Sequence of returns risk: why order matters near retirement

The thing averages hide

Suppose someone tells you their portfolio averaged 7 per cent a year over twenty years. You now know something real about their investments. You know almost nothing about whether their money lasted.

That sounds like a contradiction. It is not. It is a consequence of a specific and slightly unintuitive fact: when money is flowing in or out of a portfolio, the _order_ of returns changes the final outcome, even when the average is identical. Compounding is order-independent only when nothing enters or leaves. The moment you add contributions or take withdrawals, order starts to matter, and near retirement it can matter more than the average itself.

This is sequence of returns risk, and it is the most important idea in retirement finance that most people have never had explained mechanically.

Why order is irrelevant until it isn't

Start with a portfolio nobody touches. You invest 100,000 and leave it entirely alone.

Say returns over three years are +20 per cent, then -10 per cent, then +15 per cent. The final value is 100,000 × 1.20 × 0.90 × 1.15, which is about 124,200.

Now reverse the order: -10 per cent, then +15 per cent, then +20 per cent. The final value is 100,000 × 0.90 × 1.15 × 1.20 — the same 124,200. Multiplication is commutative. The order cannot possibly matter.

Now change one thing. Withdraw 10,000 at the end of each year.

Sequence one (+20, -10, +15):

• Year 1: 100,000 grows to 120,000, withdraw 10,000, leaving 110,000.
• Year 2: 110,000 falls to 99,000, withdraw 10,000, leaving 89,000.
• Year 3: 89,000 grows to 102,350, withdraw 10,000, leaving 92,350.

Sequence two (-10, +15, +20):

• Year 1: 100,000 falls to 90,000, withdraw 10,000, leaving 80,000.
• Year 2: 80,000 grows to 92,000, withdraw 10,000, leaving 82,000.
• Year 3: 82,000 grows to 98,400, withdraw 10,000, leaving 88,400.

Same three returns. Same average. Same withdrawals. A gap of nearly 4,000 after only three years — around 4 per cent of the ending balance, from nothing but ordering.

Three years and a small withdrawal produce a modest gap. Thirty years and a lifetime of withdrawals produce a chasm.

The mechanism, stated plainly

The reason is simple once you see it. A withdrawal converts a percentage loss into a permanent reduction in the number of units you own.

If a portfolio falls 30 per cent and you take nothing out, you own the same units; they are simply priced lower. A recovery restores you fully. If a portfolio falls 30 per cent and you sell units to fund spending, you have permanently retired those units. When the recovery arrives, it lifts a smaller portfolio. The units you sold do not come back.

Worse, a fall makes you sell _more_ units for the same money. If units were worth 100 and you needed 40,000, you sold 400 units. After a 30 per cent fall the units are worth 70, and the same 40,000 costs you 571 units. Bad markets do not merely reduce your wealth; they accelerate the rate at which you consume what is left.

Stack these effects and you get the defining asymmetry of retirement investing: **a bad market early in withdrawal is far more damaging than the same bad market later**, because the early one attacks the capital base that everything afterwards depends on.

The mirror image is true during accumulation. If you are contributing rather than withdrawing, a fall early on is arguably good for you — your contributions buy more units at lower prices, and those units participate in the eventual recovery. This is why "buy the dip" instincts formed over a working life are actively dangerous once you cross into drawdown. The arithmetic reverses, and the habits that served you become the habits that hurt you.

The same market fall that helps a 35-year-old contributor can permanently damage a 62-year-old withdrawer. Nothing about the market changed. The direction of your cash flow did.

The fragility window

Sequence risk is not uniformly distributed across your life. It concentrates in a window roughly spanning the decade before you first withdraw and the decade after.

**Before the window.** Your portfolio is small relative to your future contributions. Market falls are mostly an opportunity. Sequence risk is low and mostly favourable.

**Entering the window (roughly ten years out).** Your portfolio is now large relative to what you can still contribute. A 30 per cent fall may destroy more value than you can save in five years. You cannot easily contribute your way out of a hole any more, but you also have not started withdrawing, so you have options: work longer, save more, adjust the target.

**The sharp edge (the first decade of withdrawals).** Your portfolio is at its maximum and your contributions have stopped. Every fall now compounds against you through the selling mechanism described above. This is where sequence risk does its worst damage. Historically, retirements that failed usually failed because of what happened in these years, not because of anything that happened decades later.

