By Jack Forehand (@practicalquant) —
Factor Investing has proven itself over long periods of time. But there is a price that has to be paid for its outperformance. The individual factors such as value, momentum and quality all can have extended periods where they underperform the market.
One way to smooth out the returns from the individual factors is to use them together in a multi-factor portfolio. When constructed properly, multi-factor portfolios can produce comparable returns to the individual factors but do so with less risk. However, the process of combining the factors is much easier in theory than it is in practice. Building multi-factor portfolios is a challenging process and if it isn’t done properly, you can easily dilute all the individual factors and end up with factor exposures that aren’t very different from the market as a whole.
For this week’s interview, we are going to take in depth look at the process of building multi-factor portfolios. My guest is Liqian Ren, the Director of Modern Alpha at WisdomTree. Prior to working at WisdomTree, she worked in Vanguard’s Quantitative Equity Group and managed their newly launched factor funds. She received her MBA and Ph.D. in Economics from the University of Chicago Booth School of Business.
WisdomTree’s suite of multi-factor funds include access to U.S., emerging markets and international equities. These funds each take one of the more interesting approaches to multi-factor investing I have seen. In this interview, we will break down the process of building multi-factor strategies and some of the challenges it presents.
Jack: Thank you for taking the time to talk to us, and congratulations on the new position at WisdomTree.
One of the first decisions that has to be made in constructing a multi-factor portfolio is which factors to include. Value and Momentum are commonly used by most practitioners, but outside of that, there can be wide variation in the factors that are used. I noticed that you utilize a unique combination of factors, including a correlation factor that I haven’t seen used elsewhere. Could you talk about the process you use to determine which factors should be utilized in a multi-factor portfolio and how you look at the balance between return and risk in selecting them?
Liqian: Thank you! I am very honored to join the WisdomTree team which is at the forefront of innovation, both in using the low-cost and tax-efficient ETF structure, as well as strategies designed to outperform the cap-weighted market.
There are three main innovations in the WisdomTree multi-factor strategy that stand out from typical multi-factor strategies. One is the usage of our correlation factor. Our international multi-factor strategies also use dynamic currency hedging as well as an active construction approach.
The guiding principal of factor selection and combination must start with common sense economic theory and human behavior roots, bolstered by empirical evidence. As recent papers such as Replicating Anomalies and Digesting Anomalies had shown, the number of enduring factors is limited and the ones that have economic rational and behavioral roots tend to survive the tests of time and investment universes. We never want to take the route of only data mining.
There is some consensus that value, momentum and quality are factors, which are usually constructed as a composite score of underlying stock characteristics. These three factors are widely researched and well-known both in academic and industry literature. Among the three, quality is relatively new and upcoming. A lot of shops that have stuck to traditional factors like value have found themselves left behind in recent innovations and in performance.
For example, WisdomTree’s multi-factor strategies construct the quality factor using static observations and trends such as Return on Equity, Return on Assets, Gross Profits over Assets and Cash flow over Assets. For the value factor, we use Sales-to-Price, Book-to-Price, Earnings-to-Price, Estimated Earnings-to-Price, EBITDA-to Enterprise Value and Operating Cash Flow-to-Price. For the momentum factor, we use risk-adjusted total returns over historical periods (6 and 12 months). WisdomTree strives to be transparent and innovative in generating what we call “Modern Alpha” for investors. The characteristics used and how we construct them are available on our website.
We believe in the power of utilizing factors that, when combined, help reduce volatility without sacrificing alpha, but we didn’t use traditional low volatility in an alpha stock selection sense as many other shops typically do. Over longer periods, low-volatility stocks have not consistently outperformed the market – though they do typically deliver on the promise of lower risk. But low volatility stock selection also may lead to uncompensated and undesirable interest rate factor sensitivity as well.
The low correlation factor we used also has an endogenous tilt toward the size factor as larger companies tend to have larger correlations. This allows us not to take an explicit size bet as the validity of the pure size factor has been somewhat challenged. We believe our approach shows that we don’t shy away from size exposure when the factors we pick lead us there.
