Safe Withdrawal Rate: What 150 Years of Data Actually Says

In 1994, financial planner William Bengen tested fixed inflation-adjusted withdrawals against every US retirement start year since 1926. He found that at a 4% initial withdrawal rate, a 50-75% stock portfolio survived every historical 30-year retirement — even for people who retired right before the 1929 crash or into the brutal stagflation of the mid-1960s. The 1998 Trinity study reframed the same idea as a table of success rates, and "4%" became shorthand for retirement safety.

Both studies embed three assumptions worth making explicit: a 30-year retirement, US market returns, and a spending pattern that never responds to markets. Change any of the three and the safe rate changes with it.

The 4% rule was calibrated for a traditional retirement at 65. Someone retiring at 40 or 50 needs the portfolio to survive 40-60 years, not 30. Longer horizons expose the portfolio to more bad sequences, and the sustainable rate drifts down — though not linearly, because portfolios that survive their first 15 years usually grow away from danger. In long-horizon simulations the historically safe rate tends to land closer to 3-3.5% than 4%.

This is why serious planning tools let you set the horizon explicitly rather than assuming 30 years. A 4% rate that is comfortable over 30 years can carry materially more failure risk over 50.

The United States had one of the best-performing stock markets of the last 150 years — and it never lost a war on home soil, never saw its market closed for years, and never hyperinflated. Studies that apply the same fixed-withdrawal test to other developed countries' local market histories find 4% failing far more often. Our simulator can pool return histories from 16 developed countries (the Jordà-Schularick-Taylor macrohistory dataset, 1870 onward) precisely to avoid treating the US experience as the only possible future. At the same withdrawal rate and horizon, success rates under pooled international data typically come out 10-20 percentage points below US-only data.

Fees compound the problem. Historical return series contain no fund costs, but real portfolios do. An 0.5% annual expense drag — cheap by mutual-fund standards, expensive by index-ETF standards — lowers the sustainable withdrawal rate by roughly 0.2-0.4 percentage points over long horizons. Any simulation that ignores costs overstates what you can spend.

The classic SWR test assumes you mechanically withdraw the same inflation-adjusted amount even while your portfolio collapses around you. Real retirees cut discretionary spending in bad markets. That rigidity is what makes the fixed-withdrawal framework a useful worst-case benchmark — and also what makes it pessimistic as a plan.

Dynamic strategies such as guardrails let you start at a higher withdrawal rate in exchange for accepting modest spending cuts when markets turn against you. In our simulations, a guardrail strategy targeting the same failure risk typically supports a meaningfully higher starting withdrawal than a fixed rule, because flexibility absorbs the bad sequences that break fixed spending.

There is no universal number — there is a trade-off curve between withdrawal rate and failure risk, and the honest approach is to look at the whole curve for your horizon, allocation, and data assumptions. A 4% rate over 30 years with US data sits in very different risk territory than 4% over 50 years with globally pooled data.

Our withdrawal-rate analysis tool sweeps rates from conservative to aggressive and shows the success rate at each point, so you can pick the rate whose risk level you can genuinely live with — and see exactly how much safety each additional 0.25% of spending costs.