Is the 4% rule still safe?
Sequence of returns risk, honestly.
Two retirees with identical average returns can end up with dramatically different outcomes based purely on the order those returns arrived in. Understanding this one concept changes how you should think about the first five years of retirement, and why a cash buffer and a flexible spending plan matter more than chasing another fifty basis points of return.
This page explains sequence-of-returns risk and how to reduce it with a cash buffer, a bond tent, or an income floor. For the 4% rule's origins and today's recommended starting rate, see The 4% Rule; for taxes, RMDs, and healthcare once you are actually retired, see In Retirement (65+).
The first 5–10 years of retirement determine whether your portfolio survives. A 30% crash in year 1 does far more damage than the same crash in year 15. This "sequence risk" is why the 4% rule fails in the worst ~4% of historical periods, and why a cash/bond buffer for early drawdown years matters.
- Sequence risk: the order of returns matters as much as the average return when you're withdrawing.
- Two retirees with identical average 7% returns can have dramatically different outcomes if one got the bad years first.
- The buffer: hold 2–3 years of expenses in cash/bonds, so you don't sell equities in a downturn.
- Bond tent strategy: shift to ~60% bonds at retirement, then gradually back to equities over the first 5–10 years.
- Bitcoin's 50–80% drawdowns make it especially dangerous as an early-retirement drawdown asset without a traditional buffer.
The average return on your portfolio does not determine whether your retirement survives. The order of those returns does. Bad years early combined with withdrawals are fatal because you sell depreciated assets to live on, and those shares are not there to recover when markets rebound. This risk is worst in the first five years of retirement. It is the single most underestimated threat in the whole retirement-planning system.
On this page
The concept
Recurring contributions also create sequence effects. With no interim cash flows, rearranging a fixed set of returns leaves the same compounded result for an initial lump sum. With contributions or withdrawals, the amount exposed to each return changes. Starting from zero and adding $100 at the start of each year: +10% then −10% ends at $189; reversing the returns ends at $209. The timing convention is part of the example.
During withdrawals, specify whether money leaves before or after each period’s return. This example applies the return first, then subtracts the withdrawal. Assets sold to fund spending no longer participate in later returns, but a sale during a decline can still be appropriate when needs or risk capacity change.
Two retirees experiencing the same set of returns can have different results once withdrawals are included. Losses late in retirement can also matter; their impact depends on remaining assets, spending and lifespan.
A worked example
Two retirees, both start with $1,000,000, both withdraw $40,000 per year, both experience an identical set of annual returns over five years. The arithmetic mean of those returns is 4% per year in both cases. The only difference is the order.
| Year | A return | A balance | B return | B balance |
|---|---|---|---|---|
| Start | — | $1,000,000 | — | $1,000,000 |
| 1 | -15% | $810,000 | +25% | $1,210,000 |
| 2 | -10% | $689,000 | +15% | $1,351,500 |
| 3 | +5% | $683,450 | +5% | $1,379,075 |
| 4 | +15% | $745,967 | -10% | $1,201,168 |
| 5 | +25% | $892,459 | -15% | $980,992 |
Balances use a return-then-withdraw convention: closing balance = opening balance × (1 + annual return) − $40,000. Only display values are rounded. Both paths have a 4% arithmetic mean. No inflation, tax or fees are included.
Retiree B ends approximately $88,533 ahead after five years under this particular return set and withdrawal rule. This calculation is not a prediction or a survival probability; different return paths and cash flows can change the result.
