8.3 The “old” investor Part 2 – Bucket investing
Part 1 about the “old” investor (Article 8.2) concluded that adding some type of ballast shortly before and early in the spending phase (let’s call it the “vulnerable period”) is a mindful approach for the “old” investor. Now in Part 2 we explore the potential types of ballast, how much to use, and how best to deploy it.
The information in Article 8.2 showed that no one approach to adding ballast in the vulnerable period appears clearly superior to the others. Consequently, I think the concept of “bucket investing” created by Harold Evensky is a mindfully humble and simple starting place for using ballast. Bucket investing is based on when you will need to spend the money in your portfolio. It categorizes a portfolio in terms of the risk of permanent loss, which is more likely the sooner spending needs to occur. Many well-respected retirement researchers support the bucket or “time segmentation” approach to retirement investing. For example, Wade Pfau notes that, although bucket investing is one of the least studied retirement investing approaches, it nonetheless appears to provide a good compromise between other popular retirement approaches.
There are detractors of bucket investing out there, and their main criticisms seem to be that:
- It’s just a clever way of describing the same old ballast methods
- It can be complicated to implement.
Given we are trying to pick among many potential ballast approaches, the first criticism should not deter us from at least investigating the merits of bucket investing. Likewise, the second criticism about implementation challenges is something we all need to gauge regardless of the specific method involved. Perhaps bucket investing is complicated in some situations and relatively simple in other situations.
Bucket basics – There are many bucket templates out there, most of which rely on the concept of investor “risk tolerance”. Some examples are provided here at Morningstar. However, in Article 4.3 we determined that risk tolerance is all about irrational emotional reactions like panic selling that can be avoided using mindfulness. So, we won’t pursue a risk tolerance approach to devising investing buckets. Instead we will devise buckets based on empirical data from past Mindfully Investing articles regarding historical stock, bond, and cash volatility and returns.
Most bucket approaches have 2 or 3 buckets, although some may have more. Thus, we can categorize typical buckets as short, mid, and long term.
The long-term bucket is probably easiest to define, and this bucket should clearly be filled with stocks. Based on the Article 8 stock data, any money you plan to spend more than 5 to 10 years in the future should go into the long-term bucket. Whether you should pick 5 or 10 years, or something in between, is discussed more below.
The short-term bucket is the next easiest to define. This is money you intend to spend soon, which must be less than 5 years to avoid the short-term bucket overlapping with the long-term bucket. The exact number of years for the short-term bucket threshold is a decision that should factor in the details of your situation, expected spending needs, and the possibility of certain types of spending emergencies. For example, a person that has advanced heart disease probably wants to make the short-term bucket time frame longer due to potential additional health care costs not covered by insurance. But a 55-year-old vegan marathoner may reasonably conclude a shorter time frame is appropriate. There are many other potential factors specific to your situation that might also impact your short-term bucket time frame.
We saw in Article 7.3 that small amounts of cash can provide very stable portfolio ballast as compared to larger amounts of bonds. Consequently, the prime candidate for the short-term bucket is cash. We know the real value of cash will erode overtime (usually slowly) due to inflation, but cash has the advantage of being relatively uncorrelated with stock market movements. Further, if history is any guide at all, inflation is unlikely to substantially impact the value of cash in just the few years it resides in the short-term bucket. Inflation effects are much more pronounced when compounded over many years.
That leaves the mid-term bucket. Logically, the mid-term bucket would bridge any gap between short term and long term. If you define your short-term bucket as up to 5 years and your long-term bucket as beyond 5 years, you don’t need a middle bucket. What to put in the mid-term bucket is probably the most nuanced decision, which I will discuss more below.
Bucket summary – To review, we’ve developed the following general description of investing buckets for the “old” investor:
- Short-term = Spending that is less than 5 years away. Fill this bucket with cash.
- Mid-term = Spending duration that bridges the gap between short and long term. We have yet to determine the exact contents of this bucket.
- Long-term = Spending that is more than 5 to 10 years away. Fill this bucket with low-cost stock index funds.
