The 4% Rule: Where It Came From and How to Use It Carefully

In 1994, a financial planner named William Bengen published a paper in the Journal of Financial Planning asking a specific question: if a retiree withdrew a fixed percentage of their portfolio in year one, then adjusted that dollar amount for inflation every year after, what withdrawal rate would have survived every rolling 30-year period in U.S. market history back to 1926? His answer, describing what's often called his SAFEMAX research, was close to 4%. Four years later, three professors at Trinity University (Philip Cooley, Carl Hubbard, and Daniel Walz) published a related, broader analysis of withdrawal rates across different stock/bond mixes and time horizons, in what's now known simply as the Trinity study. Between the two, "the 4% rule" became the most quoted number in retirement planning.

Here's the important part: neither paper said 4% always works. Both described how a specific, mechanical withdrawal plan performed against specific historical data, with specific limitations the authors themselves flagged.

What the rule actually says

The mechanics are simple. In year one of retirement, withdraw 4% of your starting portfolio. In every year after, take that same dollar amount, adjusted upward for inflation: not 4% of the portfolio's new value, which would rise and fall with the market. A $1,000,000 portfolio produces a $40,000 first-year withdrawal; if inflation runs 3% that year, year two's withdrawal is $41,200, regardless of what the portfolio did in between.

That inflation-adjustment step is what makes the rule a spending plan rather than just a percentage. It's also what exposes a retiree to sequence-of-returns risk: because the dollar withdrawal doesn't shrink in a down market, a bad stretch of returns early in retirement pulls a larger percentage out of a shrinking portfolio than the same bad stretch would late in retirement. That risk is the single biggest reason a rule built on historical averages can still fail for someone whose real returns show up in the wrong order, the subject of sequence-of-returns risk in more depth.

What the original research actually covered (and didn't)

Both studies share the same structural limits, and both sets of authors were explicit about at least some of them:

• U.S. market data only. Bengen's data and the Trinity study both draw on historical U.S. stock and bond returns. Neither claims the same withdrawal rate would have held up using, say, historical Japanese or European market data over the same periods, and international markets have in fact shown worse outcomes for some historical retirement start dates.
• A roughly 30-year horizon. The "success" in these studies means the portfolio survived 30 years without hitting zero. Someone retiring at 45 with a 45- or 50-year horizon is extrapolating past what the original data actually tested.
• No fees, no taxes, no advisor costs modeled cleanly. The original analyses generally used gross returns. A 1% annual fee is not a small rounding error against a 4% withdrawal: it meaningfully raises the odds of running out of money, and taxes on withdrawals (especially from traditional 401(k)s and IRAs) aren't part of a clean 4%-of-portfolio calculation either. See taxes in early retirement for how account type changes the real number you need to withdraw.
• A specific, fixed rebalancing and allocation assumption. Bengen's original modeling assumed intermediate-term government bonds and a stock/bond split in a fairly narrow range; the Trinity study tested several allocations from 100% stocks to 100% bonds and found the "safe" rate varied by mix, generally landing near 4% for the more balanced-to-stock-heavy portfolios.
• Past performance, not a guarantee. This is the caveat every source repeats and the one worth taking most seriously: a rule that worked across all rolling 30-year periods in one country's market history is a strong data point, not a promise about the next 30 years.

Why two separate studies converged on a similar number

It's worth noting that Bengen's paper and the Trinity study weren't the same research repeated twice: they were different methodologies that happened to land in a similar place, which is part of why the 4% figure carries as much weight as it does. Bengen modeled a single, fairly specific stock/bond allocation using intermediate-term government bonds and tested it against every rolling 30-year start date in his historical data. Cooley, Hubbard, and Walz took a broader approach, testing multiple stock/bond mixes (from 100% stocks down to 100% bonds) against multiple withdrawal rates and multiple time horizons, then reporting the percentage of historical periods each combination survived. Bengen's single scenario and the Trinity study's more balanced-portfolio results both pointed to a starting rate near 4% holding up across nearly all 30-year historical windows, which is a meaningfully different kind of evidence than either paper alone, but still evidence built entirely on one country's historical returns.

That convergence is also why later researchers, including Bengen himself in more recent work, have revisited the number rather than treating it as settled. Forbes' retrospective on the rule, written roughly 25 years after the original paper, notes that market conditions, bond yields, and portfolio construction in any given decade can push the "right" number for a specific retiree meaningfully away from 4% in either direction. The rule aged into a benchmark for comparison, not a number that later research left untouched.

Decision framework: using 4% without misusing it

1. Use it to size a starting point, not a promise. Multiply your annual spending by 25 (the inverse of 4%) to get a first-pass FI number. See how to calculate your FI number for that calculation in full.
2. Adjust the rate for your actual time horizon. A 30-year retirement fits the original research reasonably well. A 40- to 50-year retirement (common in early-retirement planning) argues for something closer to 3% to 3.5%, which is a meaningfully bigger portfolio for the same spending.
3. Net out fees and taxes before you compare your plan to the historical success rates. A 4% withdrawal with a 1% advisory fee behaves more like a 5% withdrawal in terms of what the portfolio actually has to produce.
4. Build in flexibility rather than treating the inflation-adjustment as mandatory in a downturn. Plans that allow spending to flex (skipping an inflation raise after a bad market year, for instance) have historically supported meaningfully higher starting withdrawal rates than the rigid version of the rule. That's the subject of safe withdrawal rates beyond 4%.
5. Revisit the plan periodically against your actual portfolio, not just the plan you built on day one. A withdrawal rate is a starting assumption, not a policy you set once at retirement and never look at again.

Worked example

Take a $1,200,000 portfolio and a retiree spending $48,000 in year one, exactly a 4% initial withdrawal.

• Year 1: withdraw $48,000 (4.0% of $1,200,000).
• Year 2: inflation runs 3.5%; withdraw $48,000 × 1.035 = $49,680, regardless of what the portfolio is worth that year.
• If markets fell 15% in year one (a plausible single bad year), the portfolio might sit near $978,000 net of the withdrawal and market loss, and that same $49,680 withdrawal now represents about 5.1% of the portfolio — a materially higher rate than the plan started with, purely because of when the loss occurred.

This is the entire sequence-risk problem in miniature: the rule doesn't know or care what order the returns arrive in, and the order matters as much as the long-run average.

Limits and exceptions

The 4% framework fits a retiree with a roughly 30-year horizon, a diversified U.S.-heavy stock/bond portfolio, and spending that can theoretically stay flat in real terms. It fits less well for very early retirees with 40+ year horizons, for portfolios concentrated outside U.S. large caps, for anyone who can't tolerate any downside flexibility in a bad market year, and for anyone whose real spending need includes lumpy, non-inflation-linked costs like long-term care. Read the 4% rule as the historically grounded starting assumption it was designed to be, then adjust for your own horizon, costs, and tolerance for a spending cut in a bad decade.

A rule of thumb, not a retirement plan

It's easy to see why 4% became shorthand for "how much can I safely spend": it converts a complicated question into one multiplication problem. But the number is doing a lot of quiet work that a single figure can't show: it's an average outcome across many historical starting points, not a guarantee for the one starting point that happens to be yours. Some historical retirees following the rule finished 30 years with several times their starting balance still intact, because the sequence of returns happened to break in their favor. A smaller number finished right at zero, or close to it, because it didn't. The 4% figure describes the rate that survived the worst historical sequences in the data set — which is exactly why it's a reasonable starting point, and exactly why "reasonable starting point" and "guarantee" aren't the same thing.