A Simple Dynamic Strategy for Portfolios Taking Withdrawals: Using a 12-Month Simple Moving Average

by Michael M. Garrison, CFP®; Carlos M. Sera; and Jeffrey G. Cribbs, CFP®


Executive Summary

  • This paper examines the long-term effects of using a dynamic investment strategy based on a 12-month simple moving average for portfolios in both the accumulation and withdrawal phases. It compares the results of this dynamic strategy to standard static portfolio allocations based on Modern Portfolio Theory to determine whether such a strategy is optimal.
  • We created six portfolios using historical data (1926–2008) from two asset classes: U.S. large-cap equities and U.S. intermediate-term government bonds. The six portfolios were examined for 3, 4, 5, 6, and 7 percent withdrawal rates over 30- and 40-year annual rolling periods from 1927–2008.
  • For the first year, we took the beginning portfolio value and subtracted the first year withdrawal, then added its return for the year. For each subsequent year we adjusted the withdrawal for inflation. All withdrawals occurred at the beginning of the year.
  • During the withdrawal phase, the dynamic portfolio produced the highest initial safe withdrawal rate, greatest probability of success across all withdrawal rates, and consistently highest terminal values compared to the static portfolios.
  • The differences in terminal values were considerable.
  • For long-term investing periods (more than 10 years), the dynamic portfolio earned comparable rates of return to a 100 percent equity portfolio with less risk, as measured by maximum drawdown. As the investing period increased, the dynamic portfolio tended to outperform the 100 percent equity portfolio.
  • This paper concludes that a dynamic asset allocation using a 12-month simple moving average is a consistently better strategy over statically allocated portfolios for investors in both the accumulation and withdrawal phases.


Michael M. Garrison, CFP®, is a principal at Chicago Wealth Management Inc. and a portfolio manager of the CWM Fund LP. He holds a finance degree from Indiana University's Kelley School of Business.

Carlos M. Sera is a managing principal at Chicago Wealth Management Inc. and a portfolio manager of the CWM Fund LP. He holds an undergraduate degree from Johns Hopkins University and an MBA from the University of Rochester.

Jeffrey G. Cribbs, CFP®, is a managing principal at Chicago Wealth Management Inc. and a portfolio manager of the CWM Fund LP. He holds an undergraduate degree from Johns Hopkins University and an MBA from Carnegie Mellon University.


The life of an investor can typically be broken up into two distinct phases: the accumulation phase and the withdrawal phase. For individuals, the withdrawal period typically starts at retirement or soon after. For institutions, there is no phase distinction, because withdrawals are an ongoing concern.
 
While investing for accumulation has been studied for years, withdrawal planning is a relatively recent topic. With a large number of investors worried about how much they can withdraw from a portfolio without running out of money, maximizing this initial withdrawal rate has been the subject of much new research.
 
In the last few years, the first wave of baby boomers has started to retire. Unfortunately for them, their retirement has coincided with a more than 50 percent drop in the S&P 500 from November 1, 2007, through February 28, 2009 (based on month-end data). In addition, over the last 83 years, the 10-year period ending in 2008 was the worst 10-year rolling annual period for the S&P 500, with a cumulative return of –13 percent. Notwithstanding the direct effect of losses on their portfolios, many investors cannot psychologically stand these losses and make poor timing decisions, leading to potentially lower returns.
 
Financial advisers, investment consultants, and institutional portfolio money managers have started to really focus on withdrawals and the question asked by many investors: How much can I safely take? Our investment advice and strategies are one of the primary reasons we are hired in the first place. Ideally, the investment and withdrawal strategies we develop will provide our clients with the ability to live comfortably in retirement and afford them peace of mind.
 
Market losses have different consequences in the accumulation and withdrawal periods. In the accumulation phase, market losses can be mitigated by longer investing time horizons, as long as these losses do not cause an investor to abandon an effective long-term strategy. However, in the distribution phase, market losses can be devastating. As William Bengen pointed out in his article "Determining Withdrawal Rates Using Historical Data" (1994), basing a withdrawal rate on a portfolio's historical average rate of return instead of looking at the actual order of historical returns can inflate the safe withdrawal percentage to the investor's detriment. An initial period of negative returns coupled with high inflation is likely to cause a portfolio withdrawal strategy based on averages to run out of money much sooner than one based on the historical order of returns. Investors in the withdrawal phase should keep this in mind when developing a withdrawal strategy.
 
