January 2021
Save
Key Points
• The recent, decade-plus returns of diversified portfolios have disappointed relative to mainstream US stocks and bonds.

• A long-term examination, dating back to the mid-1970s, reveals such a shortfall is not unprecedented and the long-term case for diversification is still powerful.

• Extrapolating the recent superior performance of the 60/40 mix into the future requires the dangerous assumption that US stock market valuations will march to nearly unprecedented levels. Diversification is needed now more than ever.

In 1952, Harry Markowitz published “Portfolio Selection” and helped usher in the era of modern portfolio theory. Harry’s work showed us that a portfolio’s risk is not defined by the riskiness of its individual assets, but by the extent the portfolio’s assets’ price movements correlate, or move together. Harry demonstrated that a technique, now known as mean-variance optimization, derives an efficient frontier of portfolios, each of which maximizes expected return at any given level of risk.1 At a Research Affiliates’ forum, Harry told attendees that “the capital markets offer two free lunches: diversification and mean reversion.”

Diversification is a theoretically powerful concept, but as our friend Jason Hsu is fond of saying: “Diversification is a regret-maximizing strategy.” In a roaring bull market, such as 2009–2020 and 1991–1999, we investors regret every penny we put into diversifiers, while in a secular bear market, such as 2000–2009 and 1968–1982, we regret every penny we did not put into diversifiers.

In a rerun of the 1990s, the 2010s were brutal to diversifying strategies. As it was in 1999, the wisdom of diversification is again under attack. The pressure to ignore diversification grew even stronger last year when the COVID-induced bear market of February–March 2020 was no less savage to diversifying asset classes than to 60/40 investors in the US market, and the stimulus-induced rebound again rewarded the “naïve” 60/40 investor for shunning diversification.

We affirm the wisdom expressed in a quote often misattributed to Yogi Berra: “In theory, there is no difference between theory and practice, while in practice there is.”2 In theory, diversification is a “free lunch” for the patient investor. In practice, an investor was substantially better off if she had invested solely in the classic US 60/40 portfolio—60% in the S&P 500 Index and 40% in the investment-grade bond market—at a near-zero management fee over the last dozen years. Diversified investors were far less rewarded for the same level of risk! This outcome has led many advisors and fiduciaries, on the receiving end of complaints from unhappy clients, to abandon diversification.

## A Longer Look

To gain a good understanding of the merits of diversification, let’s take a patient, long-term survey of the performance characteristics of diversified portfolios compared to the mainstream blend of 60% US stocks and 40% US bonds. We look back to the mid-1970s, when diversifying asset classes first became investible and tracked by market benchmarks.

This long-horizon exercise is important for three reasons:

1. First, most investment portfolios serve future spending needs that extend far longer than the next 10 years. Even a newly minted retiree at age 65 has nearly 20 years of additional life expectancy, and most institutional investment horizons are longer than individuals’ life expectancies: Endowments are ostensibly intended to serve perpetual spending needs, and pension obligations typically extend for a century or more until the last check is cut.3 Human nature, which behavioral finance seeks to model, makes it very difficult, however, for investors to behave as if these long horizons are relevant to their decision making.
2. Second, even an entire decade may not encompass a full market cycle. For example, US investors faced neither a recession nor a bear market (as defined by a 20% drawdown on the S&P 500 using month-end data) during the 2010s.4 Peak to peak, this last cycle extended for over 12 years from October 9, 2007, to February 16, 2020.
3. Third, most importantly, past is not prologue. Whereas 5 or 10 years can seem quite a long time in many aspects of our lives, such a stretch is relatively short in the capital markets, where a recent trend can seem perpetual. Indeed, any long-horizon examination of historical returns shows mean reversion; that is, recent 5- or 10-year winning strategies tend to do poorly in the future and vice versa. We have noted that future 10-, 20-, or 30-year stock market returns are negatively correlated with the preceding 10-, 20-, and 30-year returns, and that the tendency to extrapolate the most recent seemingly long-horizon experience to the future is the “most dangerous shortcut in financial planning” (West and Ko, 2017).

