The 4% Rule Isn't Dead — Here's What Actually Works in 2026

The headline that travels — “the 4% rule is dead” — is the error. William P. Bengen did not publish a law of spending in 1994. He published a 30-year worst-case starting rate: take 4% of the portfolio in year one, then last year’s dollars plus inflation, and ask whether a balanced U.S. portfolio survived every historical sequence he had. It did. Cooley, Hubbard, and Walz did not publish a 95% probability that your retirement works. They published Table 3: the share of 41 overlapping U.S. windows, 1926–1995, in which an inflation-adjusted withdrawal did not hit zero.

What actually works in 2026 is that method on the horizon you have, plus a spending rule you will obey when the first decade is ugly: a 3.5% FIRE floor, a 4.0% 30-year reference, and a higher Guyton–Klinger start only if you will take the cut. No member survey. The cells below come from the papers.

What Bengen measured

Bengen, “Determining Withdrawal Rates Using Historical Data,” Journal of Financial Planning (October 1994), used Ibbotson Associates’ Stocks, Bonds, Bills and Inflation: 1992 Yearbook. Stocks are common stocks. Bonds are intermediate-term Treasuries. Inflation is CPI. He assumed annual rebalancing, no investment fees, a 1 January start, and withdrawals at year-end. He plotted portfolio longevity: years until a first-year withdrawal, then inflation-adjusted, exhausted the account.

On a 50/50 mix:

• 3%, and “approximately 3.5%,” never produced a life under 50 years. He called 3% “absolutely safe” to the extent history is a guide.
• 4% never exhausted the portfolio before 33 years. Most cohorts lasted 50 years or longer. That is the sentence the “4% rule” compressed.
• 4.25% could exhaust in as little as 28 years if past conditions repeated.
• 5% left late-1960s and early-1970s retirees with about 20 years. He called 5% “risky” and 6% or more “gambling” for a new retiree.

After year one, the 4% figure is not reapplied to the new balance. The withdrawal is last year’s dollars, adjusted for that year’s inflation. A 4% start that meets a down market becomes a higher current rate on a smaller pile. That is the mechanism, not a footnote.

The worst sequence was not 1929. Bengen’s “Big Bang” was 1973–74: a stock crash plus high inflation, which raised the next year’s dollar withdrawal while the portfolio was down. The Depression (“Little Dipper”) was cushioned by deflation. He advised a stock allocation as close to 75% as the client could hold, and in no case less than 50%. Holding too few stocks shortened the worst-case life more than holding too many.

For a client age 60–65, he wrote, the highest rate that satisfied a 30-year minimum life was “usually about 4 percent.” He already flagged early retirees (age 60 or younger) to stay at no more than 4% in the early years. He did not say 4% was dead. He said it was the worst-case start for a 30-year clock.

Trinity Table 3 is 41 cohort counts

Cooley, Hubbard, and Walz, “Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable,” AAII Journal (February 1998), counted portfolio success rates: the percentage of overlapping payout windows in which the ending value exceeded $0. Data: Ibbotson 1996 Yearbook, 1926–1995. Stocks are the S&P 500. Bonds are long-term, high-grade corporates — not Bengen’s intermediate Treasuries. They did not adjust for taxes or transaction costs. They studied 15-, 20-, 25-, and 30-year windows. There are 41 overlapping 30-year periods in that sample.

Table 3 is the inflation-adjusted scoreboard. Withdrawals rise or fall with CPI each year. Numbers are rounded to the nearest whole percentage.

Read a cell as a count. A 95% success rate on 41 windows is 39 of 41. A 98% cell is 40 of 41. It is not a 95% probability that a 2026 FIRE plan works, and it is not a Monte Carlo confidence level.

The authors were explicit. On Table 3, 3% to 4% “continue to produce high portfolio success rates for stock-dominated portfolios.” In the conclusion those rates “represent exceedingly conservative behavior.” They also wrote: “Early retirees who anticipate long payout periods should plan on lower withdrawal rates.” Selection of a withdrawal rate “is not a matter of contract but rather a matter of planning,” so “mid-course corrections likely will be required.”

Do not mix those cells with Guyton–Klinger’s 99% Monte Carlo confidence or with a variable-percentage rule that cannot hit zero before its table horizon. Those are different claims.

