Retirement Bucket Strategy: Cash, Rebalancing, and When to Refill

This analysis draws on Michael Kitces’s work at Kitces.com on bucket strategies and rebalancing, including his analyses of buckets as an asset allocation, of cash reserves and rebalancing in retirement, and of reporting a total-return portfolio in buckets, and on the rising equity glidepath research he and Wade Pfau published. We’d recommend reading it in full.

What a Bucket Strategy Is

Every bucket strategy starts from the same idea: money needed soon should not be exposed to the market’s short-term swings, and money not needed for a decade or more can be. The classic version, often called the three-bucket strategy, splits the money three ways.

Bucket one: cash. One to several years of withdrawals, held in a bank account, a money market fund, or short-term Treasury bills. This is the bucket a retiree actually spends from.

Bucket two: bonds. The following several years, usually in high-quality short- and intermediate-term bonds. It earns more than cash, swings less than stocks, and exists to refill the first bucket.

Bucket three: stocks. Everything else, invested for the long run. It is expected to be volatile, and the structure is built so that it rarely has to be sold in a bad year.

The modern form is often credited to Harold Evensky, a financial planner whose firm began keeping a cash flow reserve beside a long-term portfolio in 1985, by its own account. His version had two buckets, the reserve and the portfolio. Planners call the broader approach time segmentation, and it has spread in many versions: two buckets or five, buckets measured in years of spending or in dollars, and a long list of rules for moving money from one bucket to the next. The usual alternative is a systematic withdrawal approach, which draws a set amount each year from one diversified, regularly rebalanced portfolio. In a 2011 Financial Planning Association study, about three quarters of advisors said they frequently or always used systematic withdrawals and 38% said they frequently or always used time segmentation, so some were using both.

There is a second kind of bucket that has little to do with markets: a pool of money set aside for one known purpose. A down payment fund is a bucket in this sense, and so is an account labeled for a trip planned for next year. We use both kinds, and they solve different problems. Time buckets manage when money comes out of a long-term portfolio. Purpose buckets keep known, irregular expenses from landing on that portfolio at a bad moment. Most of this article is about the first kind; a later section covers the second.

Why Buckets Work on People

The case for buckets is mostly behavioral, and it is a strong case.

A paycheck again. For forty years, money arrived on a schedule. Retirement replaces that rhythm with a balance that moves every day. Kitces has told the story of a prospective client, early in his career, who sat through 20 minutes on investment philosophy and then asked again, exasperated, how the money would actually get into his checking account so he could pay his bills. A cash bucket that sends the same deposit to checking every month answers that question, and it separates two questions that feel tangled together: what can we spend this month, and what did the market do this month. The first can have a steady answer even when the second does not.

Mental accounting, used on purpose. People sort money into mental accounts and treat it differently depending on the label, an idea the economist Richard Thaler developed into a body of research. In most of financial planning, mental accounting is something to correct: a dollar is a dollar whatever account it sits in. Buckets put it to work. Money labeled “the next few years” can stay calm, and money labeled “the long run” can be allowed to swing.

Cash on hand feels like security. In a study of 585 bank customers in the United Kingdom, Peter Ruberton, Joe Gladstone, and Sonja Lyubomirsky found that the money people had readily available in checking and savings was linked to higher life satisfaction through a stronger sense of financial well-being, apart from their income, investments, and debts. A reserve in cash is not just arithmetic. People feel it.

It protects against the most expensive mistake. The worst outcome in a bear market is rarely a withdrawal taken at a poor price. It is a retiree who abandons the plan and sells the long-term portfolio near the bottom. When the next few years of spending are already in cash, that decision becomes much easier to avoid. Kitces, writing about research that found cash reserves cost money, made the point himself: a reserve that keeps a household invested beats panic selling, or giving up on stocks altogether. We cover this side of the reserve in our article on sequence-of-returns risk.

Not everyone needs it. Preferences differ, and they seem to last. In a national survey of 2,863 adults aged 50 to 80, sponsored by an annuity industry group, Alex Murguía and Wade Pfau found that about 17% preferred time segmentation for their essential expenses: safety for near-term spending without locking money into a contract. The rest split among drawing from an invested portfolio (about a third), a floor of guaranteed income from simple annuities (about a third), and guarantees wrapped around an invested portfolio (15%), and the preferences held steady across ages and through the 2020 crash. Some households find buckets deeply reassuring; others are just as comfortable seeing one portfolio. The structure should fit the people living with it.

A Bucket Strategy Is an Asset Allocation

Here is the part that changes how buckets should be run. Add the buckets together and they describe one portfolio with one mix of stocks, bonds, and cash.

