Bucket Strategies for Retirement Income

Bucket strategies do not, on average, beat a simple systematic-withdrawal portfolio with the same overall asset allocation. The published evidence is fairly consistent on this, and the most rigorous head-to-head comparisons — Pfau and Murguia (2022), Fox (2012) — show that two of the three popular bucket variants actually under-deliver mathematically.¹² What buckets buy you is behavioral: a structure that lets you stop selling stocks during a bear market, because the cash and bond buckets are explicitly there to be drained. That is a real benefit, but it is a benefit you have to earn by following the refill rules. It is not free alpha.

Get that framing right and the rest of this article is useful. Get it wrong and you'll oversize cash, drag your long-run return, and wonder why you outlived your portfolio.

The Sequence-of-Returns Problem (Why This Conversation Exists)

Two retirees with identical 30-year average returns can have wildly different outcomes if those returns arrive in a different order. Withdrawals during a bear market lock in losses on shares that won't be there to recover.

Bengen's 1994 paper put a number on this. Using historical U.S. data from 1926 forward and a 50% S&P 500 / 50% intermediate-term government bond mix, he found a maximum sustainable initial withdrawal rate of 4.15% — what he later named SAFEMAX — for a 30-year retirement.³ The binding constraint was the 1966 cohort, who retired into a 16-year stretch of negative real stock returns and a stagflationary bond market. Importantly, Bengen's 1966 retiree at 4.15% did not deplete the portfolio early; they exhausted it right around year 30, which is what makes 4.15% the worst-case maximum, not a cliff.³

The Trinity Study (Cooley, Hubbard & Walz, AAII Journal, February 1998) tested a similar framework with S&P 500 and long-term high-grade corporates and reached a similar conclusion: a 4% inflation-adjusted withdrawal succeeded in roughly 95–98% of historical 30-year periods depending on allocation, with success rates dropping sharply above 5%.⁴

The point is: the famous "4% rule" is not Bengen's; it's a Trinity-style success-rate framing built on top of Bengen's SAFEMAX work. Citing one when you mean the other is a tell that someone hasn't read the papers.

What Sequence Risk Actually Looks Like (Real Numbers)

The historical episodes that drive the 4% constraint are surprisingly specific. They are worth knowing in their actual shape — the popular retellings tend to drift.

1929–1932 (Great Depression). The S&P composite total return on a calendar-year basis was −8.3%, −25.1%, −43.8%, and −8.6% — a cumulative loss of roughly −64.8% measured calendar-to-calendar.⁵ The peak-to-trough drawdown (September 1929 to June 1932) was deeper, around −86% with dividends reinvested.⁶ Nominal price-only recovery to the 1929 high took until 1954 on the Dow; on a dividend-reinvested, real-terms basis the S&P composite recovered to its 1929 peak by late 1936, about 4.5 years from the trough — not 25.⁶ Anyone telling a 25-year recovery story is quoting Dow price index, not a portfolio.

1973–1974 (stagflation bear). S&P 500 calendar returns of −14.7% and −26.5% (cumulative ~ −37%); intermediate-term government bonds returned +3.66% and +1.99% nominal — they did not "lose 15%" as is sometimes claimed.⁵ In real terms bonds lost ground because CPI inflation ran 8.7% and 12.3% — but nominal bond returns were positive, which matters for someone funding withdrawals from the bond bucket.

2000–2002 (dot-com). S&P 500 total return cumulative loss of roughly −38% over three years; the index recovered nominally by late 2006 / early 2007.⁵

2008 (GFC). S&P 500 calendar return −37.0%; intermediate Treasuries +13.1%; full nominal recovery by early 2013.⁵

The pattern that holds: equity drawdowns this severe are rare, but when they hit early in retirement they impose a permanent withdrawal-on-losses tax. The bucket framework's job is to give you bonds and cash to spend from for long enough that the equity bucket can recover before you're forced to touch it.

