Turn savings into retirement income Lesson 3 of 9
Retirement · Turning savings into income
The 4% rule: what Bengen found, what it assumes and where it breaks
Bengen’s 1994 study and the Trinity study, the assumptions behind 4%, its critics, and what 4% did to a 60/40 SPY and AGG portfolio from 2007 and 2010.
On this page
The 4% rule says that a retiree can withdraw 4% of a portfolio in the first year of retirement, raise that dollar amount with inflation every year after, and expect the money to last 30 years. It is the most quoted number in retirement planning and one of the most misread. It is not a forecast, not a rate to take from each year's balance, and not a promise. It is the answer to a narrow historical question: what is the highest inflation-adjusted withdrawal that survived the worst 30-year stretch in the US record?
The lesson on Safe withdrawal rates: how much can a portfolio pay out? explains the general idea and tests five rates on a portfolio that retired at the start of 2000. This lesson stays with the rule itself: where it came from, exactly what it assumes, what its critics have found, and what it actually did for two retirements that began in January 2007 and January 2010 on a portfolio of the SPY and AGG funds. The second half is about the critique that gets least attention: in most periods the rule did not run out of money, it left a great deal of it unspent.
Where the rule came from
In October 1994 William Bengen, a financial planner, published "Determining Withdrawal Rates Using Historical Data" in the Journal of Financial Planning. His question was practical: clients wanted a steady, inflation-adjusted income, and it was tempting to assume they could spend something close to the portfolio's long-run average return. Bengen ran every retirement start year the US data allowed through the actual sequence of stock returns, bond returns and inflation that followed. He found that rates near the long-run average return failed badly for people who retired before bad decades, and that a first-year rate of about 4%, adjusted for inflation, was the highest that lasted at least 30 years for every start year he tested. The worst starting points were the late 1960s, when years of high inflation followed a strong market.
Four years later three professors at Trinity University, Philip Cooley, Carl Hubbard and Daniel Walz, published what became known as the Trinity study in the AAII Journal. Instead of a single worst case they reported the share of historical periods in which each withdrawal rate and mix of stocks and bonds lasted 15, 20, 25 or 30 years. The framing stuck: "a 4% withdrawal from a stock-heavy portfolio succeeded in nearly every 30-year period" is a Trinity-style sentence. Both studies measured the same thing: survival of the money, not how much was left, and not how happy the retiree was with the spending.
What the rule actually assumes
The rule is a bundle of assumptions. Change any of them and the number moves.
- A fixed real withdrawal. Year one is 4% of the starting balance. Every later year is the previous year's dollar amount raised by inflation. The portfolio's value after retirement plays no part. At $1,000,000 the first withdrawal is $40,000; if prices rise 3%, the second is $41,200 whether the portfolio is worth $1.3 million or $700,000.
- A 30-year horizon. The money has to last 30 years and nothing beyond that is counted. A retiree at 55 or a healthy couple at 62 may need longer; see the survival odds in Annuities explained: SPIAs, deferred income, QLACs and their costs.
- A balanced, rebalanced portfolio. Bengen favoured a stock share between 50% and 75%, using large US stocks and intermediate-term Treasury bonds. Too few stocks and the portfolio could not outrun inflation; too many and early crashes did too much harm.
- No fees and no taxes. Returns are index returns. A withdrawal from a traditional IRA or 401(k) is taxable income, so $40,000 withdrawn is less than $40,000 to spend; the taxes section covers how much less.
- US history repeats in kind. The future must contain no 30-year stretch worse than the worst in the sample.
- Spending never adjusts. The retiree takes the same real amount through a 50% crash and through a doubling, and nothing is ever cut or added.
