Asset Allocation Over Your Lifetime

The Importance of Asset Allocation

It is well-known that asset allocation accounts for over 90% of the variability in returns among portfolios. It has been shown that for well-diversified portfolios (which are the only ones that I would recommend), the major systematic risks of portfolios, not their stock or bond selections, account for nearly all of the return. My research has found that four broad risk factors will capture most of the systematic risk in any portfolio:

• Stock market risk (MKT), as measured by the S&P 500 Index
• Interest rate risk (LTB), as measured by the 10-Year Treasury Benchmark Index
• Currency risk (DLR), as measured by the U.S. Dollar Index
• Commodity risk (OIL), as measured by the West Texas Intermediate Crude Oil Index

Of these four, stock market risk is by far the most important, typically accounting for over 90% of portfolio risk. As I often tell clients, the single most important decision that every investor must make is how much of their portfolio to allocate to stocks. 

Given the monumental importance of that decision, how do you make it? That is the subject of this article. 

Target Date Funds

Most people accumulate most of their retirement savings in defined contribution (401k, 403b, 457) plans sponsored by their employers. And most of them use target date funds (TDFs) that automatically allocate their investments broadly for them on a “glidepath” chosen based on their anticipated retirement date. Vanguard, the largest manager of retirement plans, reports that 96% of their plans offer such funds for their five million plan participants. 83% of Vanguard’s plan participants use TDFs, and 71% have their entire account invested in a single fund. TDFs have exploded and taken over the lion’s share of retirement investments, not just at Vanguard, but across the board. 

Target date funds have become the preferred default investment option by most retirement plan sponsors. The federal government has strongly encouraged this. The Pension Protection Act of 2006 created a “safe harbor” for employers if they use an approved “Qualified Default Investment Alternative” (QDIA). This investment alternative is described as a portfolio that provides “long-term appreciation and capital preservation through a mix of equity and fixed income exposures based on the participant’s age.” That language could hardly be more specific in describing TDFs. No wonder that almost all employers include them in their retirement plan offerings as the automatic default investment for employees.

When I work with new clients, I use the largest target date fund providers as an asset allocation benchmark. The recommended asset mixes of the TDFs among the largest asset management companies are actually quite similar for each target date cohort (2020, 2025, 2030, etc.). I consider the average of the largest TDFs to be the “center of gravity” for the average investor for each retirement date cohort. I compare the new client’s asset mix and risk exposures with the TDFs, and if they are far away from the average, I generally try to persuade them to move closer to the average in the absence of reasons to do otherwise. 

Glidepaths

Target date funds follow a “glidepath” that starts with a very high allocation to stocks (often 80%-90%) for younger investors. The glidepath reduces equity allocations each year, often starting at around age 40 or 45. At retirement, often assumed to be age 65, the equity allocation tends to be close to 50%. TDFs continue to reduce equity allocations past retirement until reaching a minimum, often 25% or 30%. 

Human Capital

Reducing stock market exposure as age increases makes sense because of changes in “human capital” that take place over time. Early career individuals have a lot of “human capital,” which is the sum of their expected future earnings. This human capital is bond-like in that most people earn a fairly steady income, more-or-less indexed to inflation, over their working life. This allows them to invest a high percentage of their investment capital in risky assets, particularly stocks, early in their career. Human capital decreases with time, and investment capital increases, so lowering the riskiness of invested capital would keep overall capital (personal capital + investment capital) stable and balanced.  

Retirement is a critical point. Until retirement, it may be possible for people to increase their earnings by working harder (taking a second job), working longer hours (overtime), or delaying retirement. This may allow them to adapt to a period of poor investment returns and thus enable them to accept more investment risk before retirement. At retirement, this adaptability disappears (for the most part). Thus, at retirement and thereafter, a lower investment risk profile is prudent to reflect this decreased flexibility.

