The 4% rule was a worst case, not a target: two papers, read closely
Bengen's 1994 paper never used the word SAFEMAX and called 4% the rate that survived the worst 30 years on record. The Trinity tables add the caveats.
- What happened
- The “4% rule” rests on two papers: William P. Bengen's October 1994 article in the Journal of Financial Planning, and the February 1998 AAII Journal article by three Trinity University professors. Read in full, they say something narrower than the slogan: 4% was the highest first-year withdrawal that survived every 30-year start date in US data from 1926 in Bengen's 50%–75% stock portfolios; the Trinity table, with inflation adjustment, shows 95% of 30-year starts surviving at 50/50 and 100% at 75/25, before taxes and fees.
- Why
- Both papers replay a fixed, inflation-adjusted withdrawal through every historical start year and ask how often the money lasted. The 4% figure is set by the single worst start (the late 1960s, into the 1973–74 bear market and its inflation), not by an average outcome.
- The numbers that matter
- 4%Bengen's 30-year “should be safe” first-year withdrawal, 199433 yearsShortest portfolio life he found at that rate100%30-year success, 4% inflation-adjusted, 75/25 mix, 1926–19974.7%Bengen's own updated worst-case figure, May 26, 2025
- Why it matters
- A household applying the rule to $1,000,000 withdraws $40,000 in year one and then raises the dollar amount by inflation, whatever the portfolio does. At 3.4% inflation that is $41,360 in year two and $42,766 in year three.
- What happens nextas we saw it on 28 Sep 2026
- The next inflation figure such a household would use, the September 2026 CPI, is scheduled for release by the Bureau of Labor Statistics on October 14, 2026 at 8:30 a.m. ET.
Every 30-year start from 1926 to 1968 survived a 3% or 4% inflation-adjusted withdrawal; at 5% one start in seven failed, and at 6% more than a third did.
| 3% withdrawal | 100% |
|---|---|
| 4% withdrawal | 100% |
| 5% withdrawal | 86.0% |
| 6% withdrawal | 63.0% |
The word “SAFEMAX” does not appear in William P. Bengen's 1994 paper. Neither does the phrase “4% rule”. The article that launched the most repeated number in retirement planning, “Determining Withdrawal Rates Using Historical Data”, published in the Journal of Financial Planning in October 1994, contains one sentence on the subject. It says that, assuming a client needs the portfolio to last a minimum of 30 years, “a first-year withdrawal of 4 percent”, followed by inflation-adjusted withdrawals in later years, “should be safe”.
That is the whole rule. The name, the certainty and the idea of 4% as a spending target rather than a survival floor were added later by others. We read both papers and the authors' follow-ups to set down what they tested and what they left out.
What Bengen tested in 1994
Bengen took annual returns and inflation from Ibbotson Associates' Stocks, Bonds, Bills and Inflation: 1992 Yearbook. His illustrative portfolio was 50% common stocks and 50% intermediate-term Treasury notes, an allocation he called arbitrary, and he repeated the exercise at 0%, 25%, 75% and 100% stocks. For each withdrawal rate and each retirement beginning on January 1 of a year from 1926 onward, he counted how many years the portfolio lasted, charting up to 50 years.
The withdrawal is set once, as a percentage of the starting balance, then adjusted each year for actual inflation. The portfolio's later value never enters the calculation, which is why the results are so sensitive to the order of early returns.
A 3% first-year withdrawal was, in his phrase, absolutely safe to the extent history is a guide: no historical start left the portfolio with less than 50 years of life. At 4% with the 50/50 mix, he wrote that no past case had exhausted a portfolio before 33 years. The binding case was a retirement beginning in the late 1960s, which ran into what he called the “Big Bang”: the 1973–74 bear market, made worse by the high inflation around it. He listed 1937, 1946, 1969, 1973 and 1974 as the other “black hole” start years.
Bengen wrote that the 50/50 mix looked near-optimal for the minimum portfolio life, but his stated recommendation was a stock allocation as close to 75% as possible and in no case below 50%. At 4%, the choice between the two moved the 1966 retiree's portfolio life between 33 and 32 years and the 1969 retiree's between 36 and 34.
