Dig Deeper When Measuring Diversification (Part 1)

The more the merrier?

Adam McCullough, CFA 17 April, 2017 | 17:02
Facebook Twitter LinkedIn

An index fund that follows a well-constructed bench­mark and levies a low fee is usually a winning proposition. Market-cap-weighted indexes are particu­larly attractive, given that they outsource the work of price discovery to active market participants and tend to have low turnover. Market-cap weighting typically works best in those instances where an index is broadly diversified—as in the case of a total U.S. stock market index, for example. But market-cap weighting stocks within a smaller investment opportunity set, such as a sector, can create a concentrated port­folio and may introduce a greater level of stock-specific risk. How should investors evaluate a concen­trated fund’s level of diversification or lack thereof?

Here I take a look at a variety of metrics one might use to size up a fund’s level of diversification. I apply them to a sample of equity sector ETFs listed in the U.S. These funds’ underlying indexes market-cap weight stocks from narrow investment opportunity sets (U.S. equity sectors). State Street Global Advisors and Vanguard offer sector ETFs with track records spanning at least 10 years. Vanguard offers an ETF covering each of the 11 Global Industry Classification Standards sectors. State Street offers just 10 sector ETFs, as it combines the information technology and telecom sectors into one fund: Technology Select Sector SPDR ETF (XLK). Also, given its brief track record, I exclude Real Estate Select Sector SPDR (XLRE) from this analysis.

Common metrics investors might use as a proxy for a fund’s degree of diversification include (1) its average number of holdings and (2) the percentage of the fund’s assets invested in its top 10 holdings. But these are blunt instruments. To address their shortcomings, I also assess these funds’ level of diversification by looking at the average pairwise correlations of their top 10 holdings. All of the data I used was sourced from Morningstar Direct and spans the 10-year period ended December 2016.

The More the Merrier?
Many investors are familiar with the rule of thumb that an equity portfolio is well-diversified if it holds at least 30 stocks. This concept can be traced back to a paper published by Lawrence Fisher and James Lorie in 1970, which concluded that a 32-stock portfolio’s returns are 95% less volatile than those of a single stock. So, if a fund holds more stocks, it should be less risky, right? Not so fast. Simply counting the number of holdings in a portfolio does not consider the weighting of each security, and market-cap-weighted funds by definition do not hold positions in equal weightings. Like most rules of thumb, the number of holdings in a fund provides a starting point to measure diversification, but we need to dig deeper.

The percentage of a fund’s assets represented by its top 10 holdings is another lens through which to view its level of diversification. Unlike a fund’s holdings count, this measure does incorporate stock weight­ings. The largest stocks tend to dominate market-cap-weight funds that track narrow opportunity sets. By this metric, if a fund’s top 10 holdings represent a smaller percentage of its portfolio, then it should be—in theory—better diversified than those funds whose 10 largest holdings soak up the bulk of their assets. However, this measure also has shortcomings: Specifically, it doesn’t consider whether the prices of a fund’s top 10 holdings tend to behave similarly.

Be Different
Analyzing stocks’ correlations with each other provides a clearer sense of the true level of diversification in an equity portfolio. To assess the interaction between the largest holdings of the funds in my sample, I measured the pairwise correlations of the monthly returns of each fund’s top 10 holdings during the 10-year period ended December 2016. I then took the average of these values to gauge each fund’s degree of diversification. A fund should be better diversified if its holdings are less correlated. Conversely, a higher level of corre­lation amongst the stocks in a portfolio implies a lesser degree of diversification—irrespective of how many stocks the fund holds or how large a portion of its portfolio is allocated to its 10 largest holdings. Exhibit 1 shows the average number of holdings, the percentage of the portfolio represented by the top 10 holdings, and the average of the pairwise correla­tions for the top 10 holdings for each of the 20 equity sector funds I examined through the end of 2016.

170413 diversification(en)

In part 2 of this article, we will continue to look at diversification deeper. 

Facebook Twitter LinkedIn

About Author

Adam McCullough, CFA  Adam McCullough, CFA, is an Analyst on Morningstar’s Manager Research Team, covering passive strategies.

© Copyright 2023 Morningstar Asia Ltd. All rights reserved.

Terms of Use        Privacy Policy         Disclosures