Sequence of Returns Risk

For most people, the first day of retirement is the riskiest day of their life from a financial standpoint. They have probably just shut the door on their ability to earn income. Their retirement nest egg is probably the biggest it will ever be. For most people, there is no pension to support them. They must live on their retirement savings and their Social Security benefits. Consequently, much of their standard of living in retirement will depend on their investment returns.

What matters is not only the average return of their portfolio over, say, the 30 years of their retirement. Most important are the crucial first years, perhaps the first 5, 10, or 15 years, of portfolio returns. If their returns are poor during that critical early period, their portfolio may never recover. This risk is known as “sequence returns risk.” 

Most retirees need to draw upon their retirement portfolio to cover their living expenses. They must draw down their portfolio over time, shrinking its size. Even if returns improve after the first 10 or 15 years, the portfolio may be a lot smaller by then. And by then, if they have experienced poor investment returns, they likely would have had to cut back on their living expenses to lower the risk of running out of money later on. 

Consequently, there are two very solid and theoretically well-justified reasons why people should reduce the risk in their portfolios at retirement: 1) because their human capital will plummet to zero, and 2) because their exposure to sequence of returns risk will be at its maximum.

Age and Risk Aversion

There is another reason why many people prefer to reduce their risk as they get older: because they become more risk-averse psychologically. Based on both survey data and actual investor behavior, it is clear that age related increases in risk aversion is a reality for many people. The question is the extent to which it is justified. How much is prudent caution and how much is irrational fear? No doubt both are present.

Target date funds’ glidepaths imply an extremely high level of increased risk aversion for those about to retire. TDFs reduce their stock allocations dramatically from early career to retirement. Studies indicate that in order to justify the degree to which TDFs reduce stock percentages, investor’s risk aversion would have to more than double between early career and retirement. While there is some evidence for increased risk aversion with age, its magnitude is unclear. 

Expected Holding Period

There is a final reason why many people decrease their portfolio risk with age: the notion that as their investment time horizon gets shorter, they will have less time for the good return periods to offset the bad return periods. Many people have a sense that with a long enough holding period, stock returns will rise to achieve some sort of “historical average” return. Many people have in their head that stock returns “ought” to average around 10% per year over the long term. 

There are two problems with this notion. 

First, the basic idea that above-average returns will eventually cancel out below-average returns rests on the idea that stock returns are "mean-reverting." That is, that returns are not purely random, but follow a time-series behavior known as negative serial correlation. Although historical returns have indeed been somewhat mean-reverting over the long-term in the past, there are strong arguments to be made that they may not be mean-reverting in the future. Capital markets have become much more efficient over the past 100 years. That means that stock prices increasingly tend to reflect all available information very quickly. Past returns are clearly and easily known to all market participants. If the stock market is even weakly efficient, there should not be any actionable information based merely on past returns.   

Second, we do not know what the future average stock market return will be. Future average returns could very well be quite different than past average returns. Just because U.S. stocks achieved a compound return of about 10% per year since 1926 (according to the Ibbotson database) does not mean that 10% is a reasonable forecast going forward. For one thing, both inflation and interest rates were much higher on average over that period than they are now. The average inflation since 1926 was 2.9%, so the real return on stocks (above inflation) was 7.3%. And there are very good reasons to believe that future real returns for U.S. stocks are going to be much lower.

For one thing, the U.S. experience was a fortunate outlier. The 2011 landmark study by Dimson, Marsh and Staunton (DMS) of the London Business School showed that the U.S. has had one of the world’s highest stock market returns. “This makes it dangerous to generalize from that country’s historical returns” according to DMS. Over the 111-year period studied (which is updated annually in the Credit Suisse Global Investment Returns Yearbook), the authors found that whereas the U.S. market achieved a real return (over inflation) of 6.3% per year, the average global stock market experienced a 5.5% real return.

Perhaps more importantly, DMS found that even the 5.5% average real return was well above what could have reasonably been expected at the start of that time period. They estimate that the expected equity risk premium was 3% to 3 1/2%. The primary factor that caused actual returns to be above expected returns was a large and unexpected upward revaluation of the market: the prices paid for stocks as multiples of dividends and earnings greatly expanded over time. This is not a repeatable event. P/Es cannot expand forever.

