What a safe withdrawal rate actually is
A safe withdrawal rate is a spending rule, not a yield. You set the first year's withdrawal as a percentage of the portfolio, and from then on you raise that dollar amount by inflation and ignore the balance entirely. Take 4% of a million dollars and you withdraw $40,000 in year one; at 3% inflation you withdraw $41,200 in year two whether the portfolio rose to $1.1 million or fell to $800,000.
That rigidity is the point. William Bengen's 1994 study in the Journal of Financial Planning asked what constant, inflation-adjusted withdrawal a retiree could have taken from a stock-and-bond portfolio through every historical 30-year window in the United States without running out, and found the answer was slightly above 4%. The Trinity study a few years later reframed the same question in terms of success probabilities across portfolio mixes. Neither claims 4% is optimal or guaranteed; both establish that a fixed real income near that level survived the worst sequences their data contained.
What makes the question hard is not the arithmetic. It is that returns arrive in an order. A portfolio in drawdown that loses 30% early has to sell far more shares to fund the same real spending, and it may never recover even if the average return over the full period is fine. That asymmetry between accumulation and decumulation is why a sustainable rate is meaningfully lower than the portfolio's expected real return.
The two calculations on this page
The first is trivial and the second is the one that teaches. Your first-year withdrawal is the rate times the portfolio, and every later withdrawal is the previous one multiplied by (1 + π). The schedule then runs against your assumed return: withdraw at the start of the year, let the remainder earn the return, repeat.
The second calculation inverts it. Ask what rate would leave exactly zero at the end of n years, and you are asking for the reciprocal of an annuity-due factor evaluated at the real return. Work in real terms — divide (1 + r) by (1 + π) rather than subtracting — and the inflation-indexed withdrawal becomes a level one, so the portfolio is simply an n-year annuity. The rate that exhausts it is 1 ÷ ä(n, i).
That single expression explains most of what people argue about. At a 3% real return over 30 years it gives 4.95%; at 2% real it gives 4.38%; at 1% real, 3.84%; at zero real return it is exactly 1/30, or 3.33%. Every one of those is above 4% or below it depending only on the real return you assume, which is why the debate over the 4% rule is really a debate about expected real returns.
Investment costs enter here with more force than people expect, because they come straight off i. Moving from a 0.05% index fund to a 1.05% actively managed portfolio takes a full point off the real return, and at a 30-year horizon that drops the exhausting rate from about 4.95% to about 4.38%. On a million-dollar portfolio that is roughly $5,700 a year of income handed to fees; the expense ratio drag calculator makes the same point over an accumulation horizon.
Worked example: $1,000,000 at 4% with a 4% real return
Take a $1,000,000 portfolio, a 4% initial withdrawal rate, a 4% return, zero inflation and zero fees, over 30 years. Zero inflation keeps the arithmetic visible; the real return is what matters, and here it is 4%.
- Year-one withdrawal. $1,000,000 × 0.04 = $40,000, or $3,333.33 a month.
- Grow the portfolio. After 30 years the untouched balance would be $1,000,000 × 1.0430 = $3,243,397.51.
- Accumulate the withdrawals. Thirty payments of $40,000 taken at the start of each year and compounded at 4% come to $40,000 × 58.328335 = $2,333,133.41. The factor 58.328335 is the accumulated value of an annuity-due, (1.0430 − 1)/0.04 × 1.04.
- Subtract. $3,243,397.51 − $2,333,133.41 = $910,264 left after 30 years.
- Find the exhausting rate. ä(30, 4%) = 17.98371, so the rate that would leave exactly nothing is 100 ÷ 17.98371 = 5.56%. Your 4% sits well below it.
Now hold the nominal return at 4% and set inflation to 4% as well. The real return becomes zero, the exhausting rate falls to 1/30 = 3.33%, and the same 4% withdrawal empties the portfolio during year 25 instead of leaving $910,000. Nothing about the nominal return changed.
Withdrawal rate that exactly exhausts a portfolio
| Horizon | 0% real | 1% real | 2% real | 3% real | 4% real | 5% real |
|---|---|---|---|---|---|---|
| 20 years | 5.00% | 5.49% | 6.00% | 6.53% | 7.08% | 7.64% |
| 25 years | 4.00% | 4.50% | 5.02% | 5.58% | 6.15% | 6.76% |
| 30 years | 3.33% | 3.84% | 4.38% | 4.95% | 5.56% | 6.20% |
| 35 years | 2.86% | 3.37% | 3.92% | 4.52% | 5.15% | 5.82% |
| 40 years | 2.50% | 3.02% | 3.58% | 4.20% | 4.86% | 5.55% |
These rates assume returns arrive smoothly. Historical testing produces lower safe rates for the same average return, because a poor first decade in drawdown cannot be made up later.