The Three-Bucket Framework (As It's Actually Practiced)

Most modern bucket implementations follow a Christine Benz / Harold Evensky-style segmentation: a short-horizon cash bucket, an intermediate-horizon bond bucket, and a long-horizon equity bucket.⁷ Sizing varies, but a common starting point is:

Two things to note. First, sizing is net of guaranteed income. If you spend $80,000 and Social Security covers $35,000, your portfolio gap is $45,000 — that's the number you multiply by 1, 2, and 8. Second, the bucket sizing is a labeling exercise for an underlying allocation. If your buckets sum to 70/25/5, you are running a 70/25/5 portfolio. Buckets do not change risk; they change the rules under which you'll spend it.

Pfau, Fox, and the Honest Finding on Bucket Performance

This is the part most retirement-income articles get wrong. The published comparisons tell a consistent story.

Fox's 2012 study in the AAII Journal compared three-bucket strategies to systematic withdrawals from a target-date glide path and concluded that bucket strategies "do not provide superior mathematical returns" — and in some scenarios under-delivered because funds only ever moved from risky to conservative buckets, never refilling equities, leaving the portfolio increasingly conservative as it aged.²

Pfau and Murguia (2022, SSRN) tested four classes of strategy at an 85% funding success rate and found the cash flow reserve and two of the three-bucket variants underperformed the systematic withdrawal approach on viable retirement spending. One variant — the time-segmentation strategy — exceeded the systematic-withdrawal benchmark on viable spending ($6,206 vs. $5,898 monthly) but with the largest variance in outcomes.¹ In other words: when buckets win, they win by accepting more dispersion, not by reducing risk.

The fix isn't to abandon buckets. It's to use them for the right reason. The point is: bucket strategies are a behavioral commitment device, not a return enhancer. They keep you from selling stocks at the bottom because there's an explicit rule that says "sell from Bucket 2 first." If you would otherwise panic and sell equities in a bear market, buckets have measurable value. If you wouldn't, the simpler approach is to maintain a target asset allocation and rebalance.

Refill Rules That Survive Scrutiny

The literature converges on a small number of rules that actually do something. Use these; ignore the longer lists.

Rule 1 — Keep Bucket 1 at 1–2 years of the spending gap, no more. Cash above 24 months of need drags long-run return without buying additional crisis protection. Christine Benz's bucket research at Morningstar uses a 12–24 month range for exactly this reason.⁷

Rule 2 — Refill from the bucket that's "winning." In a normal year, sell from Bucket 3 (stocks) to refill Bucket 1. In a year when stocks are down materially (a common operational threshold is −15% from the prior year's portfolio value), refill Bucket 1 from Bucket 2 (bonds) and skip the equity sale. This is the rule that captures whatever sequence-risk benefit buckets actually provide.

Rule 3 — Use the Guyton-Klinger guardrails for spending, not buckets. Guyton & Klinger (2006) showed that two simple decision rules — a capital-preservation rule (cut withdrawal by 10% if the current-year withdrawal rate climbs more than 20% above the starting rate) and a prosperity rule (raise withdrawal by 10% if the current-year rate falls more than 20% below) — supported sustainable initial rates of 5.2–5.6% at the 99% confidence level over 40 years, materially higher than the static 4% rule.⁸ Note that these guardrails operate on the withdrawal rate, not the portfolio return — easy to get wrong.

Rule 4 — Don't drift below ~50% equities even late in retirement. Bengen's 1996/2017 follow-up work and the Trinity Study updates both find that equity allocations under 50% tend to fail on inflation over 30-year horizons, even though they look "safer" on a 1-year volatility basis.³⁴

That's the honest list. Skip the "Mistake 1 / Mistake 2 / Mistake 3" listicles you'll find elsewhere — most of them are restatements of these four.