To date, the majority of research on withdrawals has focused on two areas: the optimization of static asset-class allocations to improve a sustainable withdrawal rate and the incorporation of withdrawal decision rules based on either portfolio performance or market indicators. These are two of the three obvious approaches that may improve the withdrawal rate. The three are:

  1. Improve the asset allocation strategy by incorporating additional asset classes and/or optimizing portfolio weightings
  2. Improve the withdrawal strategy by setting up distribution rules for the investor to follow during the withdrawal phase
  3. Improve the asset allocation strategy by using a dynamic investment approach based on a formula or technical market indicators

Investment research has shown that for the S&P 500, using a simple moving average methodology can produce comparable returns to a buy-and-hold strategy with lower maximum losses (Faber 2007). Given the effect large losses have on portfolios in the withdrawal phase, would an investment strategy mitigating these losses be beneficial during the distribution period?
 
The purpose of this paper is to introduce a dynamic asset allocation strategy using a simple moving average system as an alternative to the static investment approach for both the withdrawal and accumulation phases of an investor's life. The simple moving average system is explained in detail later in this paper.

Classic Research

The pioneer of withdrawal rate research is William Bengen. As noted above, his article "Determining Withdrawal Rates Using Historical Data" examined asset allocation strategies based on actual historical returns, not averages. His article sheds light on the fact that the order of returns is a very important component for determining whether a withdrawal rate is sustainable. Using average returns can inflate the actual initial safe withdrawal rate, leading an investor to run out of money. He found a withdrawal rate of approximately 4 percent, increased subsequently by inflation, would be appropriate for most retirees. If they keep equity allocations between 50 percent and 75 percent, there is little fear of running out of money.

Asset Allocation Improvement. In additional research, Bengen (1997) determined that adding small-cap U.S. equities to a portfolio experiencing withdrawals was beneficial, increasing the safe withdrawal rate from 4.1 percent to 4.3 percent for portfolios in the distribution phase. The question of whether international equities improved portfolios in the distribution period was addressed by Cooley, Hubbard, and Walz (2003), who found that "retirees who prefer portfolios of at least 50 percent equities benefit modestly from including EAFE stocks as 25 percent of the market value of their portfolios."
 
Withdrawal Strategies. Adding to Bengen's withdrawal rate findings, research by Guyton (2004) and by Guyton and Klinger (2006) focuses on enhanced withdrawal strategies. They implemented a series of decision rules based on factors such as portfolio value, portfolio returns, and inflation. Using their guidelines, they have been able to increase the initial safe withdrawal rate. However, we find some of their rules to be potentially restrictive and wonder if clients would be willing (or able) to receive lower distributions during bad market periods. We also question the effects of adding even more withdrawal decision rules. Is it realistic to assume these would be followed?
 
Kitces's 2008 paper looks at macroeconomic factors such as current market P/E ratios to determine the safe withdrawal rate for a recent retiree. It's our hope that after a huge drop in the market—resulting in a low P/E ratio and suggesting a higher initial withdrawal rate—a new retiree or the retiree's financial adviser would feel comfortable with some of the withdrawal rates suggested, but we're concerned that this may be hard to achieve because of the psychological impact that a market drop, and the resulting loss in portfolio value, has on an investor.
 
Dynamic Allocation Research. Blanchett's article on glide paths (2007) presents four dynamic allocation approaches with different starting equity and bond percentages. He concluded that a static 60 percent equity and 40 percent bond allocation is appropriate for most retirees. However, his dynamic approaches are not based on market conditions or technical indicators.

The Portfolios

Using month-end historical return data from the Ibbotson SBBI 2009 Classic Yearbook starting in 1926, we examined six portfolios:

  • 0/100 portfolio: The portfolio is allocated 100 percent to U.S. intermediate-term government bonds.
  • 40/60 portfolio: 40 percent allocated to U.S. large-cap equities and 60 percent allocated to U.S. intermediate-term government bonds. The portfolio is rebalanced annually on January 1.
  • 60/40 portfolio: 60 percent allocated to U.S. large-cap equities and 40 percent allocated to U.S. intermediate-term government bonds. The portfolio is rebalanced annually on January 1.
  • 80/20 portfolio: 80 percent allocated to U.S. large-cap equities and 20 percent allocated to U.S. intermediate-term government bonds. The portfolio is rebalanced annually on January 1.
  • 100/0 portfolio: The portfolio is allocated 100 percent to U.S. large-cap equities.
  • 12-month simple moving average (SMA) portfolio: The portfolio is allocated 100 percent to U.S. large-cap equities in every monthly period in which the last month's ending price is above the average price for U.S. large-cap equities for the previous 12 months, based on month-end prices. For example, if the closing price of U.S. large-cap equities on January 31, 2000, is 100, and the average month-end price for the period February 1, 1999, through January 31, 2000, is 98, then the portfolio would be 100 percent allocated to U.S. large-cap equities for the month of February.

