The Failure of Portfolio Return Diversification

The Failure of Portfolio Return Diversification

by Jon Southard for CBRE

With a substantial number of markets starting to see prices increase, we can now look back on the real estate downturn and begin to draw some conclusions and lessons from the experience. One of the casualties of this past downturn was the notion that diversification is the best way to approach real estate investment. Ironically, while some have leveled the complaint that too few experts were able to anticipate the depth of the downturn, the alternative strategy of spreading out bets in lieu of trying to anticipate returns has even less to say for it after recent performance.

In recent decades, we have seen more and more finance theory from other asset classes creep into real estate investors’ thinking. One of the formerly popular tenets of stock market investing was to focus less on “stock-picking” and more on portfolio diversification. As a result, advocates of the primacy of portfolio construction can put as much focus on correlation between assets as on attempting to understand the future prospects of companies. In the parlance, low or negative “beta” can be just as important as seeking “alpha”.

A problem arises, however: in building portfolios of diversified assets, a historical correlation must be chosen. Generally, a long-term correlation is used as a predictor for how real estate markets might be expected to correlate in the future. A low or negative correlation is seen as good, in that not all markets are expected to decline at the same time. Looking back on the last downturn—even tracing out quarterly returns across markets as the figure below does—given what correlations wound up being, it is readily visible that history was a poor guide. A highly-correlated downturn meant that there was literally no market among the 19 largest office markets in which to hide from the downturn.


The difference in average correlations across markets can be quantified in the table below. If returns average 0.6 when looking at correlations across the entire time series, the average correlation skyrockets to 0.85 in the latest recession. Interestingly, the two prior recessions actually see less correlation across markets, as some markets turned down. So while these prior recessions would reinforce the idea that focusing on constructing uncorrelated portfolios is worthwhile, any confidence gained leading up to 2008 was promptly shattered as correlations rushed towards 1.0.


So, while it’s true that no one can perfectly forecast both the timing and severity of downturns before they arrive, it is likely that one way to do worse than relying on inexact forecasts is not to try to forecast at all. With all markets turning down, the way to reduce recent losses was to focus on “alpha” in the portfolio. That is, to focus investments on the markets that would perform best relative to the whole, rather than spreading out investments just for the sake of diversification.

To the extent that confidence in portfolio diversification can be taken too far, the worst case would be to believe that a diversified portfolio would allow for more leverage to be applied than would be allowed by individual buildings. The significance of returns being steeply negative across all office markets is that a highly leveraged portfolio that amassed a lot of assets would be exactly no better off than small portfolios—both would have seen their equity eliminated. This is the definition of failure for proponents of diversification as primary importance.

Those who complain about research should remember the failure of diversification, as the alternative to forecasting (whether econometric or pro forma in approach) has usually been to tout portfolio construction or even attempts to approximate market portfolio shares. With these being apparently not much of an alternative, the proper course of action would seem to be to do a better job of focusing on the selection of individual real estate properties, with only some regard to how they fit into the portfolio. With that said, asset selection approaches could also be improved by bringing in a consideration of risk.

Be Sociable, Share!
Leave a Reply