His practical counsel was that clients should withdraw at no more than 4% in the early years of retirement, especially if they retire at 60 or younger: a ceiling for early retirees, not a universal spending rate.
What the Trinity study added in 1998
Four years later, three professors at Trinity University in San Antonio — Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz — published “Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable” in the February 1998 AAII Journal. (The AAII web page credits the piece to Hubbard alone; all three are named on their 1999 and 2011 papers.) They replaced years of portfolio life with the portfolio success rate: the share of past payout periods in which the portfolio ended with a positive balance.
The 1998 study used annual data from the Ibbotson 1996 Yearbook, the S&P 500 for stocks and long-term high-grade corporate bonds, and tested five allocations (100% stocks, 75/25, 50/50, 25/75, 100% bonds) over 15, 20, 25 and 30 years: 56 overlapping 15-year periods and 51 20-year periods. It ran two versions, fixed-dollar withdrawals and withdrawals raised each year by the CPI, and did not adjust for taxes or transaction costs.
The often-quoted line — that 3% and 4% withdrawal rates were extremely unlikely to exhaust any portfolio of stocks and bonds in the periods shown — refers to the first table, the one with fixed dollar withdrawals. On the inflation-adjusted version the authors were blunter: retirees who insist on CPI-adjusted withdrawals must accept a substantially lower rate.
On the AAII page the four tables are picture files, which we could not read into text. The same authors published the analysis, extended by two years and run with monthly withdrawals, as “Sustainable Withdrawal Rates From Your Retirement Portfolio” in Financial Counseling and Planning, vol. 10(1), 1999. It covers January 1926 to December 1997, 43 overlapping 30-year periods, with text tables. The figures below are from it.
| First-year withdrawal | 100% stocks | 75/25 | 50/50 | 25/75 | 100% bonds |
|---|---|---|---|---|---|
| 3% | 100 | 100 | 100 | 100 | 79 |
| 4% | 98 | 100 | 95 | 74 | 19 |
| 5% | 81 | 86 | 70 | 26 | 16 |
| 6% | 65 | 63 | 51 | 19 | 12 |
| 7% | 56 | 47 | 19 | 7 | 0 |
Two things stand out. The 4% row is not uniformly 100%: an all-stock portfolio failed in 2% of 30-year starts, 50/50 in 5%, 25/75 in 26%. It is a stock-heavy rule. And the bond column collapses: all bonds supported a 4% inflation-adjusted withdrawal for 30 years in only 19% of starts. Playing it safe in bonds at 4% was, on this record, the opposite.
In “Portfolio Success Rates: Where to Draw the Line”, Journal of Financial Planning, April 2011, the same authors used monthly data from January 1926 to December 2009 (55 overlapping 30-year periods). They report that a 75/25 portfolio supported a 30-year retirement at a 7% fixed withdrawal in 91% of periods, 69% at 8%; with inflation adjustment the 7% figure drops to 45%, and a 5% inflation-adjusted withdrawal succeeded in 82%. They suggested requiring at least 75% success, and for inflation-adjusted plans pointed to 4% to 5%.
What neither paper covered
- Taxes. Bengen assumed every asset sat in a tax-deferred account; the Trinity authors made no tax adjustment. In December 1997 Bengen estimated a taxable portfolio at a 35% tax rate had a safe rate roughly a fifth lower.
- Fees. Neither paper mentions fund expenses, advice fees or trading costs; the returns are index returns. A household paying 1% a year in expense ratios and advice is running a different experiment.
- Markets outside the United States. Every series is American: the S&P 500 or Ibbotson large-company stocks, US Treasuries or US corporate bonds, US CPI.
- Horizons beyond 30 years. Bengen's 4% conclusion is explicitly conditional on a 30-year minimum, and the Trinity tables stop at 30 years. An early retiree planning for 45 years is outside both.
- What “success” means. A portfolio ending the period with one dollar counts as a success. Neither study measures how close the calls were, or includes a cash buffer, annuity or Social Security.