The portfolio is allocated 100 percent to U.S. intermediate-term government bonds in every monthly period in which last month's ending price is below the average price of U.S. large-cap equities for the previous 12 months, based on month-end prices. Continuing the above example, if the closing price of U.S. large-cap equities is 100 on February 29, 2000, and the average month-end price for the period March 1, 1999, through February 29, 2000, is 102, then the portfolio would be 100 percent allocated to U.S. intermediate-term government bonds for the month of March.

Methodology

The six portfolios were examined for 3, 4, 5, 6, and 7 percent withdrawal rates over 30- and 40-year annual rolling periods from 1927–2008. We did not start in the year 1926 because the 12-month SMA portfolio needs one year of returns to calculate the 12-month moving average. We also did not include withdrawal rates of more than 7 percent, because we have not seen any credible research showing this is a sustainable withdrawal rate for any portfolio strategy over a 30-year (or more) time frame.
 
Our methodology for determining year-end portfolio values, and thus the ability of a portfolio to survive over a given period, is similar to that of other researchers in this area who have focused on distributions. For the first year, we took the beginning portfolio value and subtracted the first year withdrawal, then added its return for the year. For each subsequent year we adjusted the withdrawal for inflation. All withdrawals occurred at the beginning of the year.
 
As an example, given a $1 million, 100 percent equity portfolio, 15 percent return, and a 5 percent withdrawal rate, the first year ending portfolio value is $1,092,500. ($1 million less the 5 percent withdrawal ($50,000) yields $950,000, which is multiplied by the 15 percent return, yielding $1,092,500.)
 
No stochastic analysis (for example, Monte Carlo analysis) was performed. The order of returns, as it applies to using a simple moving average system, is very important. Past returns are the cornerstone of a moving average system. Faber (2007) demonstrates that a simple moving average system picks up trends in the marketplace and, although it's not a perfect indicator of when to be invested in the market (wouldn't that be nice!), it provides reasonable signals on which an investor can act. It is implied that market returns may not be random. There is a growing body of evidence showing an investment's past performance can somewhat predict future performance relative to other investments. Dimson, Marsh, and Staunton (2008) have shown this persistency to occur with individual stocks. If stocks and markets do have persistency, we wonder how this will affect the use of Monte Carlo analysis in the investment management realm.
 
As this paper was written toward the end of 2009, we have only included distribution and return information through 2008. However, the last two signals of the 12-month SMA approach were a move to bonds in February of 2008, returning to equities in August of 2009. Although not giving a perfect signal, the 12-month SMA approach would have avoided the massive market losses from 2008 through the beginning of 2009.
 
Fees and Expenses. We excluded transaction costs and underlying investment expenses. From 1927 through 2008, there were a total of 102 signals generated by the 12-month SMA approach (including the initial trade). This equates to approximately two-and-a-half trades per year (every signal after the initial trade generates two trades—one in or out of the S&P 500 and one in or out of bonds). In comparison, there were 82 rebalancings (including the initial allocation) for an annually rebalanced portfolio over this same period, creating 2 trades per year. With the availability of exchange-traded funds (ETFs) that can be bought and sold today for less than $20 per trade from many custodians and brokerage firms, an investor with a $1 million portfolio would average less than 1 basis point per year in transaction costs.
 
SPY (an ETF for the S&P 500) has an underlying investment expense ratio of 10 basis points and is very liquid with the bid/ask spread averaging 1 basis point. The frequency and low costs of trading combined with the low ETF expense ratio make these expenses immaterial.
 
Management fees will lower the sustainable distribution rate, decrease the probability of any given portfolio strategy being successful, and decrease portfolio terminal values. Clark and Hood (2009) have shown that decreasing the percentage of equities in a portfolio has a disproportionately negative effect on the probability of portfolio success and portfolio terminal values. Of the portfolios examined, the 12-month SMA portfolio produced the highest and most consistent returns, and will be the least-affected portfolio for any given management fee. While we recognize that management fees will result in lower outcomes across all portfolios, they can vary over a broad range based on investment adviser and account size, thus they have been excluded.
 