Kitces worked through the arithmetic in 2012. Sized the way bucket strategies usually size them, with cash and short-term bonds for the first several years, more conservative holdings for the middle years, and stocks for spending 15 or more years away, the buckets add up to the same balanced allocations a systematic withdrawal approach would choose anyway: down near 30% or 40% in stocks for a conservative household and up to 60% or 70% for a more growth-minded one. The buckets change how a portfolio looks far more than what it holds.

That cuts both ways. If the total allocation is sensible, the behavioral benefit of buckets costs very little. But buckets can also hide what the allocation is doing, because the total mix stops being anyone’s explicit decision. It becomes the leftover of whatever rules move money between the buckets.

Kitces’s answer, in a 2013 piece, was to separate investing from reporting: manage one total-return portfolio (a single diversified mix, drawn on for spending whether the money comes from interest, dividends, or selling shares), and show it to the household in buckets. Take a couple with a $2,000,000 household portfolio, 60% in stocks. Managed as one portfolio, it holds $1,200,000 in stocks, $560,000 in bonds, and $240,000 in cash, and it is rebalanced to those targets. Shown as buckets, the $240,000 of cash is the paycheck bucket, the $560,000 of bonds is the next bucket in line, and the $1,200,000 of stocks is the long-run bucket. The holdings are identical. Only the labels changed.

That is how we use buckets. The investing is one household portfolio, managed and rebalanced to the mix the plan calls for. The buckets are how that portfolio is organized, explained, and paid out.

Rebalancing: The Engine Under the Buckets

If buckets are the picture, rebalancing is the machinery. It deserves more credit than it usually gets, because it quietly does most of what buckets advertise.

What a rebalance does when money is coming out

A rebalance resets a portfolio to its target mix by selling what has grown past its target weight and buying what has fallen below it. In retirement, with withdrawals coming out at the same time, that has a useful side effect: the money for spending comes from whatever did best.

Take the couple above, with $1,200,000 in stocks, $560,000 in bonds, and $240,000 in the paycheck bucket. Over six months they draw $55,000 in monthly paychecks from the cash. Here is what topping up the cash and rebalancing to target does after six months in three different markets.

A bad six months for stocks. Stocks fall 25%, bonds gain 3%, and cash earns 1%. Stocks are worth $900,000, bonds $576,800, and cash $187,400 after the paychecks, for a total of $1,664,200. Resetting to 60% in stocks and a full $240,000 in cash means selling $151,120 of bonds, using $52,600 of it to refill the cash and the other $98,520 to buy stocks. No stocks are sold. The rebalance buys them, near the low.

A good six months for stocks. Stocks gain 15% and bonds lose 1%. Stocks are worth $1,380,000, bonds $554,400, and cash $187,400, for $2,121,800. Now the rebalance sells $106,920 of stocks: $52,600 refills the cash and $54,320 goes to bonds. Gains come off the table and become the next year’s paychecks.

A bad six months for both. Stocks fall 20% and bonds fall 10%, close to what the first half of 2022 delivered. Stocks are worth $960,000, bonds $504,000, and cash $187,400, for $1,651,400. The paychecks came from cash, which did not fall. The rebalance sells $83,440 of bonds, the asset that fell less, refills the cash with $52,600 of it, and buys $30,840 of stocks.

In none of the three did the couple sell stocks after a fall. That is the heart of Kitces’s 2012 argument that a cash reserve is not needed to avoid selling stocks in a decline: the arithmetic of rebalancing already sees to it. In 2014 he went further. He tested a retiree starting in 1966 with three buckets (60% in stocks, 30% in bonds, and 10% in Treasury bills) and rules that spent from stocks only after stocks rose, from bonds when stocks fell, and from bills when both fell, as they did in 1969. With a yearly rebalance on top, the bucket rules produced exactly the same results, year by year, as taking each withdrawal across all three and rebalancing. Without the rebalance, the same rules did far worse: nothing ever bought stocks after they fell, and inflation-adjusted withdrawals gradually drained the stock bucket.

How often to rebalance

There are two basic ways to decide when. Calendar rebalancing resets the portfolio on a schedule, such as once or twice a year; threshold rebalancing acts when the mix drifts past a set band around its target. A calendar is simple but can sit through a large move waiting for the date. A band reacts when the move happens, at the cost of watching the portfolio more often.

The research favors checking often and trading rarely, though the gains are modest either way. In a 2008 study in the Journal of Financial Planning, Gobind Daryanani tested a 60/40 portfolio spread across five asset classes from 1992 through 2004. Checking about every two weeks, but trading only the holdings that had drifted past a band of 20% of their target weight (so a holding meant to be 20% of the portfolio would trade only outside 16% to 24%), added about half a percentage point a year on a risk-adjusted basis over never rebalancing, and about a fifth of a point more than checking only once a year with the same bands. For a retiree, though, the bigger benefit of rebalancing is not the extra return. It is keeping the risk where the plan put it, and making sure the paychecks come from whatever did best.

