that a 4 percent inflation-adjusted withdrawal lasted thirty years in roughly 95 percent of historical periods for portfolios with a substantial stock allocation. Two details of the rule are lost in most retellings. The 4 percent applies to the starting balance only; after year one you withdraw a fixed, inflation-adjusted dollar amount, not 4 percent of whatever the portfolio happens to be worth. And the rule was engineered against the worst history on record, not the average — it is a floor built from a stress test, which is a different thing from a forecast.

The 25-times rule is the same research read backward. If a portfolio can sustainably pay out 4 percent of its starting value each year, then funding a given amount of annual spending requires twenty-five times that amount: 4 percent of twenty-five is one. A household that needs its savings to supply forty thousand dollars a year needs, by this arithmetic, a million dollars. The multiplier is not a separate discovery and carries no independent authority — it inherits every assumption of the rule it inverts, including the thirty-year horizon, the stock-and-bond mix, and the silence on taxes and fees. Used carefully it is a fast way to convert a spending estimate into a savings target. Used carelessly — applied to gross income instead of to the spending the portfolio must actually fund — it produces targets far larger than the underlying research ever demanded, because almost nobody’s portfolio has to replace their entire paycheck.

That last point is where the second tradition in retirement math begins: replacement-rate thinking. Instead of starting from a portfolio, it starts from a paycheck and asks what fraction of it retirement spending actually consumes. The fraction is below a hundred for structural reasons. A retiree no longer pays payroll taxes, no longer saves for retirement, and no longer spends on commuting, work clothing or raising children; against that, healthcare spending rises and the mortgage may or may not be gone. Planning practice commonly lands on a replacement target somewhere in the range of 70 to 80 percent of pre-retirement income, with the honest caveat that it is a convention, not a measurement — a household that retires with a paid-off house and modest tastes can live well below it, and one carrying rent and debt into retirement can blow through it. The two traditions reconcile once you see that they are measuring different things: the replacement rate estimates the spending, and the 25-times rule then prices the share of that spending your savings must carry. They give different answers only when they are fed different definitions of the same household’s life.

What shrinks the required portfolio, often dramatically, is the income that shows up on its own. For most American households the large item is Social Security, and its scale is set by formula rather than by saving. The Social Security Administration’s actuaries calculate that the program replaces roughly 40 percent of career-average earnings for a medium earner who claims at full retirement age — a higher share for lower earners, a lower share for higher ones, by deliberate design. In cash terms, the average retired worker’s benefit was $2,092 a month in October 2026, about twenty-five thousand dollars a year, according to the SSA’s Monthly Statistical Snapshot. Run that through the two frameworks and the effect is concrete. A household whose retirement spending totals sixty thousand dollars a year, and whose Social Security covers twenty-five thousand of it, needs its portfolio to fund thirty-five thousand — which the 25-times rule prices at about $875,000, not the million and a half that multiplying the full sixty thousand would suggest. Pensions, a spouse’s benefit, part-time work in the early years and rental income all enter the same way: every dollar of dependable income is a dollar the portfolio never has to produce, and at twenty-five-to-one leverage, every thousand dollars a year of it removes twenty-five thousand from the target.

What remains is everything the rules leave out, and the list is the real content of retirement planning. Taxes sit first: withdrawals from traditional pre-tax accounts are taxed as income on the way out, so a dollar in a 401(k) is not worth a full after-tax dollar of spending, which is one reason the tax treatment compared in our explainer on Roth versus traditional IRAs changes the size of the number you need, not just its location. Longevity sits beside it — the research horizon is thirty years, but a retirement begun at sixty can run longer, and a couple retiring at sixty-five faces a meaningful chance that at least one of them is still spending at ninety-five. Healthcare is the expense that grows fastest and arrives latest, from Medicare premiums and out-of-pocket costs in the healthy years to the possibility of paid care at the end. Inflation compounds quietly underneath everything: at even moderate rates, the cost of a fixed lifestyle roughly doubles across a thirty-year retirement, which is why any serious version of the calculation is done in inflation-adjusted terms. And the spending itself is not flat — most households spend more in the active early years, less in the slower middle, and more again if health fails, a curve no single multiplier pretends to capture.

None of this makes the rules of thumb useless; it makes them starting points with visible seams. Bengen himself later published higher sustainable rates after broadening the portfolio beyond his original mix, and researchers working from forward-looking return assumptions rather than history have published lower ones — the disagreement is about the future’s resemblance to the past, and it cannot be settled by arithmetic. What arithmetic can do is show you which inputs your own number is sensitive to. The retirement calculator on this site runs the two halves of the problem against each other — what your savings could grow into, and what your desired income costs after inflation — so you can watch the target move as the assumptions change, which is the honest way to hold a number that is, in the end, a range wearing a point estimate’s clothing. Build it from your spending, subtract the income that arrives anyway, price the gap, and then argue with the assumptions. That argument, not the multiplier, is the plan.

Educational coverage — not tax, financial, or investment advice.

The Global Close, in your inbox

Markets, briefed. What moved, why it matters, and what to watch next — every weekday morning. Free.

See Market Briefer first in Google

Add to Preferred Sources
Financial disclaimer: This article is for information and education only. It is not investment, legal, tax or accounting advice and does not recommend any transaction. Market data is delayed by approximately 15 minutes.