In line with other research on this topic, we have excluded the effect of taxes. For taxable accounts, the SMA approach may create higher realized gains in some periods, compared to the static portfolios. However, there are three factors that mitigate the potential taxes of the 12-month SMA approach versus the static portfolios:

  1. The 12-month SMA portfolio has higher average pretax investment returns than the other portfolios
    (from 1927–2008)
  2. All portfolios have tax consequences when portions of the portfolios are being sold for distributions
  3. There are periods in which the 12-month SMA portfolio is 100 percent invested in stocks and will generate long-term capital gains, resulting in lower tax rates than a portfolio with a fixed income allocation

Withdrawal Phase Analysis

Before we get into our analysis, let's address the question of what defines a successful portfolio strategy. We considered these four questions:

  1. What was historically the highest maximum withdrawal rate a portfolio strategy could experience without running out of money in a given period?
  2. Given a withdrawal rate, which portfolio strategy had the highest probability of success?
  3. In periods during which a portfolio strategy failed, what was the minimum number of years the portfolio survived?
  4. What was the terminal value of the portfolio after the withdrawal period ended?

A portfolio strategy with the highest historical maximum withdrawal rate, the highest probability of success given a withdrawal rate, the longest minimum number of years lasted in periods of failure, and the highest terminal values is considered optimal.

Safe Withdrawal Rate

For purposes of this paper, we define the safe withdrawal rate as the initial maximum withdrawal rate adjusted annually for inflation that has historically not run out of money over a given period. Since life expectancies are increasing, we look at both 30- and 40-year periods. It is important to note that unfortunately, safe does not mean guaranteed. There are many events that can affect the actual safe withdrawal rate in the future. We do, however, think the historical safe withdrawal rate is a reasonable starting point to begin determining distribution rates.
 
As we can see from Table 1a, our research for a 60/40 portfolio is in line with that of others who have researched the subject, with a withdrawal rate being historically safe at approximately 4 percent for the balanced portfolios (40/60, 60/40, and 80/20). In this analysis, the 60/40 portfolio did not have the highest historically safe withdrawal rate. This distinction goes to the 12-month SMA portfolio, which provided the maximum safe withdrawal rate in both the 30-year (4.37 percent) and 40-year (4.11 percent) periods. Over 30-year periods, the 12-month SMA portfolio allowed for a withdrawal rate 8.4 percent greater than the 60/40 portfolio, and over 40-year periods, the 12-month SMA portfolio's withdrawal rate was more than 9.9 percent higher than the 60/40 portfolio. Although not shown, the 12-month SMA portfolio also had the highest safe withdrawal rate among portfolios with fixed income allocations (or the ability to have fixed income) if we look for the portfolio that allowed for the highest safe withdrawal rate during any annual 30-year period.


 
In the modern era (since 1945), the difference between the 12-month SMA portfolio and the other portfolios is even more pronounced. Comparing Table 1a to Table 1b, we see the maximum safe withdrawal rate of all the other portfolios either stayed the same or increased modestly, while the safe withdrawal rate for the 12-month SMA portfolio increased by more than 16 percent and 18 percent, for 30- and 40-year rolling periods, respectively.


 
Although we consider maximum safe withdrawal rate to be an important consideration to determine a portfolio strategy, by itself it does not make a portfolio strategy optimal.

Probability of Success

The portfolio success rate is the probability a portfolio does not run out of money given the withdrawal rate and time frame.
 
Table 2a illustrates portfolio success rates for 30 scenarios based on six different portfolios and five different withdrawal rates. The portfolios highlighted in green are the portfolios that provided the greatest probability of success given each withdrawal rate. As we can see, all portfolios in our sample with stock exposure survived a 3 percent withdrawal rate from the period 1927–2008. At 4 percent, the two portfolios with static stock exposure of at least 80 percent were not 100 percent successful.