Show the data as a table
| Year of retirement | Actual: return year | Actual: real return | Actual: balance after year | Reversed: return year | Reversed: real return | Reversed: balance after year |
|---|---|---|---|---|---|---|
| 0 | – | – | $1,000,000 | – | – | $1,000,000 |
| 1 | 1966 | −6.25% | $900,012 | 1995 | +26.79% | $1,217,183 |
| 2 | 1967 | +7.77% | $926,827 | 1994 | −5.80% | $1,108,858 |
| 3 | 1968 | +2.73% | $910,998 | 1993 | +8.83% | $1,163,187 |
| 4 | 1969 | −11.49% | $770,895 | 1992 | +5.27% | $1,182,346 |
| 5 | 1970 | +4.21% | $761,701 | 1991 | +17.67% | $1,344,229 |
| 6 | 1971 | +7.28% | $774,273 | 1990 | −3.61% | $1,257,084 |
| 7 | 1972 | +7.35% | $788,276 | 1989 | +18.88% | $1,446,855 |
| 8 | 1973 | −10.85% | $667,072 | 1988 | +7.95% | $1,518,755 |
| 9 | 1974 | −20.68% | $497,392 | 1987 | −3.06% | $1,433,436 |
| 10 | 1975 | +10.27% | $504,368 | 1986 | +19.12% | $1,659,884 |
| 11 | 1976 | +13.33% | $526,277 | 1985 | +24.01% | $2,008,829 |
| 12 | 1977 | −8.77% | $443,608 | 1984 | +5.41% | $2,075,293 |
| 13 | 1978 | −4.40% | $385,847 | 1983 | +9.27% | $2,224,030 |
| 14 | 1979 | −1.53% | $340,549 | 1982 | +19.22% | $2,603,870 |
| 15 | 1980 | +0.77% | $302,866 | 1981 | −7.75% | $2,365,130 |
| 16 | 1981 | −7.75% | $242,489 | 1980 | +0.77% | $2,343,055 |
| 17 | 1982 | +19.22% | $241,414 | 1979 | −1.53% | $2,267,774 |
| 18 | 1983 | +9.27% | $220,092 | 1978 | −4.40% | $2,129,740 |
| 19 | 1984 | +5.41% | $189,830 | 1977 | −8.77% | $1,906,372 |
| 20 | 1985 | +24.01% | $185,806 | 1976 | +13.33% | $2,115,193 |
| 21 | 1986 | +19.12% | $173,686 | 1975 | +10.27% | $2,288,323 |
| 22 | 1987 | −3.06% | $129,589 | 1974 | −20.68% | $1,783,366 |
| 23 | 1988 | +7.95% | $96,715 | 1973 | −10.85% | $1,554,173 |
| 24 | 1989 | +18.88% | $67,422 | 1972 | +7.35% | $1,625,535 |
| 25 | 1990 | −3.61% | $26,431 | 1971 | +7.28% | $1,701,033 |
| 26 | 1991 | +17.67% | $0 | 1970 | +4.21% | $1,731,042 |
| 27 | 1992 | +5.27% | $0 | 1969 | −11.49% | $1,496,692 |
| 28 | 1993 | +8.83% | $0 | 1968 | +2.73% | $1,496,395 |
| 29 | 1994 | −5.80% | $0 | 1967 | +7.77% | $1,569,544 |
| 30 | 1995 | +26.79% | $0 | 1966 | −6.25% | $1,433,966 |
Method: $1,000,000 split 50/50 between the S&P 500 and 10-year Treasury bonds, rebalanced every January. $40,000 in constant dollars (raised with inflation each year) is withdrawn at the start of each year, before that year's return. Returns are deflated by CPI-U, so every balance is in first-year dollars. No taxes or fees. In actual order the 1966 retiree has $761,701 after year 5 and $26,431 after year 25, and runs out in year 26 (1991). Reversed, the same returns leave $1,344,229 after year 5 and $1,433,966 after year 30. Withdrawing at year-end instead, as in the table above, moves the depletion to year 27; measuring inflation December to December moves it to year 25.
Why this matters more if you hold Bitcoin
Bitcoin drawdowns of 50 to 80% from a prior peak are historically normal. Stocks rarely do that outside of 1929 or 2008. If your retirement starts in a BTCBitcoin (BTC)The ticker symbol for Bitcoin, used on exchanges and in price quotes.Full definition bear marketBear marketA long slide in prices, usually defined as a drop of 20% or more from a recent high in a broad market index such as the S&P 500. Bitcoin's four completed cycle bear markets bottomed 77% to 93% below their peaks, and so far it has gone on to new highs after each one. and you are forced to sell BTC at the lows to cover living expenses, the eventual rebound cannot undo the damage, because those coins are already spent.
The practical implication: Bitcoin is not a retirement withdrawal vehicle without a substantial non-BTC cushion. A retiree with 90% of assets in BTC and no cash buffer is one bear market from a permanent capital losscapital lossThe dollar amount you lose when you sell an investment for less than you paid for it. The IRS lets you use these losses to cancel out investment profits and save on taxes.Full definition that no bounce can repair.
A cash reserve can reduce forced selling, but it is finite. Set a depletion trigger and reassess spending, asset sales, allocation and retirement timing before it runs out. Do not rely on a rule that forbids selling Bitcoin in every drawdown or assumes a rebound will arrive in time.
Mitigations that actually work
- Cash and bond buffer. Hold 2 to 5 years of living expenses in safe assets (short-term Treasuries, high-yield savings, I-bonds, bond ladder). When stocks or BTC are down, you spend the buffer and wait. That stops forced selling while the buffer lasts; it does not raise the safe withdrawal rate (see the box below).