The Pre-Retirement Glidepath

The reduction in human capital in the years leading up to retirement, and its dramatic reduction to zero at the point of retirement, certainly justifies a decrease in equity allocations prior to retirement. In addition, the “sequence of returns” risk in the early years of retirement also calls for lowered equity risk at retirement. Finally, the increased levels of risk aversion felt by many as they head into retirement is not something to be ignored. All three confirm the wisdom of reducing stock market exposure at retirement. The strong consensus among the TDF asset managers is that a 50% equity allocation is appropriate at retirement. 

Look again at the “Target Date Funds MKT Betas” graph above. The pre-retirement glidepaths for all of the asset managers are extremely smooth and gradual. Given the non-linear nature of the change in human capital, this strikes me as odd. The biggest change in human capital is abrupt and sudden: it drops to zero at retirement, from what might possibly still be a relatively high level. Many people have their highest earning years immediately prior to retirement. Their possible option to continue to work may give them the opportunity to continue to monetize their high level of remaining human capital. It strikes me that the reduction in equity exposure at retirement should be similarly abrupt. 

In a somewhat similar vein, sequence of returns risk does not start until retirement. Prior to retirement, people often still have the option to delay retirement if their retirement savings is dramatically reduced by adverse markets. To the extent that sequence of returns risk is the reason for reducing stock market risk, it should coincide with retirement. 

Increased risk aversion is probably not merely a phenomenon tied to aging, increasing steadily year by year. It would be more logical for it to be a response to an increase in the risk of life circumstances. The major change for most people is when to go from earning a steady paycheck to having no earnings at all. That change occurs in one fell swoop.

All three of the reasons for reducing risk at retirement are not gradual changes over 20-30 years. They are sudden changes that occur close to or at the point of retirement. Consequently, a more logical glidepath would maintain a much higher level of equity exposure leading up to retirement, and only very near or at the point of retirement would it be dramatically reduced, as shown in the “Suggested Optimal” glidepath in the graph below.    

The Post-Retirement Glidepath

Whereas there are very good reasons to reduce portfolio risk by lowering equity exposure at retirement, there are only weak reasons for continuing to reduce stock allocations after retirement. Human capital is already at zero. Maximum exposure to sequence of returns risk has already passed. Continued age-related increases in risk aversion may be present in some people, but it is probably not very pronounced in most people. And although the expected holding period for stocks is shrinking (very gradually) during retirement, and as noted above, the argument that stock allocations should be influenced by length of expected holding period is weak at best.

Although continuing to reduce equity exposure after retirement is still the conventional wisdom and remains the glidepath used by target date funds, more recent studies have repudiated this practice. One of the most respected recent studies (Pfau and Kitces, 2013) recommends gradually increasing stock market exposure after retirement, not decreasing it. Their research reveals that continuing to reduce stocks after retirement is the worst approach: it reduces expected return for nearly all outcomes, even those in the bottom 5%. The problem is that if equity allocations start at about 50% and continue to be reduced down to 20%-30% over the next few years (as with most TDF glidepaths), the damage from an early period of poor returns cannot be overcome with better returns later on because the absolute size of the equity portfolio has been reduced so much. They find that gradually ramping up stock allocations for the first 10-15 years of retirement (towards 70%-80%) is a “heads I win, tails I don’t lose” approach. If early returns are good, increased risk later when the portfolio is smaller won’t matter. If early returns are bad, gradually increasing stock exposure is a kind of “dollar-cost averaging” approach to take advantage of better returns when they finally arrive. 

Gradually increasing equity exposure after retirement is exactly the opposite of the approach taken by TDFs, as shown in the graph above. The divergence is predicated on the assumptions used to forecast future equity market returns. The traditional approach to lifetime asset allocation—the downward sloping equity glidepath throughout life used by target date funds—rests heavily upon research that assumes market efficiency, random return patterns, and lower than historical returns. On the other hand, studies that use long-term historical returns, which have exhibited both high returns and mean-reverting behavior, tend to favor higher equity allocations and/or allocations that rely upon long-term mean-reverting behavior, such as gradually increasing stock market exposure during retirement (as in Pfau and Kitces).

