Swedroe: REITs Aren’t Special

Many investors think of real estate investment trusts (REITs) as a distinct asset class, because in aggregate they have historically had relatively low correlation with stocks and bonds, and their returns were not well-explained by the single-factor CAPM model.

For example, during the period January 1978 through September 2016, the monthly correlation of the Dow Jones U.S. Select REIT Index with the S&P 500 was 0.58, and with five-year Treasurys, it was just 0.07.

Looking At REITs

With the advancement of modern financial theory and the development of more sophisticated multifactor asset pricing models, my colleagues at Buckingham Strategic Wealth and The BAM Alliance, Jared Kizer and Sean Grover, decided to take another look at REITs in their May 2017 paper, “Are REITs a Distinct Asset Class?”

Their research was motivated by the observation that many studies treat REITs as a distinct asset class based on correlation alone, and that many investors overweight REITs in their portfolios on a market-capitalization basis. Their data sample covers the period January 1978 through September 2016.

Kizer and Grover began by establishing criteria an asset class must meet to be considered distinct:

1. Low correlation with established asset classes, such as broad market equities and government bonds.
2. Statistically significant positive alpha with respect to generally accepted factor models.
3. Inability to be replicated, on a co-movement basis, by a long-only portfolio holding established asset classes.
4. Improved mean-variance frontier when added to a portfolio holding established asset classes.

Prior research had shown that, in terms of equity risk, REITs have significant exposure not only to market beta, but to the size and value factors. In addition, they have been shown to have exposure to the term premium.

In their analysis, Kizer and Grover employed a six-factor model comprising the market, size, value and momentum equity factors as well as the term and credit fixed-income factors. The credit factor (referred to as IGDEF) subtracts the return of a duration-matched portfolio of Treasuries from the total return of the corporate bond index.

Their regression analysis included not only REITs, but 12 other industries available on Ken French’s website. Following is a summary of their findings:

• Demonstrating the explanatory power of the six-factor model, virtually all industries are well-explained by four equity factors and two fixed-income factors. Only one industry category (a catchall that included mining, construction, transportation, entertainment and hospitality, among other sectors) had statistically significant annualized alpha, and the estimate was negative.
• The annual alpha estimate for REITs was -0.89% with a t-stat near zero (-0.3).
• REITs showed statistically significant exposure to market beta (0.61 with a t-stat of 10.2), size (0.44 with a t-stat of 6.1) and value (0.77 with a t-stat of 9.9), as well as a small negative (-0.08) and statistically insignificant (t-stat of -1.7) exposure to the momentum factor, a large (0.70) and statistically significant (t-stat of 3.8) exposure to the term premium, and a large (0.92) and statistically significant (t-stat of 3.9) exposure to the credit (default) premium.
• While the R-squared ratio was relatively low for REITs (0.51), this was also true for other industries, including energy, utilities and health care.

These findings led Kizer and Grover to conclude that, while the low R-squared ratios indicate diversifiable risks present in each industry, they do not indicate uniqueness in underlying return drivers.

The authors state: “While the relatively low correlation with the S&P 500 Index and 5YT was encouraging, the four- and six-factor regression models indicate that REITs are likely not a distinct asset class, especially when compared to the results of other industries.”

Their evidence demonstrates that, while REITs may meet the first of the four criteria they established (low correlation), they fail to meet the second (significant alpha).

Distinct Asset Class?

Kizer and Grover next tested REITs against the third criteria—a distinct asset class should not be easily replicated by a long-only portfolio of established asset classes. Given the factor exposures they had found, and using returns for U.S. small-cap value stocks (SV) from Ken French’s data library and the Barclays long-term corporate bond index (CORP), they attempted to replicate REIT returns with these two returns series.

The following table shows results from a portfolio allocating about 67% to SV and 33% to CORP. This optimal replicating portfolio has a monthly correlation with REITs of 0.72. The table also presents other statistics that compare the optimal replicating portfolio to REITs over the period January 1978 through September 2016.

As you can see, the replicating portfolio dominates REITs in almost every way—it earns higher compound returns, has lower volatility, achieves a higher Sharpe ratio, has lower kurtosis and wins on most historical risk characteristics.

A skeptic might note the replicating portfolio has a 33% allocation to long-term corporate bonds during a period in which interest rates have declined significantly, but the regression results show the term factor loading for the replicating portfolio is lower than the term factor loading for REITs. Thus, interest rate risk exposure can’t account for the results.

Kizer and Grover then tested REITs against their fourth criteria, and concluded REITs fail to improve the mean-variance frontier, on a statistically inferred basis, when added to a portfolio holding established asset classes.

Summary

After first establishing a pragmatic list of criteria for considering asset classes, Kizer and Grover found that, while REITs do indeed exhibit relatively low correlation with traditional equity and fixed income, a deeper dive into their returns reveals shortfalls in their qualifications for asset class distinction.

They found that multifactor regression analyses revealed no statistically reliable alpha generation in REIT returns and that REIT returns are well-explained by traditional risk factors. They also found that a long-only replication of REIT returns with small value equities, and long-term corporate bonds produces a portfolio that co-moves well with returns to REITs and exhibits historical return and risk characteristics generally better than REITs.

Finally, they found REITs do not reliably improve the mean-variance frontier when added to a benchmark portfolio of traditional stocks and bonds. These results, and the associated failure to satisfy their asset class criteria, led Kizer and Grover to conclude that REITs are not a distinct asset class.

It’s important to note the results did not lead Kizer and Grover to recommend excluding REITs from equity portfolios.

Instead, the results of this study led them to suggest that REITs, as an equity security with marginal diversification benefits, shouldn’t receive a weighting in investor portfolios that significantly deviates from market-capitalization-based weights. Data from Morningstar shows REITs represent approximately 3.5% of the iShares Russell 3000 ETF (IWV) on a market-capitalization basis, which is a valid starting point for a REIT allocation in a diversified portfolio.

Larry Swedroe is the director of research for The BAM Alliance, a community of more than 140 independent registered investment advisors throughout the country.