Kicking the bucket strategy

This course is no longer eligible for CE credits. Go to cecorner.ca to find eligible courses.

Retirement plan design criteria

When I entered financial planning from a career in engineering I was stunned by the prevalent practice: assume an average growth rate for the portfolio and then make a forecast for retirement assets.

That’s just like building the Eiffel Tower to withstand the average wind speed in Paris (let’s say it’s 6 mph). When the wind kicks up to 100 mph, the tower will likely collapse.

In engineering, we design for the worst-case scenario—and then some. The same applies when I design a retirement plan. While I agree past events will not be repeated in the future, knowing that a plan can cover the worst cases of the past elevates my confidence that it can withstand future black swan events.

Methodology

We only use pure market history, starting in 1900 and ending at the close of 2012. We call this aftcasting, as opposed to forecasting. We do not use Monte Carlo simulators.

Aftcasting displays the outcome of all historical asset values of all portfolios since 1900 on the same chart, as if a scenario starts in each one of the years between 1900 and 2000. It gives a bird’s eye view of all outcomes, and provides success and failure statistics with exact historical accuracy because it includes actual historical equity performance, inflation and interest rates, as well as historical sequencing of all these data sets.

Standard Single Bucket

Scenario 1: Single bucket with fixed asset allocation

This is our base scenario. Bob and Jane, both 65, are retiring. They have a $1 million portfolio with an asset mix of 50% equities and 50% fixed income/cash. As the equity proxy, we use the Dow Jones Industrial Average (DJIA) total return (index plus dividends), less 2% for portfolio expenses, fees and other costs. On the fixed-income side, we use the historical 6-month CD rate plus 0.5% yield premium, net after expenses. This approximates a bond ladder with about a five- to six-year average maturity, held to the end date (no capital gains/losses).

Bob and Jane plan on withdrawing $45,000 annually, indexed to CPI. So their initial withdrawal rate is 4.5%. Their primary concern is the sustainability of their income stream—they want it to last until age 98. The probability of one of them living beyond that age is about 7%, which covers longevity risk.

Figure 1 shows the aftcast of asset values, represented by thin, grey lines. There is one line starting at the left vertical axis for each year since 1900. We define the bottom decile of all outcomes as the “unlucky” outcome, represented by the red line. The top decile is the “lucky” outcome, represented by the green line. The blue line is the median outcome, where half the scenarios are better and half worse.

Figure 1: Aftcast of Bob and Jane’s assets, Scenario 1

Our primary focus is income, which is shown on the “income carpet” in Figure 2. The horizontal scale represents all starting years between 1900 and 2000. The vertical scale indicates age. Each “knot” of the income carpet shows the level of income received at a given age and starting year, as a percentage of total income required in real, inflation-adjusted dollars. Different colors indicate different ranges of percentages as indicated in the legend to the right of the chart. In a nutshell, green is good, red is bad.

Figure 2: Income carpet for Bob and Jane, Scenario 1

The aftcast shows there’s a 23% chance Bob and Jane run out of income by age 97. This doesn’t meet our design criteria.

“What?! How can this happen?” you might say. “We have a 4.5% initial withdrawal rate, and the plan is still inadequate?”

The answer is yes, and the reason is simple. The major studies on sustainable withdrawal rates are based on calculated historical returns, but they miss the cost factor. Portfolio management costs, advisor fees, trading costs and all other expenses to hold and manage the portfolio increase the probability of failure. For this reason, we believe the famous “4% rule” is too optimistic. In our aftcast, we account for these costs to generate more realistic results.

There’s another factor affecting the outcome: we use a 32-year retirement time horizon, while the “4% rule” uses a 30-year timeframe.

Scenario 2: Single bucket with age-based asset allocation

This is a variation of the buy-and-hold strategy we saw in Scenario 1. Some think they can reduce the risk of running out of money during retirement by holding more fixed income as they get older. For example, instead of holding 50% equity in the portfolio for the rest of their lives, Bob and Jane hold a percentage of equity that is equal to 100 minus their age. So, they hold 35% equity at age 65, 20% equity at age 80, and so on.

In this case the probability of depletion at age 97 is 32%—worse than our base scenario.