FIRE changes the horizon, not the method

Bengen’s 1996 follow-up, “Asset Allocation for a Lifetime,” extended the clock. As summarized by Kitces (“Adjusting Safe Withdrawal Rates To The Retiree’s Time Horizon,” 2012): the historical safe initial rate fell from 4.1% at 30 years to 3.5% at 45 years, and rose to 5.1% at 20 years. Blanchett (2007) also showed 3.5% for a 40-year horizon. Pfau (2012), via the same Kitces piece, estimated 3.3% at a 95% Monte Carlo confidence level for 40 years.

Kitces’s reading of the longer-horizon research: the safe rate does not keep falling past about 40–45 years. The constraint is surviving the first one to two decades of poor real returns. If the withdrawal is low enough to clear that stretch, later bull cycles rebuild the pile, and extra years stop being the binding risk. That is why 3.5% functions as a floor in the U.S. data we have — not because anyone proved a 60-year law.

For FIRE, Kitces restated the arithmetic in “Flexible Spending Rules To Avoid FIREing At 4%.” A 40- to 50-year clock is a 3.5% problem if you insist on a constant real withdrawal. 25× spending (the inverse of 4%) becomes 28×, or 30× if you want a thicker buffer. On a $1 million 60/40 portfolio at 3.5% for 50 years, he reports a 90% chance of finishing above $3 million and a 50% chance of finishing above $9.3 million. The left tail is why the rate is 3.5%. The right tail is why a fixed 3.5% leaves most paths unspent.

That is not “4% is dead.” It is “4% was built for 30 years; your clock is longer; the same method gives a lower start.”

Sequence risk is the failure mode

A 30-year average return does not set the safe rate. The order does. Bengen’s planner who used 10.3% stock and 5.1% bond compound averages, subtracted 3% inflation, and spent the 5% “real return” is the cautionary tale that opens the 1994 paper. Averages hide the year the withdrawal and the market move against you together.

Forward-looking papers that cut the safe rate because starting valuations are high are describing the environment 4% was built to survive: a bad first 10–15 years. CAPE is useful for that stretch. It is a weak 30-year return forecast. We are not inventing a 2026 CAPE, a 2026 Treasury yield, or an 18–24% “current failure risk.” Those cells are not in the papers we opened. A constant 4% on a 50-year clock is the wrong tool for a sequence problem. A written spending rule is the right one.

Three systems that actually work

Pick one. Write it down. Do not mix their success numbers.

1. Horizon-matched constant-dollar

Use Bengen’s method: a first-year percentage, then last year’s dollars plus inflation. Change only the percentage to match the clock.

• 30-year traditional retirement, 50–75% equities: 4.0% of the starting balance. That is Bengen 1994 and Trinity Table 3 at 95–98% of 41 U.S. windows.
• 40- to 50-year FIRE, same mix, no planned cut: 3.5%. That is Bengen 1996 / Kitces / Blanchett as a historical floor, not a new EarlyFIRE study.

On $40,000 of annual spending, 4.0% is a $1,000,000 target (25×). 3.5% is $1,142,857 (about 28.6×). 30× is $1,200,000. Those are inverses of the rate, not a third paper.

The failure mode is depletion in the left-tail sequence. The cost is the unspent right tail Kitces quantified. If you will not cut spending, this is the system. Do not announce a 5% lifestyle and hope the bull market arrives first.

2. Guardrails, only if you will take the cut

Guyton and Klinger, “Decision Rules and Maximum Initial Withdrawal Rates,” Journal of Financial Planning (March 2006), tested Guyton’s 2004 rules with Monte Carlo returns over 40 years (assumptions from 1928–2004 and 1973–2004). Two guardrails:

• Capital preservation. If the current withdrawal rate — this year’s dollars divided by this year’s portfolio — rises more than 20% above the initial rate, cut this year’s dollar spending by 10%. The rule expires 15 years before the maximum age you plan to.
• Prosperity. If the current rate falls more than 20% below the initial rate, raise dollar spending by 10%.

The 2004 withdrawal rule still applies: skip the CPI raise after a year with a negative portfolio return. There is no make-up.

