Safe Withdrawal Rate (SWR)

Every safe-withdrawal study runs the same experiment. Pick a starting withdrawal rate, raise the dollar amount with inflation each year, and see whether the portfolio lasts a set horizon, usually 30 years, under many different market histories. What differs is how the results are summarized.

The worst-case approach asks for the highest rate that survived every starting year on record. William Bengen’s 1994 paper in the Journal of Financial Planning took this route and called the answer the maximum safe withdrawal rate; with US data from 1926, it came to about 4%, the origin of the 4% rule.

The success-rate approach reports the share of periods that survived instead. The Trinity study of 1998 counted overlapping historical windows, and Monte Carlo simulations count many randomly generated market paths. Under this approach, the safe rate is the highest one whose success rate clears a bar you choose, such as 90% or 95%.

Either way, “safe” has a narrow meaning: the balance stayed above $0 until the horizon ended. It says nothing about how close a run came or how much was left.

The safe rate is an output of the test, not a constant: every assumption fed in moves it, even when the market history stays the same. That is why careful studies can disagree without either being wrong. The Trinity study’s tables, built on the S&P 500 and long-term corporate bonds from 1926 to 1995, show how far the main levers move the result. Three things the classic studies left out move it too:

The best-known estimates cluster around 4%, but they come from different tests, and the differences explain most of the spread. Historical studies replay the actual order of past returns, so the few worst decades on record set the answer. Simulations generate new paths from return assumptions, so their answers depend on those inputs. FINRA’s investor guidance sums up the spread: expert opinion tends to cluster between 3% and 5% a year.

Treat much higher figures with suspicion. A FINRA fact sheet for investors describes a broker who pitched starting withdrawals of 7.5% to 9% as sustainable for more than 30 years, based on assumed returns of 11% to 14%. The rates proved unachievable, and many employees who followed the plan lost a large share of their savings.

Most research stops at 30 years, which suits someone retiring in their mid-60s. A 45-year-old taking early retirement may need the money for 45 years or more, and the evidence points lower. In Bengen’s data, first-year rates of 3% to about 3.5% lasted at least 50 years from every starting point. The Trinity authors concluded that early retirees expecting long payouts should plan on lower rates, and in the simulations GAO cited, a 4% start’s chance of lasting fell from 94.0% at 30 years to 89.4% at 35.

The horizon is not the only difference. Early retirees often have years of part-time or Barista FIRE income, and most will collect Social Security later, which cuts the withdrawal once benefits start. Money in a 401(k) or IRA can also carry a 10% additional tax if taken before 59½, unless an exception applies. Modeling those phases year by year gives a better answer than one safe rate applied to the whole span.

A safe rate summarizes the past, and the past it summarizes is thin. From 1926 to 1995 there were only 41 overlapping 30-year periods, and they share most of their years, so the same few bad stretches decide many results at once. Poor returns in the first years of retirement do the most damage, a pattern known as sequence of returns risk. All of it comes from one country’s markets, and nothing guarantees the next 30 years will resemble any 30 years on record.

The rigid spending rule cuts both ways. It assumes you keep raising withdrawals through a crash, which few people do, and that you never spend more after a boom. Bengen found that a 4% start usually lasted 50 years or longer, so most historical retirees following it would have left a large balance unspent.

So treat a safe rate as the starting point of a plan you review every year. The Trinity authors themselves expected retirees to make mid-course corrections as markets unfolded.