Figure 3: Income carpet for Bob and Jane, Scenario 2

Scenario 3: Single bucket with a 5-year asset dedication

This is the same as the Scenario 1, except we have a new rule about asset allocation. Bob and Jane’s asset mix is still 50% equities and 50% fixed income/cash. However, we never allow the fixed income/cash portion of the portfolio to hold less than 5 years of income required from the portfolio.

The aftcast says the probability of depletion at age 97 is 27%—worse than our base scenario.

Figure 4: Income carpet for Bob and Jane, Scenario 3

Multiple buckets

Assets are divided into a number of buckets at the beginning of retirement. The first bucket is usually the most conservative, holding only cash. The last bucket is the most aggressive. Withdrawals are taken from the first bucket until it depletes; they’re then taken from the next bucket until its empty, and so on. This is also called ‘time segmentation of withdrawals.’

There are two types of buckets: dynamic and static. With dynamic buckets, money can flow from one to another—usually from a bucket with a more aggressive portfolio to one that holds cash. The second type is static—no money flows between buckets.

Scenario 4: Two static buckets

Let’s look at a simple, static two-bucket strategy. The first bucket holds five years of income—$225,000 in cash and CDs. The second bucket holds the remaining $775,000, with an asset mix of 50% equities and 50% bonds.

The income carpet is depicted in Figure 5. The probability of depletion at age 97 is 21%, slightly better than our base scenario. But in practice, the improvement is insignificant.

Figure 5: Income carpet for Bob and Jane, Scenario 4

Scenario 5: Two dynamic buckets

This is the same as Scenario 4, with the following exception: In any given year, if the equity portion of Bucket #2 grows by more than 8%, we take that growth and add it to the cash bucket.

The income carpet is depicted in Figure 6. The probability of depletion at age 97 is 24%, slightly worse than the standard buy-and-hold.

Figure 6: Income carpet for Bob and Jane, Scenario

Scenario 6: Three static buckets

In this scenario, Bucket #1 holds five years of income; Bucket #2 holds an asset mix of 30% equities and 70% fixed income; and Bucket #3 holds 70% equities and 30% fixed income. There is $225,000 in Bucket #1, and $387,500 in each of buckets #2 and #3. All buckets are static, i.e. we deplete Bucket #1 first, Bucket #2 second and Bucket #3 last, and no money flows between them.

The income carpet for this scenario is depicted in Figure 7. The probability of depletion at age 97 is 22%, almost the same as our base scenario.

In the article where I found this strategy (“Comparing a Bucket Strategy and a Systematic Withdrawal Strategy,” by Noelle E. Fox, AAII Journal, April 2012), it’s argued that “systematic withdrawals lack the structure of a bucket strategy.” This is certainly true, but whether or not a strategy has structure is mathematically irrelevant. The only pertinent question should be: “Which strategy can provide lifelong income?” The aftcast shows this particular bucket strategy does not yield a better result than our base scenario.

Figure 7: Income carpet for Bob and Jane, Scenario 6

Scenario 7: Three dynamic buckets

This is the same as Scenario 6, except the buckets are dynamic. In any year, if the equity portion of the portfolio in Buckets #2 and # 3 grows by more than 8%, half of that growth is added to the cash bucket.

The probability of depletion at age 97 is 21%—not much different than our base scenario.

Figure 8: Income carpet for Bob and Jane, Scenario 7

Scenario 8: Six static buckets

This is the most complicated bucket strategy I’ve come across. Bucket #1 is the most conservative; it holds cash and is first in line for income. Bucket #6 is the most aggressive and used last.

Bucket #1 holds five years of income—$225,000 in CDs. Bucket #2 has $200,000 invested in conventional bonds. Bucket #3 has $175,000 invested in a 50% equities, 50% fixed income asset mix. Bucket #4 has $150,000 invested in a 60%/40% asset mix. Bucket #5 has $135,000 with an 80%/20% asset mix. Bucket #6 has $115,000, all in equities. All buckets are static.

The probability of depletion at age 97 is 18%, slightly better than our base scenario, but still unable to provide lifelong income in all market extremes.