For portfolios with at least 65% equities, they concluded 5.2–5.6% initial withdrawal was sustainable at a 99% confidence standard over 40 years. At 50% equities the maximum initial rate dropped as low as 4.6%. Those are simulated confidence levels under the rules, not Trinity rolling-period success rates. The higher start is the reward for agreeing, in writing, to take the 10% cuts. Kitces has warned that cuts can stack in a long bad sequence. If you will not take them, you do not have this system. You have a 5.2% constant-dollar plan that neither paper supports.

On $1,000,000 at 5.2%, year one is $52,000. The cut trigger is a current rate of 6.24% (portfolio about $833,000 if the dollar withdrawal is still $52,000). The raise trigger is 4.16% (portfolio about $1,250,000).

3. Floor plus upside

Kitces’s ratchet: keep a Bengen-style floor. Raise real spending 10% only when the portfolio is more than 50% above its start, and at most once every three years. In the bad sequences the 50% threshold never hits, so the floor never moves. In the common sequences you spend some of the right tail. He calls this weakly dominant versus a frozen 4%: historically not worse, usually better.

Pair it with a spending split. Fund food, housing, insurance, and the basic life at 3.5% of assets. Let travel and upgrades float with the portfolio and with post-FI earnings you have already earned. Kitces’s capitalization example: $20,000 a year of side income, at 4%, is $500,000 less portfolio — if that income persists. Do not underwrite the floor with a consulting gig you have not landed.

On $1,000,000, a 3.5% floor is $35,000. The first 10% ratchet is $38,500, and it waits until the portfolio is above $1,500,000.

One spending number, three first-year checks

$40,000 of planned spending, 60/40, no pension in the first decade.

The 5.2% start needs $230,769 less capital than the 4.0% 30-year target for the same $40,000 year one. That gap is not free. It is prepaid in the cuts. If the first decade looks like 1966–1974 or 2000–2009, you take them, or you have broken the only thing that made 5.2% defensible.

What 4% is still for

Use 4.0% as the 30-year reference on EarlyFIRE Target FIRE. Use 3.5% as the FIRE floor. Compare the two dates. That is the band.

4% is still the right conversation number for a 60–65-year-old with a 30-year clock, 50–75% equities, and no plan to cut. Trinity Table 3 says 39 or 40 of 41 U.S. windows survived it on stock-dominated mixes. Bengen says no 50/50 path died before year 33. Declaring that dead requires a future worse than every 30-year U.S. sequence in the sample. We do not have that evidence, and we will not invent it.

4% is the wrong single number for a 40-year-old who will not cut and who needs the portfolio for 45–50 years. The same papers, plus Bengen 1996 via Kitces, put that person at 3.5% unless they adopt a variable rule.

Fees, taxes, and a non-U.S. equity/bond mix are outside both original studies. They do not “kill” 4%. They mean your personal start should be at or below the historical U.S. cell, not above it.

This week

1. Write the failure you will not accept: depletion (3.5–4.0% constant-dollar, optional ratchet) or a 10% lifestyle cut (guardrails at 5.2%). If you cannot say the cut out loud, you do not get the 5.2% start.
2. Compute both portfolio targets on last year’s actual spending: spending ÷ 0.040 and spending ÷ 0.035.
3. Put the rule in one sentence on a page: “Year-one withdrawal is X% of the 1 January balance; thereafter last year’s dollars plus CPI, unless [trigger].”

Then open the EarlyFIRE Target FIRE calculator on earlyfire.quest and rerun the same spending at 3.5% and 4.0%. Use the band. Do not pick the date you like and ignore the other.

Sources

• Bengen, “Determining Withdrawal Rates Using Historical Data,” Journal of Financial Planning (October 1994). Opened PDF.
• Bengen, “Asset Allocation for a Lifetime,” Journal of Financial Planning (August 1996), via Kitces, “Adjusting Safe Withdrawal Rates To The Retiree’s Time Horizon” (1 May 2012). Opened Kitces page; 1996 PDF not opened.
• Cooley, Hubbard, and Walz, “Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable,” AAII Journal (February 1998), Table 3. Opened PDF.
• Guyton and Klinger, “Decision Rules and Maximum Initial Withdrawal Rates,” Journal of Financial Planning (March 2006). Opened PDF.
• Kitces, “Flexible Spending Rules To Avoid FIREing At 4%.” Opened page.
• Pfau (2012) and Blanchett (2007), via the opened Kitces 2012 horizon piece.