We are seeing very large institutional clients we interact with shifting their mentality to use multi-factor indexes as benchmarks for mandates during the past three years and expect this trend to further catch on with financial advisors and retail clients.
WisdomTree designed its multi-factor strategies to go after higher active share with 85% active tilts to cap-weighted benchmarks, 5-6% tracking error, and higher outperformance targets than many of the original ‘smart beta’ factor strategies.
Jack: In global investing, the issue of whether and how to hedge currency risk can be a very important one. I have heard some argue that hedging currency risk involves unnecessary cost since over time currency movements may balance out, but others take the position that not hedging currency introduces uncompensated risk into a portfolio. WisdomTree is known for its work on currency hedging and I was wondering how you look at the issue of hedging currency in a global multi-factor portfolio?
Liqian: There are a lot of misunderstandings of currency risk and hedging, even in the professional finance community.
The typical narrative is that hedging is too expensive, and you won’t get paid. And indeed, hedging within emerging markets could cost dearly due to higher interest rate differentials in those markets. However, just as factor investing in equities has proven to add value, so too has factor-based dynamic currency hedging. We see a factor strategy for hedging emerging market currencies can lower the volatility of the emerging market strategy without sacrificing alpha.
For developed markets, WisdomTree has been a long-time advocate of strategic currency hedging as well. In particular, if you believe in the FX Carry Anomaly, one is “paid”to hedge in current macro environment due to the interest rate differentials vs. eurozone or Japan but this was also true over the last 30 years. With interest rate differentials as high as 2.5% a year today and growing, collecting the interest rate differentials by hedging these currencies makes removing uncertain currency moves a very useful and profitable proposition.
Companies that operate globally, whether U.S. or non-U.S., face inherent currency risk in their operations, thus in stock returns. A currency hedging strategy can help to enhance alpha and potentially reduce uncompensated risk. The more people stop and think about it, the more people may realize that the approaches WisdomTree has taken is ahead of the curve.
Jack: Once the factors are selected, another major challenge of building a multi-factor portfolio is how to blend them together. There seem to be two major approaches that managers utilize. One involves selecting stocks individually with each factor and then combining those individual sleeves together to form a portfolio. The other looks for stocks with exposure to multiple factors simultaneously. I have seen many arguments for each approach with proponents of each arguing that their approach offers more direct exposure to the factors and that the other approach dilutes the factors and results in a portfolio with limited exposure to any of them. Which approach do you utilize and what do you think its advantages are?
Liqian: We took the bottom up multi-factor approach, not the individual sleeves approach.
In the equal-weighted individual sleeve approach, suppose you have $1 and you have four single factor strategies (value, momentum, quality and low correlation), you first construct each strategy separately, then you invest 25 cents in each strategy.
WisdomTree’s multi-factor strategies start with ranking stocks according to the characteristics chosen for each factor. These ranked characteristics are equally-weighted to give each stock a composite single factor score, e.g. separate scores for value, momentum, quality and low correlation. Then a multi-factor composite is generated by equal-weighting these factor composites. For the international and emerging market multi-factor strategies, we generally follow, but don’t limit ourselves, to equal-weighting the factor composite for the multi-factor composite. We want to allow continuous research to lead us on whether and how to over/under weight a factor in the multi-factor score.
If you want to win a basketball game, you don’t want a team where one person is only great at 3-point shooting, the other only great in dribbling, and the other only great at passing the ball. You want a team with players who stand out at shooting but are also good at dribbling and passing the ball. An individual sleeves approach will invest in stocks that are high value and very low momentum or other factor dimensions.
As a result of this multi-factor construction, value stocks within multi-factor don’t necessarily behave the same as a stand-alone value strategy. We believe both philosophically and empirically that this approach is the better approach.
Jack: Factor timing has been one of the biggest debates in factor investing. If an investor could add exposure to a factor when it is out of favor and benefit from mean reversion when it bounces back, it could offer a nice boost to returns. But factors can go in and out of favor for long periods of time and timing their turn is very difficult, so this ends up being much more challenging in practice than it is in theory. Some of the biggest names in factor investing have gotten involved in this debate with Rob Arnott of Research Affiliates believing that tilting toward out of favor factors can boost performance and Cliff Asness of AQR advocating against it. Do you think factor timing can be used to enhance returns in a multi-factor portfolio?
