# The bucket strategy: the promise, and the case against it No retirement strategy gets sold harder than this one. Cut your wealth into three buckets. The short-term bucket holds one to two years of spending in cash. The middle bucket holds three to eight years in bonds. The last one leaves the rest in stocks. You live off the first, you refill it from the others, and you let the third compound without ever having to touch it at the wrong moment. The picture is so vivid that it has won over the personal finance press, much of the advice industry (Christine Benz, Morningstar's voice on retirement, built her brand on it) and the imagination of savers. And yet the research that has put it to the test comes back with an uncomfortable verdict (ERN, Parts 48 and 55; Kitces; Thaler's academic work on mental accounting). Simulated properly, a bucket strategy does **not** beat the equivalent portfolio run as one total return account with rebalancing. It does the same, or slightly worse. Because it is an allocation in disguise, plus a rule for the flows between the buckets. And that rule, depending on the version, turns out to be either rebalancing under another name or market timing that will not say its name. This chapter runs the full trial, and a fair one. It walks through the machinery and its variants, then takes the case against them exhibit by exhibit. It looks at what **survives** the criticism, which is the psychological framing, and that part is genuinely valuable. Then it lays out the clean version, the one that keeps everything buckets give you and drops what they hide. ::: cle The verdict in one sentence A bucket portfolio (2 years of cash, 6 years of bonds, the rest in stocks) is a ~7/21/72 portfolio with a cash buffer. Nothing more, nothing less. Everything it earns will come from that allocation and from the rules that move money between the buckets. And those rules, once written down in black and white, turn out to be either plain rebalancing (in which case say so and simplify) or a decision made case by case (in which case it is timing, with the track record we know). Buckets are an excellent **language** and a mediocre **machine**. ::: ## The promise, and where it comes from The idea goes back to Harold Evensky in the 1980s. He ran two sleeves, five years of spending in short bonds and the rest in growth assets, to free his clients from the fear of a crash. The modern three-bucket version, popularized by Christine Benz at Morningstar, adds the middle bucket and a set of refill rules. The promise has three parts. **Never** sell stocks in a hole, since bucket 1 covers the bad years (the anti-sequence argument, [[sequence-of-returns]]). Give every dollar a **horizon**, the 2027 money in cash and the 2045 money in stocks, an alignment of horizon and risk that feels flawless. And make the plan **tellable**, because everyone understands their buckets while nobody understands "a 65/35 rebalanced to bands, funding the withdrawal by selling the overweight". Give the promise its due right away. All three ideas are sound, and this book defends every one of them under its proper name: a cash buffer with written rules ([[cash-buffer]]), liability matching for dated needs ([[bond-ladders]]), and a plan the household can follow ([[couples-and-family]]). The problem with buckets is not the ingredients. It is what the packaging hides. ## The case against, exhibit by exhibit **Exhibit 1: buckets are an allocation, so judge the allocation.** Two years of spending in one bucket, six in the next, the rest in stocks: at a 3.5% withdrawal rate that comes to ~7/21/72% of your wealth. It is one point on the allocation plateau ([[stock-bond-allocation]]), no better and no worse than the total return portfolio sitting next to it at the same weights. The head-to-head simulations (ERN Part 48) confirm it. At **equal** allocation, buckets and a plain portfolio produce the same distribution of outcomes, give or take the flow rules. All the magic people feel in buckets ("my stocks have time") is the real magic of the allocation underneath. You can have it without the buckets. ::: figure buckets-allocation The canonical recipe, two years of spending in cash and six years in bonds, gives 5/15/80 at a 2.5% withdrawal rate and 10/30/60 at 5%. The equity share is never chosen: it falls out of the withdrawal rate. Pure arithmetic, no market data. ::: **Exhibit 2: the mechanical refill is rebalancing that does not know its own name.** The first flavor of flow rule is the systematic waterfall. Every year bucket 3 refills bucket 2, and bucket 2 refills bucket 1. Write the accounting down. Selling stocks every year to refill the buckets downstream means drawing on stocks every year, down years included. The founding promise, never sell at the bottom, is broken by the rule itself, with one extra pipe in the way. The variants that suspend the refill in a down year only refill in the good years. Formalize them and they collapse into a single rule: sell the overweight when stocks have gone up, sit tight when they have gone down. That is, line for line, the rebalancing draw of a plain portfolio ([[fixed-inflation-adjusted-withdrawal]]). Good. But then buckets add **nothing**. They rename. **Exhibit 3: the discretionary refill is market timing, with the track record that goes with it.** The second flavor, and the common one in practice, puts the refill off ("we will refill once the market feels better"). Judging it as you go is market timing, whole and entire, and it fails in both directions. In a **long** decline, buckets 1 and 2 run dry and you sell stocks at the worst moment anyway. A crossing, the years spent inside a bear market, can run past ten years when the buckets cover six to eight, and 1966 to 1981 empties them ([[market-regimes]]). In a **long** rise, you never refill ("it keeps going up, why sell?") and the equity share drifts toward 85 to 90%. The bucket retiree of 2007 was, in fact, **more** exposed than the plan said, at the worst possible time. Discretion with no rule delivers the opposite of the promise in both regimes. **Exhibit 4: bucket 1 costs what a buffer costs, without a buffer's rules.