Worked Sequencing: A 65-Year-Old with $1.5M and a $45,000 Spending Gap

You retire at 65 with $1,500,000, expect to spend $80,000/year, and Social Security covers $35,000. Your portfolio gap is $45,000/year. Using a 1-year cash / 5-year bond / equity-residual sizing:

• Bucket 1 (cash): $45,000 — about 3% of portfolio
• Bucket 2 (Treasury ladder + intermediate fund): $225,000 — about 15%
• Bucket 3 (equities): $1,230,000 — about 82%

That's an 82/18 portfolio. Aggressive — but recall that you are 65 and likely retiring into a 30-year horizon, and you have Social Security covering 44% of spending. The right comparison is not 82/18 vs. 60/40; it's whether 82/18 is consistent with your full balance-sheet allocation, which now includes Social Security as an inflation-linked annuity worth roughly $700,000 in present value terms.⁹

Year 1, normal market: Spend $45,000 from Bucket 1. End of year: Bucket 1 is empty. Refill it by selling $45,000 from Bucket 3.

Year 2, S&P −22% (a 2008-style year): Spend $45,000 from Bucket 1. End of year: Bucket 1 empty, Bucket 3 down ~$270,000. Refill Bucket 1 from Bucket 2, not Bucket 3. Bucket 2 drops to ~$180,000 (still ~4 years of gap). You have not realized any equity losses. The equity bucket, untouched, is free to recover.

Year 3, S&P +28% (a typical post-bear bounce): Spend $45,000 from Bucket 1. Equities recover materially. Refill Bucket 1 from Bucket 3, and use the strong equity year to top Bucket 2 back to $225,000. This is the "sell-from-the-winner" rule operationalized.

The mechanical alternative would have you rebalance to 82/18 every year regardless. In Year 2 that means selling enough equities at depressed prices to fund the withdrawal and restore the bond weight — exactly what the bucket framework is designed to prevent. Whether this matters to your 30-year outcome depends on how big and persistent the bear is. In the historical record the difference is real but smaller than the marketing implies — the Pfau-Murguia and Fox studies are the ones to read on the magnitude.¹²

Where to Hold Each Bucket (Account Location Matters)

A bucket framework intersects with the account-location problem. Generally:

• Cash (Bucket 1) is most efficient in a taxable account — you want immediate access without RMDs or 10% penalties, and money-market interest is fully taxable anyway.
• Bonds (Bucket 2) belong in tax-deferred accounts (Traditional IRA / 401(k)) where their ordinary-income coupons compound untaxed. Holding intermediate Treasuries in taxable wastes the tax-deferred wrapper.
• Equities (Bucket 3) are most tax-efficient in Roth (no future tax on growth) or in taxable accounts (qualified dividends, long-term cap gains, step-up at death).

This is just standard asset location applied to bucket labels. Done right it's worth roughly 20–60 basis points per year of after-tax return over a long horizon — meaningful but not transformative.¹⁰

The Detection Signals (You're Probably Misusing Buckets If…)

• Your "Bucket 1" holds three or more years of cash. You're paying a real return drag for protection you don't need.
• You rebalance buckets on a calendar (every January 1) regardless of what markets did. The whole point is to not sell equities in down years.
• Your buckets sum to a more conservative allocation than you'd otherwise hold. Buckets are a labeling exercise, not a justification for de-risking.
• You're citing Pfau or Kitces for the claim that buckets reduce sequence risk by 20–30%. Neither said that; both have written that the benefit is mostly behavioral.¹¹¹
• Bucket 2 has drifted to long-duration corporates because they yield more. You've defeated the purpose — Bucket 2 needs to be resilient when stocks fall, which means short-to-intermediate Treasuries, not credit.

Your Next Step

Open your retirement account statements and write down two numbers: your annual spending gap (spending minus guaranteed income), and the dollar value of your cash + short-term bond holdings. Divide the second by the first.

• Less than 1 year of gap in cash + short-bonds: you have no real Bucket 1 and will be selling equities in any drawdown. Build it.
• 1–5 years: you're in the operational range. Pick a refill rule (the −15% portfolio-drop trigger is a clean default) and write it down.
• More than 7 years: your "Bucket 1+2" has eaten your equity allocation. You're running a more conservative portfolio than you probably intend. Re-aggregate the buckets, decide on a target overall allocation, and resize.

That's the test that separates a real bucket strategy from a story about one.

