Continuing from part 1 of the article, in Exhibit 1, I have sorted the funds in ascending order by the correlation figure. Intuitively, those funds whose top 10 holdings face similar risk factors should have higher correlation figures and, thus, a lesser degree of diversification. Vanguard Telecommunications Services (VOX) had the third-fewest average number of holdings and highest percentage of its fund represented by its top 10 holdings. As sized by more rudimentary measures of diversification, this fund doesn’t stack up well. But exploring how its top 10 holdings interact with each other paints a different picture. VOX’s top 10 holdings had the lowest average pairwise correlations (28.4%) among the 20 sector funds in the sample. Conversely, Vanguard REIT ETF (VNQ) ranks well on more-basic measures, but closer analysis reveals that its top 10 holdings tend to flock together.
.png)
In Exhibit 1, we see that real estate, utilities, energy, industrials, and basic materials sector funds all had correlation figures above 0.50. This makes economic sense, as firms operating in these sectors tend to share similar fundamentals. For example, fluctuating oil prices affect most energy stocks, and interest-rate changes affect nearly all real estate stocks. On the other hand, telecom, consumer discretionary, consumer staples, healthcare, and information technology sector funds’ correlation figures were all below 0.50. Again, this owes to fundamentals. These sectors’ largest stocks are less correlated to the extent that they have more diverse business lines facing a more-varied set of risk factors.
Let’s take a closer look at Vanguard Consumer Staples ETF (VDC)) and VNQ. VDC’s top 10 holdings represented nearly 60% of its portfolio, but they weren’t very correlated. The fund’s top 10 holdings are spread across five GICS industries (household products, beverages, tobacco, food and staples retailing, and food products). So although the fund’s 10 largest holdings make up the majority of its portfolio (an effect of market-cap weighting), its intrasector diversification is better than most. Exhibit 2 displays the pairwise correlations of VDC’s top 10 holdings during the 10-year period ended Dec. 31, 2016.
-02.png)
Data for Philip Morris (PM) was not included because it doesn’t have 10 years of performance data (it was spun off from Altria Group (MO) in March 2008). Not surprisingly, the returns for Coco-Cola (KO) and PepsiCo (PEP) were the most correlated.
On the flip side, VNQ’s top 10 holdings represent about 35% of its portfolio, but these stocks have tended to move in the same direction more often than not. The real estate sector is not well-diversified. There is only one GICS industry in the sector, and it’s— you guessed it—the real estate industry. Exhibit 3 shows the pairwise correlations amongst VNQ’s top 10 holdings.
-03.png)
Equinix’s (EQIX) returns stand out as being the least correlated with the rest of VNQ’s top 10 holdings. This is because Equinix is actually a data center operator that wasn’t structured as a REIT until early 2015. Prior to restructuring, this stock was a member of the information technology sector.
Dig Deeper
In 1952, Harry Markowitz introduced the efficient frontier that laid the groundwork for Modern Portfolio Theory. One of his key insights was that an investor can reduce risk by holding a combination of assets that are not perfectly correlated. Investors eyeing concentrated market-cap-weighted funds should apply this framework at the fund level. Because market-cap weighting a small investment opportunity set can create concentrated portfolios, understanding how the fund’s top holdings interact with one another can shed more light on a fund’s true level of diversification.
:quality(80)/cloudfront-us-east-1.images.arcpublishing.com/morningstar/5FNGF7SFGFDQVFDUMZJPITL2LM.png)
:quality(80)/cloudfront-us-east-1.images.arcpublishing.com/morningstar/EOGIPTUNFNBS3HYL7IIABFUB5Q.png)