- The bond tent. Higher bond allocation in the first five retirement years, then gradually declining. Michael Kitces has written extensively on this. It concentrates risk reduction where it is most needed.
- Flexible spending. Plan to cut discretionary spending 15 to 20% in bad years. Retirees who are willing to flex survive at dramatically higher rates than those who spend a fixed dollar amount.
- Part-time income for the first two years. Any earned income in the early window reduces portfolio withdrawals when they are most damaging.
- Delay Social Security to 70. A larger guaranteed income floorincome floorGuaranteed retirement income that covers non-negotiable expenses regardless of market performance. Sources include Social Security, pensions, annuities, and bond or TIPS ladders. means the portfolio has to cover less, which reduces the withdrawal rate and the sequence risk directly.
A cash buffer protects your behavior, not your withdrawal rate. While it lasts, spending from cash in a crash means you do not have to sell stocks or Bitcoin at the bottom to pay the bills, and it lowers the odds that you panic and sell everything. It does not make a portfolio last longer. Javier Estrada's 2019 paper The Bucket Approach for Retirement: A Suboptimal Behavioral Trick? tested a two-year bucket of short-term government bills across 21 countries from 1900 to 2014: plain stock-and-bill mixes rebalanced every year beat it on all four of the paper's measures, and in the US data, holding five years of withdrawals in bills pushed the bucket failure rate at a 4% withdrawal from under 10% to over 25% verify×DON'T TRUST, VERIFYClaim: Estrada (2019) found that simple stock-and-bill mixes rebalanced every year outperformed two-year cash buckets on four performance measures across 21 countries from 1900 to 2014, and that a five-year bucket raised the US failure rate at a 4% withdrawal from under 10% to over 25%.Verify at: Estrada, J. (2019) "The Bucket Approach for Retirement: A Suboptimal Behavioral Trick?" Journal of Investing 28(5) ↗ Source 1 in the list belowThe author's copy of the published paper: the abstract states the four-measure result, and Exhibit 4 gives the US failure rates with one, two, three and five years of withdrawals in bills.. Kitces's review of the cash-reserve research reaches the same answer: years of spending held in cash drag down long-term returns, and yearly rebalancing already sells whatever is up to pay the bills verify×DON'T TRUST, VERIFYClaim: Research reviewed by Michael Kitces found that holding several years of spending in a cash reserve lowered retirement success rates because of the cash drag, and that rebalancing alone usually ensures the right assets are sold.Verify at: Kitces: Retirement cash reserve strategies don't work (2012) ↗ Source 2 in the list belowKitces summarizes a Journal of Financial Planning buffer-zone study: success rates fell across the board, though a reserve still beats panic selling or staying out of stocks.. If a buffer keeps you invested through a crash, hold one, but keep the safe assets inside an allocation you rebalance every year rather than in a separate pile you only refill.
The Floor and Enjoy framework
A structural alternative to the rigid 4% rule, formalized by retirement researcher David Blanchett and others. The portfolio is split into two functional pieces:
- Floor: guaranteed income that covers non-negotiable expenses regardless of market performance. Sources: Social Security, employer pension, single-premium immediate annuity (SPIASingle Premium Immediate Annuity (SPIA)An annuity bought with one lump sum that starts paying within about a year, for life or a set period like 20 years. A lifetime payout removes the risk of outliving that money, but the purchase generally cannot be undone, payments depend on the insurer staying solvent, and fixed payments lose buying power to inflation.), TIPS ladderTIPS ladderTreasury Inflation-Protected Securities (TIPS) bought so some mature each year, giving yearly income that rises with CPI inflation, often for a retirement income floor. Held to maturity, price swings do not matter, but income ends with the last bond, and in taxable accounts inflation adjustments face federal tax yearly., Treasury bond ladder.
- Enjoy: the remainder invested in a diversified portfolio (stocks, bonds, optionally Bitcoin) that funds discretionary spending and long-term growth. Volatility on this piece is acceptable because the floor still covers essentials.
Why this addresses sequence of returns risksequence of returns riskThe risk that bad returns early in retirement permanently damage a portfolio supporting withdrawals. Two retirees with the same average return can have very different outcomes depending on the order of returns. directly: in a year-1 market crash, the floor income covers groceries, housing, healthcare, and insurance regardless of portfolio value. The retiree does not have to sell equities at depressed prices to fund essential spending. The growth portfolio can recover without forcing distributions during the worst possible time.
