We Put the Bucket System Through a Longer Stress Test

"You have made all of us spreadsheet guys very happy!" 

That was one reader's comment about my recent stress test of the bucket system for retirement-portfolio management, which included a link to a spreadsheet detailing the cash flows in and out of a model bucket portfolio. The model portfolio performed well over the challenging, but admittedly arbitrary, time period of 2007-12, and readers said they found it useful to see the logistics of retirement-portfolio management from year to year. 

Still, commenters said they'd like to see additional variations, a few of which I addressed last week. I featured a portfolio with withdrawals starting on day 1 of retirement, and I also explored how stripped down the bucket system could get, employing a simple balanced fund in lieu of the multifund portfolio featured in my original stress test. 

This week, I'll take a look at two more variations on retirement-portfolio distribution systems. The first portfolio simulation forgoes a dedicated cash bucket and instead makes do with a portfolio consisting of bond and stock funds. With cash yields as low as they are right now, the simulation demonstrates the opportunity cost of having a dedicated cash component. (Financial-planning expert Michael Kitces made the case for not carrying too much in retirement in this video.) 

The second variation received multiple reader requests after my first stress test: It features a longer time frame, simulating the portfolio's results from 2000 through 2012. As such, it demonstrates how the multifund portfolio would have fared during not just one but two major bear markets: the dot-com meltdown and the 2007-09 financial crisis. 

Without a Bucket
Starting Portfolio Balance: $1,500,000
Ending Portfolio Balance: $1,622,332
Amount Withdrawn: $378,549 

This portfolio features the same long-term holdings as in my original bucket portfolio, and the simulation runs during exactly the same time frame--2007-12. However, in lieu of holding a dedicated cash component--bucket 1 in the portfolio featured in my initial stress test--this portfolio steers those dollars to the short- and intermediate-term bond funds instead. That means that this portfolio has a limited cushion: If neither the stock nor bond holdings generate a return substantial enough to cover living expenses, the retiree will have to raid the principal in the short-term bond fund. You can view the simulation in this spreadsheet (Microsoft Excel required) or this PDF.

I used a similar system for withdrawals and rebalancing that I employed in my initial stress test. Specifically, I harvested gains from positions that exceeded 110% of their starting values and used those assets for living expenses; additional rebalancing proceeds (above and beyond income for living expenses) were used to top up depressed long-term positions and moved into the short-term bond fund, in that order. If rebalancing proceeds were insufficient to meet living expenses, I took withdrawals from the short-term bond fund. 

That happened early in the life of the stress test. Although harvesting gains from long-term holdings easily met living expenses and allowed for reinvestment in year 1--2007--none of the holdings was 10% above its starting value at the end of 2008, forcing me to sell some of the short-term bond fund. 

That said, the short-term bond position didn't get terribly low throughout the exercise; at its lowest ebb, in early 2011, it was still close to two years' worth of living expenses. (I also could have boosted the short-term bond fund stake more rapidly after raiding it by using a different rebalancing protocol, steering rebalancing proceeds first into that fund and then moving any excess cash into the depressed long-term holdings.) In nearly all other years, using rebalancing proceeds from appreciated long-term holdings helped meet the desired living expenses; we tapped the short-term bond fund for living expenses in just two of the six years. (The second time, in 2010, the distribution was only partial; we were able to raise the rest of our living expenses by selling appreciated securities that had hit our rebalancing threshold.) 

Moreover, the big benefit of doing without a cash bucket was that the portfolio's ending value was higher than was the case in my first stress test, which included a dedicated cash bucket in the simulation. Clearly, this portfolio benefited from not having the drag of cash, especially as money market fund yields all but vanished during the simulation period. 

Additionally, the bond holdings in this portfolio benefited from downward-trending yields, which helped boost bonds' principal values and gave this stock/bond portfolio a leg-up on the one that featured cash. Cash instruments such as money market funds, meanwhile, only experience the downside when yields drop; they're forced to invest in lower-yielding securities and don't enjoy rising net asset values. 