 
The 5 percent withdrawal rate is where Table 2a gets interesting. The 40/60, 60/40, 80/20, and 100/0 portfolios all had huge drop-offs in success rates, with each of these portfolios being successful less than 75 percent of the time. Although its probability of success did drop, the dynamic portfolio using a 12-month simple moving average fell considerably less, down to only 96 percent. Additionally, as withdrawal rates increased to 6 percent and 7 percent, the static portfolios with the most equity exposure, as well as the 12-month SMA portfolio (which has the ability to go to 100 percent equity), were the most successful. In every historical withdrawal rate scenario, the 12-month SMA portfolio was dominant.
 
Analyzing 40-year periods, the 12-month SMA portfolio again demonstrates it had the highest probability of success for all withdrawal rate scenarios. As the period examined increased from 30 to 40 years, the 4 percent withdrawal rate was where the 12-month SMA portfolio starts to differentiate itself, whereas in 30-year periods this did not manifest until the 5 percent withdrawal rate. Again, we can see how the portfolios with the higher percentages in stocks (or the ability to have a high percentage in stocks) were more successful as withdrawal rates and time periods increased, suggesting that investors with longer time horizons or higher withdrawal needs should limit bond exposure.

Years Until Failure

While historical portfolio success rate can be a good measure to evaluate different portfolio strategies, it does not tell the entire story. Investors are also concerned about how long their money will last in periods of failure. For example, it may not be an ideal strategy if a portfolio is successful 95 percent of the time in lasting 30 years, but only lasts five years in the 5 percent of the time when the portfolio strategy fails—especially when compared to a portfolio strategy with an 80 percent success rate of lasting 30 years, with money lasting a minimum of 27 years in the periods the portfolio fails. Investors concerned about running out of money would more likely adopt a strategy in which, even when it fails, they would have 27 years before their portfolio is exhausted.
 
As 100 percent equity and 100 percent bond portfolios have been shown by previous research to be inefficient strategies during the withdrawal phase, we have excluded them from our analysis. The figures that folllow focus on the dynamic and balanced portfolios. From Table 2a, we saw the most striking difference in probability of success between balanced portfolios and the dynamic portfolio occurred at the withdrawal rate of 5 percent, and thus this is the withdrawal rate illustrated.
 
Figure 1a shows the years money lasted at a 5 percent withdrawal rate for the 12-month SMA portfolio. The first bar on the left represents an investor starting the withdrawal phase in 1927 and ending in 1956. As we can see in Figure 1a, the 12-month SMA portfolio failed in the 30-year periods beginning in 1937 and 1939. This makes intuitive sense to those who have studied monthly market data, because there was extraordinary monthly volatility during the beginning of this era. There were 21 months during the period from January 1937–June 1940 in which U.S. large-cap equities were either up or down more than 5 percent, and in nine of those months, large-cap equities were either up or down more than 10 percent (four months down over 10 percent and five months up more than 10 percent). This led to poor signals for the 12-month SMA portfolio. In spite of this poor signaling early on, the 12-month SMA portfolio still lasted a minimum of 17 years for the period starting in 1939.


 
Similar graphs of 40/60, 60/40, and 80/20 portfolios at the 5 percent withdrawal rate illustrate what has been shown in Table 2a: The 40/60 portfolio was successful approximately 57 percent of the time, while the balanced portfolios with greater than 50 percent equity had success rates in excess of 70 percent. Illustrated by Figure 1b, 1c, and 1d, the minimum number of years the 40/60, 60/40, and 80/20 portfolios survived was 20, 19, and 18, respectively.


 
An important observation from Figures 1a–1d is that, although the 12-month SMA portfolio minimum survival was 17 years, it only ran out of money in 20 years or fewer once, while this occurred more than four times for each of the balanced portfolios. Focusing on the postwar era (starting in 1945), there have been no annual 30-year rolling periods in which the 12-month SMA portfolio did not survive for at least 30 years.

Terminal Value

Many investors would like to leave a legacy to people or institutions they care about. So although the primary goal of many investors is to not run out of money, the ability to leave the maximum amount of wealth possible to heirs or charitable organizations is also of great concern. If two strategies can make an investment last during the withdrawal phase, but one of the strategies has no terminal value and the other has double the initial investment remaining, which strategy will the investor choose?
 
The upper-left section of Table 3 shows the 30-year annual rolling average ($1,547,361), median ($921,086), minimum ($0), and maximum ($5,766,712) ending values for the 0/100 portfolio at a 3 percent withdrawal rate. For each withdrawal rate, the numbers highlighted in green are the highest average, median, minimum, and maximum values. Again, the 12-month SMA portfolio came out on top having the highest average, median, minimum, and maximum ending values compared with the other portfolio strategies across all tested withdrawal rates. We were somewhat surprised that the 100/0 portfolio did not have the highest average, median, or maximum terminal values for any withdrawal rate scenario.