The framework's design implication: maximize the floor (delay Social Security to 70, consider a SPIA for the remaining gap) before optimizing the growth portfolio. The size of the floor determines how much equity volatility the retiree can structurally absorb verify×DON'T TRUST, VERIFYClaim: Floor-and-upside frameworks formalize the case for using guaranteed income to cover essential expenses and a risk portfolio for discretionary spending.Verify at: Blanchett, D. (2014) "Exploring the Retirement Consumption Puzzle," Journal of Financial Planning ↗ · Boston College Center for Retirement Research ↗ Source 3 in the list belowThe two-bucket / floor-and-upside framework has multiple academic precedents including Bodie, Merton, and Samuelson on lifecycle finance and Blanchett's empirical retirement-spending research..
Building the floor
- Social Security at 70: the cheapest, most reliable, inflation-adjustedInflation-adjustedAdjusted for rising prices. Some amounts move with inflation, like IRS tax brackets, TIPS principal, or I-bond rates. Others are restated in today's dollars so different years can be compared: by the BLS consumer price index, $30,000 in 1985 bought about what $93,000 buys in 2026. income annuity available. Each year of delay from full retirement age to 70 adds roughly 8% to the monthly benefit, paid for life verify×DON'T TRUST, VERIFYClaim: Delayed retirement credits add approximately 8% per year between FRA and age 70.Verify at: SSA: Delayed Retirement Credits ↗ Source 4 in the list belowSSA's delayed retirement credit is 8% per year for those born in 1943 or later..
- TIPS ladder: Treasury InflationinflationA general increase in prices over time, meaning each dollar buys less than it did before.Full definition-Protected Securities matched to retirement years 1-30 produce a real-dollar income stream with US-government default risk only. The Bogleheads-style 30-year TIPS ladder is the cleanest non-annuity floor.
- Single-premium immediate annuity (SPIA): a one-time purchase that converts a lump sum into lifetime monthly income. Strips out longevity risk on a piece of the portfolio. Trade-off: you give up control of the principal.
- Pension (if applicable): already a defined-benefit floor. Counts toward the floor calculation.
For a household whose Social Security at 70 plus a small SPIA covers essential expenses, the equity portfolio can sit at 80-100% stocks for decades without sequence risk being the binding constraint. For a household where Social Security alone leaves a $30,000/year essential gap, the right answer is either to bridge the gap with a TIPS ladder or to keep the portfolio more conservative.
Ben Felix, portfolio manager at PWL Capital and host of The Rational Reminder podcast, has produced the clearest public research on sequence of returns risk, free on YouTube at youtube.com/@BenFelixCSI. His work shows how two retirees with identical average returns can have completely different outcomes depending on when their drawdowns arrive. If you are planning a Bitcoin-inclusive retirement, his sequence-risk content is required reading.
What an income floor does not guarantee
Check the terms and limits of every income source. Benefits and insurance contracts, bond ladders and dividend portfolios have different credit, inflation, tax, liquidity and longevity exposures. A ladder can end before a person’s lifetime; a dividend can be cut. A plan needs contingencies rather than treating every expected payment as guaranteed.
Where to go from here
Sources
Verify badges in the text link to these sources. Figures, rates and rules change: check each source’s date before acting on it.
- Estrada, J. (2019) "The Bucket Approach for Retirement: A Suboptimal Behavioral Trick?" Journal of Investing 28(5) ↩
- Kitces.com, Kitces: Retirement cash reserve strategies don't work (2012) ↩
- Financial Planning Association, Blanchett, D. (2014) "Exploring the Retirement Consumption Puzzle," Journal of Financial Planning ↩
- Social Security Administration, SSA: Delayed Retirement Credits ↩
- Kitces, Michael. "Understanding Sequence of Return Risk." Nerd's Eye View - kitces.com.
- Pfau, Wade. "Safe Savings Rates: A New Approach to Retirement Planning over the Lifecycle." Journal of Financial Planning.
- Bengen, William P. (1994). "Determining Withdrawal Rates Using Historical Data."
- Guyton, Jonathan and William Klinger (2006). "Decision Rules and Maximum Initial Withdrawal Rates." Journal of Financial Planning.
- Social Security Administration. Delayed retirement creditDelayed retirement creditThe increase in your Social Security benefit for each month you delay claiming past full retirement age, up to age 70: 2/3 of 1% a month, or 8% a year, for anyone born 1943 or later. With a full retirement age of 67, waiting to 70 raises the monthly benefit by 24% for life. Waiting past 70 adds nothing. tables - ssa.gov.
- Introduction to Investing | Investor.gov
Last updated . Not financial advice. Disclosures · How pages are reviewed