Retirement withdrawal rates in United States: what 125 years of market history say
Every number on this page is a property of United States's own market history from 1900 to 2025 — real (inflation-adjusted) stock and bond returns in local currency, its worst 30-year stretch, its inflation tail — plus one Monte Carlo result: the withdrawal rate that would have succeeded 90% of the time when those returns are resampled. Nothing here depends on your plan; it exists so you can see what United States's past implies before you run your own numbers.
Every withdrawal-rate figure on this page is a Monte Carlo result under one fixed scenario, so the 16 countries stay comparable: $1,000,000 portfolio · 30% United States equity / 30% global equity / 40% United States bonds · 0.05% fees on every asset · fixed real (inflation-adjusted) withdrawals · block bootstrap of United States's 1900–2025 returns, 5,000 paths × 5 seeds. Success means every planned withdrawal was paid in full. 30/30/40 is this page's reference blend and is not the simulator's default allocation, which is 40/40/20.
Key numbers
- United States, same scenario: 4.25%
- Median of the 16 countries: 3.03%
- United States ranks #2 of 16 countries by 30-year withdrawal rate at 90% success.
- United States, same scenario: 93.2%
- Median of the 16 countries: 78.4%
- United States, same scenario: 6.81%
- United States, same scenario: 1.41%
- United States, same scenario: 2.88%
Withdrawal rate at 90% success, by retirement length
| Retirement length | Rate at 90% success | Range across seeds |
|---|---|---|
| 30 years | 4.25% | 4.23%–4.26% |
| 40 years | 3.65% | 3.63%–3.67% |
| 50 years | 3.33% | 3.32%–3.35% |
The range is the spread across 5 random seeds — sampling noise of the bootstrap, not a confidence interval for United States's true safe rate.
How often each initial withdrawal rate succeeded over 30 years
| Initial withdrawal rate | Success rate |
|---|---|
| 3.0% | 99.2% |
| 3.5% | 97.5% |
| 4.0% | 93.2% |
| 4.5% | 86.1% |
| 5.0% | 76.8% |
Success means every planned real withdrawal was paid in full for all 30 years; a final year that could only be partly funded counts as a failure. Rates are read off the same simulation that produced the 90%-success figure above.
The worst 30-year window
A portfolio of 60% United States equities and 40% United States bonds had its worst 30-year stretch starting in 1955: 2.82% per year in real terms. With the reference blend — 30% United States equity, 30% global equity, 40% United States bonds — the worst window, starting 1902, was 2.76% per year.
The gap between those two numbers is the value of global diversification for a United States-based retiree: most of what makes a national catastrophe catastrophic is that it is concentrated at home. United States equities alone had a worst 30-year window of 3.44% per year (from 1903), and the worst 10-year inflation stretch, starting 1973, multiplied the price level by 2.20×.
What happened: 24 events, 1900–2025
The crises, wars, bubbles and policy shifts marked on United States's charts in the simulator. Global events are included because they hit United States too.
- 1907Panic of 1907Crisis
- 1914–1918World War IWar
- 1929–1932Great DepressionCrisis
- 1939–1945World War IIWar
- 1950–1953Korean WarWar
- 1962Cuban Missile CrisisCrisis
- 1971Bretton Woods CollapsePolicy
- 1973–1974Oil CrisisCrisis
- 1979Second Oil CrisisCrisis
- 1979–1982Volcker Shock / High InterestPolicy
- 1985Plaza AccordPolicy
- 1987Black MondayCrisis
- 1994–1995Mexican Peso Crisis (Tequila Crisis)Crisis
- 1998LTCM / Russian CrisisCrisis
- 2000–2002Dot-com Bubble BurstBubble
- 20019/11 AttacksCrisis
- 2008–2009Global Financial CrisisCrisis
- 2013Fed Taper TantrumCrisis
- 2015–2016Chinese Stock Market TurbulenceCrisis
- 2018–2019US-China Trade WarPolicy
- 2020COVID-19 PandemicCrisis
- 2022–2023Global Inflation / Rate HikesCrisis
- 2022Russia-Ukraine WarWar
- 2023Regional Bank Crisis (SVB)Crisis
Data notes for United States
Figures come from the Jordà-Schularick-Taylor Macrohistory Database (JST) for 1900–2020, in local currency and adjusted for United States's own inflation, plus an unofficial 2021–2025 extension built from IMF, OECD and market data. Real returns are geometric averages over the full window. Before 1950 the global-equity leg uses a purchasing-power-parity fallback in the years where administered wartime exchange rates broke the currency conversion.
19 value(s) in United States's 1900–2025 series are modelled rather than observed — imputed bond returns, market closures or wartime exchange-rate substitutions. 19 are declared by the upstream pipeline and 0 are inferred from the published data, which does not preserve the observed/imputed distinction. The years concerned are 1917, 1918, 1919, 1920, 1922, 1923, 1937, 1939, 1940, 1941, 1942, 1943, 1944, 1945, 1946, 1947, 1948, 1949, 1950; compare them with the worst-window start years above to judge how much weight those figures can carry. They are disclosed here rather than averaged in silently.
Frequently asked questions
Did the 4% rule work in United States?
Under the reference scenario, a 4% initial withdrawal over 30 years succeeded 93.2% of the time when United States's 1900–2025 returns are resampled; reaching 90% success required an initial rate of 4.25%. The 4% rule was derived from United States data, where the same scenario gives 4.25% — United States's figure is close to that.
Why does United States have a different safe withdrawal rate from the United States?
Because a withdrawal rate is set by the bad tail, not by the average. United States's real equity return averaged 6.81% per year against 6.81% for the United States, but the number that matters is the worst 30-year window: 2.82% per year for a home 60/40 portfolio in United States versus 2.82% for the United States. One destroyed decade early in retirement is what ends a plan.
Should I plan with United States's history or with the 16-country pool?
Plan with the market you will actually spend in. If you retire in United States with a United States-heavy portfolio, United States's own history is the relevant stress test; the 16-country pool is a broader prior that includes catastrophes no single country has had yet. Run both in the simulator and plan for the more demanding one.
Simulate a United States-based retirement
Opens the simulator with United States selected as the market history. The simulator starts from its own defaults — a 40/40/20 allocation and a 55-year horizon — not from this page's reference scenario; set your own portfolio, spending and allocation and see how United States's history treats them.
Open the simulator with United States selected →