**Leaving the window.** If you get through the first decade with your portfolio intact, the remaining horizon is shorter and the portfolio has usually grown. The same percentage fall is now survivable, because there are fewer years of withdrawals left to fund.

The practical implication is that the risk you carry should not be constant across these phases. Many people set an allocation in their thirties and never revisit it, then walk into the sharp edge of the window carrying exactly the risk that a long horizon justified — at the moment the horizon is shortest in effect.

This matters more now than it did for previous generations. As retirement provision has shifted from arrangements where an employer or state bore the investment and longevity risk toward arrangements where the individual holds a pot and must convert it into income, sequence risk moved from being an institution's actuarial problem to being a household's personal oneSource^source.

Three mirror scenarios on a 500,000 portfolio

Take a hypothetical portfolio of 500,000 at retirement, withdrawing 25,000 in year one and increasing that by 3 per cent a year for inflation. All three scenarios use exactly the same five returns, in different orders. The returns are invented for illustration.

The five returns: -20 per cent, -12 per cent, +8 per cent, +22 per cent, +18 per cent. The arithmetic average is +3.2 per cent a year.

**Scenario A — bad years first.**

• Year 1: 500,000 × 0.80 = 400,000; withdraw 25,000 → 375,000
• Year 2: 375,000 × 0.88 = 330,000; withdraw 25,750 → 304,250
• Year 3: 304,250 × 1.08 = 328,590; withdraw 26,523 → 302,067
• Year 4: 302,067 × 1.22 = 368,522; withdraw 27,318 → 341,204
• Year 5: 341,204 × 1.18 = 402,621; withdraw 28,138 → **374,483**

**Scenario B — good years first.**

• Year 1: 500,000 × 1.18 = 590,000; withdraw 25,000 → 565,000
• Year 2: 565,000 × 1.22 = 689,300; withdraw 25,750 → 663,550
• Year 3: 663,550 × 1.08 = 716,634; withdraw 26,523 → 690,111
• Year 4: 690,111 × 0.88 = 607,298; withdraw 27,318 → 579,980
• Year 5: 579,980 × 0.80 = 463,984; withdraw 28,138 → **435,846**

**Scenario C — alternating.**

• Year 1: 500,000 × 1.22 = 610,000; withdraw 25,000 → 585,000
• Year 2: 585,000 × 0.80 = 468,000; withdraw 25,750 → 442,250
• Year 3: 442,250 × 1.18 = 521,855; withdraw 26,523 → 495,332
• Year 4: 495,332 × 0.88 = 435,892; withdraw 27,318 → 408,574
• Year 5: 408,574 × 1.08 = 441,260; withdraw 28,138 → **413,122**

Identical returns. Identical withdrawals. A spread of over 61,000 between the best and worst ordering after just five years — more than twice the first year's withdrawal, created by nothing but sequence.

Extend this over a full retirement and the gap does not merely persist, it compounds. Scenario A enters year six with 14 per cent less capital than Scenario B, and that deficit must fund the same spending for the same number of remaining years.

Note also what the averages tell you: nothing. All three retirees would truthfully report the same average annual return. Only one of them has a comfortable margin.

Five defences, ordered by how much control you have

Not all responses to sequence risk are equally available or equally reliable. Here they are ranked by the degree of control you genuinely exercise over each.

1. Flexibility in spending — highest control

You decide this. Nobody else does. If you can reduce withdrawals by 10 to 15 per cent in a bad year without your life materially changing, you have neutralised much of the mechanism: you stop selling extra units at depressed prices at precisely the moment that does the most damage.

Making this concrete beforehand is what separates flexibility from wishful thinking. Write down which specific categories get cut, by how much, and what trigger sets off the cut. A decision made in advance, in calm conditions, is a plan. The same decision attempted mid-crash, while watching the value fall, is usually not made at all.

2. Timing of the start — high control, until you use it

Working an additional year is unusually powerful because it does four things at once: it adds a contribution, it removes a withdrawal year, it shortens the horizon the portfolio must fund, and it may increase any salary-linked pension or end-of-service entitlement. Few single decisions have that much leverage.