The term does not appear in the 1994 paper. Bengen's December 1997 “Conserving Client Portfolios During Retirement, Part III” speaks of the maximum safe withdrawal rate and puts it at 4.08% for large-company stocks and intermediate Treasuries; adding small-company stocks lifted the peak to 4.3%, over a narrow range of allocations. His May 2001 “Part IV” calls it MAXSAFE and reports about 4.14% across 55 retirees starting 1926 to 1980. On his own Substack on May 26, 2025, he writes of a “4.7% Rule”, tied to the SAFEMAX of a retiree starting October 1, 1968, adding that it should be safe although it cannot be certain not to be broken.
What it means for your money
Whatever your starting balance, the rule fixes a dollar amount in year one and then indexes it to inflation. The balance is not consulted again.
Inputs: starting balance $1,000,000; first-year rate 4%; inflation 3.4% a year, the 12-month CPI change to August 2026 published by the Bureau of Labor Statistics on September 11, 2026, held constant for illustration. Three-year total: $124,126, against $120,000 had prices been flat.
The balance does not enter: if the portfolio fell to $800,000 by the end of year one, the year-two withdrawal is still $41,360, which is 41,360 ÷ 800,000 = 5.17% of what is left. That is the sequence-of-returns problem in one line.
| Year | Calculation | Withdrawal |
|---|---|---|
| 1 | 0.04 × 1,000,000 | $40,000 |
| 2 | 40,000 × 1.034 | $41,360 |
| 3 | 41,360 × 1.034 = 42,766.24 | $42,766 |
| Total | $124,126 |
Used honestly, the 4% figure answers a narrow historical question: at what fixed, inflation-indexed rate did a US stock-heavy portfolio, before taxes and costs, outlast every 30-year start from 1926 to the late 1960s? It does not say what returns come next, and neither paper claims it does.
Three ways to use it without over-reading it. Treat the 30-year, 50–75% stock, pre-tax, pre-fee conditions as part of the number. Subtract your costs and tax drag before comparing your plan with the tables. And test the plan against specific bad starts rather than the average: the freedom-number tool replays a withdrawal plan through historical sequences, and Market Memory (the Market Memory tool on the tools page) shows how the drawdowns that broke the worst cases compare with those in your own investing life.
Sources
- Determining Withdrawal Rates Using Historical Data (October 1994; 2004 anniversary reprint PDF)Journal of Financial Planning / Financial Planning Association · Sep 28, 2026 · Primary source
- Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable (February 1998)AAII Journal, American Association of Individual Investors · Sep 28, 2026 · Primary source
- Sustainable Withdrawal Rates From Your Retirement Portfolio, Financial Counseling and Planning vol. 10(1), 1999Association for Financial Counseling and Planning Education · Sep 28, 2026 · Primary source
- Portfolio Success Rates: Where to Draw the Line (April 2011)Journal of Financial Planning / Financial Planning Association · Sep 28, 2026 · Primary source
- Conserving Client Portfolios During Retirement, Part III (December 1997) PDFJournal of Financial Planning / Financial Planning Association · Sep 28, 2026 · Primary source
- Conserving Client Portfolios During Retirement, Part IV (May 2001) PDFJournal of Financial Planning / Financial Planning Association · Sep 28, 2026 · Primary source
- Potpourri on the 4.7% Rule (May 26, 2025)Bill Bengen, author's own Substack · Sep 28, 2026 · Primary source
- Consumer Price Index Summary — August 2026U.S. Bureau of Labor Statistics · Sep 11, 2026 · Primary source
- Schedule of Releases for the Consumer Price IndexBureau of Labor Statistics · Sep 28, 2026 · Primary source
Related analysis
What 3.4% inflation actually does to money sitting in a savings account
The arithmetic is simple, rarely done, and slightly uncomfortable. Here it is, with every assumption shown.
Should you pay down your mortgage or invest? Here is the arithmetic.
A 6.95% mortgage is the only guaranteed 6.95% return most households can buy. Every risk-free alternative on offer in September 2026 pays less than that. Four assumptions decide the rest.
How to check whether a financial number you just read is still true
Every economic figure has two dates, a publisher, a revision history and an error bar. Four of those five are usually missing from the article you read it in. Here is how to find them in about a minute.