Sequence of Returns Risk: Why the First Years Matter Most

Take two people who retire with $1,000,000 each. Both withdraw $40,000 at the start of the first year and raise that amount by 3% a year for inflation. Both earn exactly the same 30 annual returns, which average 6%. Retiree A receives them from worst to best, starting with a 9% loss. Retiree B receives the identical returns from best to worst. This is the stress test built into our retirement withdrawal calculator, run on its opening plan.

Retiree A runs out of money in year 15. Retiree B never comes close: after 30 years of the same withdrawals the account holds $2.87 million. Neither result has anything to do with skill, costs or the average return, which are identical.

The table shows where the damage is done. After ten years Retiree A has $237,742 left and Retiree B has $3,487,460. The eleventh withdrawal is the same dollar amount for both, but for A it is now 22.6% of the remaining balance. The strong returns A has been waiting for arrive from then on, and they arrive on a portfolio too small to benefit from them.

Because multiplication does not care about order, and withdrawals break the multiplication. Leave the $1,000,000 alone, with nothing going in or out, and run it through the same two sequences: both end at $5.16 million. A 9% loss followed by a 21% gain leaves exactly what a 21% gain followed by a 9% loss does.

A withdrawal changes that. Each one removes a fixed number of dollars whatever the portfolio is worth, so after a fall it removes a larger share. Those dollars have left the account for good, and when prices recover, the recovery is earned on a smaller base. A retiree who sells shares at a low to pay for groceries has turned a paper loss into a permanent one for that slice of the portfolio. Repeat that for several years in a row and the arithmetic that rescues a buy-and-hold investor no longer applies.

The mirror image while you are saving

Contributions break the multiplication in the opposite direction. A saver wants the poor years early, when the balance is small and each new contribution buys at low prices, and the strong years late, when there is a large balance for them to work on. Put $20,000 into an account at the start of each year for 30 years and run the same two sequences. With the worst years first, the saver ends with $3.30 million. With the best years first, the saver ends with $846,000, on identical contributions and an identical average return.

Put the two halves together and the pattern is clear. The returns that matter most are the ones earned when the balance is largest, and for most households that is the stretch from roughly ten years before retirement to ten years after. A crash at 35 is an opportunity. The same crash at 63 or 67 falls on the largest balance the household will ever have. Our guide to asset allocation by age covers how that changes the mix people hold as retirement approaches.

The two-retiree example is an extreme. A fairer test is to simulate thousands of plausible retirements and ask how much of the outcome the first decade explains. Our Monte Carlo retirement calculator does this with 10,000 paths. Its opening plan is a $1,000,000 portfolio held 60 / 35 / 5 in stocks, bonds and cash, $40,000 of spending a year in today's dollars, a 0.25% fee and 30 years, with return assumptions taken from US history since 1928. Across all paths, 90.5% pay the full spending for 30 years.

Now sort the paths by their return over the first 10 years only, and split them into five equal groups. Every group draws from the same assumptions for all 30 years. They differ in nothing but luck in the opening decade.

In the worst fifth, where the portfolio earned a median of 0.2% a year above inflation for a decade, 58.6% of retirements lasted. In the best fifth, 100% did. Of the 947 paths that failed, 87% belong to the bottom group. Among failed paths the median ran out in year 26, long after the decade that caused it.

That lag is the awkward feature of sequence risk. The cause is early and the consequence is late. A retiree whose first ten years go badly is not broke at 75. The portfolio is smaller than planned and the withdrawals are a larger share of it, and the plan fails quietly over the following decade unless something changes. It also means the warning arrives while there is still time to act on it.

The starting withdrawal rate sets how much a bad decade can hurt. At a low rate the portfolio can absorb a poor start and wait. At a high rate the same poor start is fatal in most cases. The table runs the same 10,000 paths at four starting rates and reports the result for everyone, for the unlucky fifth and for the lucky fifth.

Read the middle column. At 3%, even the unlucky fifth survives 92.7% of the time: the withdrawal is small enough that a poor decade can be waited out. At 5%, the unlucky fifth survives only 17.1% of the time. The right-hand column tells the same story from the other side. At low rates nearly every failure is a sequence failure. At 6%, failures spread into average decades too, because the rate is too high for the portfolio to support even with ordinary luck.

For where the 4% figure comes from, what it assumes and how it holds up over 20 and 40 years, see our guide to the 4% rule.

Simulations draw each year independently. History did not, so it is worth running the real record. We took Aswath Damodaran's annual US returns from NYU Stern (Historical Returns on Stocks, Bonds and Bills, 1928 to 2025), restated after inflation, and ran four retirements through the same drawdown engine as the example above: $1,000,000 held 60 / 35 / 5 in the S&P 500, 10-year Treasury bonds and Treasury bills, rebalanced each year, with $40,000 withdrawn at the start of every year in constant buying power. No fees or taxes are deducted. All figures are in start-of-retirement dollars.

