Is the 4% Rule Still Safe?

Where the 4% rule actually comes from

The 4% rule isn't a rule of thumb someone made up — it's the headline number from real research. Financial planner Bill Bengen published the original analysis in 1994, running historical U.S. market returns back to 1926 through every rolling 30-year retirement period he could construct, looking for the highest withdrawal rate, adjusted annually for inflation, that survived all of them. 4% (technically closer to 4.15% in his original work) was the answer for the worst-case historical starting point.

A few years later, three professors at Trinity University ran a related but distinct study — testing various withdrawal rates and stock/bond allocations against the same historical data and reporting the percentage of 30-year periods that survived at each rate. Both studies converged on roughly the same number, which is a large part of why "4%" became the number everyone quotes.

Both studies were also built around the same shape of retirement: a fixed 30-year horizon. That was a reasonable assumption for someone retiring in their mid-60s in the 1990s. It is not automatically a reasonable assumption for someone retiring in their 40s today.

Why a 40-50+ year horizon doesn't automatically get the same number

Retire at 45 with a target of living to 90 or 95, and you're planning for a 45-50 year retirement — 50-65% longer than the window the original research tested. That extra length matters because a longer horizon has to survive more of the bad sequences that show up somewhere in market history. A 30-year study never had to test what happens when a rough decade lands in year 35 of a retirement, because there was no year 35.

This doesn't mean 4% is "wrong." It means it's an answer to a different question than the one a FIRE-length retirement is asking. A withdrawal rate tuned to survive every historical 30-year stretch isn't guaranteed to survive every historical 45-year stretch, because the 45-year stretch has strictly more opportunities for a bad sequence to land somewhere in the exposure window.

A rate that held up for 30 years was never tested against 45 — the extra decades are exactly where the additional risk lives.

Sequence-of-returns risk: why the order matters more than the average

Here's the part a single "expected return" number hides: two retirees can experience the exact same average return over their retirement and end up in completely different financial positions, purely because of the order those returns arrived in. A steep downturn in the first five to ten years of retirement — while you're actively drawing the portfolio down and it has the least time to recover before you need the money — does dramatically more damage than the identical downturn arriving twenty years later, after growth has built a larger cushion and fewer years of withdrawals remain.

This is called sequence-of-returns risk, and it's the core reason a single "safe withdrawal rate" isn't really a fixed number at all. It's a function of several things at once: how long the money needs to last, how much flexibility you have to cut spending if the first decade goes badly, your actual asset allocation, and — critically — how the plan is tested. A rate that's safe against an average-return projection can be dangerously unsafe against a bad early sequence, which is exactly the scenario a single-number rule can't distinguish from a good one.

Why a simulation-based rate beats borrowing someone else's percentage

Once you accept that the safe rate depends on horizon, flexibility, and sequence — not just an average return — borrowing a fixed percentage from a study built for someone else's retirement stops making sense. The more useful question isn't "is 4% safe?" It's: run against your actual horizon and portfolio, using many randomized orderings of market returns rather than one average, what withdrawal rate holds up with real confidence?

That's what a Monte Carlo-style simulation does. Instead of assuming one tidy average-return path, it runs the plan through hundreds or thousands of different possible sequences of returns and reports how many of them the money survives. The output isn't a borrowed rule — it's a rate calculated specifically for your time horizon, which is the one variable the original 4% research held fixed at 30 years.

Want this as a standalone tool, or to embed it on your own site? Try the Safe Withdrawal Rate Calculator →

A worked example: 30 years vs. 45 years

The calculator above defines your retirement horizon directly from two numbers you enter: retirement age and life expectancy. Set retirement age to 65 and life expectancy to 95, and the horizon it simulates is 30 years — the same length the original Bengen and Trinity research tested. Set retirement age to 45 and leave life expectancy at 95, and the horizon becomes 50 years — a realistic FIRE-length window.

The mechanics don't change between the two runs — same simulation, same confidence target, same inflation assumption — only the number of years the portfolio has to survive. Directionally, this is consistent with the broader retirement-research literature on longer horizons (work following on from Bengen and Trinity, including later analysis by researchers like Michael Kitces and Wade Pfau on extended and FIRE-length retirements):

These are illustrative, not a promise about your specific plan — your real number depends on your actual portfolio size, allocation, and spending pattern. That's the entire point: run your own horizon through the calculator above rather than trusting a table built for someone else's numbers, including this one.

What this doesn't capture

Being straight about the limits matters as much as the number itself. This calculator is deliberately portfolio-only — it forces Social Security to zero internally so the comparison to the static 4% rule stays apples-to-apples, since the original Bengen and Trinity research was also portfolio-only. In practice, Social Security (or a pension) can meaningfully raise how much you can actually spend, which this specific tool won't show you. It also can't price in a tax law that doesn't exist yet, a health event with no historical precedent, or a market regime genuinely unlike anything in the historical record. Simulations model the range of what's historically plausible — they don't predict what will happen. Projection, not prediction.

How this is calculated

The rate above comes from a 1,000-scenario Monte Carlo simulation using randomized market returns based on historical data, targeting an 85% confidence level over your specific retirement horizon (retirement age to life expectancy), with a 3% inflation assumption applied throughout. It's compared directly against the static 4% rule's own framing — roughly 95% confidence over a fixed 30-year window — so you can see both numbers side by side rather than picking one on faith. It's the same engine behind the full app, free to run, no account required. The complete math, including the specific assumptions and data sources, is documented in the methodology.

Common questions