We lean toward rebalancing on drift: whenever the mix moves past a set band around its target, whatever the date, so a sharp decline or rally does not have to wait months for a review. The paycheck bucket runs on the calendar instead. Its target is updated and the cash topped up twice a year, at the plan reviews.

One portfolio across several accounts

Most retired households own several accounts: a taxable brokerage account, IRAs, Roth IRAs, sometimes an old 401(k). We lean toward managing the mix across all of them as one household portfolio, with each kind of investment placed where its tax treatment fits best, as our article on asset location explains. That changes how rebalancing works in practice. Trades inside an IRA or a Roth create no taxable gains, so much of the rebalancing can happen there, while the taxable account is used when it is cheap to do so: when new cash or dividends can be directed to whatever is underweight, in a year with room in a low capital gains bracket, or after a decline that offers losses worth harvesting.

When stocks and bonds fall together

The rebalancing arithmetic works best when one side of the portfolio holds up while the other falls. Bonds usually do that. Not always. In 1969 stocks and bonds both lost money. In 2022 the S&P 500 lost about 18%, a broad index of high-quality U.S. bonds about 13%, and long-term Treasury bonds more than a quarter of their value, while three-month Treasury bills earned a small positive return. Correlation measures how two investments move together, on a scale from negative 1 (always opposite) to positive 1 (always together). Edward McQuarrie’s long history of stock and bond returns, which Larry Swedroe summarized at Kitces.com in 2024, found that the correlation between stocks and bonds has ranged from about negative 0.7 to about positive 0.9 over 20-year periods. Bonds have sometimes moved with stocks for decades at a time.

That is one reason the paycheck bucket is held in cash rather than in bonds. Its one job is to hold its value until it is spent, and a year when stocks and bonds fall together is when that job matters most.

What the Research Says About Refill Rules

If rebalancing does most of the work, the next question is what happens when bucket strategies run on their own refill rules instead. The research is consistent: the rules matter more than the buckets.

Cash held on top of the portfolio has cost money. Walter Woerheide and David Nanigian of The American College tested “buffer zone” strategies in research published in the Journal of Financial Planning in May 2012: one to four years of withdrawals held in cash beside portfolios ranging from no stocks to all stocks, run through actual market history from 1926 to 2009, spending from cash when the portfolio was down and refilling it when the portfolio was up. Failure rates were generally higher with the buffer, most of all for long retirements and high withdrawal rates. In the paper’s 560 comparisons, the portfolio without a buffer did better 226 times and the buffer did better 35 times. The reason is simple: at a 4% withdrawal rate, four years of spending is close to a sixth of the portfolio sitting in cash.

Refilling after the first bounce can mean selling near the bottom. Kitces pointed out a trap in rules like these. Picture a three-year reserve and a three-year bear market, with returns of down 5%, down 10%, and down 20%, followed by a gain of 10%. A rule that refills after the first up year sells three years of spending at once, barely off the bottom, before the recovery has really begun. Selling a little each year along the way would have gotten better prices.

Rules that wait for good years have trailed a steady mix. Javier Estrada of IESE Business School tested three common bucket rules, each holding two years of withdrawals in Treasury bills and never rebalancing, across 21 countries from 1900 through 2014, with 30-year retirements and inflation-adjusted withdrawals that started at 4%. He compared them with simple portfolios of stocks and Treasury bills rebalanced every year. In the U.S. data, the steady 60% stock portfolio never ran out of money and the steady 50% and 70% portfolios failed in 1.2% of the periods, while the bucket rules failed in 3.5% to 4.7%. Across all 21 countries, the steady 70% portfolio failed less often than every bucket rule; the best bucket rule failed slightly less often than a steady 50% portfolio but trailed it on his other measures. His conclusion, published in the Journal of Investing in 2019, was that simple static strategies, which by definition rebalance, clearly outperformed the bucket strategies.

A small reserve can pay part of its way, mostly through costs and taxes. Shaun Pfeiffer, John Salter, and Harold Evensky tested a one-year cash reserve in a 2013 simulation study in the Journal of Financial Planning. The reserve was refilled during rebalancing when it fell below two months of spending, and left alone after a year in which stocks and bonds had both lost money. With no costs or taxes counted, it lasted about as well as drawing from the portfolio every month: 76.0% of simulated 30-year retirements succeeded at a 4% withdrawal rate, against 76.9% without the reserve. With trading costs counted, the reserve came out ahead, 75.7% against 70.8%, and it led in taxable accounts too. Most of that edge came from trading less often at $30 a trade on a $200,000 portfolio, and the authors noted it shrinks as trading costs fall relative to the size of the portfolio. With most trades now free of commissions, the case for a reserve rests more on taxes and behavior than on trading costs.

