The Safe Withdrawal Rate Was Built to Survive the Worst Case. Die With Zero Is a Different Equation Entirely.

The 3.9% figure Morningstar published for 2026 retirees, and the traditional 4% rule it's compressed from — originating in William Bengen's research and popularized through the Trinity Study — gets talked about in FIRE circles as though it answers the question "how much can I spend." It doesn't. It answers a narrower question: how much can I withdraw, adjusted for inflation, off my starting balance, without that fixed dollar amount ever exceeding what my portfolio could have sustained through the worst 30-year sequence of returns in the historical record. That's a survival-probability calculation. It is not the same computation as "what is the spending path that gets the most out of this money before I die," which is the actual question Bill Perkins's Die With Zero framework is trying to solve, and the two produce genuinely different numbers.

Here's the mechanism underneath the fixed-rate approach. You take your starting balance, multiply by a rate calibrated so that a fixed, inflation-adjusted withdrawal from it would have survived the worst historical stretch a retiree with your time horizon could have drawn into, and you spend that dollar amount every year regardless of what the balance actually does afterward. The rate never looks at your age, your remaining life expectancy, or your current balance relative to plan — it only looks at the balance and the calendar on day one. That's the entire design intent, and it's a good one for the problem it's solving: it gives you a number you can commit to without having to guess your own mortality or forecast markets year to year. What it does not do is target a terminal balance of zero. It targets a terminal balance of "not negative," which, because it's built around the worst historical sequence rather than the median one, means that in the majority of actual outcomes the retiree ends up with substantially more money than they started with, not less.

An optimal spend-down curve is solving a different equation, and it's worth being precise about what that equation actually is. Instead of a fixed rate off a fixed starting balance, it recalculates the withdrawal every year as a function of the balance you actually have left and the years you actually expect to have left — the same math used in an annuity payout or a mortgage amortization schedule, just run in reverse. The formula is the standard annuity-payment calculation: withdrawal equals balance multiplied by [r ÷ (1 − (1+r)⁻ⁿ)], where r is your assumed real rate of return and n is your remaining time horizon in years. Run that on a $1 million balance with a 2% assumed real return and a 30-year horizon, and the amortized method authorizes roughly $44,650 in year one — noticeably more than the roughly $39,000 that a fixed 3.9% rate authorizes on the same balance. The gap exists because the amortized number is explicitly solving for a balance of zero at the end of the horizon, while the fixed-rate number is solving for "doesn't fail even in the worst case," which by construction leaves room unused in every case that isn't the worst one.

That gap is the entire content of the Die With Zero critique, made computable rather than philosophical. Perkins's framework includes what it calls a Net Fulfillment Curve — the idea that the utility you get from a dollar declines as you age past the point where you can physically use it, and that dying with a large unspent balance is a measurable loss against that curve, not a neutral outcome. One popular breakdown of the framework puts the measurable loss from unspent savings at roughly $225,390 for the 70-to-74 age band alone; the exact figure comes from a secondary analysis rather than Perkins's own published work, but the underlying point survives without it — money that existed, that could have funded travel, care, or family transfers at a point in life when it would have done more, and instead sat in a brokerage account solving a problem that had already been solved.

The honest objection to the amortized approach is the one fixed-rate withdrawal was specifically designed to avoid: it requires you to forecast your own remaining lifespan, and if you guess wrong in the optimistic direction, the math that guaranteed zero at year thirty gives you nothing in year thirty-one. Fixed-rate withdrawal buys certainty against a known worst case at the cost of leaving money on the table in every better case. Amortized spend-down buys a materially higher standard of living for the years you're most likely to actually spend it, at the cost of needing a real fallback — Social Security, a paid-off house you can downsize, family, or a late-life annuity purchase — if you outlive the number you amortized against. Neither the FIRE community nor the retirement-income research field has settled which failure mode is worse to design around, and that's a genuine, live disagreement, not a gap in the math.

What this means in practice isn't a choice between the two models — it's recognizing that most retirees are unconsciously running the fixed-rate math their entire retirement, checking only whether the balance survives, when the actual question they care about is closer to the amortized one: am I spending in a way that matches what I have left to the years I have left, or am I just making sure I don't run out. A retiree five years into a fixed-3.9% plan who checks their actual current balance against an amortized recalculation for their actual remaining horizon will usually find they're authorized to spend more than the fixed rule tells them — sometimes considerably more — because the fixed rule is still pricing in a worst-case sequence that, five years in, may have already not happened. Whether to act on that gap by raising spending is a separate decision covered in "The Withdrawal Rate Isn't Fixed. The Rule Is." — this piece is only about seeing that the gap exists, and that it's arithmetic, not sentiment.

The 4% and 3.9% rules were never trying to spend your money well. They were trying to make sure it didn't run out. Those are not the same design goal, and the difference between them is exactly the size of the balance most retirees leave on the table.