1966. The S&P 500 lost 13.0% after inflation in 1966, which is unremarkable. What followed was not a crash but a grind: for fifteen years the portfolio returned −0.9% a year after inflation, as consumer prices rose by as much as 13.3% in a single year and bonds lost buying power alongside stocks. The next fifteen years returned 9.2% a year. Averaged over all thirty, the real return was 4.8% a year, a perfectly ordinary figure. The retiree drawing 4% still ran out in year 27. The highest starting rate that would have lasted the full thirty years was 3.76%.

To see how much of that was order, run the same thirty returns backwards, so that the strong years of the 1980s and 1990s come first and the 1970s come last. The same retiree, with the same withdrawals and the same average return, finishes with $1.61 million in real terms.

1973. This is the crash retirees fear. The S&P 500 lost 21.2% and then 34.0% after inflation, 48% over two years, while inflation ran at 8.7% and 12.3%. The balanced portfolio fell 14.4% and 23.8% in real terms. Yet the 1973 retiree lasted the thirty years, ending with $246,000, about a quarter of the starting value. The fall was deeper than anything the 1966 retiree saw in a single year, but it came at once, and the long recovery that began in 1982 arrived in year ten, not year seventeen. William Bengen, whose 1994 study gave rise to the 4% rule, called 1973 to 1974 the most damaging event in his data for the same reason: a market fall combined with high inflation (Journal of Financial Planning).

2000. Stocks lost 42% after inflation from 2000 to 2002, but Treasury bonds rose in each of those years, so the balanced portfolio fell only 2.6%, 6.4% and 10.0%. The harder part was that 2008 arrived in year nine. Over its first decade the portfolio earned 0.2% a year in real terms. Through 2025, 26 years in, this retiree holds $580,000 of the original $1,000,000 in real terms, having touched a low of $522,000. Diminished, and still solvent.

2008. The S&P 500 lost 36.6% after inflation in the retiree's first year. The 10-year Treasury gained 20.0%, and the balanced portfolio was down 14.9%. The recovery began the following year. After 18 years of withdrawals this retiree has $1.28 million in real terms, more than they started with.

The lesson of the four is that a sharp crash followed by a recovery is survivable at moderate withdrawal rates, and a long stretch of poor real returns is the true danger. Inflation is as much the enemy as falling prices, because it raises the withdrawal while the portfolio stands still. Two cautions apply. These runs use one bond series and whole calendar years, so other studies of the same periods report slightly different numbers. And four retirements are anecdotes, which is why the simulated table above carries more weight.

Nothing removes sequence risk. Each approach below reduces how much of the plan depends on the first decade, and each has a cost. They are described here as information about how each one works. Which, if any, suits a household depends on its spending, its other income and its tolerance for change, and that is a conversation for the household and its advisers.

A lower starting withdrawal rate

The table above is the case for it. Starting at 3% instead of 4% lifts the unlucky fifth from 58.6% to 92.7%. In the historical record, the 1966 retiree who drew 3.5% instead of 4% finished thirty years with $231,000 left. The cost is paid in every outcome: $10,000 a year less spending on a $1,000,000 portfolio, or more years of work to build a larger one, including in the large majority of retirements where the poor decade never comes. In the simulations the median path at 3% ends with $2.08 million unspent.

A cash or short-term bond reserve

Often called a bucket strategy: one to three years of planned withdrawals are held in cash, Treasury bills or a CD ladder, and spent after a market fall so that shares do not have to be sold at depressed prices. The reserve is refilled in good years. It works best against sharp falls with quick recoveries, such as 2008, and it makes a bad year easier to live through. It cannot outlast a decade like the 1970s.

The cost is the return given up on money held in cash. Our simulation holds a fixed mix, so it cannot credit a reserve for the choice of what to sell, but it can price the drag. For a $1,000,000 portfolio at a 4% withdrawal, ten points of cash is about 2.5 years of spending.

A reserve funded from bonds costs little. A reserve funded by selling stocks costs more, and more still at higher withdrawal rates, because it lowers the return the withdrawals depend on.

Flexible spending and guardrails

A fixed real withdrawal is the harshest rule a plan can follow, because it takes the same amount the year after a crash as the year after a boom. Guardrail rules cut spending when the portfolio falls behind and restore it when the portfolio recovers. The best-known published version, by Jonathan Guyton and William Klinger in 2006, cuts withdrawals by 10% when the current withdrawal rate has risen 20% above its starting level and raises them by 10% when it has fallen 20% below (Kitces.com, which also documents how deep the cuts under those rules can become).

Our Monte Carlo calculator models a simpler rule: in any year that starts with the portfolio below 80% of its starting value in real terms, spending is cut by 10%. At a 4% starting rate that lifts the overall success rate from 90.5% to 94.2%, and the unlucky fifth from 58.6% to 73.4%. At 5% it lifts the overall rate from 72.7% to 80.3%.

The cost is the cuts themselves. At 4% the rule is triggered at least once in 52% of paths, and in those paths the median time spent at reduced spending is 13 years. Flexibility converts some of the risk of running out into a larger chance of living on less for a long stretch. It suits a budget with real discretionary room and does little for one that is mostly fixed bills. Morningstar's research for 2026 reaches the same trade from the other direction: its highest safe starting rate for fixed inflation-adjusted spending over 30 years is 3.9%, and retirees willing to tolerate fluctuations in spending could start at nearly 6% (Morningstar).

