The 4% Rule: What the Original Study Actually Said

What the original study actually did

The figure comes from William Bengen, a financial planner who published "Determining Withdrawal Rates Using Historical Data" in the Journal of Financial Planning in October 1994.

Bengen's question was narrow and well-posed. Given actual US market history, what is the highest initial withdrawal rate that would have survived every historical starting year, including the worst one? He tested every 30-year period in the data, using a portfolio of large-cap US stocks and intermediate-term US government bonds, with an allocation between 50% and 75% in stocks.

The method he tested is precise, and the precision is the part people lose:

• Withdraw a fixed percentage of the portfolio in year one only.
• In every subsequent year, withdraw the same dollar amount adjusted for inflation — not a percentage of the current balance.
• Rebalance to the target allocation annually.
• Continue for 30 years.

He found that just over 4% survived all historical periods, including the worst starting year in the dataset — a retiree beginning in 1966, who faced a brutal combination of poor real returns and high inflation through the 1970s. He called this figure SAFEMAX. It was explicitly a worst-case result, not an expected one. In the large majority of historical periods, a 4% withdrawal left the retiree with more money after thirty years than they started with.

The Trinity Study — Cooley, Hubbard and Walz, published in 1998 — extended this with corporate bonds and a range of horizons, and introduced the language of "success rates" that is now standard.

The assumptions that get dropped

Each of these was explicit in the research and is routinely omitted in the retelling.

A 30-year horizon. The number was calibrated to thirty years, roughly a retirement from 65 to 95. Someone retiring at 45 is planning for fifty years or more, and the safe rate falls as the horizon lengthens — though not proportionally, because a portfolio that survives the first fifteen years has usually grown enough to survive indefinitely.

Rigid inflation-adjusted spending. The model assumes you withdraw the same real amount every year regardless of what markets do. No real retiree behaves this way. The assumption makes the test conservative, because it removes the most powerful defense available — spending less in bad years.

A specific allocation. Fifty to seventy-five percent stocks. A portfolio held at 20% stocks does not support a 4% withdrawal over thirty years; too little growth. The rule is inseparable from the allocation that produced it.

US historical data only. This is the most serious limitation. The twentieth-century United States was among the most successful equity markets in world history. Studies applying the same method to other developed markets have generally produced lower safe withdrawal rates. The 4% figure may encode a survivorship bias at the level of an entire country.

Zero costs and zero taxes. This one is quantitatively devastating and almost never mentioned.

The fee problem

Bengen's simulations used raw index returns. No expense ratio, no advisory fee, no trading costs.

Costs come almost directly off the sustainable withdrawal rate. A portfolio paying a 1% advisory fee plus 0.5% in fund expenses is giving up 1.5% a year, and the safe rate drops toward roughly 2.5% to 3%. Put differently, that cost structure requires something like 40% more capital to support the same lifestyle.

There is a pointed irony here. The 4% rule is most frequently cited by the advisory industry, and the version of the portfolio it describes — index funds at a few basis points, no advisory layer — is not the portfolio most of those clients hold.

Taxes compound this. Withdrawals from a traditional IRA or 401(k) are ordinary income. A retiree needing $60,000 to spend may need to withdraw $70,000 or more, and the withdrawal rate that matters is the gross one.

Four percent was a worst-case result on raw index returns with no fees and no taxes. Every layer you add comes straight off the top.

Why sequence risk is the whole mechanism

Underneath the rule sits a single phenomenon: the order of returns matters enormously once you are withdrawing.

Two retirees can experience identical average annual returns over thirty years and end in completely different places, depending on when the bad years arrived. Poor returns in the first five to ten years force you to sell more shares at depressed prices to fund the same spending, permanently reducing the base that must recover. Good early returns build a cushion that absorbs later declines comfortably.

This is why the safe rate is set by the worst historical sequence rather than the average one. The 1966 retiree did not experience unusually bad average returns across thirty years — they experienced their bad returns first.

What practitioners actually do now

The research has moved well past a fixed percentage, in three directions.

Guardrails. Popularized by Guyton and Klinger, this approach sets an initial rate and then adjusts based on what the current withdrawal rate has become. If markets fall and your withdrawal rises above an upper guardrail, you cut spending by a set percentage. If markets rise and it drops below a lower guardrail, you give yourself a raise. Because it responds to conditions, a guardrails approach typically supports a higher starting rate — often in the 5% range — at the cost of accepting variability.

Flexible spending. Even modest flexibility transforms the outcome. Foregoing the inflation adjustment in years following a market decline, or trimming discretionary spending by 10%, dramatically improves survival rates. Most retirees have substantial discretionary spending — travel, gifts, dining — that can flex without touching anything essential.

Floor-and-upside. Cover non-negotiable expenses with guaranteed income — Social Security, a pension, or an annuity — and apply a withdrawal rate only to discretionary spending from the portfolio. Delaying Social Security to age 70 is the cheapest way most people can buy inflation-adjusted guaranteed income, and it directly reduces how much the portfolio must carry.

So is the rule still useful?

Yes, for one specific job: sizing the target. Inverted, 4% means you need roughly 25 times your annual portfolio-funded spending. Someone needing $50,000 a year from investments, after Social Security and any pension, is looking at about $1.25 million. That is a genuinely useful planning anchor, and it is far better than no anchor at all.

Where it fails is as a withdrawal policy. Nobody should retire, set a fixed inflation-adjusted withdrawal, and stop paying attention for thirty years. The number was the answer to an academic question about historical worst cases. It was never designed to be a spending plan, and Bengen himself has revised his estimates upward over the years as he incorporated additional asset classes.

Use it to know roughly how much you need. Use flexibility, guardrails and guaranteed income to decide what you actually spend.