How to read your result
Compare your chosen rate against the exhausting rate, and treat the gap as your margin for error rather than as surplus. A 4% rate against a 5.56% exhausting rate is a comfortable plan under smooth returns; the same 4% against a 4.38% exhausting rate leaves almost nothing in reserve for a bad sequence. If your rate exceeds the exhausting rate, the plan is not merely tight, it fails on its own assumptions.
Read the ending balance in today's dollars, never the nominal one. Ending with $2.1 million after 30 years at 2.5% inflation sounds like a large legacy, but it is about $1.0 million of purchasing power. The chart on this page plots the real balance for exactly this reason: a rising nominal line and a falling real line describe the same portfolio, and only one of them tells you whether you are getting poorer.
Then decide what the number is for. Ending with a large real balance means you underspent, which is a real cost, not a free win — retirees who die with most of their money intact bought security they could have converted into travel, help for family, or simply working fewer years. Ending near zero means the plan has no cushion for a long life or a late health shock. Most people should aim to end somewhere in between and, more importantly, to stay flexible: cutting spending 10% in the years after a large market fall raises the sustainable rate far more than any adjustment to the starting percentage.
Finally, sanity-check the rate against how long you actually need it. A 4% rate is calibrated to 30 years. Retiring at 50 means a 45-year horizon, where the exhausting rate at a 3% real return falls to about 3.96%, below the 4.20% shown for 40 years above, and historically tested safe rates land lower still. The FIRE number calculator handles that longer horizon, and the portfolio longevity calculator answers the reverse question of how long a given withdrawal survives.
What this model does not do
- It assumes a constant return. There is no volatility and therefore no sequence-of-returns risk, which is the single largest reason historically tested safe rates come in below what smooth-return arithmetic allows.
- It ignores taxes. A withdrawal from a traditional IRA is ordinary income; one from a Roth generally is not. Enter gross withdrawals for tax-deferred money.
- It assumes you never adjust. Real retirees cut spending after bad years, and guardrail strategies that do so support materially higher starting rates than a rule that never looks at the balance.
- It excludes guaranteed income. Social Security and pensions should be netted out of your spending before you decide what the portfolio must produce.
- It says nothing about the required minimum distributions the IRS imposes from age 73 or 75, which can force withdrawals larger than your chosen rate late in retirement.
- It applies one inflation rate to everything you spend. Health care and housing costs for retirees follow their own trajectories, so a single national rate is only an approximation of your personal one.
Key terms
- Initial withdrawal rate
- The first year's withdrawal expressed as a percentage of the starting portfolio. It is applied once; later withdrawals follow inflation.
- Sequence-of-returns risk
- The risk that poor returns arrive early in retirement. Two portfolios with identical average returns can end decades apart depending only on the order those returns came in.
- Real return
- Return after inflation, computed as (1 + nominal)/(1 + inflation) − 1. It is the only rate that matters once spending is indexed.
- Annuity-due factor
- The present value of $1 paid at the start of each year for n years, written ä(n, i). Its reciprocal is the withdrawal rate that exhausts a portfolio in exactly n years.
- Guardrails
- A withdrawal strategy that raises or cuts spending when the withdrawal rate drifts outside a band, trading a variable income for a higher average one.
Alternatives to a fixed rule
The constant-real-dollar rule is a benchmark, not the only option, and it is deliberately the most rigid strategy anyone has proposed. Three families of alternatives relax that rigidity in different ways.
Percentage-of-portfolio rules recompute the withdrawal each year as a fixed share of the current balance. They can never deplete the portfolio, because you always take a fraction of what is left, but income falls with markets and can drop sharply after a bad year. Required minimum distributions work this way, and the RMD calculator shows the IRS version of the same idea, with the percentage rising every year as life expectancy shortens.
Guardrail rules sit in the middle: spend a fixed real amount, but cut it when the current withdrawal rate rises above an upper band and raise it when it falls below a lower one. Modest, rare adjustments recover a large part of the gap between the rigid rule and the flexible one.
Floor-and-upside approaches split the problem instead of solving it. Cover essential spending with income that cannot run out — Social Security, especially delayed to 70, plus an inflation-indexed annuity or a ladder of Treasury Inflation-Protected Securities — and apply a withdrawal rule only to the discretionary layer above it. This is the only approach that actually removes longevity risk rather than modelling it, and it is worth pricing before you settle on a percentage. Size the total requirement first with the retirement savings needed calculator, then decide how much of it you want guaranteed.