What survives the research. None of this says buckets fail. It says that cash held as extra cash, on top of the plan’s allocation, costs return; that refill rules keyed to the market’s last move behave like accidental market timing; and that a rebalanced portfolio has matched or beaten those rules. Kitces’s conclusion was not to abandon buckets but to keep them from distorting the allocation: explain the portfolio in buckets, and run it on rebalancing. That is how we run ours. Our cash sits inside the allocation, so it is not extra cash on top, and it is refilled on a schedule from whatever sits above its target weight, not by a rule about last year’s returns.

There is a deeper problem with any rule keyed to the market’s recent move: it asks for a judgment no one can make in real time. In the middle of a decline, nobody knows whether it will be short or long. That only becomes clear looking back. A schedule does not need to know.

The Rising Glidepath Question

There is one way bucket strategies have looked better than rebalancing, and it explains a good deal of the debate.

Suppose a retiree spends the cash first, then the bonds, and never refills either. Each year the share of the portfolio in stocks rises, because the safe money is being spent and the stocks are left to grow. That path has a name, a rising equity glidepath, and it runs against the old advice to own fewer stocks with age.

In 2013, Pfau and Kitces reported that much of the benefit researchers had credited to bucket strategies, and to buying an annuity with part of a portfolio, really came from this rising stock share. In that draft research, as Kitces reported it (the work was later published in the Journal of Financial Planning), a portfolio that started at 30% in stocks and rose to 70% had a slightly higher simulated success rate (95.1%) than one held at 60% throughout (93.2%), while averaging only 50% in stocks. The logic runs through sequence risk. If the first half of retirement, roughly 15 years, is bad, a rising path adds to stocks at lower prices, so the household owns more of them when the recovery comes. If the first half is good, the household is far enough ahead that the later stock exposure matters less.

To see what each method does on its own, we ran the three textbook refill methods from this research through the same stretch of market history: rebalancing to a fixed mix, refilling only after up years, and never refilling. They are generic methods, not a model of how we or anyone else manages a client’s money. The test follows a retiree who started in January 2000 with $1 million, 60% in stocks, and three years of withdrawals in cash, taking 4% of the starting value each year, raised with inflation and paid monthly.

The refill rule, not the retiree, decided the stock share. Rebalanced to its target every January and July, the portfolio stayed between about 51% and 66% in stocks and ended at 64% in June 2023. Refilled each January from stocks after a year stocks rose, and from bonds otherwise, it drifted the other way, falling as low as about 29% in 2022 and ending at 35%, because stocks were sold after every good year and bonds only after bad ones; Kitces described this effect in 2013, noting that holding buckets at a fixed number of years of spending can make a portfolio steadily more conservative. Never refilled, it was all in stocks (100%) by April 2013 and stayed there.

The outcomes followed the stock share. After inflation, in this hypothetical test, the never-refilled portfolio was worth about $935,000 in June 2023, the rebalanced one about $442,000, and the one refilled after up years about $284,000, each after 23 and a half years of withdrawals and before taxes and fees. A retiree who started in January 1973 saw a similar pattern: after 20 years, about $581,000 never refilled against about $398,000 rebalanced.

It is worth being precise about why. The never-refilled portfolio never bought a share of stock. Its stock share rose because the cash and bonds were spent and the stocks were never trimmed after rallies, so more of them were still there for each recovery. In the 2000 test that meant being entirely in stocks from 2013 through 2023, a strong stretch for stocks, so the end date drives much of the gap.

So why don’t we let the stock share drift up? Three reasons.

It is a decision made by accident. Spending order, not the household, would set the allocation. In the 2000 test, a retiree who was 65 at the start owned nothing but stocks from 78 on. For a household whose plan still depends on that money, a bad market late in retirement with nothing outside stocks is a risk nobody chose to take. In this hypothetical test the extra risk showed up, if less than one might expect: in the 2008 crisis, the never-refilled portfolio, by then about three quarters stocks, fell about 42% after inflation from its peak, against about 36% for the rebalanced one. In 2022, when bonds fell too, both fell about a quarter.

The evidence for it is narrower than the simulations suggest. When Kitces and Pfau tested glidepaths against actual market history in research published in 2015, a rising path spread over the full 30 years was no better than a steady mix. A faster path, rising from 30% to 60% in stocks over 15 years, helped slightly when stocks started out expensive, matched a steady mix when they started near fair value, and trailed it when they started cheap. They measured expense with Robert Shiller’s cyclically adjusted price-to-earnings ratio, which compares stock prices with ten years of inflation-adjusted earnings. January 2000 was close to its highest reading on record.

