The cash reserve (bucket) strategy: how much cash to hold in retirement and when it helps
Almost every retirement conversation eventually lands on the same comforting idea: keep a few years of spending in cash, so a market crash can't force you to sell stocks at the bottom. Financial planners call it a cash reserve or a bucket strategy, and Runway has a Cash reserve setting that models it. It's an appealing idea and a popular one. Whether it makes your money last longer is a different question, and the honest answer, from both the research and Runway's own engine, is "only in the worst crashes, and at a cost."
What is a cash reserve, exactly?
The idea is simple. Instead of drawing from the whole portfolio each year, you hold a reserve in cash worth, say, two years of spending. When the market is down, you spend the cash and leave the stocks alone, giving them time to recover. When the market has climbed back above its previous high, you top the reserve up from gains. Runway's version works exactly that way: you enter a number of years; the reserve is funded at the start from your other accounts; in a down market (the stock market below its own previous high) cash is spent first, floor included; in every other year the floor is protected and refilled. Cash earns about 0% after inflation in the engine, which is the usual assumption for a safe, liquid balance. (An earlier version of Runway protected the floor even during a crash, which left the cash idle exactly when it was supposed to be used; that is fixed.)
Does it make the money last longer?
Over a whole retirement, mostly no. Using the sample couple Runway loads (Pat 65, Alex 64, $1.46 million), here is the chance the money lasts through age 95 across 2,000 simulated markets, by size of reserve:
| Reserve (years) | Lasts at $95,000 | Median ending | Lasts at $110,000 |
|---|---|---|---|
| None | 99.5% | $1,574,254 | 87.0% |
| 1 year | 99.5% | $1,532,265 | 82.1% |
| 2 years | 98.8% | $1,471,697 | 73.2% |
| 3 years | 97.8% | $1,325,598 | 59.0% |
| 5 years | 88.0% | $561,755 | 10.0% |
More cash, lower odds, and the cost climbs quickly as the reserve grows. A reserve is money that isn't invested for growth: three years of spending is $285,000 sitting out of the market, and five years is $475,000. Over a 30-year retirement that drag outweighs the benefit of not selling after a drop.
The same is true when the engine replays real history instead of simulated markets. Across 125 starting years (1872 through 1996, each a 30-year run), the plan at $95,000 of spending never ran short with no reserve or a one-year reserve; with two years two starting years ran short (1966 and 1969), and with three years, five did. At $110,000, the number of starting years that ran short climbed from 31 of 125 with no reserve to 41 with a two-year reserve and 51 with a three-year one. Even in the worst year to retire, 1966, the $95,000 plan finishes with $211,851 and no reserve but runs out at 93 with a two-year reserve.
| Reserve | Historical success at $95,000 | Historical success at $110,000 |
|---|---|---|
| None | 100.0% | 75.2% |
| 1 year | 100.0% | 71.2% |
| 2 years | 98.4% | 67.2% |
| 3 years | 96.0% | 59.2% |
Does it at least protect me in a crash?
That is the case for it, so it deserves a direct test. Run a severe crash through the same couple's plan: stocks down 30% a year for three years, beginning in the first year of retirement. With a two-year reserve the cash is spent over the crash ($190,000 at 65, $90,130 at 66, none at 67), so the plan sells fewer depressed stocks. The balance three years later and the year the money first runs short:
| Spending | Reserve | Balance at 68 | First year money runs short |
|---|---|---|---|
| $95,000 | None | $253,352 | Pat is 75 |
| $95,000 | 2 years | $306,156 | Pat is 77 |
| $95,000 | 3 years | $374,024 | Pat is 80 |
| $110,000 | None | $208,586 | Pat is 72 |
| $110,000 | 2 years | $269,159 | Pat is 73 |
| $110,000 | 3 years | $350,881 | Pat is 75 |
That is real protection: three years of cash buys five years at $95,000 in the worst case. But look at the size of it. The reserve delays the day the money runs out; it doesn't prevent it, because a crash that deep can't be bought off with three years of cash. And in a milder crash the reserve stops helping. After stocks fall 25% a year for two years, a $95,000 plan runs short at 89 either way. After 20% a year for two years, the plan with no reserve survives to 95 with $191,441, while the plan with a two-year reserve runs short at 94: the drag outweighed the protection.
What does the research say?
Runway's result matches the literature. Michael Kitces wrote in 2013 that bucket strategies "can actually lead to more conservative portfolios, lower returns due to the drag of big cash positions, and worse retirement outcomes," and that standard rebalancing makes most bucket strategies unnecessary from an asset-allocation point of view. A 2018 working paper by Javier Estrada of IESE Business School ("The Bucket Approach for Retirement: A Suboptimal Behavioral Trick?") examined 21 countries over 115 years and found that "simple static strategies, which by definition involve periodic rebalancing, clearly outperform bucket strategies," on four different measures of performance. Both authors also take the behavioral benefit seriously. A retiree who can see a few years of spending safely set aside may be less likely to panic and sell in a crash, and a plan you actually stick with beats a better plan you abandon.
So when does a cash reserve make sense?
- If it keeps you invested. If the alternative is selling stocks in a panic, a small reserve may be worth its cost. The price is a lower expected result: a two-year reserve costs under one percentage point of success at comfortable spending and about 14 points near the edge.
- For spending you know is coming. A known near-term expense (a house, a wedding, next year's bills) is a separate matter from a standing reserve. Money you'll need soon doesn't belong in stocks.
- Not as a way to spend more. A reserve can't raise how much you can safely spend. Its effect over a full retirement was the reverse.
Test a cash reserve on your own numbers. Set the reserve on Market Assumptions or in Explore Your Plan and see what it does to your chance of success.
Try the plannerWhat does this leave out?
- Behavior. The engine assumes you do what the plan says. It can't measure whether a visible reserve would keep you from selling at the bottom, which is the main argument for it.
- Cash earns 0% real. If your cash earns a real positive return (high-yield savings or short Treasuries when rates are high), the drag is smaller than shown.
- The sample couple's mix is 60% stocks. A reserve might look less costly on a very aggressive portfolio, and more costly on a conservative one.
- Crashes that start at 65. The crash tests start on the first day of retirement, the worst time. A crash later in retirement leaves less time for the reserve to matter.
- What "down market" means. Runway treats the market as down when it is below its own previous high, and refills only after it climbs back above. After a deep crash that can take many years, during which the reserve stays empty.
How were these numbers computed?
All figures are Runway's engine on the sample couple: married filing jointly, plan through age 95, 60% stocks and 40% bonds, stocks at 7.5% and bonds at 3.5% a year after inflation, 16% and 6% volatility, cash at 0% after inflation, long-term-care cost left out. Monte Carlo runs use 2,000 paths and seed 2026. The historical backtest uses annual real returns for U.S. stocks and 10-year bonds from Robert Shiller's public dataset; each of the 125 starting years from 1872 through 1996 runs for 30 years. The crash tests are deterministic projections with stocks down 30% a year for three years (and 25% and 20% for two years, as described) from age 65. All amounts are in today's dollars.