---

Footnotes

1. Murguia, Alejandro, and Wade D. Pfau. "A Model Approach to Selecting a Personalized Retirement Income Strategy." SSRN, 2022. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4011297. The cash flow reserve bucket and two variations of the three-bucket strategy underperformed systematic withdrawals on viable retirement spending at the 85% funding-success threshold. ↩ ↩² ↩³ ↩⁴

2. Fox, Noelle E. "Comparing a Bucket Strategy and a Systematic Withdrawal Strategy." AAII Journal, April 2012. https://www.aaii.com/journal/article/comparing-a-bucket-strategy-and-a-systematic-withdrawal-strategy. Concludes that bucket strategies "do not provide superior mathematical returns" — the advantage is psychological. ↩ ↩² ↩³

3. Bengen, William P. "Determining Withdrawal Rates Using Historical Data." Journal of Financial Planning 7, no. 4 (October 1994): 171–180. https://www.financialplanningassociation.org/sites/default/files/2021-04/MAR04%20Determining%20Withdrawal%20Rates%20Using%20Historical%20Data.pdf. SAFEMAX of 4.15% with 50% S&P 500 / 50% intermediate-term government bonds; the 1966 cohort survived 30 years. ↩ ↩² ↩³

4. Cooley, Philip L., Carl M. Hubbard, and Daniel T. Walz. "Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable." AAII Journal 20, no. 2 (February 1998): 16–21. https://www.aaii.com/files/pdf/6794_retirement-savings-choosing-a-withdrawal-rate-that-is-sustainable.pdf. The original Trinity Study, with updates published in 2009 and 2011. ↩ ↩²

5. Damodaran, Aswath. "Historical Returns on Stocks, Bonds and Bills: 1928–2024." NYU Stern, accessed April 2026. https://pages.stern.nyu.edu/~adamodar/New_Home_Page/datafile/histretSP.html. S&P 500 calendar-year total returns and 10-year Treasury bond returns by year. ↩ ↩² ↩³ ↩⁴

6. Officialdata.org. "S&P 500 Returns since 1929." https://www.officialdata.org/us/stocks/s-p-500/1929. Real, dividend-reinvested S&P composite recovered to its August 1929 level by November 1936 (~7 years from peak, ~4.5 years from the June 1932 trough). The often-cited "25-year recovery" refers to the Dow Jones Industrial Average price index, which is not a comparable measure. ↩ ↩²

7. Benz, Christine. "The Bucket Approach to Retirement Allocation." Morningstar, multiple updates. https://www.morningstar.com/retirement/bucket-approach-retirement-allocation. The standard 1–2 year cash, 3–8 year bond, equity-residual framework derived from Harold Evensky's earlier work. ↩ ↩²

8. Guyton, Jonathan T., and William J. Klinger. "Decision Rules and Maximum Initial Withdrawal Rates." Journal of Financial Planning 19, no. 3 (March 2006): 48–58. https://www.financialplanningassociation.org/sites/default/files/2021-11/2006%20-%20Guyton%20and%20Klinger%20-%20Decision%20Rules%20and%20SWR%20(1).PDF. Capital-preservation and prosperity rules support 5.2–5.6% initial withdrawals at the 99% confidence level over 40 years. ↩

9. A 65-year-old with $35,000/year of inflation-adjusted Social Security, valued as a real annuity at a ~3% real discount rate over a 22-year unisex life expectancy, has a present value of roughly $570,000–$700,000. See Social Security Administration period life tables, https://www.ssa.gov/oact/STATS/table4c6.html. ↩

10. Reichenstein, William. "After-Tax Asset Allocation." Financial Analysts Journal 62, no. 4 (2006): 14–19. The standard reference for the magnitude of the location-vs-allocation effect; Vanguard's later research finds 20–60 bps for typical investors. ↩

11. Kitces, Michael. "Why a 60/40 Portfolio Is No Longer Good Enough." Nerd's Eye View, multiple posts on bucketing and sequence risk. https://www.kitces.com/. Kitces's own writing repeatedly notes that the bucket approach's primary benefit is behavioral discipline, not statistical superiority. ↩