When a rising path fits, it can be chosen on purpose. Kitces noted in 2014 that a rebalanced portfolio is a better way to follow a rising glidepath than letting buckets drain, because the rebalance keeps buying stocks after a decline. If a household’s plan is better served by owning more stocks later, the target itself can rise, and the rebalance follows it.

We lean toward setting the allocation around retirement case by case, with this research in the conversation. Rebalancing on drift and topping up the cash on schedule keeps the mix where the plan put it, so the risk the household chose and the risk its plan can carry stay in step. When a rising path fits a household, we would rather choose it on purpose and rebalance toward it than leave it to the order of spending.

Kitces has described the mirror image before retirement, a “bond tent”: building up bonds over the last years of work, when a decline would do the most damage to a portfolio at its largest, then spending that extra down in the first years of retirement. Whether a household builds one is part of the same case-by-case allocation decision.

Buckets Don’t Have to Make a Portfolio Conservative

A common worry about holding years of cash is that it makes the whole portfolio timid. It doesn’t have to, and the research on how much growth a retirement needs explains why the mix itself still deserves the most care.

The cash replaces bonds, not stocks. Because the paycheck bucket sits inside the allocation, a household with a 60% stock target still owns 60% in stocks. In our data from 1934 through June 2023, stocks earned about 7.0% a year after inflation and a constant 10-year Treasury bond about 1.2%. The paycheck bucket costs the gap between cash and bonds, about 0.18 points a year for the couple above, not the far wider gap between cash and stocks.

Long retirements need real growth. For a man and a woman who are both 65, there are about even odds that one of them reaches 90, as our article on how long a retirement plan should last explains. The safe withdrawal research, as Kitces summarized it, found the best stock share for a 30-year retirement at around 50% to 60%, rising to about 60% to 65% for horizons of 40 years or more, though Pfau’s work pointed to more conservative mixes. Bengen’s original research made a related point: in 1994 he advised holding as close to 75% in stocks as possible and never less than 50%, because portfolios with less in stocks ran out sooner, as our article on the 4% rule describes.

A plan that can adjust can carry more risk. A fixed withdrawal has to survive the worst sequence with no help, so it pushes toward caution. A plan with guardrails does not have to: in a bad stretch, spending comes down by a small, planned step instead of the portfolio running dry, and that flexibility is capacity to own stocks. Kitces has made a related point about judging retirement strategies: how large a shortfall would be, and how small an adjustment fixes it, matters as much as how often one happens.

The decade around the retirement date deserves the most care. Kitces has called the years just before and just after retiring a danger zone, because the portfolio is at its largest and has the most to lose: a 20% decline at a 4% withdrawal rate erases five years of spending. As the previous section showed, his research with Pfau found that starting more conservatively and letting the stock share rise helped slightly when stocks started out expensive. The paycheck bucket and the guardrails carry part of that early risk. The mix at the retirement date carries the rest. We lean toward treating current market valuations as part of that conversation, not as a rule for sizing the mix.

Where each household lands comes out of conversation, not a model portfolio. We weigh how much risk the plan can carry, measured by how far spending would need to adjust in a bad stretch, and how much the household is comfortable living with as balances move. Kitces has argued those two should be weighed separately, with the lower one acting as the limit. A household with little room to adjust, or little appetite for swings, is usually better served by a more conservative mix, and some households choose one for their own reasons, which is a fine answer when it is a decision rather than a default. We lean toward a mix that keeps enough growth for a long retirement, set with each household, with the retirement-date research in view.

How Long Declines Last, and What Carries a Household Through

The usual advice for buckets is to hold enough cash and bonds to ride out a bear market without selling stocks. So how long have bear markets lasted?

We measured every decline of 20% or more in the S&P composite since 1871 (today’s S&P 500 and its predecessors), with dividends reinvested and after inflation, from the market’s prior high to the month it regained that high. Through mid-2023, when the data end, fourteen declines had fully recovered. Five recovered within three years and nine within five. The median took about four years from peak to new high. Two recent drops are not on the chart for a reason. Measured the way the chart measures (monthly average prices, after inflation, with dividends counted), the brief 2020 decline came to about 19%, just short of the 20% line, and the 2022 decline was still under way when the data end.

The nine quicker recoveries took from 1.4 to 4.5 years: the declines that began in 1961 (1.4 years), 1876 (1.8), 1987 (2.0), 1902 (2.4), 1906 (2.7), 1892 (3.2), 1968 (3.9), 1911 (4.3), and 1946 (4.5). The other five took far longer: 1916 about 7.8 years, 1929 about 7.2, 1937 about 8.2, 1973 about 12.0, and 2000 about 12.8. The longest two were not single crashes. The 1973 decline was followed by a decade of high inflation that kept real returns below the old high until 1985, and the 2000 decline ran straight into the 2008 crisis before stocks regained the 2000 high in 2013.

