The Best Diversifier Has Been the Simplest

Investors in search of diversification for their equity holdings should probably hold their noses and add high-quality bonds, if Morningstar's data about asset-class correlations are any guide.

In the past five years as stocks have soared--and in the bear market when they nosedived--long-term government bonds have reliably gone the opposite direction. Meanwhile, other security types that investors frequently use to round out their portfolios--from precious metals to commodities to real estate to market-neutral funds--have had mixed success on the diversification front. Of course, the future may look different from the recent past, but these data suggest that investors in search of diversification could do a lot worse than holding a plain-vanilla stock/bond portfolio.

Charting Correlations The Morningstar Direct database--geared toward institutional investors--provides a snapshot of how closely correlated various asset classes have been over time, based on their monthly performance. Investments with negative correlations with one another have moved in opposite directions, whereas investments whose performances have moved in lockstep will have correlations close to 1.00.

I used mutual fund categories for this exercise, as well as a handful of individual funds to stand in for categories that didn't yet exist 10 years ago. I examined correlations during the past three-, five-, and 10-year periods. Because the three-year correlations table closely resembles the five-year correlations view, we're showing the three- and 10-year grids here (as PDF files).

The numbers along the horizontal axis correspond to the investment with the same number along the vertical axis. So, if you were to look to the right of the fifth item from the top on either grid--the U.S. OE (open-end) moderate-allocation category--the first column would represent its correlation to the first item (the U.S. open-end large-blend category), the second would represent its correlation to the second item (the U.S. open-end foreign large-blend category), and so on.

Bonds Get It Done As you can see, the (U.S. OE) long-government category is the only category with a strong negative correlation with both the U.S. large-blend and foreign large-blend equity categories, as well as the moderate-allocation category. That's not just a phenomenon of the bull market, either. While the correlation between long-government bonds and stocks isn't quite as strongly negative during the trailing 10-year period as it has been in the past three years, it's still the most negative relationship depicted on the 10-year chart.

A glimpse at the annual returns of

Note that the correlation between the intermediate-term bond category and U.S. stocks is not as low or even negative. That's because most intermediate-term bond funds have a healthy dose of corporate bonds and other non-government-issued bonds, which means they have more sensitivity to equity-price movements than government-issued bonds.

While the correlation between the Barclays U.S. Aggregate Bond Index and the large-blend category isn't depicted in these charts, it's lower than the correlation between the typical intermediate-term fund and the large-blend category: negative 0.18, negative 0.32, and negative 0.01 during the past three-, five-, and 10-year periods, respectively. Judging from past correlations, then, it's likely that an Aggregate Index tracker is apt to be the better diversifier than the typical actively managed intermediate-term fund. The flip side is that most active intermediate-term funds have beaten the Aggregate Index soundly during the past three- and five-year periods, in large part because of their emphasis on corporate and other non-government-bond types.

Diversification? Not So Much

The other asset classes have had varying degrees of success of delivering diversification benefits relative to equities or relative to a combination U.S. stock/bond portfolio. (Investors can look to the moderate-allocation category as a proxy for the latter.) After high-quality bonds, the next-best diversifier depicted here was

All the other asset classes deliver less impressive diversification benefits relative to both the large-blend and moderate-allocation categories. The only close contender was the market-neutral category, which had impressively low correlations with large-blend and moderate-allocation funds in the past decade. However, those correlations have jumped sharply during the past three- and five-year periods.

Real estate funds exhibited the opposite pattern. Correlations with the S&P 500 rose sharply during and immediately after the bear market, as discussed in this article, and are quite high in the five- and 10-year periods. But they've declined in the past three years. A turning point was the so-called "taper tantrum" in the second quarter of 2013, which roiled interest-rate-sensitive securities such as bonds as well as REITs. Indeed, REITs' correlation with bonds has risen in the past three years, even as their correlation with stocks has declined.

Somewhat surprisingly, given that recent performance has been poor even as U.S. stocks have soared, the commodities-tracking Oppenheimer Commodity Strategy Total Return did not deliver an extremely low correlation relative to equities or the moderate-allocation category over any time period.

Declining Foreign Correlations One of the bigger surprises in the data was that even as many investors had begun to take it as an article of faith that U.S. and foreign market performance would be closely correlated, correlations between U.S. large-blend and foreign large-blend funds declined in the past three years. Currency issues explain much of the divergence. Even as developed foreign markets have rallied, the euro has declined relative to the dollar, causing unhedged foreign funds (which is most of them) to badly underperform U.S. equity funds.