Our survey includes 16 asset classes and their index proxies. We break the asset classes into three categories, or pillars. The “first pillar” is mainstream developed-market equities, whose historical role in a portfolio has been to provide higher returns via participation in real economic growth. The “second pillar,” mainstream US investment-grade bonds, is intended to provide steady, reliable income in a portfolio—pretty anemic today!—and countercyclical dry powder available for investment when the economy and risk markets soften. The purpose of the “third pillar” is to diversify the portfolio and offer the ability to produce better returns than the other two pillars during inflationary periods. Why is inflation a specific function of the third pillar? Because inflationary bouts often lead to rising interest rates and tumbling stock market valuations, a combination that breaks down the complementary, diversifying characteristics of the first and second pillars.

We constructed an equally weighted portfolio that includes all 16 asset classes and a 60/40 portfolio consisting of 60% S&P 500 and 40% Barclays Capital US Aggregate Bond Index. Our survey begins in the first quarter of 1975 with 10 of the 16 asset classes from the first and second pillars and two of the third-pillar asset classes. We added new asset classes to the survey at the inception of their return histories. This gives us four full decades and a fifth stub decade, which comprises the second half of the 1970s. As a simple shorthand, we consider the return difference between the equally weighted portfolio and the 60/40 blend to be the return premium attributable to diversification.

Since 1975, we see comparable results for the more-diversified equally weighted portfolio versus the 60/40 blend. The equally weighted portfolio produced a return of 10.9% a year with annual volatility of 10.0% and a Sharpe ratio of 0.64. The 60/40 blend earned an annualized 10.7% return with annual volatility of 10.4% and a Sharpe ratio of 0.60. If the diversifiers in the equally weighted portfolio can boost the Sharpe ratio of the 60/40 portfolio, certainly they must work even better if an investor just concentrates in these markets. Not so fast! The third-pillar asset classes boost the Sharpe ratio because they offer diversifying and complementary risks in the 60/40 investor’s portfolio.

We can make several high-level observations about the returns of each of the three pillars over the analysis period. The first pillar (essentially developed-market equities) and the second pillar (essentially US investment-grade bonds) produce identical Sharpe ratios of 0.47. The blend of the two pillars in the 60/40 portfolio boosts the Sharpe ratio to 0.60. The remaining third-pillar asset classes did slightly better by producing an annualized return of 10.4%, annual volatility of 10.3%, and a Sharpe ratio of 0.58, essentially matching the result for the 60/40 portfolio. We call attention to the following observations: the 60/40 portfolio has a higher Sharpe ratio than either of the broad stock and bond categories, which perform similarly on a risk-adjusted basis; the third pillar performs as well as the 60/40 portfolio; and the equally weighted portfolio, which combines all three pillars, improves the Sharpe ratio relative to both the 60/40 blend and the third pillar alone.

A very important observation is that the 46-year time span of our survey works in favor of the 60/40 portfolio because our analysis begins at the start of 1975, a bear-market low for both US stocks and bonds, and concludes at the end of 2020, a market high for both of these asset classes. Yet, the diversified equally weighted portfolio still outperforms the 60/40 blend. When we examine the relative performance of the two portfolios based on cumulative wealth creation, the equally weighted portfolio exhibits relentless outperformance over the 46 years. Since 1975, in only one quarter (of the 180 total!) did the 60/40 portfolio show a cumulatively larger value-add than its more-diversified competitor. Ever so briefly in the first quarter of 1999, the 60/40 portfolio caught up the equally weighted mix before falling off substantially. In other words, less than 1% of the history we examine shows a 60/40 strategy as producing superior long-term wealth.

Suppose an asset manager approached you with the recommendation to invest in a strategy that showed such relentless long-term underperformance, simply because the strategy had fared better over the last decade. Given the long history of underperformance, and the theoretical advantage of the diversified strategy, you’d laugh them right out of your life! But that recommendation is precisely what the current crop of diversification naysayers are proclaiming today!

We also looked further to determine how consistent this risk-adjusted edge was for diversification. Let’s call this a Sharpe ratio “hit rate.” We found that in 60% of rolling five-year windows the equally weighted portfolio produced a superior Sharpe ratio relative to 60/40. Collectively, this simple analysis suggests a modestly superior risk-adjusted result for diversification, even during a span in which the 60/40 portfolio benefited from sharply falling yields and rising prices. At a minimum, it is impossible to assert that a 60/40 strategy is superior based on long-term empirical results. So far, so good, for Harry Markowitz!