How can we compare portfolio risk between competing funds based on their holdings?
How do we know what typical levels of portfolio risk are, and what is considered high or low versus competitors?
The information that investors typically use to compare risk is based on backward-looking returns. But if we are comparing funds, we need answers about the actual portfolios that we are investing in now. Let’s take a look at some practical ways that holdings-based portfolio risk measures can help answer these questions.
Portfolio Risk Rank graphic below, taken from Peer Insights™, shows a summary of holdings-based (not returns-based) portfolio risk measures for 245 global equity products in the Morningstar World Stock universe of US domiciled funds, as at the end of March 2016. We’ve selected two products for comparison, the Oakmark Global Fund (“Oakmark”) and the RBC Global Opportunities Fund (“RBC”).
Stock Diversification – Actual and Effective
Both portfolios have an almost identical number of stocks, with Oakmark at 36 and RBC at 35, both as at the end of March 2016. The typical number of holdings (i.e. the median or 50th percentile) in the Morningstar World Stock universe is 85. So, both of these products are much more stock-concentrated than the typical global equity fund in this universe. In fact, they are both at the 10th percentile in this universe. Of course, the number of holdings in a portfolio can be misleading. A portfolio could have a significant concentration in a few stocks and a long “tail” of very small positions.
Effective Number of Stocks
One measure which takes into account the concentration of holdings weights is called the Effective Number of Stocks. The median effective number of stocks in this peer group is 62. The effective number of stocks is 31 for both Oakmark and RBC products and, since this is fairly close to both funds’ actual numbers of 36 and 35 respectively, this also means that the stock weights are reasonably balanced within each portfolio.
If we take into account the weightings of stocks in the benchmark, we can calculate the “active share” of each portfolio versus the MSCI World index. This can be calculated very simply by adding up all the active weights of all the overweight stock positions in a portfolio (which will be symmetric to the negative of the underweights sum). The typical active share for this universe is 88.8%. Oakmark has an active share of 95% and RBC has an active share of 94.3%, i.e. almost identical. So these are both fairly concentrated, high active-share portfolios and they are both ranked in at least the top quartile for the measures so far.
Predicted Tracking Error
However, the challenge with active share is that it implies that each 1% of active stock difference contributes the same to portfolio risk – no matter the stock. This ignores volatilities and correlations from sectors, countries, styles, and individual stocks. We can incorporate all of that using the Style Research “Predicted Tracking Error.” This is calculated using the active weights (portfolio minus benchmark) applied to a proprietary covariance matrix and further risk calculations. The predicted tracking error, which is also known as “active risk,” is typically expressed as an annualized percentage. It is essentially a forward-looking estimate of relative volatility of the portfolio versus the benchmark.
We can see for example that, versus the MSCI World index, the typical predicted tracking error is 3.6% per annum for this particular universe. Oakmark has a predicted tracking error of 5.5% per annum, which seems consistent with its levels of stock concentration and active share in terms of peer ranking. Despite having some significant stock concentration and a high active share, it is very interesting that RBC has a typical predicted tracking error of 3.6% per annum. Given that Oakmark’s active risk is more than 50% higher than RBC, we might expect more significant relative performance versus the index from Oakmark (good or bad).
Conclusion* – Ranking Risk
Oakmark’s risk rankings are quite consistent no matter which measure is used, with a high active risk consistent with a higher stock conviction approach. It is interesting to note that RBC has a lower active risk than might initially be suggested using the number of stocks, the effective number, or even active share. In our next post we’ll examine why this is by examining the sources of risk in each portfolio. We’ll also show how combining style and risk insights can lead to more powerful fund differentiation, and therefore more efficient fund allocation for investors.
It is important to stress that we cannot say whether a fund is intrinsically “good” or “bad” based on any risk measure. As with all of our portfolio analytics, these are independent and objective results. Each of these funds could be a perfect fit for two different clients, or perhaps even for the same client diversifying within their own portfolio. The fit will depend on the risk tolerance of the investor, as well as other facets of the portfolio, such as the style orientation.