The Wall Street Journal: Why REITs may not be so special

Long before investors fell in love with Facebook

FB, +0.91%

and Amazon.com

AMZN, +0.28%

or even the hot stocks of past generations such as utilities and railroads, real estate inspired dreams of wealth. “They aren’t making any more of it,” say proponents, as if to end all argument.

Some financial advisers and pundits helped fuel the love affair, suggesting that investors could reduce volatility and boost returns by adding real-estate investment trusts, or REITs, to portfolios in an amount exceeding their representation in broad market indexes. REITs behaved differently enough, zigging when other investments zagged, and provided solid enough returns to warrant the extra exposure, they said.

But a new study by Jared Kizer and Sean Grover of Buckingham Asset Management in St. Louis disputes the old claim about REITs. It argues that REITs deserve a portfolio weighting that reflects their own weighting in broader indexes (around 3%), but not necessarily an additional weighting that a separate asset class with different return and volatility characteristics would.

Sharpe analysis

To understand the authors’ argument, it helps to consider the Sharpe ratio, a basic metric of modern finance theory that divides return by volatility. The object is to create portfolios that exhibit the highest Sharpe ratio—that is, they provide the biggest return for the lowest volatility. That means the best return doesn’t always win in the game. The best return per unit of volatility wins, so reducing volatility is as important as increasing return. It turns out that adding REITs to a portfolio above the weighting they already have in broad indexes may not do much of either.

Kizer and Grover set out to determine whether REITs are a separate and distinct asset class, as many investors believe, and thus deserving of a heavier portfolio weighting.

They established four criteria for defining an asset class. First, the group of stocks in question has to have low correlation with other established asset classes, such as stocks and bonds. Assets with low correlation to each other tend to move in different directions at the same times, potentially reducing volatility.

Second, an asset class has to have statistically significant outperformance that is unrelated to “factors” or characteristics already established as adding return to an index. Those are things like value—low price relative to book value or earnings—momentum or small size. In other words, a group of stocks would have to attribute its outperformance to something other than, say, value, which is a factor or characteristic already agreed to give a stock an edge over a broad index.

Third, an asset class has to be unique enough so that a combination of other securities can’t mimic its return and volatility characteristics. Fourth, it has to improve the volatility-adjusted performance of the portfolio when it’s added in an amount exceeding its weighting in broad indexes.

Results of the study

Kizer and Grover discovered that REITs do, indeed, have low correlation to both broad stock and bond indexes. So they meet the first criteria of being a separate asset class. However, REITs didn’t produce any significant outperformance that couldn’t be explained by factors already known to produce outperformance. It turned out that REITs’ performance can be explained by characteristics known as the momentum, term and credit premiums.

Although there are disagreements about the definition, the momentum premium is sometimes viewed as compensation for bearing risk that an appreciating asset won’t continue to rise. The term premium is compensation for risk that short-term rates will be higher than future estimates. And the credit premium is compensation for assuming credit risk. (REITs typically operate with a lot of borrowed money, and they deliver high current dividend yield to investors, giving them corporate-bond-like characteristics.) All of this means any outperformance REITs deliver can be explained by these already understood risk premiums. Kizer and Grover discovered that other sectors have similar characteristics that also can be explained by known premiums. REITs just aren’t that special when analyzed in light of these premiums.

Not only can known factors explain REIT performance, but combinations of those factors can replicate that performance. It turns out that a combination of 67% small cap value stocks and 33% long-term corporate bonds produces a security that has a 0.72 correlation to REITs. A correlation of 1 is perfect. That high correlation means REITs fail the third part of the Kizer-Grover asset-class definition.

Finally, adding REITs to the classic balanced portfolio of 60% stocks and 40% bonds didn’t improve the portfolio’s volatility-adjusted returns—what modern finance calls “efficiency.” The authors conclude that REITs shouldn’t be excluded from portfolios, but also that they shouldn’t be included in doses that exceed their representation in broad indexes.

Investors should note that there are times when REITs have done wonders for portfolios. When the technology-stock bubble peaked in early 2000, for example, REITs and small-cap value stocks in general had been left for dead by investors. Consequently, they were priced to deliver massive returns when investors were finally disillusioned with profitless “dot-coms” and regained their affection for hard assets in 2000-02.

Adding groups of stocks to portfolios based on valuation metrics is the more old-fashioned view of the world than Kizer’s and Grover’s emphasis on historical price action and correlation, but that doesn’t mean it can’t be effective.

John Coumarianos, a former Morningstar analyst, is a writer in Laguna Niguel, Calif. He can be reached at reports@wsj.com.

The article “Investors Might Want to Rethink Their Approach to REITs” first appeared in The Wall Street Journal.