 
Let's examine the details more closely. For reasons stated previously, we focus on the terminal values for the 12-month SMA portfolio, 60/40 portfolio, and the 80/20 portfolio using 4 percent and 5 percent withdrawal rates. Because of its lower terminal values as compared to the other balanced portfolios, we excluded the 40/60 portfolio.
 
Analyzing the data in Figure 2a and 2b, using a 4 percent withdrawal rate, the 12-month SMA portfolio had the highest terminal value in 48 out of 53 30-year periods, while the 80/20 portfolio had the highest terminal value in the remaining 5 periods. At the 5 percent withdrawal rate, the 12-month SMA portfolio had the highest terminal value in 48 out of 53 periods, with the 80/20 portfolio having the highest terminal value in 4 periods. There is one period, starting in 1937, in which all portfolios failed, so the terminal value of each portfolio was $0.


 
The differences in terminal values between the balanced portfolios and the dynamic portfolio are striking. When the withdrawal rate is 4 percent, the 12-month SMA portfolio's terminal value is at least $2.5 million greater than the 60/40 and 80/20 portfolios in 49 periods and 45 periods, respectively. Examining a 5 percent withdrawal rate, there are 43 periods in which the difference between the 12-month SMA portfolio and the 60/40 portfolio is more than $2.5 million, and it keeps this $2.5 million terminal value difference in 42 periods when compared to the 80/20 portfolio.
 
Out of curiosity, we also tested the 12-month SMA portfolio and compared it to the 100/0 portfolio to see how it held up. When compared to the 100/0 portfolio, the 12-month SMA portfolio had a higher terminal value in 40 of 53 30-year periods when the distribution rate was 5 percent, as shown by Figure 2c. During the 1927–2008 period, the investor with a high risk tolerance looking to potentially maximize terminal value would have been better served investing in the 12-month SMA portfolio as opposed to the 100/0 portfolio.


 
The 12-month SMA portfolio had the highest average, median, minimum, and maximum terminal values compared to all of the portfolios tested. We found that these higher terminal values were not the result of a few periods of outperformance, and the difference in terminal values for the 12-month SMA portfolio was significant. For an investor looking to maximize the ending value of his or her portfolio, we conclude the 12-month SMA portfolio is optimal compared to the other portfolios.
 
What does this mean for the investor? We recognize that targeting a safe withdrawal percentage and never deviating from it is not realistic. Withdrawal rates will probably be lumpy, with the investor needing more than the safe withdrawal rate in some years. Completely restricting lifestyle because of a targeted withdrawal rate is not ideal, because investors may need to give up life experiences that are important to them. A strategy with consistently higher terminal values gives the investor more opportunities to enjoy life and deal with unplanned situations.

Sensitivity Analysis

Does the time frame of the moving average matter? We researched withdrawal rates using 10-, 11-, 13-, and 14-month moving averages in addition to the 12-month moving average presented. The results were very similar to that of the 12-month SMA portfolio relative to maximum safe withdrawal rate, probability of success, and terminal values. The number of years the money lasted fluctuated some. However, none of the tested moving averages lasted fewer than 14 years when the distribution rate was 5 percent.

Examining Failures of the Dynamic Approach

Compared to the statically allocated portfolios, the 12-month SMA approach consistently achieved superior results regardless of whether the S&P 500 was performing well or not. A period of high monthly return volatility (from high positive monthly returns to high negative monthly returns and vice versa) leading to a whipsawing effect, not long-term market trends, is the biggest potential weakness in the 12-month SMA approach. This extreme whipsawing in returns has not occurred since the beginning of the 1940s.
 
At the 5 percent distribution level, the two most relevant 30-year periods during which the balanced portfolios outperformed the dynamic portfolio were from 1937–1966 and 1939–1968. In the period ending in 1966, the 12-month SMA portfolio ran out of money after 21 years, equaling the 40/60 portfolio, but less than the 60/40 portfolio (23 years) and the 80/20 portfolio (24 years). During this period, if the investor had reduced his or her withdrawal rate to 4 percent, the 12-month SMA portfolio would have survived with a terminal value of more than $2.1 million, which was greater than all of the terminal values of the balanced portfolios except the 80/20 portfolio ($2.3 million).
 
