Decomposing Risk: How Popular Holdings Analysis could be Misleading
by Pauline Shum Nolan, PhD, Co-founder & CEO and
Simiao Zhou, PhD, Director of Research
When it comes to evaluating a portfolio’s sector risk exposures, analysts typically take stock of a portfolio’s holdings, and then group them by sector. In the pie chart below, we show the breakdown of the S&P 500 constituents by sector. This type of holdings analysis (or breakdown approach) provides a useful point-in-time snapshot of a portfolio. However, we think that it can give an incomplete - or at times, even a misleading - picture of a portfolio’s risk exposures. In this article, we use the energy sector to explain why.
First, an obvious limitation of the methodology is that it can only be applied to portfolios where consistent holdings information is available. In a diversified, multi-asset class portfolio, investors may hold equities, fixed income, and alternative investments. For these portfolios, holdings information is not aligned. For example, equity sectors are classified according to the Global Industry Classification System (GICS), which is based on the industry that a business operates in. Fixed income sectors, however, are classified by issuer type, such as government (federal, provincial, municipal) or corporate (AAA, AA, A...etc.). Alternative investments will again have their own classification, for instance, commodity, real estate, or currency. Therefore, holdings analysis is typically applied to an individual asset class only, and not to the entire portfolio.
Second, even within an individual asset class, say equity, we know from basic finance principles that part of a security's sector risk may be diversified away. Consider the following portfolio that is made up of seven equity ETFs (ticker in brackets):
Emerging Markets (EEM) 20%
S&P 500 (IVV) 10%
US Telecom (IYZ) 10%
US Small Cap (VB) 10%
US High Dividend Yield (VYM) 30%
US Materials (XLB) 10%
US Utilities (XLU) 10%
Holdings analysis shows that 6.2% of the portfolio’s holdings operate in the energy sector. But what is the portfolio’s overall risk exposure to the energy sector? Using statistical analysis to decompose the volatility of the portfolio, we find that it is essentially zero, because the diversified portfolio experienced no statistically significant change in returns when the energy sector fluctuated over the 10-year time period, April 2007 to April 2017.
Let’s look at one more example. This time, the issue stems from industry practice. Consider this ETF portfolio:
US Materials (XLB) 25%
US Industrials (XLI) 25%
US Utilities (XLU) 15%
Commodity-Index Trust (GSG) 35%
The typical holdings analysis would show that there are no energy holdings in this portfolio, where in fact the energy sector accounted for 36.6% of the risk in the portfolio over the 10-year period, while fluctuations in the price of oil accounted for 38.8%. The discrepancy comes from the fact that GSG is an alternative investment that tracks a diversified commodity index and as such, does not conform to the GICS sector classification for equities. Even though 58% of GSG consists of energy products, the industry practice is to exclude GSG, and performs a holdings analysis on the equity portion of the portfolio only.
The preceding examples illustrate the main point of this article: there are two different ways of looking at risk exposures in your portfolio, and they can yield different conclusions. Despite the caveats discussed above, holdings analysis is based on the portfolio’s current holdings so the information is the most up to date. Statistical risk decomposition takes into account the effects of portfolio diversification and does not depend on any classification systems. It can also take into account the indirect effects of a risk factor. The disadvantage, however, is that historical portfolio returns must be available.
Hence, on Wealthscope, we provide our users with both views.