A worked example: 2007 and 2010 on SPY and AGG
To see the rule in motion, Finabase ran it on $1,000,000 invested in 60% SPY (the S&P 500 fund) and 40% AGG (the US investment-grade bond fund), with total returns including dividends, rebalanced every month. Each month starts with a withdrawal of 1/12 of $40,000, raised by the change in the consumer price index (CPI-U) since retirement; the rest earns that month's return. There are no fees and no taxes. CPI for October 2025, which BLS did not publish, is the average of the months either side. The data end in September 2026, so the two retirements have run 19.75 and 16.75 years: short of 30, which is a limit of this example.
| Retired | Years so far | Real return a year | Lowest real balance | Real balance, Sept 2026 | Ending ÷ start | Withdrawal now |
|---|---|---|---|---|---|---|
| January 2007 | 19.75 | 5.17% | $635,972 (Feb 2009) | $1,177,202 | 1.18 | $65,806 a year |
| January 2010 | 16.75 | 6.70% | $964,966 (June 2010) | $1,794,632 | 1.79 | $61,493 a year |
Real balances are in dollars of the month before retirement. In current dollars the 2007 retiree held $1,949,241 in September 2026 and had withdrawn $986,532 in total; the 2010 retiree held $2,776,810 after withdrawing $804,822.
The 2007 retiree met the worst of the global financial crisis in year two. By February 2009 the real balance had fallen 36%, and the withdrawal had become 5.44% of the balance at the end of 2008 and was still 4.79% at the end of 2012. Under the rule nothing changes in response: the retiree keeps taking the same real amount and waits. The real balance first got back to $1,000,000 in October 2017, and 2022, when stocks and bonds fell together while inflation ran high, knocked the balance back to $932,586 before it rose again. The 2010 retiree, starting after the crash, never fell more than 4% below the starting balance in real terms.
The same two dates with other rates show how much room there was, in hindsight:
| Retired | 3% | 4% | 5% | 6% |
|---|---|---|---|---|
| January 2007, real balance Sept 2026 | $1,559,280 | $1,177,202 | $795,125 | $413,047 |
| January 2010, real balance Sept 2026 | $2,087,278 | $1,794,632 | $1,501,986 | $1,209,340 |
None of these ran out. The fixed real rate that would have spent the 2007 portfolio exactly to zero by September 2026 was 7.08%; for 2010 it was 10.13%. Both are rates for periods of under 20 years, so they say nothing about a full 30-year retirement, but they show the scale of the cushion in these particular decades.
What the critics say
Fees. Bengen's returns were index returns. A retiree paying 1% a year to an adviser or in fund costs loses that 1% from every year's return. In the 2007 example a 1% fee lowers the September 2026 real balance from $1,177,202 to $862,121, and the low point from $635,972 to $621,473. Over a full 30 years the effect keeps compounding, and a rule calibrated on index returns says nothing about a portfolio that earns one point less every year.
Longer retirements. The 4% figure is for 30 years. Early retirees need money for 40 or 50 years, and a rate that survived 30 years in the worst case did not always survive 40.
Starting valuations and yields. The worst historical starts came when stocks were expensive or bond yields were about to be overtaken by inflation. Researchers have argued that a retiree facing high stock valuations or low real bond yields should expect a lower safe rate than the historical worst case implies. The Treasury page shows the real yield on TIPS, the market's own price for a safe inflation-protected return.
The US was lucky. Wade Pfau's 2010 study applied Bengen's method to the stock and bond histories of other developed countries and found that 4% would have failed in many of them. The US had one of the best equity markets of the twentieth century, and a rule calibrated to its worst case is calibrated to a good worst case.
Spending is not rigid. Real retirees cut spending when markets fall and spend more on travel early and on health care late. A rule that never adjusts describes nobody, and rules that do adjust allow a higher starting rate: see Guardrails withdrawal strategy: Guyton-Klinger rules tested from 2007.
It usually leaves too much behind. This is the critique the 2007 and 2010 runs illustrate. Because the rule is fixed to the worst case, the typical retiree who follows it ends with more than they started with, and may have lived on less than they could have afforded.
The surplus problem
To see how general that is, Finabase ran the rule from every January from 2004 to 2016. Every one of the thirteen retirements ended September 2026 with more real money than it started with.