Bucket investing encompasses many of the potential advantages of other ballast approaches but is a mindfully simple concept. Like the other approaches, it keeps some money in less risky ballast assets to help minimize portfolio declines and gives you more time to wait out any bad luck stock market crashes before having to sell any stocks. If we assume 30 years of retirement and a constant (non-inflation adjusted) 4% annual spending rate over time, about 20 to 25 years-worth (or 60 to 80%), of your portfolio would be in higher performing stocks at the start of the spending phase. And, as we have seen in the Article 7 series, more high performing stocks in your portfolio generally increases your investment returns relative to other options.
We still have some details to work out for our mindful bucket investing plan to be fully implementable. Let’s go back to the short-term bucket and see how we can mindfully refine that bucket. After that, I will flesh out similar details for the other two buckets.
For the short-term bucket, let’s look closer at the percentage of a retirement portfolio that should reasonably be held as cash.
Cash Amounts – As discussed in Article 8.2, “success rate” is a highly relevant measure for retirement portfolios. After all, if we run out of money before we die, that’s not a very good retirement plan. So what is the likely success rate for holding various amounts of cash in a retirement portfolio? FIRECalc is a handy free webtool that uses Monte Carlo analysis to examine how various investing plans would have worked in all possible periods of past stock/bond market history, in this case dating back to 1926. I entered the full range of possible stock and cash portfolio ratios into FIRECalc and then compiled the resulting success rates for those portfolios as shown in this graph. (Note that FIRECalc uses very short-term T-bills to represent cash, which fits within our general definition of cash.)
Consistent with my previous examples, I used a 30-year investing period and a starting portfolio of $1,000,000 to keep the numbers nice and round. I looked at three different inflation-adjusted annual spending rates of 4%, 4.5%, and 5%, which equate respectively to $40,000; $45,000; and $50,000 at the start of the period.
First, the graph clearly shows that the more you spend, the greater your chances of failure. (For any early retirees out there, I should also point out that if you conduct the same analysis with a longer retirement period, like 40 or 50 years, the rates of success drop considerably.) Regardless, if we examine the more prudent spending rates ($40,000 and $45,000 per year) over 30 years, we can see that the chances of success do not greatly diminish when the portfolio contains at least 80% stocks (or no more than 20% cash) in the portfolio. If history is any guide, having up to 20% of your portfolio in cash does not substantially increase the chances of running out of money. At the 4% spending rate, you could even make the argument that 30-40% cash still has no substantial impact on the success rate, but such a conclusion may rely too heavily on historical analogy.
We seem to have arrived at an answer of 20% cash, which at a constant 4% annual spending rate, equates to 5 years of spending in the short-term bucket. However, success rate is just one measure of any portfolio over time. Ending your retirement with one dollar is a “success” by this measure, but we all recognize it would be wise to have some safety margin both in terms of the rate of success and the value of your portfolio at the end of any given period. So, we should also consider how more or less cash can impact the performance of a portfolio.
Cash performance – The above analysis assumes that the cash just sits there in your portfolio and is drawn upon proportionally along with stock value to fund spending in any given year. Under these assumptions, the cash is serving no other function than ballast and has almost no positive impact on portfolio performance.
In fact, using cash in your portfolio for only this ballast function isn’t supported by simple math. To illustrate, let’s go back to our steady growth rate assumption we used in Article 8. The graph below shows a severe stock decline scenario during the vulnerable period of a 30-year retirement and assuming: a 35% stock market crash at year 5, a 5.5% annualized stock value growth the rest of the time, and a 4% annual constant spending rate.
Larger amounts of cash help moderate the initial decline at year 5. But after that, continuing to hold the cash doesn’t help portfolio value over time as compared to holding 100% stocks. The more cash you hold, the worse your portfolio value gets.
So, what should we do with the cash in our portfolios? In Article 7.3, I discussed some other potential uses of cash in a portfolio beyond just the ballast function including:
- Cash as a “reserve” – where cash is drawn upon instead of stocks in times of stock market crashes to allow time for stocks to recover.
- Cash for “buying opportunities” – where cash is invested after severe stock market crashes to help the portfolio value grow back more quickly.
It’s simple to add these two scenarios to the analysis using steady stock growth rates as shown here.