My position is between these two extremes. I believe that assuming market efficiency is a sound default position, but I also acknowledge that historical returns have exhibited mean-reverting that could continue in the future. Consequently, my “Suggested Optimal” default stock market allocation for the post-retirement period (shown in the graph above) would remain the same as it was at the onset of retirement. If it started at 50%, I would leave it at 50%. 

Equity Allocation at Retirement

However, I do not believe that 50% is the correct starting equity allocation for all retirees at all times. Even though that target stock market exposure is the strong consensus among target date fund asset managers, in my opinion there are three considerations that should influence (and alter) it:

• Risk aversion
• Risk capacity
• Stock market valuation

Risk Tolerance

Clients with higher psychological aversion to risk should have lower equity allocations, and vice-versa. However, assessing a client’s level of risk aversion is an art, not a science. There is very little real scientific validation behind the various questionnaires that are usually used to assess a client’s risk aversion level. Often these are quite lengthy and elaborate, asking about past investment behavior (e.g., did you sell in 2000, 2008, or 2020?), level of investment knowledge and experience, sources for investment information, etc. 

I have a very simple approach. I ask clients to rate their own risk aversion level relative to “the average person” on a 1-5 scale, with “1” being much more risk averse, “3” being about average, and “5” being much more aggressive. The behavioral finance literature seems to back me up in my assertion that this is as good a way as any to assess risk aversion. 

My approach is to adjust stock allocations in 5% increments based on this input. A “1” (highly risk averse) would get a -10% stock allocation adjustment. A “3” would get no adjustment. A “5” would get a +10% adjustment. 

Risk Capacity

This one is even more difficult to estimate. In my opinion, risk capacity has to do with the extent to which poor investment returns will crimp someone’s retirement lifestyle. It requires several inputs:

1. Estimated or desired annual spending level in retirement in dollars
2. Annual guaranteed income in retirement (Social Security, pension income, annuity income, etc.)
3. Estimated value of retirement savings at retirement

The formula is (1-2)/3. That is, desired spending in excess of guaranteed income divided by assets. If the figure is 2% or below, the required draw on retirement assets is quite low, and the stock allocation adjustment is +10%. At 4%, no adjustment is made. At 6% or more, any significant downdrafts in assets are likely to severely impact retirement success, and the stock allocation adjustment is -10%.

For example, let’s assume that a couple would like to spend $120,000 per year in constant (inflation-adjusted) dollars. Let’s also assume that their expected combined annual Social Security benefit level is $70,000 per year, and that they have no pension or annuity income. Finally, assume that their retirement nest egg is $1,250,000 in a combination of taxable, tax-deferred (IRA), and tax-free (Roth IRA) assets. Using the formula above:

            $     120,000     Desired real spending level

             -      70,000    Guaranteed income (Social Security)

            $      50,000    Real spending covered by investments

          /  $1,250,000    Total investments at retirement

          =               4%    Required real spending draw

At 4%, I would make no adjustment to their equity allocation based on their risk capacity. If their desired real spending level was only $95,000, then the real spending to be covered by investments would be $25,000, or 2% of their beginning retirement portfolio. At that level or less, I would increase their equity allocation by 10% because they have a lot of room for bad market events to happen without it crimping their retirement lifestyle. On the other hand, a real spending level of $145,000 would require $75,000 in real spending covered by investments, which is a required real spending draw of 6%. At that level or higher, I would decrease the equity allocation by 10% because they are more exposed to poor returns impinging on their spending plans. 

