Sequence-of-returns risk and how guaranteed income changes the picture

Two retirees can earn exactly the same ten annual returns, spend exactly the same amount each year, and finish the decade with very different amounts of money. The only difference is the order in which the returns arrived. A retiree who meets the bad years first sells more shares at low prices to fund living costs and has less left to benefit when markets recover. A retiree who meets the good years first gets the reverse. That effect is called sequence-of-returns risk, and it is one of the main reasons retirement planning is harder than saving for retirement.

This explainer uses two invented portfolios to show the effect in dollars, then adds a guaranteed income stream to see how it changes who carries the risk. The returns, the spending and the annuity payout are all made up for the illustration. They are not forecasts and not product terms.

The setup

Two retirees each start with $1,000,000. Each takes $60,000 at the start of every year and leaves the rest invested, so the first year's withdrawal is 6% of the starting balance. The money earns the following ten annual returns, which average 5.8% a year:

Portfolio A receives the returns in that order, with the losses first. Portfolio B receives the same ten returns in reverse order, with the gains first and the losses at the end. Nothing else differs.

Each year, the arithmetic is: subtract the $60,000 withdrawal from the balance, then multiply what remains by one plus the year's return. For Portfolio A in year one: $1,000,000 − $60,000 = $940,000, times 0.80 equals $752,000.

What happens to the balances

After ten years, Portfolio A holds $639,109 and Portfolio B holds $989,854. That is a gap of $350,745 between two people who earned identical returns and spent identical amounts.

Two further facts make the point sharper. First, if neither retiree had withdrawn anything, the order would not matter at all: each would end with $1,000,000 × 1.6519 = about $1,651,915, because multiplication is the same in either order. Withdrawals are what make the sequence matter. Second, look at what the next withdrawal represents. At the end of year two, Portfolio A's $622,800 must fund a $60,000 withdrawal, which is 9.6% of the balance, and at the end of year one it was 8.0% ($60,000 divided by $752,000). Portfolio B's withdrawal after year one is 5.9% of its balance. A is withdrawing at a rate that cannot be sustained for decades, even though it started at 6%.

Why the order matters

When the balance falls, a fixed dollar withdrawal represents a larger fraction of what is left. In year one, Portfolio A sells $60,000 of investments before the 20% loss and then loses 20% of the rest. In year two, it sells another $60,000 of investments at depressed prices, and those shares will not be there to recover. The portfolio has fewer shares when the good years arrive. By year three it holds $590,940, only 59% of its starting value, with seven years left to go and no ability to add new money.

Portfolio B experiences the mirror image. The early gains build a cushion, so the later losses arrive when the portfolio is large enough to absorb them. Even so, notice Portfolio B's final years: its balance falls from $1,501,464 to $989,854 over the last three years. A retiree who plans to live another 20 years is in a different position from the one the ten-year table shows, and sequence risk is not removed by good early returns; it is only postponed.

Sequence-of-returns risk is mostly a risk of the years around the start of retirement. Savers who are still contributing face the opposite effect, because poor returns early in the accumulation phase let new contributions buy cheaply. The risk flips sign when you move from adding money to taking it out.

What retirees do about it without insurance

Planners describe several ways to reduce the damage, each with a cost:

• A lower starting withdrawal rate. Starting at 4% rather than 6% reduces the strain, at the price of spending less.
• Flexible spending. Cutting withdrawals after a bad year gives the portfolio room to recover. This works only if your essential costs are covered.
• A cash or short-term bond reserve. Spending from the reserve in down years avoids selling stocks at low prices. The reserve earns less in good years.
• A more conservative mix near retirement. Lowering stock exposure reduces the size of early losses, and also reduces growth.
• Working a little longer, or part-time. Delaying the first withdrawals shortens the period of exposure.

None of these removes the risk entirely, and each requires discipline at the moment markets feel worst.

Adding a guaranteed income floor

Now suppose each retiree uses part of the money to buy guaranteed income. The guide to single premium immediate annuities explains how one works: you pay a lump sum, and the insurer pays you a fixed amount for life. Say each retiree puts $300,000 into a life-only immediate annuity with an invented payout of 7%, which is $21,000 a year. The remaining $700,000 stays invested. To keep the same $60,000 of total spending, each retiree now needs $60,000 − $21,000 = $39,000 a year from the portfolio.

Run the same ten returns:

Each year the arithmetic is the same: subtract the $39,000, then multiply by one plus the return. Year one for A: ($700,000 − $39,000) × 0.80 = $528,800.

Now compare what each retiree faces at the start of year 11. Without the floor, Portfolio A holds $639,109 and needs $60,000, or 9.4% of the balance; with the floor it holds $498,017 and needs $39,000, or 7.8%. For Portfolio B, the rates are 6.1% without the floor and 5.4% with it. The guaranteed income does not make the portfolio bigger. It lowers the withdrawal rate the portfolio must sustain, particularly in the bad-sequence case, and because the $21,000 does not depend on markets, a good part of the household's spending is protected from the sequence.

Who carries which risk

Buying the floor does not eliminate risk. It rearranges who holds it:

Read down the right-hand column and you can see the logic of the product. It reduces the two risks that are hardest to control, bad early markets and an unusually long life, and increases the ones that are easier to plan for, such as keeping a reserve for emergencies and choosing a cost-of-living feature.

What the floor costs

The floor is not free, and the table shows part of the price. Because $300,000 left the portfolio at the start, the liquid balance at year 10 is lower in both cases: $498,017 against $639,109 in A, and $726,001 against $989,854 in B. In the better sequence the retiree who bought the annuity ends with $263,853 less in the account. That money bought something else: $21,000 a year that continues for life, however long that is, and cannot be affected by markets. The balance sheet does not show the value of the payments beyond year 10, so the table understates the annuity's side. It also does not show what happens if the retiree dies in year 3, in which case a life-only annuity has paid out little and the heirs receive nothing from it.

