- All five factors have economically large and statistically significant (at the 1% confidence level) factor betas. The betas were: market beta: 0.58; size: 0.38; value: 0.72; term: 1.04; default: 0.45.
- The alpha was -2.15%. However, it was not statistically significant (with a t-stat of -0.95), indicating that common factors explain the returns of public and private REITs.
- The r-squared value was 0.67.
- These betas are analogous to a balanced portfolio allocated approximately 60% to small value stocks and the remainder to long-term, high-yield bonds. Real estate is a hybrid asset class, showing return and risk behaviors common to both stocks and bonds.
- The largest contributor to excess REIT returns is the market beta factor (which accounts for more than 50%), followed by the term (about 35%), value (about 20%) and default (about 10%) factors. The total sums to more than 100% because the alpha is subtracted from returns.
- The largest contributor to excess REIT risk is idiosyncratic real estate sector risk (accounting for about one-third), followed by market risk (approaching 30%) and default risk (more than 20%). Two-thirds of excess REIT risk is explained by compensated factor risk and one-third by uncompensated sector risk (alpha).
- All five risk factors contribute positive return to REITs. However, real estate sector risk does not contribute a positive return while still contributing risk.
Additional Conclusions On Real Estate
Mladina also conducted a horserace to determine which of the four proxies he used for real estate—REITs, NPI, Cambridge Real Estate Index or his factor benchmark—dominates the real estate allocation along the efficient frontier. He found that the factor benchmark dominated the other real estate indexes along the efficient frontier. He writes: “REITs, NPI, and Cambridge Real Estate achieve no material allocation in the presence of the factor benchmark for real estate.”
Finally, Mladina used his five-factor model to test the alphas of all live and dead active real estate mutual funds in the Morningstar database from 1986 to 2015. He found that the average alpha was negative, with fewer active real estate funds showing statistically significant positive alphas than would have been predicted by chance.
This finding is consistent with the latest SPIVA results, which demonstrated that over the prior 5-year and 10-year periods, 85% of active REIT managers underperformed ther benchmark. At 15 years, 82% underperformed. On an equal-weighted (asset-weighted) basis, over the prior 15-year period, the average active REIT fund underperformed by 0.86 percentage points (0.81 percentage points).
Mladina concluded: “We find that current and lagged REIT betas and factor benchmark betas explain the entire return premium of private real estate, suggesting that private real estate does not offer a unique source of compensated return that differs from its exposure to systematic risk factors.” What’s more, Mladina writes, the “residual risk is idiosyncratic and uncompensated real estate sector risk.”
He added: “These results suggest that REITs are no different from any other industry sector, with the notable exception of their hybrid (stock-bond-like) nature and rich factor mix.” Finally, Mladina also concluded that if one were taking a factor-based approach to asset allocation, then “real estate would not be considered a separate source of return.”