In the period ending in 1968, the balanced strategies were even more dominant, using a 5 percent distribution rate. The 60/40 and 80/20 balanced portfolios succeeded, while the 12-month SMA portfolio and 40/60 portfolio failed. The 40/60 portfolio survived five years longer than the 12-month SMA portfolio. The cause of this failure was the extreme market volatility from January 1937–June 1940. As stated previously, there were 21 monthly periods with returns of at least +/– 5 percent, and nine of those were +/– 10 percent. This caused the dynamic approach to receive mixed signals and be in and out of the market at the wrong times. However, these were the only 2 periods in which the 12-month SMA portfolio failed, while the 40/60, 60/40, and 80/20 portfolios failed in 23, 15, and 14 periods, respectively. With greater liquidity in financial markets combined with more government controls in place now than in the 1930s, a long period of high volatility may be less likely to occur. Looking at the modern era (starting in 1945), the 12-month SMA portfolio had 0 failures at a 5 percent withdrawal rate while the three balanced portfolios all failed in more than 10 30-year periods.

What About the Accumulation Phase?

Although we believe we have presented persuasive evidence for using a dynamic approach during the portfolio distribution phase, the question still remains whether it is an appropriate strategy for investors to use in the accumulation phase. For long-term investors not taking withdrawals, higher equity percentages typically lead to higher returns. How does our dynamic 12-month SMA portfolio compare to the 100/0 portfolio for long-term periods?
 
Figure 3 shows the compounded annual rate of return an investor would have earned from investing in the 100/0 portfolio compared to the 12-month SMA portfolio, depending on the year the investment was started (through 2008). Moving left to right in Figure 3, an investor starting in 1927 would have earned a 9.6 percent compounded annual return by buying and holding U.S. large-cap equities until the end of 2008 (82 years). Investing in the 12-month SMA portfolio, that same investor would have earned an 11.9 percent compounded annual rate of return over the same period. The next pair of bars on Figure 3 shows the period 1928–2008 (81 years), followed by 1929–2008 (80 years), and the final bars show the returns since 1999 (10 years). We excluded periods of fewer than 10 years.


 
The results are remarkable. Historically, in every year since 1927, an investor using the 12-month SMA portfolio until 2008 would have achieved a higher rate of return than if the investor would have bought and held the 100/0 portfolio.
 
This may be a bit unrealistic, as most investors do not have an unlimited time horizon giving them the ability to buy and hold an investment for an infinite period. In addition, the huge market drop in 2008 overstates the 12-month SMA portfolio's advantage, so we broke down the returns into five-year increments, starting in 1945.

Depending on the time frame, Table 4 shows us there are periods in which a buy and hold in the 100/0 portfolio outperformed the 12-month SMA portfolio. For example, an investment in the 100/0 portfolio earned a 10.7 percent annual return from the beginning of 1945 to the beginning of 1950, while the 12-month SMA portfolio earned 9.7 percent during the same period. However, a key observation from this table is that the compounded annual rate of return for the 12-month SMA portfolio improves versus the 100/0 portfolio as the holding period increases. For investors (such as institutions) with long to infinite time horizons, the 12-month SMA portfolio has a noticeable advantage over the 100/0 portfolio.

A Quick Word About Risk

In the investment community, it is generally assumed that investors with longer time horizons should be willing to take more risk, with risk measured by portfolio standard deviation. In our examples above, we assume the investor will have the internal fortitude to execute a buy-and-hold methodology and will be able to follow a portfolio strategy under any circumstance. We have found that this is not always the case. When the market continually decreases, investors get scared. Even if they know they are using a portfolio strategy that has proven to work over time, they just cannot stick with it.
 
Maximum drawdown (or maximum loss), is in our opinion, the most appropriate measure of risk to consider when analyzing a portfolio. Maximum drawdown is the highest percentage loss a portfolio experiences over a given time period.
 
Table 5 illustrates the maximum drawdown each of these three portfolios experienced from 1945 through 2008, using month-end data. (Please note that the maximum drawdown for the 100/0 portfolio would exceed 50 percent if this table included returns through June 2009.) The implications of maximum drawdown are far-reaching. Would an accumulation phase investor stick to the 100/0 portfolio even after experiencing a loss greater than 44 percent? And after a significant portfolio loss, would the investor be able to invest subsequent funds using the same strategy?