For a classic bucket that is spent down and refilled only after a recovery, the lesson is sobering. A three-year reserve would have outlasted 5 of the 14 declines and a five-year reserve 9. No reasonable reserve covers the longest. A household that wanted twelve years of withdrawals in cash would give up so much return that the cost of the cushion would outweigh the risk it protects against. Kitces has made the same point: the real danger in retirement is a long stretch of mediocre returns, and even three years of cash cannot outlast a bad five to ten years.

That is not how our paycheck bucket is meant to work. It is not built to outlast a decline on its own. It is topped back up at every review, even in a long decline, because no one can tell in real time how long a decline will last. What carries a household through a long decline is the conservative side of the portfolio as a whole, with the rebalance buying stocks along the way. For the couple above, that is $800,000 of cash and bonds, about seven years of their $110,000 of withdrawals. If a decline runs longer than that kind of cushion is comfortable with, the guardrails decide whether spending should change.

Sizing the Paycheck Bucket

How much cash is enough depends less on the market than on the household. We lean toward sizing the reserve on the portfolio’s job, household by household, rather than on the whole budget.

Every plan we build has two layers of spending, using the core and adaptive labels from Michael Kitces’s 2019 work at Kitces.com, which built on Jonathan Guyton’s approach. Core spending is what the household intends to keep doing in any market. It is more than groceries and shelter: for many families, in Texas and beyond, an annual trip with the children and grandchildren belongs in it, and so does their regular giving. Adaptive spending is the part that can grow in good years and shrink or wait in hard ones: the size of that trip, the larger gifts, the bigger projects. Our article on retirement spending guardrails explains how the two layers work in a plan.

From core spending, subtract the income that arrives regardless of markets, such as Social Security and any pension. What is left is the core gap: the part of core spending the portfolio has to fund. Then add a buffer from the adaptive layer.

A worked example. Consider the couple above. Their core spending is $100,000 a year, and Social Security covers $60,000 of it, so the portfolio has to fund a core gap of $40,000. On top of the core, they plan another $70,000 a year of adaptive spending: the bigger versions of the family trip, extra time away, larger gifts, and the projects they have waited years for. The portfolio pays out $110,000 a year in all.

A reserve sized on the core gap alone would be $120,000 for three years or $200,000 for five. A reserve sized on everything the portfolio pays, the core gap plus the full adaptive layer, would be $330,000 for three years or $550,000 for five. Our plans land between those. The core gap alone covers the essentials but leaves the household watching markets for the bigger trips and larger gifts; the full budget ties up so much in cash that the cost of the cushion starts to work against funding those same things later. The couple’s $240,000 is three years of the core gap plus an illustrative $40,000 a year of the adaptive layer. Measured against their full $110,000 of withdrawals, it is about two years. The buffer, and so the number, is set with each household.

All or nearly all in cash. In our plans the paycheck bucket sits in cash, such as a money market fund, Treasury bills, or a bank account, rather than in short-term bonds. The bucket has one job, holding its value until it is spent.

The cost, honestly. Cash earns less than bonds over time. In our data from 1934, when the Treasury bill series begins, through June 2023, three-month Treasury bills earned about 3.4% a year, slightly less than inflation of about 3.6%, while a constant 10-year Treasury bond earned about 4.8%, about 1.2 points a year ahead of inflation. Holding 12% of a portfolio in cash instead of bonds, as the couple does, cost about 0.18 points a year of the whole portfolio’s return in a hypothetical test on the same data, with twice-yearly rebalancing. Holding the same cash on top of a 60/40 allocation instead, so that it comes partly out of stocks, cost about 0.59 points a year, more than three times as much. The cost is measured at the start, though. In a long decline, a reserve held at a set dollar amount becomes a larger share of a smaller portfolio, so the cost grows in exactly the hard years. The target is revisited at each review, alongside the spending plan, so it moves with the household’s actual spending.

A starting range, not a rule. A household whose guaranteed income covers nearly all of its core spending may need a much smaller reserve, because the portfolio’s core job is small. A household that relies on the portfolio for most of its essentials, or that would lose sleep watching stocks fall, may want more. The number is chosen with the household, then revisited as the plan changes.

Purpose Buckets for Known, Irregular Expenses

Not every expense arrives as a monthly bill. Property taxes, insurance premiums paid once a year, a new roof, a replacement car, a milestone anniversary trip, a child’s wedding, help with a grandchild’s tuition: these are known, or close to known, but they arrive on their own schedule. Kitces has written about how often irregular expenses are left out of budgets, because people estimate their spending from the bills that come every month, and about two ways to plan for them: turn each one into a regular amount, so a car replaced every few years shows up as a monthly cost, or enter each in the plan in the year it is expected. We agree with both, and with his larger point that a budget missing its irregular expenses overstates what a household can save, and, in retirement, what it can spend and give.