The catch is that this lever is most valuable exactly when it is hardest to use — during a market fall, when you may also be facing redundancy or health limits. Its value comes from being retained as an option, which means not committing irreversibly to a date until you can see conditions.

3. Asset allocation and its path — moderate control

Reducing risk as you approach the fragility window lowers the size of the shock that can hit you at the worst moment. It also lowers expected returns, which raises the risk that you run short later. There is no free version of this trade.

One approach that has been studied is a "rising equity glidepath" — reducing risk into the start of retirement and then increasing it again over the following decade. The logic follows directly from the fragility window: hold the least risk when sequence risk is at its sharpest, then rebuild growth exposure once the dangerous years are behind you. Whether it suits you depends on whether you would actually tolerate re-adding risk in your seventies, which many people would not.

4. Cash and short-duration reserves — moderate control

Holding one to three years of withdrawals in cash or short-dated instruments lets you fund spending from the reserve during a fall instead of selling growth assets. The reserve is refilled in good years.

Be honest about the cost. Holding several years of spending in low-returning assets across a long retirement is a real drag on expected wealth. What you are buying is not higher returns; it is the ability to not sell at the bottom, plus the psychological cover that makes staying invested bearable. For many people that is worth paying for. It is still a payment.

5. Guaranteed lifetime income — low personal control, high structural effect

Income that arrives regardless of markets — a state pension, a defined-benefit entitlement, a purchased annuity — is immune to sequence risk by construction. Every unit of spending covered by such income is a unit that never has to be funded by selling assets in a bad year.

You typically cannot manufacture this yourself beyond deciding whether to buy it, and the terms available depend on interest rates and pricing at the moment you buy, which sit well outside your control. Asset prices and rates move through long cycles, so the conditions prevailing on any single retirement date are not representative of long-run averagesSource^source. That is an argument for staging such decisions across time rather than committing everything on one date.

What sequence risk is not

Some clarifications, because this concept gets stretched.

• **It is not a reason to avoid growth assets entirely.** Eliminating market risk does not eliminate risk; it converts sequence risk into inflation and longevity risk, which are slower but equally capable of ruining a plan. A portfolio that cannot outpace rising prices over thirty years fails just as surely, only more quietly.
• **It is not a market-timing argument.** Recognising that early falls are more damaging does not give you the ability to predict them. The defences above are all structural — they change how you are positioned and how you respond, not what you forecast.
• **It does not apply meaningfully to lump sums left untouched.** If money genuinely will not be withdrawn for decades, order is close to irrelevant, as the first example showed.
• **It is not only about crashes.** A long grinding period of mediocre returns during the fragility window can be just as corrosive as a sharp fall, and is harder to notice because there is no dramatic moment to react to.
• **It is not solved by diversification alone.** Diversification reduces the size of the shock. It does not change the arithmetic that any shock, during withdrawals, forces you to sell more units.

A short self-assessment

Work through these honestly. They are not scored; they are diagnostic.

1. How many years until you first withdraw from this portfolio? If it is under ten, you are approaching or inside the fragility window.
2. If your portfolio fell 30 per cent in the first two years of withdrawals, by how much could you reduce spending without changing anything that matters to you? Express it as a percentage.
3. What proportion of your essential spending is covered by income that does not depend on markets?
4. Could you realistically work, or return to work, for another year or two? Under what conditions would that stop being possible?
5. If you had to fund three years of spending without selling any growth asset, could you? From what?
6. Have you written down, in advance, what you will do in the first bad year — or are you relying on deciding at the time?

The last question is usually the most revealing. Sequence risk is an arithmetic problem, but the failures it causes are usually behavioural: people who had the flexibility to adapt did not adapt, because they had never decided in advance what adapting would look like.

The takeaway

Average returns describe a market. Sequence describes your outcome. Once withdrawals start, they are not the same thing, and treating them as interchangeable is the single most common analytical error in retirement planning.

You cannot control which sequence you get. You can control how exposed you are when it arrives, how much of your essential spending depends on it, and how quickly you adjust when the first bad year appears. That is the whole of the practical response, and it is entirely available to you before anything goes wrong.

This is general education, not advice about your circumstances. What the right defences are for you depends on your entitlements, obligations, health and horizon — but understanding the mechanism is what lets you ask the right questions about all of them.