The graph shows portfolio value changes when the cash is used 1) as a “reserve” to fund all spending after the stock crash until the cash is all used up and 2) the cash is used to buy stocks at a discount shortly after the stock market dips. In this second “buy-the-dip” scenario, after the cash is used to buy stocks, the cash bucket is not replenished, and thereafter, stocks are sold as needed to fund all spending. Using cash as either a reserve or to buy-the-dip performs better than a 100% stock portfolio. And in general, the buy-the-dip approach performs better than the cash reserve approach.
This seems like good evidence that more cash is better. But we have to remember that the stock market is unpredictable. What if that stock market crash never comes? Here’s how that scenario looks.
In this case, more cash is worse. You hold on to the cash in hopes you can ignite that “dry powder” when the stock market crash comes. But as time progresses and the market crash does not materialize, that additional cash is just causing a drag on your portfolio return. Because market direction is essentially unpredictable, we need to strike a balance between too much and too little cash. Examining the last two graphs together, the best balance appears to be the 20% cash scenario, including a plan to invest that cash in stocks when/if a severe stock market decline occurs. If the crash occurs, you will be substantially better off than if you had just held all stocks. If the crash does not occur, you will still have relatively decent overall portfolio growth and not too much cash drag.
Cash criticisms – Some observers express doubt about holding cash for long-term investment purposes. The folks at Early Retirement Now have published a wonderful series on retirement spending scenarios. I highly recommend you read it. One of their articles questions the validity of a “cash cushion” in retirement. They cite six main concerns. Three of those concerns have to do with how dividends are spent and accounted for over time and whether those dividends could be cut in the future. Our analysis is outside those concerns because everything here is done on a total return basis, which includes dividends. In subsequent articles I will conduct some calculations that assume dividends are reinvested annually, but all the portfolio growth and spending assumptions are still on a total return basis. Another concern is that inflation erodes cash too fast. As I will discuss further in Article 8.6 (coming soon), inflation impacts the real returns of all assets. To hold one asset class to a different inflation standard than the others will substantially confuse any analysis of a mixed asset portfolio as discussed more in Article 8.6 (coming soon). Another concern is about how to replenish the cash cushion. I think I’ve already made it clear, and I will discuss more in Article 8.4 , that replenishing the cash reserve is not likely a useful scenario and not one that is advocated here. The last concern is that cash has a low return. We can see in the above analysis that the low returns from cash do not necessarily mean that a portfolio with moderate cash amounts is more likely to fail.
Mindful expectations for cash – As I discussed in the Article 5 through 7 series, we should have humble expectations for our portfolio performance as compared to various hypothetical beat-the-market schemes. By the same token, no one can predict the single best portfolio that will obtain the highest value at the end of any given future unpredictable investment period. Rather, our goal is to have sufficient portfolio value for our retirement needs or similar spending plans. Looking at the last graph again as an example, if you can sleep with $2,000,000 in your account after 30 years (100% stocks scenario), you should be able to sleep just as well with $1,400,000 in your account (20% cash scenario), particularly if you take a mindful view of the situation.
We have determined that proper use of 20% cash in a portfolio provides a reasonable ability to withstand the bad luck of a stock market crash early in retirement, while at the same time performing reasonably well in other potential future stock market scenarios. Using a 4% constant spending rate, this equates to about 5 years of spending. Before finalizing our determination for the short-term bucket, let’s look at the other two buckets to ensure our overall approach is consistent.
Up to this point, I’ve only vaguely defined the mid-term bucket as any spending window that exists between the short and long-term buckets, or roughly in the 5 to 10-year time range. And I have made no suggestions yet about the potential contents of this bucket. Conventional advice is to fill the mid-term bucket with bonds. The supporting rationale is that the moderately greater return of bonds as compared to cash helps minimize the impact of inflation, which starts to cause a more noticeable erosion of your portfolio’s real value when compounded over more than a few years. At the same time, bonds still perform a similar ballast function as cash.
Bonds as ballast – Let’s examine the conventional wisdom by first looking at the historical ballast performance of bonds using FIRECalc again. Here is the same success rate chart as before, but I added intermediate bond/stock combinations that mirror the cash/stock combinations previously presented. Intermediate bonds are defined as having a duration of about 5 to 10 years.
