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

Two retirees, both starting with $1,000,000. Both withdraw $40,000/year (the 4% rule). Both earn an average annual return of 5% over 30 years. After three decades, one has $1.4M left. The other ran out of money in year 22.

Same starting amount. Same withdrawal. Same average return. Wildly different outcomes.

The difference is the order in which the returns arrived. This is sequence-of-returns risk, and it’s the single most important concept in retirement-withdrawal math that doesn’t show up in the compound-interest formula.

The mechanism

Compounding during accumulation is robust to the order of returns. If you save $1,000/year for 30 years, you can scramble the annual returns in any order and end up with exactly the same amount. The math doesn’t care because contributions and growth are additive in each year.

Decumulation is different. Each year you withdraw a fixed dollar amount. A bad year early in retirement means:

1. The portfolio drops (say, $1M → $700k)
2. You still withdraw $40k (4% of the original $1M)
3. Now you’ve drawn down to $660k
4. Even if the market recovers next year, you’re recovering on a smaller base

If the bad year had come in year 28 instead of year 2, you’d already have 28 years of withdrawals behind you and a portfolio that’s grown substantially. The same percentage drop on a $1.5M balance still leaves you with $1.05M — comfortably above what you need.

Mathematically, the difference is between additive (accumulation) and multiplicative (decumulation) effects on the principal. Bad years compound their damage during decumulation in a way they don’t during accumulation.

A concrete example

Retiree A and Retiree B, both starting with $1,000,000, both withdrawing $40k/year inflation-adjusted, both earning average 5% real over 30 years.

Retiree A — bad sequence first:

• Years 1–5: −10%, −15%, +5%, −5%, +10% (average: −3%)
• Years 6–30: Strong markets that average +6.6% to bring the 30-year average to exactly 5%

Result: portfolio crashes early. The $40k withdrawals come out of a shrinking base. By year 5, the balance is already below $700k. The strong recovery in years 6–30 helps, but on the diminished base. Portfolio is fully exhausted by year 22.

Retiree B — bad sequence last:

• Years 1–25: Strong markets averaging +6.6%
• Years 26–30: −10%, −15%, +5%, −5%, +10% (average: −3%)

Result: portfolio grows aggressively in years 1–25 while withdrawals are still small relative to the growing base. By year 25, balance is around $1.8M. The bad sequence in years 26–30 hits, but on a much larger base, and there are only 5 years of withdrawals remaining. Ending balance: ~$1.4M.

Same average return. Same withdrawals. $1.4M difference.

The risk window: years 1–10

Research on the 4% rule consistently finds that the first decade of retirement determines almost everything. If you survive the first ten years without depleting the portfolio significantly, the remaining 20 years almost always work out.

Empirically:

• A 20% drop in years 1–5 of retirement → failure rate of the standard 4% strategy roughly triples
• The same 20% drop in years 20–25 → barely affects the success rate
• A retirement that starts in a “good decade” (1982, 1995, 2009) → almost always succeeds with a 4.5% or 5% rate
• A retirement that starts in a “bad decade” (1929, 1966, 2000) → can fail even with a 4% rate

The 4% rule was specifically chosen to survive the worst historical sequence (broadly: 1966 in US data). Most retirees don’t start in the worst sequence and end with vastly more than they began with. The mismatch — small minority fail, large majority over-save — is real.

Why the 4% rule is “safe”

Bengen’s 1994 study tested every rolling 30-year window from 1926 onward. The 4% withdrawal rate survived even the worst sequence. That sequence was the late 1960s and early 1970s: high inflation, weak real equity returns, a long stretch where inflation-adjusted withdrawals ate the portfolio at exactly the wrong time.

If you can survive that sequence, you can probably survive most of what the future holds. That’s the embedded conservatism: the 4% rule isn’t tuned to the average past. It’s tuned to the worst historical past, which is a meaningful margin of safety.

But — and this is the catch — “worst historical past” isn’t a guarantee of “worst possible future.” Markets could theoretically produce sequences worse than anything seen in US history. Researchers who run the math on international data (Pfau, Big ERN) find the 4% rule failure rate is meaningfully higher when you include other countries’ equity histories.

Defenses against sequence risk

You can’t predict when bad years will come. You can structure your retirement to be less fragile to early ones.

