By Clifford S. Asness, Ph.D. and John M. Liew, Ph.D.
Let’s be blunt. Smart Beta1 is mostly re-packaged, re-branded quantitative management. That’s not to say we don’t like it or think it’s not good for inves.tors. We love quantitative management, hav.ing spent our careers pursuing these types of strategies. However, we work in a business where good ideas are constantly repackaged as something new. Smart Beta is the latest ex.ample. It takes well-established, quantitative investing styles, or factors, and implements them in a simple, transparent manner often, though not always, at lower fees than what we’ve seen in the past. That certainly sounds like a worthwhile repackaging, and it’s not surprising that Smart Beta has received great attention.
Ironically, despite Smart Beta’s aim for transparency and simplicity compared with traditional active (judgmental or quantitative) management, there remains much confusion about what Smart Beta means (sometimes the marketing can serve to obscure not enlighten2). This likely stems from the fact that there are so many related concepts and so much overlapping terminology. Furthermore, there are a wide range of styles, implementation methods and marketing techniques. For example, fundamental indexing, low-volatility investing, systematic value or momentum investing, each implemented as ’40 Act Funds, separate accounts or ETFs, all could be classified as Smart Beta.3 Moreover, some of these focus on single factors while others use multiple factors.
In this article, we try to put Smart Beta into perspective by categorizing the major approaches
and describing how each might contribute to a broad portfolio. In doing so we extend the concept from the very basics, to what we believe are logical next steps. Exhibit 1 lays out our plan of attack. We will start out with Smart Beta as it is most commonly implemented (long-only using a single factor within a single asset class — usually equities). We will then broaden out to multiple factors. Finally, for a bigger jump (in both efficacy and risk/unconventionality), we’ll go to multiple asset classes in a long/short framework.
Basic Smart Beta: Single Factor Within a Single Asset Class (Long-Only)
Most Smart Beta funds today are long-only equity strategies focused on one investment style or factor4 and these generally come from the academic and practitioner literature. This fact was underappreciated, even denied, by some of the creators of early versions of what came to be called Smart Beta products. However, today it’s generally accepted that Smart Beta is a repackaging of these well-trodden ideas.
To be considered Smart Beta, we believe that these factors must also be simple and transparent. However, they don’t have to be the same for all managers or products. One can, and many do, argue that their particular version of simple value, low risk or any other tilt is better. These are the same arguments quant managers used to have with each other. However, in the case of Smart Beta, these factors are implemented in a somewhat simpler, more understandable framework (until we all start arguing and obscure it again).
Furthermore, they are implemented either directly (e.g., tilting toward or restricting investments to value or low-risk stocks) or through portfolio construction choices that give exposure to these styles (e.g., equalweight or weighting by non-price measures of size). Some argue a lot about which of these should actually be called Smart Beta. (While we like to argue, we’re going to skip this one.5)
Let’s be clear about a few things that are sometimes obscured:
• Rebranding old school “quantitative management” into Smart Beta doesn’t change the fact that the factors must work. You want to avoid the perils of data-mined factors or strategies that were chosen to fit history, but have a little hope of future returns. To help achieve these aims, an important criterion for any viable factor is that it have a good economic story behind it — there must be a reason it is expected to deliver positive excess returns. In addition, robust out-of-sample evidence — over time, in other markets and even in other asset classes — can help assuage data-mining concerns.
• One difference between Smart Beta and traditional quantitative management is that the former is usually built from the bottom up with “tracking error” against a benchmark falling out (though outperforming a benchmark is still the goal). Another is that while traditional quantitative investing generally bets on many of the same factors, it tends to be more about the particular manager’s improvements (real or data-mined) to these factors, new factors (real or data-mined) that the manager has found, and the manager’s particular portfolio construction and implementation methodology, among other things.
