1966 is the worst year to retire in recorded history. It’s also the wrong stress test for 2026.

Harold retires in January 1966 with $1,000,000 and a textbook 60/40 portfolio. He withdraws $40,000 a year, inflation-adjusted. The chart above shows what happens: 25 years of withdrawals, and the account hits zero in 1991.

The standard narrative frames this as a market crash story. It isn’t. The 1966 portfolio was destroyed by inflation, not a crash. In the 15 years between Harold’s retirement and the 1982 bull market, inflation averaged 7.1% annually 1967–1981 (bls cpi). the portfolio lost real value every year even when nominal returns were positive. His withdrawals, indexed to that inflation, rose in dollar terms every year — while the real purchasing power of his portfolio declined even in years when stock prices were nominally up.

Then the 1982–2000 bull market arrived and rescued portfolios that hadn’t exhausted. Harold’s did not survive the wait.

One more detail worth sitting with: Harold’s starting CAPE was 24.06. Not historically cheap, but not extreme either. The market that broke the 4% rule wasn’t even that expensive going in.

Each dot above is one historical retirement cohort — every year from 1881 to 2015. The x-axis is how expensive stocks were when that cohort retired (the CAPE ratio: price divided by the 10-year average of real earnings). The y-axis is what a 60/40 portfolio actually returned over the following decade, after inflation.

The downward slope says: retire into an expensive market, and you should expect lower returns over the next ten years. The relationship is real but noisy — CAPE explains about 18% of the variation in 10-year 60/40 returns. Not the whole story, but a meaningful headwind.

Harold’s 1966 cohort (CAPE 24.06) is labeled. The 1982 cohort (cheapest market in modern history) is labeled too — that’s the retiree who got rescued by the 1982–2000 bull market. The purple line is where we are today: CAPE 42.0, well above where Harold started.

This is the core problem with using 1966 as the 2026 stress test. Harold started at CAPE 24.06and got hit by inflation. Today’s retiree starts at CAPE 42.0 and faces a different risk: compressed returns from a historically expensive starting point. Same math, different mechanism.

A common pushback to the 1966 comparison: modern markets recover faster because of the Federal Reserve, passive investing inflows, and corporate buybacks. The COVID crash recovered in 6 months. The chart above shows the trend is real.

But notice two things. First, Japan: the Bank of Japan has owned roughly 60% of the domestic ETF market and is the top shareholder in one in five Nikkei 225 companies — the ultimate concentrated institutional buyer. The Nikkei hit its 1989 peak and didn’t recover it until 2024. Concentrated capital doesn’t guarantee fast recoveries.

Second, even the US crashes that did recover quickly took years. Dotcom: 7 years. Financial crisis: 5.5 years. That’s long enough to seriously damage a 4% withdrawal strategy, especially at today’s valuations. Five years of withdrawals from a declining portfolio permanently depletes capital.

And the mechanism that shortened US recoveries — the Fed keeping interest rates near zero from 2008 to 2022 — pushed cheap capital into asset markets, which disproportionately benefited asset owners. That era is over. Rates went from near-zero to 5%+ in 2022. The structural support that drove the COVID 6-month recovery is materially weaker now than it was in 2012–2021.

The 1966 stress test asks: can your withdrawal strategy survive 15 years of inflation eroding real returns before a rescue arrives? That’s a reasonable question for an inflation scenario.

The 2026 stress test is a different question: can your withdrawal strategy survive 7–10 years of compressed growth from a high-valuation starting point, before returns mean-revert? You don’t need stagflation to answer “no” — you just need a long enough stretch of 2–3% real returns instead of the historical 7%.

Three practical implications:

• The relevant cash buffer is 5–7 years, not 15. Modern crashes tend to resolve within that window. A portfolio that doesn’t need to sell equities for 5–7 years can ride out the most likely bad sequences.
• Flexible spending matters more than the withdrawal rate. The question isn’t “does 4% work historically” — it’s “what can you genuinely sustain for 7 years if the first decade is poor?” A floor you’d actually live on beats an optimistic rate that breaks under real pressure.
• Starting valuation adjustments are more important than historical worst-case comparisons. At CAPE 42.0, the CAPE-adjusted expected return is meaningfully below the historical average. That compression is the real 2026 risk.

1966 isn’t useless. It’s still the historical worst case, and starting at CAPE 42.0 vs Harold’s CAPE 24.06 is a reason for more caution, not less. But it stops being a useful planning scenario when the specific mechanism — a multi-decade inflation regime — is unlikely to repeat. The valuation concern is real. The stagflation scenario is the wrong frame.

These are historical simulations and analytical frameworks for educational purposes only — not personalized retirement advice. Past data does not predict future results. Households considering a specific retirement plan would typically want to discuss it with a qualified financial advisor who knows their full situation.