Where the number comes from
The 4% rule has two fathers. In October 1994, the American financial planner William P. Bengen published the paper “Determining Withdrawal Rates Using Historical Data” in the Journal of Financial Planning. Just over three years later, three finance professors at Trinity University in San Antonio followed up: Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz. Their paper appeared in February 1998 in the AAII Journal and has been known ever since simply as the Trinity study.
The three did not calculate with an average return. They took the actual annual returns from 1926 to 1995, the S&P 500 for stocks and long-term high-grade corporate bonds for bonds. From these they formed every overlapping time window. For 30 years, that gives 41 windows, the first from 1926 to 1955, the last from 1966 to 1995. In each window they withdrew a fixed percentage of the starting assets in the first year and then raised this amount every year by the actual inflation. Then they counted in how many windows there was still money left at the end.
The result for 4% over 30 years, with inflation adjustment: a portfolio of 75% stocks and 25% bonds survived 98% of the windows. With 50% stocks and 50% bonds it was 95%, with stocks only also 95%. So only the 75 to 25 mix came out above the pure stock portfolio, while the 50 to 50 mix was level with it. At a 5% withdrawal, the same three figures fell to 83%, 76% and 85%. Between 4% and 5% there is no gentle step, but a cliff edge.
One line of the study is almost always left out when it is quoted. The authors state explicitly that they took neither tax nor transaction costs into account. So the 4% is a gross figure, not what you keep.
A concrete example: with a fixed withdrawal of $1,000 a month and no return, inflation, tax or costs, $300,000 lasts 25 years. In the article How long will $300,000 last in retirement? you can compare monthly amounts and calculate your own duration.
What 4% means for a real portfolio
I take $500,000. 4% of that is $20,000 in the first year, or $1,666.67 a month. This amount rises with inflation every year.
In the retirement withdrawal calculator I enter exactly that: $500,000 starting capital, $1,666.67 monthly withdrawal, a 5% return on the remaining capital, 2% inflation. The capital lasts 45.2 years. Over that time, $1,448,286.73 is paid out, and the last monthly withdrawal is $4,063.10.
45.2 years looks like plenty of room compared with the 30 years of the Trinity study. That room is a gross figure.
Taxes depend on where you live
The 4% rule is a pre-tax rule: the 4% is what you sell, not what you keep. How gains and withdrawals are taxed depends on the country you live in, so check the rules there or ask a tax advisor. If tax is due and you need a certain amount after taxes, you have to sell more, and your actual withdrawal rate rises above 4%. That step matters: in the Trinity table, the success rate of the 75/25 portfolio falls from 98% at 4% to 83% at 5%.
The 30-year horizon rarely fits
The Trinity study tested payout periods of 15, 20, 25 and 30 years, nothing more. In 1998, 30 years was a generous assumption for someone stopping work at 65.
Average life expectancy is only an average, and that is where the trap lies: it says nothing about whether you belong to the short-lived or the long-lived. If you stop at 60 and live to 92, you need 32 years. If you stop at 55, you plan for 35 years and more. For a couple, it is not the average that counts anyway, but the partner who lives longer. In the Die with Zero calculator you set your own planning age instead of a fixed 30 years.
The order of returns beats any average assumption
The retirement withdrawal calculator assumes the same return every year. That is exactly what the Trinity study deliberately did not do, and that is where its real value lies.
An example with the same monthly mechanics: $500,000, a withdrawal of $1,666.67, 2% inflation. The return is 7%, except for three years in which it is minus 15%. If these three years fall at the start, the capital lasts 21.5 years. If the same three years fall in years 21 to 23, it lasts 54.4 years. Same returns, different order, more than 30 years of difference.
What helps against this is not a percentage figure, but a buffer of two to three years of withdrawals that you draw on in weak years instead of selling units. I compared the withdrawal strategies available for this in Retirement withdrawal plan.
Then there is the development of prices. The American series from 1926 to 1995 contains ten years with falling prices, in which the withdrawal even fell in arithmetic terms. Two years of high inflation at the start, on the other hand, raise the withdrawal for the entire rest of the plan, and this higher base never goes away.
Three questions I receive most often
I receive the following three phrasings almost word for word, which is why I work through them here.
Is $500,000 enough for $2,000 a month? Not in the 4% logic. $2,000 a month is $24,000 a year. Under the rule of 25 you need $600,000 for that, before taxes. If tax is due on your withdrawals, you need more.
Do I still have to pay tax on the 4%? The rule does not answer that. The 4% is a gross withdrawal, and whether and how it is taxed depends on the country you live in. Check the rules where you live, or ask a tax advisor.
What happens if the market crashes right in the first year of retirement? In my example above, the duration shrinks from 54.4 to 21.5 years, just because the same three loss years come at the start instead of from year 21. The average return does not change. This exact pattern is behind the 2% to 5% failures in the Trinity table.
What remains of the rule
As mental arithmetic it is useful. Annual need times 25 gives you a ballpark figure for the target capital in ten seconds, and this ballpark figure is better than none. The financial independence calculator calculates your personal target from monthly expenses and additional income. The guide Living off your investments adds the cost blocks, later pensions, tax and insurance contributions that a single rule of thumb leaves out.
As a withdrawal plan it does not work. Three corrections turn the rule of thumb into a calculation that fits your portfolio. First: start from your net need and add the tax on top instead of deducting it at the end. Second: set the period by your own age, not by the 30 years from a 1998 study. Third: calculate at least three return assumptions and look at the worst one, not the middle one.
For my example this means: 4% of $500,000 lasts 45.2 years before taxes, at an even 5% return and 2% inflation. With three weak years at the start, a similar plan ends after 21.5 years. Your numbers look different because your taxes, your returns and your starting age are different. That is exactly what the calculator is for.