Other costs worth naming:

• Liquidity. The $300,000 is generally gone. A large unexpected expense must come from the remaining $700,000.
• Inflation. A level $21,000 buys less each year unless the contract includes increases, which lower the starting payment. The post on flat annuity checks and inflation works through the erosion.
• Insurer dependence. The guarantee is only as good as the company, with state guaranty association limits as a backstop.
• Pricing. The payout depends on interest rates and age when you buy. The annuity payout calculator helps you see how much it varies, but only real quotes show real prices.

Other ways to get a floor

A guaranteed floor can come from sources other than a new annuity, and in many households some of it already exists:

1. Social Security. Inflation-adjusted lifetime income, and delaying the claim raises it.
2. A pension. The post on pension lump sums and annuities discusses a related decision.
3. A deferred income annuity. The guide to deferred income annuities explains buying income that starts later, which protects the late years at a lower cost.
4. A withdrawal-benefit rider. A guaranteed lifetime withdrawal benefit on a deferred annuity lets you keep an account while guaranteeing lifetime withdrawals from a benefit base. The guide to the lifetime withdrawal benefit covers its fees and rules, and the post on benefit base versus account value explains the formula number.

How the starting withdrawal rate changes the damage

The size of the withdrawal matters as much as the order. Rerun the same two sequences with a starting withdrawal of $40,000 (4%) and of $80,000 (8%), keeping everything else the same, and compare the balances after ten years:

At 4%, even the bad sequence leaves the retiree with nearly the starting amount after ten years. At 8%, the bad sequence leaves about 30 cents on the dollar, and the retiree is withdrawing about 26.5% of the remaining balance in year 11 ($80,000 divided by $301,507), a rate no portfolio can keep up. The gap between the two sequences widens as the withdrawal rate rises, so the retiree with the least cushion is the one most exposed to being unlucky. A very low withdrawal rate is a form of protection in itself, though not everyone's savings and spending allow it.

The first three years in slow motion

Return to Portfolio A at 6% and follow the dollars for the first three years, because that is where the damage is done.

• Year 1. Start $1,000,000. Withdraw $60,000, leaving $940,000. The portfolio falls 20%, which is a loss of $188,000, and ends at $752,000.
• Year 2. Withdraw $60,000, leaving $692,000. The portfolio falls 10%, a loss of $69,200, and ends at $622,800.
• Year 3. Withdraw $60,000, leaving $562,800. The portfolio gains 5%, a gain of $28,140, and ends at $590,940.

After three years the retiree has taken $180,000 in withdrawals and holds $590,940, which is 41% less than the start. To get back to $1,000,000 from $590,940 the portfolio would need to gain about 69% ($1,000,000 divided by $590,940 is about 1.69), while still funding $60,000 a year. The later good returns of +12%, +18% and +14% do help, but they work on a smaller base. This is the arithmetic of why advisers worry most about the first five to ten years of withdrawals.

What a single example cannot show

The two portfolios are two paths out of many possible ones. Real returns are not a tidy list of ten numbers, and real retirements are longer than ten years. Planners often test a plan against thousands of simulated sequences and look at the share of paths in which the money lasts, a method sometimes called a Monte Carlo analysis. The output depends on assumptions about returns, volatility and inflation that nobody can know, so it should be read as a stress test rather than a forecast. The illustration here is the simplest version of that idea: same average, different order, different result.

It also leaves out several things that a real retiree would face: taxes on withdrawals, investment fees, inflation raising the withdrawal each year, the possibility of cutting spending after a bad year, Social Security and other income. Those would change the numbers, and in some cases the conclusion. Treat the tables as a demonstration of mechanism, not as a model of your retirement.

Questions to ask about your own plan

1. What share of my spending is covered by income that does not depend on markets (Social Security, pensions, annuities)?
2. What is my withdrawal rate in the first year, and what would it be if my portfolio fell 25% right away?
3. What would I cut, and by how much, if the first two years were poor?
4. Where would next year's money come from if stocks were down 20%?
5. If I bought guaranteed income, what would I give up in liquidity, legacy and inflation protection, and what would the insurer's failure mean for me?

An even-handed summary

The case for a guaranteed floor is that it moves the worst of sequence risk from you to a company that pools many people, and that it lets you invest the rest with less fear of selling in a panic. The case against is that you pay for that in liquidity, in legacy, in exposure to inflation and in reliance on one insurer, and that a household with large guaranteed income already, or flexible spending, or a modest withdrawal rate, may get little from adding more. The guide to are annuities a good idea? sets out both sides. A fee-only planner can run these tables with your own figures, including a more realistic range of return sequences than the one invented here. The related question of living longer than the plan assumes is covered in the post on longevity risk.

Key takeaways

• With identical returns and withdrawals, Portfolio A (losses first) ended ten years at $639,109 and Portfolio B (gains first) at $989,854, a gap of $350,745.
• The order matters only when money is being withdrawn; with no withdrawals both would end at about $1,651,915.
• A guaranteed floor of $21,000 a year lowered the portfolio's year-11 withdrawal rate from 9.4% to 7.8% in the bad sequence.
• The floor cost liquidity and legacy: the account was smaller in every case, and inflation, early death and insurer risk remain.
• Other tools, including a lower withdrawal rate, flexible spending, delayed Social Security and a cash reserve, also reduce the risk, and a planner can test them against your numbers.