Safe withdrawal rate: how good is the 4% rule?

Basics8 min read · updated 11 September 2026 · 3 sources

The safe withdrawal rate is the share of your starting wealth that you can withdraw each year, adjusted for inflation, without the money running out over a period such as 30 years. It is known as the 4% rule. It is a good starting point for planning, but not a law of nature: it depends on the data, the period, costs and how rigidly you withdraw.

Where the 4% comes from

Using US data since 1926, the financial planner Bengen (1994) tested which first-year withdrawal (increased by inflation every year thereafter) would have lasted at least 30 years in every historical period. The result was just over 4%, best with an equity share between 50 and 75%. Fewer equities did not make the portfolio safer, but made it run dry sooner, because the return was missing.

Later studies, such as the much-cited Trinity study of 1998, arrived at similar values for US portfolios using other data. For other countries, higher costs or longer periods, the rate turns out lower, because the US had one of the best stock market performances in the world in the 20th century.

Sequence risk: the order matters

Two investors with the same five annual returns, just in reverse order, each withdraw €5,000 at the start of each year from €100,000. The average of the returns is 2% in both cases.

Bad years at the beginning hit a portfolio that is being drawn down at the same time particularly hard: units are sold cheaply and are missing in the recovery. Without withdrawals, the final wealth would be the same in both cases. The order of returns only matters once money flows. That is why the maximum drawdown matters more in the withdrawal phase than in the accumulation phase. More detail in the article Sequence-of-returns risk.

How much wealth do I need?

The withdrawal rate can be turned around: required wealth = annual withdrawal / withdrawal rate. What matters is only the gap the portfolio has to fill, i.e. expenses minus pension, rental income and other secure income.

Taxes on withdrawals belong in the calculation: if you sell fund units, part of the proceeds is gain and is taxed after the partial exemption (Teilfreistellung) and the saver's allowance (Sparer-Pauschbetrag) (see flat-rate withholding tax (Abgeltungsteuer)). At the start the gain share is small; after decades of saving it can make up half of the proceeds or more.

Perpetual withdrawal rate

If your wealth is to be preserved in real terms (for heirs or a foundation, say), the withdrawal can in the long run only be as high as the real return after costs and taxes.

Depending on portfolio, costs and taxes, the perpetual withdrawal rate tends to land at 2 to 3.5% rather than 4%. If you are allowed to use up the capital, you can withdraw more; if you want to preserve it, you have to make do with less.

Flexible rules

• Guardrails: cut the withdrawal by 10% if it rises above a threshold relative to the current portfolio, and increase it if it falls below.
• Percentage of the current portfolio: the money never runs out, but the withdrawal fluctuates with the market.
• Skip the inflation adjustment: do not increase the withdrawal after a losing year. This alone raises the success rate noticeably.
• Buffer: one to two years of withdrawals in the money market, so you do not have to sell in a crisis.

For advanced readers: ruin, longevity and glide paths

The 4% rule looks at a fixed period. In reality, lifespan is unknown. Milevsky and Robinson (2000) calculate the probability of ruin for a fixed withdrawal from a portfolio with random, lognormal returns and combine it with mortality tables: the question is whether the money runs out before the end of life, not after exactly 30 years. The probability rises with the withdrawal, volatility and life expectancy, and falls with the expected return.

A second question is the equity allocation in retirement. Traditionally it falls with age. Pfau and Kitces (2013) argue for the opposite: a low equity allocation at the start of retirement that rises afterwards (a rising glide path) dampens precisely the sequence risk of the first years, when a crash does the most damage. More on this in the article Equity allocation by age.

Third, the safe rate depends on the valuation level at the start: historically, the worst starting years for retirees were those with high equity valuations followed by weak years. A Monte Carlo simulation with cautious return assumptions is therefore often more informative than the best or worst historical case (see Monte Carlo simulation).

Limitations of the 4% rule

• US data: the rule comes from a country with exceptionally good equity returns. With global data, it turns out lower.
• No costs or taxes: Bengen calculated with index returns. Fund costs, custody fees and taxes on gains lower the sustainable withdrawal.
• Rigid withdrawals: hardly anyone keeps withdrawing unchanged through a crash. Flexible rules change the result considerably.
• 30 years: if you stop working at 55, you are more likely to need 40 years and a lower rate.
• No guaranteed inheritance: in many paths a lot is left over, in some almost nothing. The rule only optimises against running out.

How Portfolio Lab calculates it

For every portfolio, the Backtest calculates the safe and the perpetual withdrawal rate over the selected period: the highest withdrawal as a percentage of the starting value per year, paid monthly at one twelfth each and adjusted for the inflation of the selected country, at which the wealth lasts until the end (safe) or still has its real starting value at the end (perpetual). The rate is found by bisection over the actual historical path. The result therefore applies to exactly this period. A different start can turn out quite differently. Monte Carlo shows the probability that a rate holds across many possible futures; Financial goals plans withdrawals together with pensions, taxes and goals. More in the backtest methodology.

This article explains a method; it is not investment or tax advice. Worked examples are simplified and rounded. All terms in the glossary.