How Diversified Are You When It Counts?

A common goal and one of our obsessions in the investment world is the concept of diversification.  Wherever you read your financial information, the discussion and advice is to build a well-diversified portfolio.  So what does it mean to have a well-diversified portfolio?  Typically one would start with including multiple equities in a portfolio to eliminate single company risk (e.g. avoid excess exposure to an Enron or Lehman occurrence).  The next step may be to include international equities and allocate portfolio capital to other developed markets.  We can also include the higher growth, higher risk emerging market economies, all in an effort to improve diversification.  Next, we can include exposure to real assets, such as real estate and commodities.   Building a portfolio with these components, one would likely think they are close to being “well diversified”.  To take a closer look at how diversified this example portfolio is over time, we will go to the data and use a correlation analysis to take a look at the historical diversification benefits during times of U.S. Stock market declines and elevated volatility. 

For clarity, we will refer to correlation as a measure of how each portfolio component moves versus U.S. Stocks – a correlation measure of 1.0 suggests every movement is identical on a daily, monthly, and yearly basis (i.e. it does not aid in diversification).  Counter to correlation being 1.0, a correlation of -1.0 suggests the two portfolio components move in perfect opposite directions within the same time frames. Lastly, a correlation of 0.0 simply suggests none of the movement of the portfolio components can be explained by U.S. Stocks, and vice-versa.  Typically, one would look for a negative-to-positive 0.60 correlation coefficient over broad periods of time for adequate portfolio diversification.  In summary, correlation is a basic way of determining portfolio component relationships, yet there are various other more sophisticated tools available.

To measure the correlation metric in times of U.S. Stock market stress (arguably the exact time an investor wants a lowly correlated asset class), we will isolate two notable time periods of U.S. Stock market volatility: i) the Great Recession of 2008 (January 2007 through June 2009); and ii) the most recent correction in February 2018 (January 2018 through April 2018).   To analyze the correlation metrics, we will use a 90-day rolling correlation between the individual portfolio components versus U.S. Stocks (measured by the Vanguard Total U.S. Stock Market Index).  We will then compare the rolling correlation metrics for each of the portfolio components to the returns of U.S. Stocks over the same period.  For meaningful diversification benefits, we would want to see correlation levels remain moderate-to-low (negative to 0.60) during U.S. Stock declines.  First, set forth below as Figure 1, we illustrate the rolling correlation statistics versus U.S. Stocks for each international stock index (emerging and developed countries), a U.S. real estate index, and a broad basket commodities index, within the top portion of the chart.  The returns for U.S. Stocks over the same time period are set forth on the bottom portion of the same graphic.   Here we can analyze the relationship of the correlation metric of each portfolio’s component to the performance of the U.S. Stocks.

Figure 1:  Correlation by Asset Class (Great Recession)

Fig 2.png

As one can observe in the graphic depicted above, the correlations among U.S. Stocks, international stocks (both emerging and developed), and real estate ranged between 0.70 and 0.95 throughout the Great Recession.  More importantly, each of these so-called diversifiers actually increased in correlation to the S&P to 0.90+ during the worst of the financial crisis, which is the exact time investors needed the diversification benefits from these assets within their portfolio.  We can also note that commodities tended to help diversification until the later stages of the financial crisis, when they too became increasingly correlated to U.S. Stocks, dampening their diversification effects.  Lastly, we can point out that U.S. Bonds remained negatively correlated to U.S. Stocks and was the only portfolio component to help diversify throughout the broader market correction.  While these portfolio components can provide a different source of return for a given level of risk, the reality is during times of market duress (or when it “hurts to be hurt”) the correlation among the majority of these portfolio components versus U.S. Stocks tends to increase towards 1.0, meaning they move in tandem with U.S. Stocks at precisely the time U.S. Stocks are declining precipitously. 

Rationale for this phenomenon is largely related to the ultimate return driver or “factor” for each of the portfolio components.  Each of the asset classes in this example ultimately depend most heavily on a single factor - global economic growth. As a result, in times of market corrections when the “risk-off trade” is implemented, all these asset classes typically correct or discount their value in unison, counteracting their potential as effective diversifiers. 

To further illustrate this phenomenon, we can take a look at the same portfolio components and rolling correlation statistics during the most recent correction in U.S. Stocks – January 2018 through April 2018.   

Figure 2:  Correlation by Asset Class (January-April 2018)

Fig 2.png

As depicted in Figure 2 above, we can see each of the portfolio components exhibited an increase in their respective rolling correlation statistics when U.S. Stocks began to decline.  Once again, we can observe within the data each of the portfolio components that were expected to help overall portfolio diversification ended up actually contributing to downside portfolio volatility.  We can also point out that while U.S. Bonds provided great diversification benefits during the Great Recession, as displayed in Figure 1, declines in U.S. Bond prices actually acted as a catalyst to the decline in U.S. Stocks in February 2018, causing the correlation between these two portfolio components to increase (Note: Please see our additional research regarding the On and Off relationship between Stocks and Bonds).  As the larger “risk-off trade” is often implemented across global asset classes, correlations to U.S. Stocks tend to increase dramatically, implying diversification benefits throughout this example portfolio can be relatively poor, or even potentially non-existent, depending on the circumstances.    

In conclusion, the common approach to diversification through passive allocations to global asset classes can have its flaws, particularly at times when one typically thinks their portfolio is properly diversified.  This phenomenon is largely driven by the heavy reliance on returns derived from the global economic growth factor.  With so much of a traditional portfolio’s risk tied to this one primary factor, when the move to the “risk-off trade” is implemented, correlations rise and diversification goes out the window.  As such, we feel investors should use this data in an effort to analyze their own portfolio components to determine an accurate assessment of diversification, particularly for times when they need it most.   

Disclaimer: These materials have been prepared solely for informational purposes and do not constitute a recommendation to make or dispose of any investment or engage in any particular investment strategy.  These materials include general information and have not been tailored for any specific recipient or recipients.  Information or data shown or used in these materials were obtained from sources believed to be reliable, but accuracy is not guaranteed.  Furthermore, past results are not necessarily indicative of future results. The analyses presented are based on simulated or hypothetical performance that has certain inherent limitations.  Simulated or hypothetical trading programs in general are also subject to the fact that they are designed with the benefit of hindsight.