Again, portfolio construction matters. When it comes to multifactor funds, the devil is in the details. Often, the details are many and they are nuanced. Here, we present a framework that will help investors to parse these funds' approaches to portfolio construction. This framework will help investors better understand these funds and (we hope) better manage their own expectation—which is critical to using them well.
1. What is the fund’s selection universe?
The selection universe, also referred to as a parent index, is the collection of potential stocks that a fund whittles down to build its investment portfolio. This is typically a broad index, like the Russell 1000. The selection universe should serve as a benchmark for the fund’s performance. It may also offer insight into the fund’s potential to outperform its parent index and/or category peers. For example, the payoff to most investment factors has historically been the greatest among the smallest stocks. This may be because they are more likely to be mispriced than larger stocks. So—all else equal—funds that start with a universe of large- and mid-cap stocks (as most multifactor funds do) likely have less potential to outperform than those that start with an all-cap universe or a group of small-cap stocks.
2. Which factors does the fund target?
There are only a handful of investment factors that truly matter. These include:
Factors that may enhance returns
- Dividend yield is arguably a subset of value, though some break it out as a separate factor
- Small size
Factors that may reduce risk
Each of these factors has been extensively and independently vetted in academic research and has tended to pay off in nearly every geographic market studied over the long term. But more important, there are reasonable economic explanations for why each of these factors has worked and will likely continue to pay off. These include compensation for risk, behaviorally driven mispricing, and institutional frictions.
In contrast to the other four factors, low volatility doesn’t aim to deliver higher returns than the market, but rather reduce risk and, in turn, potentially deliver better risk-adjusted performance than the market. While the low volatility factor can help diversify the others, it can disproportionately affect a fund’s performance (unless the fund explicitly limits active risk from this factor). It can also reduce the fund’s long-term return potential.
While there are myriad other factors, they either are not widely accepted, are not investable at scale (like illiquidity), or just repackage one or more of these core factors. It is best to stick to funds that target a combination of the core factors.1
3. How does the fund measure its targeted factors?
There are many ways to measure stocks’ exposure to each factor. For example, a fund could measure value based on each stock’s price/book, price/earnings, or enterprise value/EBITDA ratios, dividend yield, or some combination of those metrics. Sometimes one metric, or a set of metrics, will work better than another, but it isn’t clear that there is an optimal way to define value. What matters is that the chosen metrics are:
- Simple, so as to reduce the risk of data mining (where the strategies’ developers cherry-pick selection criteria to make a back-test look good)
- Transparent, so that investors understand what they are getting
- Clearly representative of the investment style to ensure the fund can achieve what it sets out to do
The specific metrics chosen tend to move the needle less than whether the fund measures each stock’s factor characteristics relative to its sector peers or the entire universe. There is a trade-off between these two approaches. A sector-relative approach leads to less pronounced sector biases than the universe-relative approach. Sector bias can be a source of uncompensated active risk that often isn’t necessary to capture the targeted factor.2,3 A sector-relative approach can also improve comparability across stocks (particularly for the value and quality factors), as firms in the same sector tend to have more similar balance sheets and profitability than firms in different sectors. The drawback is that it may reduce the fund’s factor purity, causing it to own stocks with weaker absolute factor characteristics than it would if it measured each stock against the entire universe.
One approach isn’t clearly better than the other, but funds that don’t control for sector differences would likely benefit from sector constraints, which can help improve diversification. After all, diversification is one of the core reasons to own a multifactor fund.
In part 2 of this article, we will continue to discuss the remaining two steps of the framework.
1 While the market risk factor, or beta, is an important driver of returns, we excluded it from this list because stocks with a lot of exposure to market risk have not historically offered attractive compensation for this risk.
2 Bryan, A. & McCullough, A. 2017. “The Impact of Industry Tilts on Factor Performance.” March 15, 2017.
3 Bryan, A. 2017. “Quality and Value Without the Sector Side Bets.” July 19, 2017.