Delaying Social Security as longevity insurance

A benefit claimed after full retirement age grows by two-thirds of 1% for each month of delay up to age 70, which is 8% of the full benefit a year, for anyone born after January 1, 1943 (20 CFR 404.313). The larger benefit is adjusted for inflation and lasts for life, so it reduces what the portfolio must deliver in every later year. The price is that the portfolio carries the whole budget until the benefit starts, which makes the first years heavier, the opposite of what sequence risk would suggest.

The simulation shows the net effect. Take the $1,000,000 portfolio with $70,000 of spending and a $30,000 benefit at 67. Claimed at once, 90.5% of paths last 30 years. Held to 70, the benefit becomes $37,200 and 94.0% last; in the unlucky fifth the figure rises from 58.6% to 72.1%. Over 40 years the gap is wider: 80.9% against 86.2%. The insurance is against a long life with a depleted portfolio. Health, a spouse's benefit and taxes all bear on the decision; see when to claim Social Security and the break-even calculator.

Annuitising a floor

Some retirees exchange part of the portfolio for guaranteed lifetime income, sized so that Social Security, any pension and the annuity together cover essential bills. Spending that is covered this way is no longer exposed to the order of returns, and the remaining portfolio funds discretionary spending, which can flex. The costs are that the money is no longer available for emergencies or heirs, that most income annuities pay a fixed dollar amount which inflation erodes, and that the income depends on the insurer's ability to pay. Our guide to the types of annuities explains the contracts, and the annuity payout calculator shows what a lump sum pays over a set period.

A rising equity glide path

The conventional path holds fewer stocks each year through retirement. In 2014 Wade Pfau and Michael Kitces tested the reverse in the Journal of Financial Planning: start retirement with a low share in stocks and raise it over time. Using 10,000 simulated retirements for each of 121 combinations of starting and ending allocations, they found that rising paths could reduce both the probability of running out and the size of the shortfall, and that the best starting stock allocations were generally around 20% to 40%. A portfolio that began at 30% in stocks and finished at 60% did better in their tests than one held at 60% throughout (Pfau and Kitces, 2014; Kitces's summary).

The logic is sequence risk. Stock exposure is lowest when the portfolio is largest and most vulnerable, and if the early years are poor, the retiree buys more shares at low prices on the way up. The authors' own caveats matter: the results depend on the return assumptions used, the benefit is small over 20-year horizons, taxes were not modelled, and holding more stocks at 85 than at 65 is uncomfortable for many people. The cost in good markets is the growth given up by holding fewer stocks early.

Part-time income in the first years

Earnings in the first years of retirement reduce withdrawals at exactly the time withdrawals do the most harm. Suppose the January 1966 retiree had earned $20,000 a year, in constant dollars, for the first 5 years and so withdrawn half as much, then the full $40,000 from year 6. On the same historical returns the portfolio lasts the thirty years and ends with $130,000, where the full withdrawals ran out in year 27. Lighter withdrawals in the five years when the portfolio was most exposed were enough to change the outcome. The cost is the obvious one: it is five more years of work, and not everyone can find it or wants it.

Trying to step aside before the bad years. Sequence risk invites the thought that a retiree should move to cash when markets look expensive and return when the danger has passed. That requires two correct calls, out and back in, and the history above shows why it tends to fail. The worst retirement in the record began in 1966 with a modest loss and no obvious signal. The damage accumulated over sixteen years in which cash and bonds lost buying power as well.

Selling after the fall is worse. It turns the temporary loss into a permanent one, which is the mechanism of sequence risk itself. Bengen's 1994 paper works through the case of a 1929 retiree who moved entirely to bonds at the end of 1932: the money ran out in 1946, while the same retiree holding 75% in stocks still had a portfolio in 1992. His conclusion was that retreating from stocks after a decline is the one alternative such a retiree cannot afford (Journal of Financial Planning).

A higher average return does not help much either. Reaching for more return usually means more volatility, and volatility in the first decade is the problem. Nor does a single reassuring projection at an average return: every plan in this guide looks fine on that basis, including Retiree A's.

Two of our calculators test sequence risk directly, and they answer different questions.

• The retirement withdrawal calculator runs the two-retiree test on your own figures. Enter your balance, withdrawal, expected return and volatility, and its sequence tab shows the plan with the worst years first and the best years first. If the worst-first case runs out well before your planning horizon, the plan depends on a kind first decade.
• The Monte Carlo retirement calculator runs 10,000 paths and reports the share that last, with the first-decade table from this guide computed for your inputs. Look at the worst-fifth row, not only the headline. Then switch on the guardrail rule and add Social Security from its start year to see how much each changes that row.

Three numbers are worth writing down: the starting withdrawal as a percentage of the portfolio, after subtracting Social Security and pensions from spending; the share of spending that could be cut for several years without hardship; and how many years of withdrawals are held outside stocks. If you are still several years from retiring, the coast FIRE calculator and our guide to how much you need to retire cover the saving side, where a late crash is the mirror-image risk.