1. Glide-path bond allocation

Most retirement portfolios are 60–80% stocks at retirement age. Some research (Pfau, Kitces) suggests a rising equity glide path instead: start retirement at maybe 40% stocks / 60% bonds, increase stock allocation over the first decade. The bond cushion protects against early sequence risk; the rising equity allocation captures the long-term return when sequence risk has receded.

Counterintuitive — but the math holds up under simulation. It works because the bond-heavy first decade reduces the damage from a bad start, and by year 10–15, you’re past the danger zone.

2. Cash-equivalent bucket

Keep 2–3 years of expenses in cash or short-term Treasuries at retirement. In a market downturn, withdraw from cash, not from the equity portfolio. This lets equities recover before you sell.

The cost: cash earns less than equities. The benefit: in a 30% market drop in year 2 of retirement, you’re not forced to sell at the bottom. Estimates suggest a cash bucket of 2–3 years can lift safe withdrawal rates by 0.3–0.5%.

3. Flexible spending rules

The constant-real-dollars 4% rule is rigid by design. Flexible withdrawal rules are not. The Guyton-Klinger guardrails, for instance:

• Skip the inflation adjustment in a year when the portfolio is down
• Cut withdrawals by 10% if the calculated rate exceeds 120% of the initial rate
• Increase withdrawals 10% if the rate drops below 80% of initial

These rules don’t require timing the market — they’re triggered by withdrawal-rate thresholds. Simulations show 5% with Guyton-Klinger has comparable success rates to a constant 4%.

4. Income floor

Social Security at 67, a pension, annuity income, or part-time work in retirement all reduce sequence risk by reducing the withdrawals you need from the portfolio. The math: every $10k of pension income reduces the portfolio you need by $250k (at 4%). It also reduces the dollars you must withdraw in a bad year, which is where sequence risk does its damage.

The cleanest sequence-risk defense for many early retirees: a small part-time income for the first 5–10 years of retirement. Even covering 30% of expenses with earned income cuts sequence risk substantially.

5. Variable percentage withdrawal (VPW)

Instead of a constant inflation-adjusted dollar amount, withdraw a percentage of current portfolio balance each year. The percentage starts low (~4%) and rises with age (because remaining lifespan shortens).

Mathematically, VPW cannot run out — it asymptotes toward zero. The cost: income fluctuates with the market. In a bad year, you withdraw less. Many retirees find this acceptable because their fixed costs are lower than their gross spending, so a 20% income drop in a bad year doesn’t break the budget.

What sequence risk does not affect

A few things people assume sequence risk affects, but it doesn’t:

• Pre-retirement saving. As above, accumulation is order-invariant. Saving more in your 20s during a bear market actually helps long-term outcomes (cheaper share purchases).
• The geometric average return. The 30-year average return is the same regardless of order. Sequence risk is about withdrawal timing, not return calculation.
• The IRR of contributions. Internal rate of return calculations on contributions account for timing automatically.

Sequence risk is specifically a withdrawal-phase problem. If you’re not withdrawing, you’re not exposed to it.

How to think about it pragmatically

Don’t try to time markets to avoid bad starting years. Nobody can. Instead:

• Plan for a worse-than-average sequence. Use a 3.5% SWR target instead of 4% if you’re retiring early. The extra savings is your insurance premium.
• Hold cash equivalents at retirement. 2–3 years of expenses outside the equity portfolio.
• Be willing to flex spending in bad years. The biggest leverage on sequence-risk survival is your behavior, not the asset allocation.
• Earn part-time if you can, for the first 5–10 years. Smallest of incomes makes the biggest difference because of where it falls in the risk window.

The 4% rule survives sequence risk in nearly all historical cases. The behaviors that get retirees into trouble aren’t math problems — they’re discipline problems. Sticking with the plan through the first decade is the hardest part.

For the underlying math of how compounding works (and where it fails during withdrawals), see the compound interest cornerstone. For the FIRE-number target math, see the FIRE number guide. For the savings-rate math that determines how soon you face this problem, see the savings-rate guide. For the rule this entire risk was calibrated against, see what the 4% rule actually says.

Educational content, not financial advice. Sequence-of-returns analysis depends heavily on historical data and modeling assumptions. Consult a fee-only fiduciary advisor before relying on any specific withdrawal strategy.