Is the 4% Rule Still Safe? What the Research Actually Found

The most famous number in retirement planning was never actually a rule. The original work never called it one. Here is what the research found — and what three decades of follow-ups added.

What a “safe” withdrawal rate means

A safe withdrawal rate is the first-year paycheck you can take from a portfolio, then usually raise with inflation, without a high chance of running out over a set horizon — classically 30 years. It is a planning shorthand. It is not a law.

The mechanic: take 4 percent of the starting nest egg in year one. Next year, take last year's dollar amount plus inflation. Do not rebase to 4 percent of whatever is left. One million dollars becomes $40,000 in year one. If inflation is 3 percent, year two is $41,200 — even if the portfolio just fell to $800,000.

What “safe” meant in the original tests

The original tests used United States stocks and bonds, about 30 years, no advisor fees, and spending that never cuts after a crash. Success meant money left at year 30. And the number that survived was the worst historical case in that sample — not the average case. In typical histories, people could have taken more and died with a surplus.

1994 — Bengen

William P. Bengen, a CFP in El Cajon, California, published “Determining Withdrawal Rates Using Historical Data” in the October 1994 Journal of Financial Planning (volume 7, number 4). He used Ibbotson U.S. stock and bond returns and inflation, starting a new hypothetical retiree in each year from the mid-1920s. Each path took a first-year percent of the starting pile, then raised that dollar amount with inflation.

Three percent lasted 50 years in every window he ran. Four percent lasted at least about 33 years in every window. Five percent blew up for late-1960s and early-1970s retirees. He later named the worst-case starting rate SAFEMAX. The press shortened it to the 4 percent rule.

1998 — Trinity

Philip L. Cooley, Carl M. Hubbard, and Daniel T. Walz of Trinity University published “Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable” in the February 1998 AAII Journal. Same idea, rolling windows from 1926 through 1995, tables by stock/bond mix. They emphasized portfolio success rates rather than Bengen's single worst-case minimum: for a 50/50 mix, 4 percent inflation-adjusted succeeded in about 95 percent of those 30-year periods. That paper made 4 percent classroom-standard.

What came after

Bengen kept rerunning. Broader stock holdings pushed his historical minimum toward the mid-4s. Fees and taxes, mostly ignored at first, push it back down. Wade Pfau (2010) showed the U.S. 4 percent minimum was not the world minimum: in 17 developed-country histories, 4 percent was “safe” in only four countries, and a 50/50 mix failed at some point in all 17. Japan's 1940 retiree is the textbook nightmare — high inflation, poor real returns.

Forward-looking shops printed numbers under 4 percent when starting valuations were high and bond yields were low. Morningstar's mid-2020s work landed in the high 3s for rigid, inflation-adjusted 30-year spending at a 90 percent model success rate — 3.7 percent in the 2024 vintage, about 3.9 percent in later updates — while noting that flexible spending supports a higher start. Bengen's later historical minimum with a more diversified portfolio is often quoted near 4.7 percent. Those are different questions: what a model of the future allows if you never cut, versus what the unluckiest U.S. cohort got away with.

How the Planner uses it

In the Planner, nest egg needed = annual income gap at retirement ÷ withdrawal rate. At 4 percent, a $36,000 annual gap needs $900,000. At 3.5 percent it needs about $1.03 million. At 5 percent it needs $720,000. Lower rate means more savings required — which means more slack against a bad first decade.

The history is saying one clear thing: 4 percent was a worst-case U.S. 30-year minimum from mid-20th-century data — not an average, not a 40-year promise, and not a global constant. Treat it as a starting point for the math, not a guarantee from the market.