Where that money waits matters too. We lean toward separate, labeled accounts for known expenses with a date attached: one for the roof, another for the next car, another for the trip. When a household plans to spend $20,000 on a trip to Italy next year, that money is set aside in cash, in an account that carries the trip’s name. It is not part of the paycheck bucket, and it is not part of the long-term portfolio’s normal planning. We do keep it in view when we set the portfolio’s mix, because it is part of what the household owns and of what it has committed to spend.

Why separate accounts rather than one big pool?

The big expense doesn’t raid the paycheck. A roof paid out of the paycheck bucket would leave the reserve short just when it might matter most. Paid from its own account, it changes nothing else in the plan.

The money is held for its date. Money due soon, such as a trip a year away, sits in cash, so the full amount is there when the bill arrives, whatever markets do in the meantime.

Labels change behavior, in a good way. This is mental accounting again, used on purpose. An account named for the Italy trip is harder to spend on something else, and it turns a vague hope into something the household can see itself preparing for. When a household hesitates to spend on something it has said it wants, we often step out of the money conversation and ask what it would love to do, as we describe in our article on why retirees underspend. A labeled account is where that answer starts to become real.

Watching it fill is part of the fun. A trip account that grows over the months before the trip gives a household something concrete to look forward to long before the plane leaves the ground.

Purpose buckets work for giving too: an account for a grandchild’s education gift, or for the family trip the grandparents pay for and take with everyone, the kind of gift we discuss in our article on giving while living. The same idea of naming money for a purpose shows up on a larger scale in the plan itself. For a household that plans to pay for care itself, a long-term care reserve is often its own named and modeled reserve, as our article on funding long-term care explains.

The Twice-Yearly Top-Up and the Monthly Paycheck

If refill rules are where bucket strategies succeed or fail, the refill deserves the most care. Ours has four parts.

The reserve lives inside the allocation. We count the paycheck bucket as part of the portfolio’s conservative side, not as an extra pile on top: a couple with a 60% stock target holds it within the other 40%, alongside its bonds. The total mix stays where the plan put it, so the buckets cannot drift into a rising glidepath, or a falling one, by accident, and the cost of the cash stays limited to the gap between cash and bonds.

Topped up twice a year, in every market. We set aside the cash, and update the target amount, twice a year, at the same reviews where the plan itself is updated. The top-up comes from whatever sits above its target weight at the time. After a good stretch for stocks, that usually means trimming stocks. After a decline, it usually means drawing from bonds, and a rebalance may buy stocks at the same time, as the three examples above showed. The top-up happens at every review, including in the middle of a decline, because the rule follows the targets and does not depend on guessing what markets will do next. Between reviews, the portfolio is rebalanced whenever its mix drifts past its bands, so the stock and bond mix does not wait for the calendar.

A paycheck every month. Most of our retired clients take a set monthly transfer from the cash account to checking, so they keep getting a paycheck in retirement. No holding has to be sold to pay a particular bill, and the reserve is refilled on a schedule, not in response to a bad week. Kitces has described the ways advisors turn a portfolio into paychecks: paying out only interest and dividends, topping up a cash balance with periodic rebalancing, or holding no cash at all and selling whatever is needed each month. Ours is the top-up approach, with a much larger reserve than the few months of cash he describes.

Taxes decide where it happens. Because the paycheck bucket is a slice of the household portfolio, it can sit in whichever account the year’s tax plan draws from: a taxable account in a year when realizing gains is cheap, an IRA in a year when a distribution has to come out anyway, or both. Where the refills happen is part of the same yearly withdrawal sequencing decision as everything else. For households taking required minimum distributions, the distribution itself can be paid into a cash account outside the IRA, which satisfies the requirement and refills the paycheck bucket at once.

Where Buckets End and Guardrails Begin

A cash reserve answers one question: how do we keep paying ourselves without selling stocks into a decline? It does not answer the larger one: how much can we afford to spend and give?

That job belongs to the spending plan. We lean toward a plan with guardrails set in advance: a starting level of spending and giving, and portfolio values at which the household would take a raise or a small, temporary cut, all shown in dollars before anything happens. In a short decline, the top-ups and the rebalancing carry the household through, and the guardrails usually are not reached. In a long decline, the top-ups keep the paychecks coming, the rebalancing keeps buying stocks, and the guardrails decide whether spending should change. If a cut does come, it is small and planned, made on lines agreed in advance rather than in a moment of fear. In a strong market, the same guardrails can signal a raise, and the paycheck bucket grows with the new budget at the next review.