One action that can and probably should be taken for those who approach or even exceed non-guaranteed spending of 6% of their retirement nest egg is to purchase an income annuity, as I discussed in “Your Spending Policy in Retirement.” Buying an income annuity provides guaranteed income of a certain amount each month for the rest of your life (and perhaps for your spouse’s life). In most cases, the annual yield for an income annuity will be above the yield available from ordinary fixed income investments. There are two reasons for this. First, unlike with fixed income investments, there is no return of principal to the investor. There is no maturity for this type of annuity other than the death of the annuitant. The other reason has to do with the uncertainty of the annuitant’s life expectancy. The annuitant may pass away before they collect many annuity payments. This possibility is offset by the possibility that they may live for a very long time. The economic loss to those who pass away early is paid to those who live longer in the form of a higher yield, known as “mortality credits.” 

Stock Market Valuation

Many studies point to the importance of valuation in setting equity allocations. These use historical data in various ways to calibrate valuation. They often have very large swings in target allocations based on estimates of valuation. If 50% is the neutral point, they might recommend 30%-40% on the low end for overvalued markets and 60%-70% on the high end for cheap markets. 

I prefer to not be quite that heroic. The studies that recommend making large moves in equity allocations based on valuation tend to rely on historical market return data, usually with a heavy emphasis on U.S. stock market returns since 1926. Of course, we now know with absolute certainty what the average stock market return has been since 1926. (It was much higher than could reasonably have been expected.) We also now know the distributions of returns around the mean, and that they were somewhat “mean-reverting.” That is, periods of weak returns were followed by periods of strong returns. Actual returns were not purely random, but had “runs” of low returns and high returns. This behavior strongly rewarded tactical asset allocation models focused on valuation, which rely entirely upon mean-reversion to work. Armed with 20/20 hindsight, it is quite easy to construct tactical asset allocation parameters that would have added considerable value, avoiding the weaker periods and emphasizing the best periods. 

The problem is that there are strong arguments in favor of the assumption that future returns are more likely to be random (and lower) than was true historically. As noted above, over the past 100 years, markets have become much more efficient. That means that they have increasingly reflected in their prices all available information. Globally, markets have become much more “integrated.” The free flow of capital and information has increased dramatically over the past 100 years. The conditions which led to mean-reverting returns are likely no longer present to the same extent, if at all.

Consequently, I prefer to limit the influence of valuation to +/- 10%. 

My clients invest primarily in U.S. stocks and bonds, so I focus my valuation measurement on those two asset classes. 

The most famous and popular method for forecasting the long-term expected return for the U.S. stock market is the famous Shiller CAPE (“cyclically adjusted price/earnings”) ratio based on current price divided by the inflation-adjusted earnings over the past 10 years. However, I prefer a slightly different methodology to smooth earnings. Like Shiller, I use the trailing 10 years of earnings, but rather than use a simple average I use the slope of the earnings growth line (unadjusted for inflation) to get a forecast of the next year’s earnings, which I then divide by current price. This is an “earnings yield” rather than a price/earnings ratio, which is more amenable to comparison with bond yields. By using the slope of the 10-year history to forecast earnings for the following year, I am taking into account how fast earnings are growing, whether due to inflation or from corporate profitability. Both of these have been very important in the recent past.

Shiller himself has migrated away from his simple CAPE ratio to something he calls the “Excess CAPE Yield.” It uses inflation-adjusted 10-year earnings divided by price for stocks (same as for the CAPE) and compares that to the 10-year Treasury bond yield. He was able to construct a time-series of long-term U.S. Government bond yields back to 1871 to coincide with his measure of U.S. corporate earnings. Like most academics, Shiller prefers to use very long-term data. 

However, in my opinion there is a problem with comparing stock market earnings yield with bond market yield. It is an apples to oranges comparison. Stocks are able to increase their earnings to more-or-less offset future inflation. Not so nominal Treasury bonds. In fact, long-term Treasury bonds are particularly ravaged by inflationary surprises, whereas stocks tend to be somewhat immune.  