Historically, intermediate bonds have performed a nearly identical ballast function as cash, particularly for the two lower withdrawal rate scenarios. It’s interesting that the success rates for these two scenarios when holding 40% or less in bonds is the same or a little worse than when holding similar amounts of cash.
Bond performance – The above analysis potentially leads to the conclusion that bonds have no use in a retirement portfolio at all, but such a conclusion ignores terminal portfolio values. As discussed in Articles 6.1 and 6.2, intermediate bonds are expected to perform poorly in the next few years, but they are still expected to perform a bit better than cash. Because of compounding growth (Article 3), we know that the slightly higher returns of bonds in a bond/stock portfolio will cause a substantially higher terminal value than a portfolio with a similar balance of cash and stocks in most historical periods. Put another way, the success rates of similar amounts of cash or bonds in a mixed portfolio will be about the same, but you will usually have a higher final account value with the bonds.
Bond risks – Because the whole point of bucket investing (and ballast in general) is to minimize the risk of permanent losses, we need to also consider the risks associated with bonds. Again, the idea is that by the time it’s necessary to sell your bonds to fund spending, you want to minimize the chances that the bonds lost money over the period you held them.
I discussed in Article 8 that we can avoid stock losses with a holding period longer than about 5 to 10 years. We should make the same determination for bonds. This graph shows historical annual 10-year bond return data compiled by Aswath Damodaran an New York University Stern School of Business.
You can see there are very few years where 10-year bonds provided negative returns. More importantly, bond returns were negative for two years in a row in only two instances (both in the 1950s) . Through a close examination of the data, I can further state that in every instance of a prior negative return, the positive returns in the next year nearly or completely offset (if not greatly exceeded) the prior negative return. In other words, investors who held onto their bonds saw a recovery in any bond losses just one year later. This is echoed by a chart from Wade Pfau using the same data set. He found that intermediate bond returns have never been below zero for a holding period of about 2 to 3 years. (Note for later discussion that the graph also shows the worst stock return in 5 years was only about -12%, and the worst return in 10 years was only about -1%.)
Therefore, the worst-case time span for a bond recovery has been about 3 years (two years of losses and one year to recover from those losses), and more often, the recovery has been in about 2 years.
Bond amounts – I presented in Articles 6.2 and 7.2 some worst-case scenarios and evaluations that showed intermediate bond returns are likely to be at least slightly positive if held over the duration of those bonds. I also presented in Article 6.2 that the “sweet spot” for bond durations is around 7 years, because it balances between decent yields and manageable potential price declines. And as we just reviewed, intermediate bond returns have historically recovered within about 2 to 3 years.
We can conclude from all this that intermediate bonds are a reasonable selection for ballast in the mid-term bucket, if held for durations of about 2 to 7 years. Using our 4% constant spending assumption, this means anywhere from 8% to 28% of your portfolio would be in bonds. However, to select a percentage of bonds within this range we must also factor in the amount of ballast held as cash in the short-term bucket. We want the total amount of ballast in the short and mid-term buckets to be appropriately balanced with the amount of stocks in the long-term bucket. We’ll explore this inter-bucket balancing act in Article 8.4 (coming soon).
Bond types and TIPS – Before moving on to the long-term bucket, we should also consider the types of bonds that might be appropriate for the mid-term bucket. All the above discussion assumes we are using relatively safe U.S. government bonds or similar bonds from other developed countries. Because bonds are performing the safety function of ballast in the mid-term bucket, it makes sense to use relatively lower risk government bonds. This specifically leaves out things like high yield corporate (junk) bonds and similar higher risk bonds for reasons described in Article 7.2.
Treasury Inflation Protected Securities (TIPS) are one type of government bond that may be well suited to the mid-term bucket. TIPS are discussed more in Articles 6.2 and 7.2. In summary, TIPS are U.S. Treasury bonds issued in 5, 10, and 30-year durations that are designed to provide protection against inflation. They provide this inflation protection by adjusting the principal and interest rates of a regular U.S. Treasury bond by the annual inflation rate, measured by the Consumer Price Index (CPI). TIPS will typically outperform similar duration Treasury bonds when inflation is positive, and under perform Treasuries when inflation is negative (called deflation).