Liqian: Yes, I believe that factor timing can add value, but only when the noise of the timing signal is managed. Like individual stocks, factors have their own momentum and value. Factor momentum is highly correlated with stock momentum, but it is not just stock momentum. And as factor investing may become more prevalent in the next 10 years, this effect could potentially strengthen.
There are generally two approaches in factor timing: fundamental-based factor timing or exogenous information-based factor timing. Fundamental-based factor timing relies on fundamental measures such as value, momentum, or quality of a factor. Factor momentum and factor value that use valuation spreads between top and bottom deciles of the factor strategy are the usual starting points, since it translates naturally from stock momentum and stock valuation. Some also use quality spreads instead of valuation spreads to time factors.
The other factor timing approaches rely on exogenous information, such as macro variables for economic cycle to time factors. I usually start from the fundamental-based factor timing but keep my mind open to macro or exogenous variable-driven factor timing. Empirically, the relationship between factor returns and macro conditions have been not very stable. Whether and how macro variables leads factor returns is murky in theory and empirical evidence. But this is also where the reward could be larger because the source of alpha is less correlated with existing stock selection.
But factor timing signals are very noisy. You also only have 5 or 6 factors to choose from, unlike cross sectional stock picking where you have a couple hundred even thousands of stocks to choose. The noise should be managed to harvest the value of factor timing.
Jack: Portfolio sizing can also present a difficult challenge in building a multi-factor portfolio. On one hand, focused portfolios can offer the most direct factor exposure, but on the other, they can may lead to significant tracking error, which makes it difficult for investors to stick with the strategy. How do you look at the balance between factor exposure and tracking error when building a multi-factor portfolio?
Liqian: Generally speaking, as a fund grows in size, trading costs increase more than linearly, and this could limit the amount of alpha yielded from a multi-factor strategy. But capacity analysis I have done using past transaction cost assumptions usually put the upper capacity of a given strategy in the $50-100 billion range.
Besides, transaction costs have been decreasing consistently over the last 20 years, and it may be a long time before we need to worry about this limit. This is true not just for a multi-factor strategy but any strategy that requires turnover. Even large index funds can face significant transaction costs during index reconstitution.
A highly concentrated strategy could potentially have very high factor exposure but under-perform for a very long time and is why we try to maintain a healthy balance of factor tilts across all our factors. We put in risk control constraints that try to balance between factor exposure and tracking error. For example, we limit active country and sectors weights to under 5% in our international multifactor strategy while applying sector neutral constraints in our U.S. multifactor strategy.
Within our emerging market and international multifactor strategies, our active approach gives us flexibility in continuously researching every aspect of the methodology, which makes us unique as well. A lot of multi-factor products on the market right now are constrained to index limitations as soon as the products get launched. I am happy that as Director of Modern Alpha, our team will be continuously working to enhance and improve our investment process.
Jack: Thank you again for taking the time to talk to us today. If investors want to find out more about you and WisdomTree’s approach to multi-factor investing, where are the best places to go?
Liqian: Our website is the best place to start. Here are a few links to some of our recent blogs that highlight our thoughts and research on multi-factor investing:
You can follow WisdomTree on Twitter at @WisdomTreeETFs.
 Source: Kewei Hou, Chen Xue, Lu Zhang. “Replicating Anomalies.” The Review of Financial Studies, 10 December 2018, https://academic.oup.com/rfs/advance-article-abstract/doi/10.1093/rfs/hhy131/5236964?redirectedFrom=fulltext
 Source: Kewei Hou, Chen Xue, Lu Zhang. “Digesting Anomalies: An Investment Approach.” The Review of Financial Studies, Volume 28, Issue 3, 1 March 2015, pp. 650–705, https://academic.oup.com/rfs/article/28/3/650/1574802
Photo: Copyright: dacasdo/ 123RF Stock Photo