The lowest ratio, 1.18, belongs to the retirements of 2005 and 2007, both of which lived through 2008 early on; the highest, 2.07, to January 2009, which began at the bottom. This is a short and favourable stretch of market history, and the periods are not independent: they share most of the same years. It is not evidence that 4% is too low in general. It is evidence of what the rule is built to do. Protecting against the worst sequence means over-saving for the ordinary one, and the money left over goes to heirs or to a late-life cushion rather than to spending.
That trade-off is a choice, not a mistake. Someone who values certainty and a bequest may be content with it. Someone who would rather spend more while healthy can look at three alternatives: adjusting spending with markets (Guardrails withdrawal strategy: Guyton-Klinger rules tested from 2007), buying guaranteed income for part of the need (Annuities explained: SPIAs, deferred income, QLACs and their costs), or raising guaranteed income by delaying Social Security (When to claim Social Security: 62, 67 or 70, with break-even ages).
Pitfalls in using the rule
- Taking 4% of each year's balance. That is a different rule, with income that rises and falls with markets and never runs out. The 4% rule fixes the dollar amount.
- Applying it to the whole household budget. The rate applies to what the portfolio must supply after Social Security, pensions and annuities.
- Ignoring the current rate. By the end of 2008 the 2007 retiree was withdrawing 5.44% of the balance. The current rate, this year's withdrawal divided by today's balance, says more about the plan's health than the original 4%.
- Treating success as binary. A plan that "succeeds" with $5,000 left and one that succeeds with $2 million are counted the same in a success rate.
- Forgetting taxes. Withdrawals from tax-deferred accounts are taxable; the order in which accounts are drawn changes the bill.
The 4% rule beside other spending rules
| Rule | How this year's spending is set | Spending stability | Can run out? | Typical unspent money |
|---|---|---|---|---|
| 4% rule (fixed real) | 4% of the start, then raised with inflation | Fixed in real terms | Yes, in bad sequences | Often large |
| Percent of portfolio | A set share of each year's balance | Moves with markets | No, but income can fall far | Moderate |
| Guardrails | Fixed real, cut or raised when the current rate drifts too far | Mostly stable, with steps | Rarely | Smaller |
| RMD-style | Balance ÷ remaining life expectancy | Moves with markets, rises with age | No | Small |
| Life annuity | Insurer pays for life | Fixed (nominal unless indexed) | No | None |
How Finabase does it
Finabase is a research app: it does not manage a retirement account, run a withdrawal plan or know a reader's taxes. The test above uses the same monthly price and dividend history that the symbol pages show. The SPY comparison page puts SPY and AGG side by side over any range, the Dividends tab of each fund lists what it paid, and the Treasury page shows today's Treasury and TIPS yields. On the Portfolio page, the current withdrawal rate is a sum a reader can do: planned yearly spending from the portfolio divided by its total value.
Frequently asked questions
What is the 4% rule in simple terms?
Withdraw 4% of savings in the first year of retirement and raise that dollar amount with inflation every year. In US history back to 1926, as tested by Bengen in 1994, a portfolio of roughly half to three-quarters stocks lasted at least 30 years on that rule in every start year.
Who came up with the 4% rule?
William Bengen, a financial planner, in a 1994 Journal of Financial Planning article. The Trinity study by Cooley, Hubbard and Walz (1998) popularised it by reporting historical success rates for different withdrawal rates and portfolios.
Does the 4% rule include taxes and fees?
No. Both studies used index returns before fees, and the withdrawal is before income tax. In this lesson's 2007 example a 1% yearly fee cut the ending real balance by about 27%.
Is the 4% rule still valid?
It remains a reasonable first estimate for a 30-year US retirement and nothing more. It has failed in other countries' histories, may be too high for long retirements or expensive markets, and in most US periods it left a large surplus. Whether it fits depends on the reader's horizon, costs and willingness to adjust spending.
What happens if I retire just before a crash?
The rule keeps withdrawals fixed, so the current withdrawal rate rises. The January 2007 retiree in this lesson's example saw the real balance fall 36% by February 2009 and did not regain the starting real balance until October 2017, but did not run out.