** Those two years of cash carry the same opportunity cost as any cash buffer ([[cash-buffer]]). The bucket frame presents them as "the short-term money" and usually leaves out the two rules that make a buffer work: a written trigger for drawing on it, and a refill at the highs. You pay the price and get no instructions. ::: science What happens when it is put to the test ERN's head-to-head simulations run buckets against equivalent portfolios over every historical cohort (Part 48, extended by the discussion in Part 55). The conclusion is clean. No bucket setup beats its total return twin. The academic test exists too, and it says the same thing: Javier Estrada (2019) runs the strategy against its static twins over the historical cohorts and finds the same ranking, at best a tie, most often slightly behind, the idle cash paying its toll over time. The discretionary versions do worse, through allocation drift. The mechanical ones do the same, with more moving parts. Kitces gets to the same place through the accounting of the flows, and his phrase stuck: buckets are an asset allocation mirage, better at explaining a plan to a client than at running one. The academic work on mental accounting (Thaler, Shefrin) explains **why** the strategy appeals anyway. Separate mental accounts lower anxiety and cut panic errors. That benefit is real, behavioral, and independent of the machinery. So the literature does not say buckets ruin a plan. It says buckets add nothing mechanically and can help psychologically. That precise verdict is the one you can act on. ::: ## What survives: buckets as a language Take the verdict seriously in both directions. If the benefit of buckets is behavioral, it is **real** for anyone who needs it. The anti-panic effect of mental accounts is documented. A household that **sticks** with a plan written in buckets, where it would have capitulated on a 65/35, has won the thing that matters ([[the-psychology-of-spending]], [[cash-buffer]]). The practical conclusion is simple. **Keep the buckets as the presentation, and insist that under the hood the plan is an explicit portfolio.** Three requirements, and one judge. - **Allocation first.** Pick your point on the plateau, and what goes in each sleeve, on the real criteria ([[stock-bond-allocation]], [[defensive-assets]]). **Then** tell the story in buckets if it helps. The cash buffer becomes "bucket 1", the dated bond sleeve "bucket 2", the engine "bucket 3". - **Written flow rules, not moods.** The trigger and the refill for the buffer ([[refilling-the-buffer]]), and selling the overweight to fund the rest. The buckets then have the plumbing of a plain portfolio, and all of their poetry. - **Bucket 2 under contract when the floor demands it.** If the idea that "the 2028 to 2035 money is safe" matters to you, make it **true**. Build a ladder of bonds held to maturity, linkers for preference, matched year by year ([[bond-ladders]], [[inflation-linked-bonds]]). That is bucket 2 with the ambiguity taken out, a contract in place of an intention. - **And simulation as the judge.** Once your version of the buckets is written as an allocation, a buffer and a set of flow rules, all of it can be simulated. Two outputs settle the question. The income delivered year after year says what the setup makes you live on, and the same plan run without the buffer says what bucket 1 really costs or earns. If your version beats your current plan, it is the allocation doing the beating. At least you will know why. ::: exemple The same buckets, first vague, then clean The brochure version. "$1.5M, of which $100k in cash (bucket 1), $350k in bond funds (bucket 2) and $1.05M in stocks (bucket 3); we refill when the markets allow." Ten years later, depending on the mood behind each refill, the real allocation could have landed anywhere between 60 and 88% stocks. There was no plan. Now the clean version, with the **same** numbers. Target allocation written down: 7% cash, 23% bonds, 70% stocks, bands of plus or minus 5 points. A $100k buffer in a money market fund, trigger at an 18% real drawdown, refill at the highs. The bond sleeve in two pieces, $150k in a ladder of linkers maturing 2028 to 2033 (the floor for the next six years, matched) and $200k in a rebalanced 5 to 8 year fund. Monthly draw from the overweight. Tellable in buckets at a family dinner, testable in a simulator on Monday, runnable by a spouse. The buckets have become a plan. ::: ## The essentials - Buckets are an allocation (~7/21/72) plus a set of flow rules. Judge the allocation on the real criteria, and insist the rules be written down. Mechanical, they turn out to be plain rebalancing renamed. Discretionary, they are market timing that drifts: too much equity in a bull market, empty buckets in a long decline. - The tests (ERN 48 and 55, Kitces) agree. No mechanical benefit against the total return twin. The real benefit is **behavioral**, the anti-panic effect of mental accounts, valuable to anyone who needs it, at the right price. - The clean version rests on three pieces: an explicit allocation, a buffer with rules ([[cash-buffer]], [[refilling-the-buffer]]), and a ladder under contract for the floor years ([[bond-ladders]]). Tell it in buckets if it helps. Poetry out front, plumbing underneath. - The one trap to avoid at all costs is the discretionary refill. It turns a harmless framing into unmanaged allocation drift, the exact opposite of a plan. - Test your version in a simulator: the allocation, the buffer and the rules all fit inside one. If "the buckets" beat your plan, it is their allocation doing the beating. Take the allocation, keep the story. --- ## Going further - Early Retirement Now, Parts 48 and 55: buckets simulated against their twins, cohort by cohort ([[the-ern-series]]). - Kitces, "Managing Sequence Of Return Risk With Bucket Strategies Vs A Total Return Rebalancing Approach": the accounting of the flows, laid out plainly. - Christine Benz (Morningstar), "The Bucket Approach to Retirement Allocation": the best version of the other side, so you can judge it for yourself. - Thaler, "Mental Accounting Matters": why buckets soothe, the real root of the debate. - Javier Estrada, "The Bucket Approach for Retirement: A Suboptimal Behavioral Trick?" (2019): the academic test of the verdict, cohort by cohort. - In this book: [[cash-buffer]] (bucket 1 with its rules), [[bond-ladders]] (bucket 2 under contract), [[refilling-the-buffer]] (the flows), [[stock-bond-allocation]] (what actually decides).