Retire in 1982 and ride a generational bull market in stocks and bonds alike, and a 6% withdrawal rate looks trivially safe across thirty years. Retire in 2000, and even a “conservative” 4% draw against a 60/40 portfolio can leave a retiree sweating out a lost decade. The arithmetic separating those two outcomes is not skill or portfolio construction — it is sequence-of-returns risk, the fact that losses absorbed in the first five to ten years of decumulation do outsized, often irreversible damage because withdrawals are funded from a shrinking capital base rather than a stable or growing one. A 10% loss followed by a 10% gain nets out to roughly the same annualized return whether it lands in year one or year twenty-five of retirement, but the dollar consequences diverge completely once capital is being drawn down along the way.
The 1966 Cohort and Why Averages Lie
Bill Bengen’s foundational 1994 Journal of Financial Planning research modeled rolling 30-year retirement periods using U.S. stock and bond returns back to 1926, and the single worst starting year in that dataset was 1966 (Retirement Researcher). A 1966 retiree holding a 50/50 stock-bond mix compounded roughly 9.56% annualized over the following three decades — a headline number that would satisfy almost any lump-sum investor. But because a decade of stagflation and weak equity returns front-loaded into the first fifteen years, with bonds losing roughly 2.6% a year in real terms even as nominal yields climbed, the maximum sustainable withdrawal rate for that specific cohort came in barely above 4%, at roughly 4.15% (Forbes Retirement). Shift the same retirement date a few years in either direction and the safe rate swings by more than a full percentage point — not because the total return changed much, but because of when within the sequence the drawdowns landed.
The 2000-2002 window produced a similar case study. A retiree starting with $1 million in a 60/40 portfolio and drawing $50,000 a year walked directly into consecutive S&P 500 declines of roughly 10%, 12%, and 22%; by 2010, even after a partial market recovery, the account had fallen below $500,000 (TheStreet). Researchers decomposing historical failure rates find that a market shock arriving at the moment of retirement lifts the probability of eventual portfolio failure to roughly 50.6%, versus about 33.5% for an identical shock landing fifteen years into retirement (TheStreet) — the same magnitude of loss, a starkly different survival outcome, purely as a function of timing.
From Bengen’s 4% to Morningstar’s 3.9%
The 4% figure that emerged from Bengen’s original work was never meant as a target; it was the worst historical case, rounded down from roughly 4.15% for conservative planning purposes (Retirement Researcher). Bengen himself has since revised his own view upward: incorporating a wider opportunity set of small-cap stocks and international diversification, he now argues something closer to a 4.7% starting rate is defensible in many environments. Morningstar’s annual retirement-income research — now the industry’s most closely watched running update to Bengen’s original methodology — takes a more conservative view for 2025-2026 conditions specifically: its base-case safe starting withdrawal rate for a 30-year horizon at 90% portfolio-survival probability, for an allocation holding 30%-50% equities, is 3.9%, up modestly from 3.7% the year before, with the increase driven mostly by capital-markets-assumption methodology changes rather than a brighter return outlook. The gap between 3.9% and 4.7% is not a rounding error: on a $1.5 million portfolio it is the difference between $58,500 and $70,500 of first-year income, compounding over a thirty-year plan.
Guardrails: Trading Certainty for a Higher Starting Rate
Static withdrawal rules — the classic inflation-adjusted 4% rule — ignore what the portfolio is actually doing, which is precisely why sequence risk is so dangerous under them: a retiree mechanically inflating a fixed dollar withdrawal every year regardless of market performance is the retiree most exposed to a bad opening decade. Jonathan Guyton and William Klinger’s 2006 dynamic framework addresses this directly through a small set of decision rules rather than a fixed percentage, and it permits a materially higher starting rate as a result — research on the approach suggests roughly 5.2%-5.6% initial withdrawal rates can still succeed for the overwhelming majority of historical cohorts (Kitces), because the rate is not held fixed. The capital-preservation rule cuts withdrawals by 10% if the current withdrawal rate drifts more than 20% above the initial rate, signaling the portfolio has shrunk faster than spending has adjusted; the prosperity rule raises withdrawals by 10% if that ratio falls more than 20% below the initial rate, signaling the portfolio has grown enough to support a raise. In practice, guardrails convert sequence risk from a binary “did I retire in a good year or bad year” gamble into a series of small, mechanical corrections made along the way — a retiree who opens into a 2000-style market gives up an inflation adjustment or two rather than discovering, twenty years on, that the plan quietly failed years earlier.
The Bucket Solution
Where guardrails address sequence risk through the withdrawal formula, the bucket approach — most associated with planner Harold Evensky’s “five-year mantra” of never holding money needed within five years in volatile assets — addresses it through asset location instead (Morningstar). The mechanics are simple: a cash-and-short-term bucket covers one to two years of living expenses, an intermediate-bond bucket funds years three through ten, and a growth bucket of equities is left to compound undisturbed for the long end of the plan. The point is behavioral as much as mathematical: a retiree drawing from cash reserves during a 2008- or 2022-style equity drawdown never has to liquidate depressed shares to cover ordinary expenses, which sidesteps the exact mechanism — selling low, locking in losses, permanently impairing the capital base — that turns an ordinary bear market into a sequence-risk catastrophe. Morningstar’s own testing of the approach finds it behaves less like a distinct asset-allocation strategy in its own right and more like a psychological discipline layered on top of a conventional glide path, since bucket refilling in practice tends to resemble ordinary rebalancing (Morningstar).
The 2026 Calculus
None of this is academic for someone retiring this year. The Shiller CAPE ratio for the S&P 500 sat in the 40-41 range in July 2026 — a level the U.S. market has touched only once before, in the run-up to the 2000 peak — while the 10-year Treasury yield hovered around 4.6%-4.7%. That combination is precisely the environment in which sequence risk bites hardest: rich equity valuations compress expected forward returns just as a retiree begins drawing down, and a respectable bond yield does not fully offset that risk if inflation or a valuation reset arrives within the first five years. It is also exactly the backdrop behind Morningstar’s more conservative 3.9% figure relative to Bengen’s more optimistic long-run 4.7%. For someone starting a withdrawal plan today, the practical takeaway is not to anchor on a single number, but to combine the tools: size the initial draw closer to the valuation-aware estimate rather than the historical-average one, build a two-to-three-year cash buffer to avoid forced selling into a repeat of 2000-2002 or 2022, and adopt guardrail-style rules so that if markets do open badly, the correction is a small, planned trim rather than an unplanned, terminal one.
