Different ways to construct ETF portfolios

May 16, 2014 | Last updated on May 16, 2014
4 min read

When my colleagues and I started building ETF portfolios in 2001, we used a conventional core-satellite approach that’s popular with institutions and large pension funds. These institutions would build a passive core portfolio at low cost and seek active alpha generating strategies around that core.

ETFs made that approach accessible to individual investors and we were excited to offer it. The portfolio reflected the investor’s strategic asset mix. The satellite sought returns above the core.

We reasoned that the core would always be working towards the investor’s long-term goal, while the satellite should add value to the core return. Selection was simple because there were only 110 ETFs in North America and conventional capitalization-weighted indices dominated. While there is nothing wrong with this original approach, our methodology has changed to reflect our research into how groups of securities act when combined with other groups of securities. Today, we focus on where returns come from, and how value can be added. Here’s our new approach.

Read: How to manage a bond portfolio

Stock picking

As an analyst/portfolio manager scouring the U.S. for undiscovered small- to mid-cap companies, I enjoyed walking plant floors, visiting manufacturing and research facilities, going down mineshafts, and poring over financial statements and five-year-plans with company executives.

Getting an informational edge was possible then, but less so today, given tougher disclosure rules.

And something always bothered me about the portfolio construction process (see “Impact of stock selection,”). Intuitively, buying a portfolio of great companies sounded intelligent. But diversifying away the specific risk associated with any one of them seemed counterproductive.

In a one-stock portfolio, company-specific factors like growth rate, margins, market share and pricing power account for 78% of the return variance. Stock picking skills dominate.

However, when 50 stocks are combined, only 15% of the return variance is explained by stock-specific factors; the market explains 85%. Stock selection’s impact is diminished. If three 50-stock funds are held, the market explains 97% of the return variance, and stock selection accounts for only 3%.

So if you think you’re good at stock picking, concentrate your portfolio. Both Steadyhand and Brandes, for instance, promote concentrated portfolios. With only 25 stocks in Steadyhand’s North American equity portfolio, and 15 to 35 in Brandes’ Canadian Equity Fund, each has outperformed its respective benchmarks handily for one and three years (periods ending March 31, 2014). Capturing a market’s return through indexing is an efficient use of resources for many asset classes. Private equity investing, on the other hand, acknowledges the risk-reward payoff of concentrating on few holdings and watching them carefully.

Read: Analyze a company’s board before investing

Picking sectors

Foregoing stock selection for sector selection with ETFs, and picking the ones expected to do best, is a way to outperform the competition. But the same conclusion for stock selection applies. Concentrate your portfolio. But this is more difficult with ETFs because most are already broadly diversified.

“Impact of sector picking on returns” shows the contribution to return variance from sector-specific factors and the market for 10 U.S. industry sectors. The weight of each sector in the S&P 500 is shown, progressing from low (Telecom) to high (Information Technology). Utilities (47%), Energy (55%) and Healthcare (62%) have the lowest market impact on returns. Industrials’ returns, on the other hand, are largely explained (89%) by the market.

One active method of ETF portfolio construction is to select sectors that are expected to do well during different phases of an economic cycle. Something predictable happens when sectors are combined—the market’s influence over return variance increases.

We grouped three sectors to address each phase of an economic cycle (see “Impact of sector selection,” this page). Consumer Staples, Healthcare and Utilities were used as an example of a defensive cluster. When combined, only 13% of the cluster’s return variance was explained by sectors and 87% by the market.

Other groupings owed even more of their return variance to the market—expansion at 96%, and aggressive at 93%. We assigned different sectors to the economic phases, but any combination leads to a similar market-dominating result.

Read: Manage portfolios to fixed risk levels

In practice, a portfolio would likely include a broad-based index in addition to these sectors, making the market’s influence even more pervasive. So how can advisors add value when the market dominates return variance?

  • Pick the right markets and asset classes. As asset classes become more correlated, this becomes less effective.
  • Find a reliable method to determine risk-on and risk-off timing. Market timing is difficult but remains the most effective way to actively manage ETF portfolios.

Stay tuned for next month’s column, where we’ll look at the timing method we prefer.