Figure 9: Income carpet for Bob and Jane, Scenario 8

Scenario 9: Six dynamic buckets

This is the same as Scenario 8 except the buckets are dynamic. Each year, if equities in any bucket grow by more than 8%, half of the growth is added to the cash bucket.

The probability of depletion at age 97 is 16%, somewhat better than our base scenario.

Figure 10: Income carpet for Bob and Jane, Scenario 9

Scenario 10: Six dynamic buckets with age collars

Philip G. Lubinski claims his six-bucket strategy is structured to provide five years of income in each bucket. For example, Bucket #1 is supposed to provide income between ages 65 and 69 (inclusive), Bucket #2, between ages 70 and 74, and so on. I tested this claim as part of my due diligence on bucket strategies.

I generated Scenario #10, which is identical to Scenario #9 with one exception: Each bucket has age collars for withdrawals, i.e. withdrawals can be made only between the specific ages for that bucket. All buckets are still dynamic: Each year, if equities in any bucket grow by more than 8%, half of the growth is added to the cash bucket for that particular asset.

Can each bucket provide five years of income, as claimed? The income carpet (see Figure 11) shows that in many instances there’s little or no income in the 5^th year of each bucket (non-green pixels). In some cases this shortfall appears in the fourth year. Therefore, the claim that “each bucket should be able to provide five years of income” is incongruent with historical experience.

In reality, the retiree will take this shortfall from the next bucket. Then, he will eventually end up with the similar outcomes as Scenario 8 or Scenario 9. Neither of these scenarios provided lifelong income in all extremes, failing to meet the design criteria we set out at the beginning of the lesson.

Figure 11: Income carpet for Bob and Jane, Scenario 10

A simpler explanation

You’re not alone if you find some of this confusing at first. Here’s a simple analogy we can all relate to.

You and three of your buddies are playing poker. Everyone’s hungry and you order a large pizza. But it arrives unsliced, so you take a knife and suggest slicing it like this:

One of your buddies smartly suggests, “if we make the slices like this, we’ll each have more pizza:”

Which pattern will better satisfy your appetite? It makes no difference. The pizza doesn’t get larger because you slice it this way or that.

The same goes for retirement assets and the bucket strategy: For a given total asset size, once you exceed the sustainable withdrawal rate, it doesn’t matter how many buckets you divide the assets into. The probability of running out of money will remain just about the same.

The luck factor overwhelms any of the potential benefit the bucket strategy’s time segmentation might provide. The only reliable way of having lifelong income is through pooling risk through annuities.

One of the bucket strategy’s assumptions is if you delay withdrawals from the equity-laden buckets as long as you can, it will eventually grow and provide a sufficient asset base for withdrawals. This isn’t always true. There were three secular bullish trends during the last century, occupying about 43 years in total.

I observed in my aftcasts that after the two-thirds mark of a typical secular bull market, it is almost always better to draw down income from the more aggressive equity buckets. This way, you ring the cash register while the going is hot, minimize money left on the table and reduce the eventual risk of adverse sequence of returns. This strategy also reduces the eventual portfolio volatility and negative effects of reverse-dollar-cost averaging when the bear market eventually arrives.

Another benefit is you would be preserving your fixed income holdings for the eventual joyride on the yield curve, when the Federal Reserve reduces interest rates after the equity market correction. Luckily, you don’t need any convoluted bucket strategies or sophisticated market timing for that. A single bucket with proper asset allocation and occasional rebalancing will accomplish all of these objectives just as well.

And what if your client’s withdrawals are below sustainable? We define this as the green zone. Here, you don’t need any complicated strategies or any annuities—almost anything will work.

Conclusion

History clearly shows that bucket strategies do not provide a noticeable improvement over the basic, single bucket withdrawal strategy. To add insult to injury, much of the material I’ve examined uses higher withdrawal rates than the 4.5% used in the scenarios discussed above.

Now that you’ve finished reading, complete the exam to receive your CE credits. If your score is 85% or higher, send an e-mail to jim@retirementoptimizer.com with your name and proof of your score (a screen shot will do) to get a free retirement calculator based on aftcasting, and a free, read-only pdf copy of Jim Otar’s 525-page book, Unveiling the Retirement Myth.