The two work together. Buckets make the plan livable from month to month. Guardrails make it sustainable over decades. Neither one alone is a retirement income plan.

How We Think About It

We lean toward using buckets the way they help most: as a way to see, live with, and draw from a portfolio, not as a separate way to invest it. The investing is one household portfolio, its mix set case by case in conversation with the household. The paycheck bucket, sized with each household and held in cash, pays a monthly paycheck and is topped up twice a year, while the portfolio is rebalanced whenever its mix drifts past set bands. Purpose buckets hold money for known expenses on their own schedule, from the roof to the trip to Italy. We call them buckets with clients, because the picture works, and we let the rebalance and the guardrails do the steering.

This approach keeps what the research says buckets do well and drops what it says they do poorly. The behavioral benefit is real: a known paycheck, a visible reserve, a clear answer to where next year’s money is coming from, and a named account for the things the household is looking forward to make it far easier to stay with the plan in a bad market and to enjoy it in a good one. The costs come from the rules, not the buckets. Cash held on top of the allocation costs return, refill rules tied to last year’s market behave like accidental market timing, and a reserve that is never refilled hands the allocation to the order of spending. Keeping the cash inside the allocation and refilling it on a schedule avoids all three.

The honest trade-offs. Cash earns less than bonds, so the paycheck bucket gives up some return, about 0.18 points a year in our test for a reserve the size of the couple’s, and more in a long decline unless the target is revisited. Letting the stock share drift up came out well ahead of a steady mix in both hard starts shown here, by roughly double after 2000, and by refilling on schedule and rebalancing to target we give up that accidental version on purpose. When a rising stock share fits a household’s plan, we lean toward choosing it deliberately and rebalancing toward it, in an amount the household and its plan actually chose. And a household that is comfortable seeing one total-return portfolio and drawing from it directly gets less from the buckets than one that finds them reassuring. For some households the structure is mostly reporting, and we size it accordingly.

Common Questions

What is the bucket strategy for retirement?

A bucket strategy divides retirement savings by when the money will be spent: cash for the next few years, bonds for the years after that, and stocks for the long run. The retiree spends from the cash bucket, refills it from the others, and leaves the stocks to grow. Its main benefit is behavioral, a steady paycheck and a clear answer to where near-term spending comes from. We lean toward keeping the buckets inside one total allocation, refilling the cash on a schedule, and rebalancing the whole when its mix drifts.

How many years of cash should a retiree keep?

There is no single right number, and the measure varies by household. We lean toward sizing the cash on the portfolio’s core job (core spending minus guaranteed income such as Social Security) plus a buffer, rather than on the whole budget, often around three to five years of the spending it protects, and holding it in cash. Of the 14 stock declines of 20% or more since 1871 that had fully recovered by mid-2023, 5 recovered within three years after inflation and 9 within five, so a reserve in that range, spent down and never refilled, would have outlasted many declines but not the longest ones. Ours is topped up on a schedule instead, with the rest of the conservative side behind it.

When should you refill the cash bucket?

We lean toward a schedule rather than a signal: topping the reserve up twice a year, at the plan reviews, in good markets and bad. The top-up comes from whatever has grown past its target weight: usually stocks after a good stretch, usually bonds after a decline. Rules that refill only after good years have trailed simple rebalancing in historical tests, and no one can tell in the middle of a decline whether it will be short or long.

How often should a retirement portfolio be rebalanced?

Often enough to keep the mix near its target, but not so often that trading costs and taxes eat the benefit. We lean toward rebalancing on drift, whenever the mix moves past a set band around its target, rather than only on the calendar; the cash bucket is topped up separately, twice a year at the plan reviews. Across several accounts, we manage the household as one portfolio, with the allocation assigned account by account through asset location, and trades inside IRAs and Roth accounts trigger no capital gains tax.

Is a bucket strategy better than rebalancing?

Buckets and rebalancing do different jobs, so the choice is less of a contest than it sounds: the buckets organize the money and pay the paycheck, and rebalancing keeps the overall mix where the plan wants it. What has trailed in historical tests is buckets run on their own refill rules with no rebalancing: in one study of U.S. data from 1900 through 2014, with 30-year retirements starting at a 4% withdrawal, a steady 60% stock portfolio rebalanced every year never ran out of money, while three common bucket rules failed in 3.5% to 4.7% of periods. Never refilling has sometimes done better, because spending the cash and bonds first leaves the portfolio owning more stocks over time, a riskier mix that is better chosen on purpose than by accident. The buckets add no return on their own; their value is behavioral, a steady paycheck and a visible reserve that make a plan easier to stay with. We use both: a paycheck bucket of cash, counted inside the allocation and topped up twice a year, and rebalancing whenever the mix drifts, so the whole portfolio stays on the mix the plan chose.