This article was written by Invesco PowerShares Vice President, ETF Product Management, John Feyerer.
Diversification. As the old saying goes, it’s the only free lunch in finance. Construction of investment portfolios involves mixing low or uncorrelated asset classes with varying risk/return profiles in order to attain the desired balance of risk and return potential. Correlation is a statistical measure of how securities move in relation to each other. For investors to benefit from diversification, they must combine assets that are less than perfectly correlated. The lower the correlation, the greater the diversification benefit. (While low correlations are good, negative correlations are even better.) Of course, diversification does not guarantee a profit or eliminate the risk of loss.
Style investing as a means of diversification
Traditional style investing, commonly used in the asset allocation process, is one means of diversifying a portfolio. A simple example of traditional style investing involves employing both growth and value investment styles across a full spectrum of company sizes – large-, mid- and small-cap. This approach is designed to improve performance, while reducing portfolio risk. Style allocations can also be used to tilt an investment portfolio based on an investor’s market outlook.
Historically, investor implementation strategies have focused on finding active managers who could deliver “alpha,” or risk-adjusted performance, to fill each style box allocation. In recent years, passive strategies have also been widely adopted, as evidenced by $160 billion in net flows into style-based index mutual funds and $54 billion in net flows into style-based index exchange-traded funds (ETFs) over the past three years. During this time, style-based ETFs have grown to represent approximately 10% of the $2 trillion in US ETF assets under management.1
Constituent overlap can increase portfolio risk
Given the desired outcome of improving performance while reducing portfolio risk, we believe it is important for ETF investors to “look under the hood” of the more popular style indices to understand the index construction methodology. Russell and Standard & Poor’s (S&P) provide some of the most widely used style indices, and both have at least 30% constituent overlap between growth and value allocations, as shown in the graphic below.
What do we mean by this? An examination of index holdings illustrates that a number of companies have their weight apportioned between both the growth and value indexes based upon their strength of style. Index providers typically structure their style indexes in this way to provide exhaustive coverage (so that all parent index stocks are included in these benchmark indexes) and cost-efficient* exposure to the broad style market. But the overlap of constituent companies within these indexes typically results in higher correlations, which can help reduce diversification and increase portfolio risk.
Given that nearly one-third of both the S&P and Russell style indexes contain the same stocks, investors should evaluate the degree to which they can benefit from diversification through a reduction in correlation.
Pure style investing eliminates the issue of constituent overlap
Recently, Russell introduced a suite of pure style indexes designed to include only stocks with pure growth and pure value characteristics. Unlike traditional style indexes, the Russell pure style methodology eliminates overlap and weights constituents based upon relative style attractiveness. This approach not only eliminates the issue of constituent overlap, but also focuses the exposure on companies that exhibit the greatest style strength.
The table below illustrates these differences. Notice that the Russell growth style and the Russell value style both include four of the same large companies, while the Russell pure growth style and the Russell pure value style have no constituent overlap.
Now let’s consider the impact on correlations. In the table below, note how highly correlated Russell’s traditional value and growth style indexes are across the various indices: 0.79, 0.73 and 0.84 (a level of 1.00 reflects perfect correlation). This isn’t entirely surprising when you consider that a portion of each index consists of the exact same companies.
Now observe the lower correlations between the various Russell pure value indexes: 0.58, 0.44 and 0.64. Russell’s pure style indexes show an average reduction in correlation of 30% when compared to Russell’s traditional value and growth indexes — driven by the fact that there isn’t any constituent overlap between the pure style indexes, as well as the fact that constituents are weighted based on style strength, rather than on market capitalization.
In order to benefit from the “free lunch” that is afforded by diversification, we believe investors need to pay close attention to the correlations between the allocations within their portfolio. Some of the most commonly used style indices have significant constituent overlap – resulting in higher correlations, and thus hampering the ability of investors to reduce portfolio risk. By employing a strict construction discipline, Russell pure style indexes isolate companies that exhibit stronger style characteristics – resulting in a sharper, more focused and stylistically pure index.
Learn more about Invesco PowerShares’ suite of ETFs based upon Russell’s pure style methodology.
1 Source: Morningstar, March 31, 2015
* Since ordinary brokerage commissions apply for each buy and sell transaction, frequent trading activity may increase the cost of ETFs.