Summary
The 4% rule is Bengen's 1994 worst-case answer for a 30-year US retirement: 4% of the starting balance, raised with inflation, from a portfolio about half to three-quarters in stocks, with no fees, no taxes and no changes to spending. The Trinity study turned it into success rates. Its critics point to fees, longer retirements, high starting valuations, weaker markets abroad and rigid spending, and the data here add the opposite worry: from every January start between 2004 and 2016, a 60% SPY and 40% AGG portfolio ended September 2026 with more real money than it started with, from 1.18 times for 2005 and 2007 to 2.07 times for 2009. The rule protects against the worst case by underspending in the ordinary one.
Check your understanding
4 questions on this lesson. Each answer comes with the reason, and nothing is scored.
A retiree follows the 4% rule on $1,000,000. Prices rise 3% in the first year and the portfolio falls to $800,000. What is the second year’s withdrawal?
- $32,000, 4% of the new balance
- $41,200, the first year’s $40,000 raised by inflation
- $40,000, unchanged
Show the answer
B. $41,200, the first year’s $40,000 raised by inflation The rule fixes the dollar amount in year one and raises it with inflation. The balance after retirement plays no part, which is what makes it simple and rigid.
What question did Bengen’s 1994 study answer?
- The average return a balanced portfolio will earn in retirement
- The highest inflation-adjusted first-year withdrawal that lasted 30 years in every historical US start year he tested
- The withdrawal rate that maximises what is left to heirs
Show the answer
B. The highest inflation-adjusted first-year withdrawal that lasted 30 years in every historical US start year he tested The 4% figure is the worst-case survivor across historical start years. It describes the past and the worst sequence in it, not an expected outcome.
In this lesson’s test, a 2007 retiree with a 60/40 SPY and AGG portfolio ended September 2026 with $1,177,202 in 2006 dollars. What does a 1% yearly fee do to that figure?
- Almost nothing, because fees are small
- Lowers it to $862,121, about 27% less
- Empties the portfolio
Show the answer
B. Lowers it to $862,121, about 27% less A fee comes out of every year’s return, so it compounds. The rule’s 4% was derived from index returns with no costs.
Every January start from 2004 to 2016 ended September 2026 with more real money than it began with at 4%. Why does that not prove the rule is too cautious?
- Because the periods are short, overlap heavily and fall in a favourable stretch of US history
- Because SPY and AGG are not real investments
- Because inflation was zero in those years
Show the answer
A. Because the periods are short, overlap heavily and fall in a favourable stretch of US history Thirteen overlapping periods of 11 to 23 years from one market say little about a future 30-year retirement. What they show is the rule’s design: built for the worst case, it leaves a surplus in ordinary ones.
Exercise
One task to do in Finabase. Nothing is graded or saved.
Open the comparison of SPY with AGG and set the range to show 2007 to 2009. Read SPY’s change from early 2007 to its March 2009 low, and AGG’s change over the same stretch. Weight them 60% and 40% to estimate the lesson’s portfolio. Then take $1,000,000 started in January 2007 with $40,000 a year withdrawn: about what would the withdrawal rate be on the reduced balance at the low?
Compare SPY with AGGWhat you should find
SPY fell roughly half while AGG rose a little, so a 60/40 mix fell about a quarter; after two withdrawals the balance is near $700,000, and $40,000-plus is close to 6% of it. That is the sequence risk the 4% rule was built to survive, and in the lesson the 2007 start still ended September 2026 at 1.18 times its real starting value.
Sources and further reading
- William P. Bengen, “Determining Withdrawal Rates Using Historical Data,” Journal of Financial Planning, vol. 7, no. 4 (October 1994), pp. 171–180
- Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz, “Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable,” AAII Journal, vol. 20, no. 2 (February 1998) — the Trinity study
- Wade D. Pfau, “An International Perspective on Safe Withdrawal Rates: The Demise of the 4 Percent Rule?”, Journal of Financial Planning, vol. 23, no. 12 (2010)
- U.S. Bureau of Labor Statistics, Consumer Price Index for All Urban Consumers (CPIAUCSL), via FRED; no index was published for October 2025
- Finabase chart data — monthly prices and dividends of SPY and AGG, January 2004 to September 2026; simulations computed by Finabase