TIPS are likely to perform particularly well when inflation rises unexpectedly and rapidly. Although rapid inflation in the next few years is not expected for reasons discussed in Article 6.2, the inflation adjustment in TIPS provides an explicit hedge against the erosive effects of inflation on your ballast. In other words, we’ve seen that cash is a reasonable form of ballast, but only if it’s held for short durations, when the compounding effect of inflation is relatively small. But if inflation does unexpectedly soar during the vulnerable period, the real value of that cash could diminish rapidly in just a few years. So, TIPS directly protect ballast against such a scenario. Given we are looking for bonds in the 2 to 7-year duration range, 5-year TIPS or a 5-year TIPS ETF (such as the fund using the symbol TDTF) would fit well in the mid-term bucket.
I established in Article 8 that holding stocks for 5 years reduces your chances of a permanent loss to about 20% based on historical data. And holding stocks for 10 years further reduces the chance of a loss to around 12%. Further, as presented above, Wade Pfau found that the worst historical performance for stocks over a 5-year period was about -12% and over a 10-year period was just -1% (hardly any loss at all). I also reviewed in Article 8 some popular wisdom indicating stocks should be held for no less than 10 years. I can see why people quote this 10-year threshold, because I think most of us would be reluctant to give advice that has a 1-in-5 chance of failing but would be relatively comfortable giving advice based on a 1-in-10 chance of failure. A 10% chance also starts to get into the realm of statistical certainty.
Stock Risks – The above observations suggest expanding the short and mid-term buckets so they encompass up to 10 years of spending. But once again, mindfulness provides a fresh perspective on these failure probabilities in general. Looking at the recovery data for stocks, just like we did for bonds, starts to shed some mindful light on the 10-year threshold for stocks. This graph uses the same New York University data compiled by Aswath Damodaran for stocks as was presented above for bonds.
For stocks, we can’t make either of the two assertions we made for bonds. First, stocks underwent multiple instances of back-to-back annual negative returns spanning 2 to even 4 years. Second, the data don’t support that the subsequent years’ positive returns always quickly negated the prior losses.
I examined stock history more closely using Robert Shiller’s stock data, which covers a period all the way back to 1871. Here are the six worse stock market collapses during that entire period and the time it took for those stocks to return to their former value (without inflation adjustment) and assuming dividends are reinvested annually.
|Start of Stock Crash||
|Mar. 1937||May 1946||Jan. 1973||Mar. 2000||
|Years to Full Recovery||
Each of these calculated recovery periods assumes that stocks were bought at the peak just before the crash, which is the worst possible bad luck in each case. The average recovery period across the six worst stock events in history is 8.2 years. If you take out the whopper crash of 1929, the average is 6.8 years. These recovery periods would be shorter if we instead assumed that dividends were reinvested quarterly, which is the typical dividend payout period in the U.S. Conversely, the recovery periods would be somewhat longer if the stock values were inflation adjusted.
Stock Amounts – Even assuming perfectly bad-luck timing and less aggressive dividend reinvestment, holding stocks for 10 years guards against all but the very worst historical stock catastrophes. And as the graph from Wade Pfau discussed above shows, the worst historical return for holding stocks for 10 years was a mere -1%. From a mindful perspective, using a 10-year threshold seems like more than a fair bit of needless worrying about a small potential loss. Using the 4% constant withdrawal assumption, a 10-year threshold for the long-term stock bucket means only about 60% of your portfolio would be in stocks. That relatively small amount of stocks will severely under perform a 100% stock portfolio during most long-term scenarios, as I repeatedly demonstrated throughout Articles 6 through 8. And as I pointed out in Article 8.2, evidence like the Estrada study indicates that holding 100% stocks is a very competitive retirement strategy both in terms success rates and terminal portfolio values. Consequently, a mindful perspective leads us to a long-term bucket threshold at something like 5 years, rather than 10 years.
Now that we know the mindful contents and approximate amounts (time spans) for each bucket, Article 8.4 puts it all together into a mindful bucket investing plan for the “old” investor.