But the opposite could also hold true: If bond yields trend up, the bond holdings in a stock/bond (no-cash) portfolio could get slammed. For the year to date through mid-August, for example, short-term bond funds are actually underperforming good-quality money market vehicles. Maintaining a dedicated cash component could help provide peace of mind in such an environment, but doing so will likely carry an opportunity cost over a longer period of time if bonds outperform cash. 

Incorporating 2 Bear Markets
Starting Portfolio Balance: $1,500,000
Ending Portfolio Balance: $2,029,138 
Amount Withdrawn: $925,313 

Retired investors well know that encountering a bear market early in retirement can be hugely problematic. What would have happened if a retiree began taking withdrawals during the first year of a bear market and encountered an even worse bear environment less than a decade later? The 2000-12 period allows us to simulate the outcome. 

For this simulation, I used nearly the same holdings as in my initial simulation but had to swap in a few different funds as replacements for offerings that weren't in existence in 2000. I used the sturdy  T. Rowe Price Equity Income (PRFDX) in lieu of  Vanguard Dividend Growth (VDIGX), and  Oppenheimer Commodity Strategy Total Return in place of  PIMCO Commodity Real Return (PCRAX). Otherwise, however, I essentially employed the same rebalancing and withdrawal rules as in my initial simulation, with some minor adjustments. You can view the simulation in this spreadsheet (Microsoft Excel required) or this PDF.

The headline is that a well-allocated and consistently rebalanced portfolio would have done quite well during the period, supporting our desired withdrawal rate (4% of the portfolio's starting balance, with a 3% inflation adjustment per year) and ending 2012 well above its initial level. Another major takeaway is that you didn't need to blow up your portfolio plan just because the markets blew up. Keeping a strategic portfolio rather than tinkering to try to preserve wealth allowed the portfolio to recover on its own.

Even as the portfolio endured two bear markets, including the worst market since the Great Depression, it also featured exceptionally strong recoveries, during 2003-07 and again during 2009-12. Our bond holdings also thrived for most of the period, benefiting from declining yields. In up years, rebalancing the portfolio handily met living expenses and enabled us to top up all of the long-term positions. To fully participate in those rallies, however, this portfolio had to reinvest in losing positions during difficult times. At the end of 2008, for instance, that meant withdrawing more than $300,000 from the safety of a short-term bond fund and reinvesting it in funds that had lost 30%, 40%, or even 50% of their value compared with the preceding year. The panicked media environment in late 2008 would have made this doubly difficult. But the simulation shows that having the fortitude to rebalance to a planned allocation in tough times can make a big difference to the bottom line.

Furthermore, our protocol called for adding any additional rebalancing proceeds to cash and our short-term bond fund; the net effect was that the portfolio ran with a big surplus in both holdings (well above the original position sizes) for much of the period. However, it's worth noting that a less conservative investor could have deployed those extra dollars in a different manner. One option simply would have been to spend any additional monies once all of the positions in the portfolio--including cash, short-term bonds, and all long-term holdings--were back to their initial values. Alternatively, one could deploy those extra dollars into the portfolio's long-term holdings to ramp up its growth potential. 

As encouraging as the results of the test case were, however, it's important to approach them with a healthy dose of skepticism; they shouldn't be used as a justification for a substantially higher withdrawal rate than what's featured here. For starters, the simulation is biased toward a very positive end date--2012, after stocks had rocketed to great gains during a four-year period. As recently as early 2009, the portfolio wasn't far above its starting value. 

Perhaps more important, the conditions that have prevailed in the past 13 years simply aren't repeatable, especially for bonds. Due to declining yields, the high-quality bonds in the portfolio contributed an unbroken string of solid gains throughout the entire period. Should yields trend up in the future, as they're apt to do, the bond holdings could be a drag on returns in certain years. That doesn't mean that retiree portfolios should do without bonds--they're still likely to lose less than stocks during weak equity markets--but it does mean that bonds are unlikely to contribute as much on the return front during the next decade.

See More Articles by Christine Benz