Consequently, for the bond yield, I prefer to use the 10-year TIPS (Treasury Inflation Protected Securities) bond. This is a measure of the real (inflation-adjusted) yield available in the Treasury bond market. Therefore, it is more comparable to the forward earnings yield of the stock market. Both are real yields, and both are forward-looking. The publicly available monthly yield history for 10-year TIPS starts in January 2003, so my stocks-bonds valuation comparison starts then.

The higher the earnings yield of the S&P 500 (red line above) relative to the 10-year TIPS bond yield (green line above), the more it makes sense to emphasize stocks over bonds, and vice versa. I use a simple approach based on the fact that the long-term average spread of stock earnings yield over TIPs yield has been about 4% (blue line above). For every 1% increase in the stocks-bonds yield spread, I add 5% to the stock allocation (up to +10%). And for every 1% below the average in the spread, I subtract 5% from the stock allocation (up to -10%). Recently, the stocks-bonds spread has been about 2%, which calls for a reduction in equity allocations of  -10%. The reason for this is not so much that stocks have become particularly overvalued, but rather than since 2022 the dramatic increase in interest rates has made bonds, include the 10-year TIPS bond, much more attractive. 

Ongoing Adjustments After Retirement

I use three factors to adjust the equity allocation at retirement: 1) risk aversion, 2) risk capacity, and 3) stock-bond valuation. Each of them can move the strategic allocation to stocks by +/- 10%. Since the starting default equity allocation at retirement is 50%, that means that even at the extremes, equity allocation will not fall below 20% nor rise above 80%. And as a practical matter, it rarely diverges outside of a band between 35% and 65%. 

Ideally, strategic equity allocations should be revisited if there are major changes to a client’s situation. For example, the receipt of a material inheritance could very well increase their risk capacity enough to require an adjustment. And of course, stock market valuation conditions change every day. Large changes in the stocks-bonds spread are usually infrequent. My 9-factor tactical MKT model reacts very quickly (it is updated every day), and captures quite a lot of shorter-term fluctuations in valuation. However, significant long-term changes in stock market valuation may require some adjustment to strategic equity allocations. 

Summary and Conclusions

• Asset allocation is vitally important, determining over 90% of portfolio return variability.
• Target date funds (TDFs) are by far the most popular way that most people determine their asset allocation.
• TDFs follow an asset allocation “glidepath” that reduces equity allocations over time, starting at 80%-90% for young investors and gradually lowering it to 20%-30% well after retirement. 
• The gradual reduction of “human capital,” monetized through earned income, is one reason for reducing equity allocations prior to retirement, when human capital falls to zero.
• “Sequence of returns risk” is another reason for lowering stock market risk prior to retirement since poor returns early in retirement can seriously crimp retirement spending plans.
• Risk aversion levels tend to increase with age, adding further impetus to risk reduction in the years leading up to retirement. 
• Many people have the impression that length of the expected holding period should heavily influence asset allocation—the longer the holding period, the higher the equity percentage.
• This notion of “time diversification” rests upon the idea that the stock market “mean-reverts” over long time periods—that poor returns in one period will be followed by higher returns in another, leading to stock returns that achieve the historical average return of about 10% per year.
• Future stock market returns may not mean-revert, however. Assuming that stock market returns are random and follow no particular time-series behavior is a sound default position.
• Also, future returns are likely to be a lot lower than they were for the past 100 years. The U.S. stock market was an outlier compared to other countries, and over the past 100 years nearly all stock markets experienced substantial increases in price/earnings ratios, which cannot go on forever.
• Rather than gradually and continuously lowering equity exposure over the 25-30 years before retirement, I suggest that a more optimal path would be keeping equity allocations high until right before retirement, and dramatically reducing stock risk right before or at the point of retirement. 
• 50% in stocks at retirement is the strong consensus recommendation among TDFs.
• I suggest adjusting that figure based upon three client-specific considerations:
1. Risk aversion
2. Risk tolerance
3. Stock-bond relative valuation
• Each of the three could change default equity allocations by up to +/- 10%.
• In retirement, stocks should never be less than 20% of a portfolio, nor more than 80%, regardless of the three criteria.