 
Table 4 shows that, in some periods, the 100/0 portfolio has a better compounded annual rate of return than the 12-month SMA portfolio. However, a 100/0 portfolio is clearly not superior, and probably will experience a loss of almost double that of the 12-month SMA portfolio. We argue the 12-month SMA portfolio would help mitigate the risk of an investor switching his or her investment strategy at inopportune times because of the lower maximum losses it has historically incurred. When incorporating risk into the discussion for the accumulation investor, a dynamic 12-month simple moving average strategy should definitely be considered.

Implementation

It is very easy for investors to use a simple moving average in their portfolio management. The only rule that must be followed is to be in the market when the last month's ending price is above the average price of the market over the last 12 months, and be in the dedicated "safe" asset class when the previous month's ending price is below the 12-month moving average. There are no complex withdrawal rules or tactical allocation procedures to follow.

Further Research

We believe the investment community could benefit from future research related to dynamic allocations using simple moving averages. Using this technique across other asset classes may lead to better portfolio performance.
 
Depending on additional research documenting investment return persistency, the use of stochastic analysis or random return generators for investment return results should be further questioned. The results we found just using a simple moving average method suggest that the order of returns may not be random.
 
There is still a lot of space for research using other technical market indicators and dynamic portfolio approaches, as well as combinations of multiple strategies. We look forward to reading other research that has yet to be done in these areas.
 
The use of target-date retirement and lifecycle funds has grown considerably over the past few years, with assets expected to be more than $1 trillion by 2015 (Marcks 2008). We question how, by nature of an investor's birthday, these funds can force an investor into a lower equity percentage that may occur after a large market loss. This eliminates the possibility of recovery from the equity assets that have been shifted to bonds. We have shown how, as the equity percentage decreases, the probability of being able to maintain higher withdrawal rates also declines, potentially destroying an investor's retirement. We call on researchers in the investment management and financial advising community to illustrate the potential for wealth-destroying effects of these funds.

Conclusion

From 1927–2008, the 12-month moving average investment method has been a better strategy for investors requiring distributions compared with statically allocated portfolios. For many periods, it has been superior for the accumulation phase investor as well. Financial advisers who provide investment management for their clients should examine ways to employ this methodology in their recommendations. In doing so, they likely will better serve their clients.


References

Bengen, William P. 1994. "Determining Withdrawal Rates Using Historical Data." Journal of Financial Planning (October): 171–180.

Bengen, William P. 1997. "Conserving Client Portfolios During Retirement, Part III." Journal of Financial Planning (December): 84–97.

Blanchett, David M. 2007. "Dynamic Allocation Strategies for Distribution Portfolios: Determining the Optimal Distribution Glide Path." Journal of Financial Planning (December).

Clark, John M. and Matthew Hood. 2009. "Policy Implications for Modeling the Next Generation of Target Date Funds." Journal of Investing (Fall): 53–61.

Cooley, Phillip L., Carl M. Hubbard, and Daniel T. Walz. 2003. "Does International Diversification Increase the Sustainable Withdrawal Rates from Retirement Portfolios." Journal of Financial Planning (January).

Dimson, Elroy, Paul Marsh, and Mike Staunton. 2008. ABN AMRO Global Investment Returns Yearbook 2008. ABN AMRO Bank NV in association with The Royal Bank of Scotland and London Business School.

Faber, Mebane T. 2007. "A Quantitative Approach to Tactical Asset Allocation." Journal of Wealth Management (Spring): 69–79.

Guyton, Jonathon T. 2004. "Decision Rules and Portfolio Management for Retirees: Is the 'Safe' Initial Withdrawal Rate Too Safe?" Journal of Financial Planning (October).

Guyton, Jonathon T. and William Klinger. 2006. "Decision Rules and Maximum Initial Withdrawal Rates." Journal of Financial Planning (March): 50–58.

Kitces, Michael P. 2008. "Resolving the Paradox—Is the Safe Withdrawal Rate Sometimes Too Safe?" The Kitces Report (May): www.kitces.com/assets/pdfs/Kitces_Report_May_2008.pdf. Retrieved 9/4/09.

Marcks, Christine C. "Strengthening Target-Date Funds with Guarantees to Enhance Retirement Security." www.prudential.com/media/managed/Target_Date_Fund_with_Income.pdf. Retrieved 9/4/09.