• Even though we call this “Smart Beta,” it’s not really beta in the traditional sense (we considered “Clever Beta,” “Ingenious Beta,” “Pensive Beta,” “Perky Beta” and, of course, “Ennui-Stricken Beta,” but they didn’t make it past the drawing board). Smart Beta is an exposure to market-capitalization-weighted indices (traditional beta) plus an attempt to outperform these indices (remember when we used to call this alpha?). Perhaps the simplest equation in the world is:
Smart Beta = Cap Weight Index + [Smart Beta – Cap Weight Index]
• The second part, [Smart Beta – Cap Weight Index], is usually a very well-defined simple tilt in the direction of known factors. For instance, take the case of an index where you weight stocks not by their market cap, but instead by their sales. In this Smart Beta index (this of course follows for any such measure), the ultimate weight of a stock will be its weight in the Cap Weight Index multiplied by the price-to-sales ratio of the market (P/SM) divided by the price-to-sales ratio of the stock (P/Si). It’s that simple. In other words, it’s not “kind of like a price-to-sales tilt.” In this case it is, and always is, only a specific price-to-sales tilt. This index owns precisely the below amount of stock :
Weight in Smart Beta Portfolio = [Weight in Cap Weight Index]
The value-add in Smart Beta is that we, and many others including a legion of researchers looking in a legion of places, believe certain factor tilts work reliably over time, and that simple versions such as a one-factor Smart Beta strategy can be understandable, transparent and available at a reasonable fee. We believe there can be tremendous value add to this undertaking and we’re here to praise, not bury Smart Beta — but only after an honest understanding of it.
Click to see the full sized version.
Among the various investment styles, let’s consider the efficacy of three that are well known (and admittedly chosen as near and dear to our hearts): Value, Momentum and profitability. Based on our analysis and a large body of academic and practitioner literature, each has produced long-run, hypothetical excess returns with low correlation to traditional markets over multiple decades, in multiple geographies and asset classes, and each is well-supported by economic theory and research. We believe that each of these, at the proper fee, would be an attractive proposition in a single factor, simple Smart Beta format and should add value to most traditional portfolios.
Furthermore these styles’ excess returns tend to be lowly correlated with one another, with performance often coming at different times. This can be really important. Consider Exhibit 2, which separately shows the worst three-year hypothetical excess returns for each style along with the hypothetical performance for the other two styles during that same period. In each case, the worst performance for any one style is mitigated by the other two. Diversifying across various Smart Betas can and has provided a more consistent way of beating a traditional benchmark.
If multiple styles are better than just one, how should investors combine them? There are two broad choices (ignoring the many narrow choices like how best to get a value tilt) while staying with traditional long-only investing: (1) invest in single-style funds a la carte (e.g., a separate value fund, a momentum fund, etc.), or (2) invest in a single fund that integrates multiple styles simultaneously. Assuming fair fees for the risk taken, both approaches can be useful. However, we believe that the second choice — multi-style—is naturally more efficient for at least three reasons:
Lower Transaction costs: The multistyle approach can net different style signals before trading, avoiding needless turnover and its associated costs — and for taxable investors, potentially lowering their tax burden.
Interaction benefits: A stock that might barely fall below the threshold for being included in separate, single-style funds may be among the most attractive when viewed across multiple styles. Fur.thermore, if deemed warranted, more weight can be given to stocks that are attractive on multiple style dimensions.
Maintaining active risk: Combining separate single-style funds into an in.vestor’s portfolio lowers “active risk” (versus the relevant benchmark) be.cause the style tilts will tend to partly offset each other. This can be a good thing, except if the expected excess return comes out too low (because the active risk gets too low). In an integrat.ed, multi-style fund, the amount of ac.tive risk can be chosen so this does not occur.
We’d be remiss in not pointing out that the advantages to integrating multiple styles in one fund are not free. They generally involve less diversification across managers and less explicit end-investor control of factor exposures. Perhaps most importantly, while fixable with transparency, the multi-style tilt does lose some of the beautiful simplicity that perhaps first attracted investors to Smart Beta. We think it’s a natural evolution and believe the hard quantifiable advantages outweigh the soft disadvantages, but we mention them and leave them for the reader to decide.
Click to see the full sized version.
The Next Step: Multiple Factors Within Multiple Asset Classes (Long/Short)
Now we take it to its logical, though admittedly more-provocative and less-conventional conclusion. Remember our simplest equation in the world? Smart Beta = Cap Weight Index + [Smart Beta – Cap Weight Index] This equation tells us that long-only Smart Beta is a tie-in sale. When you buy it, you actually get two things. One is a traditional index fund (Cap Weight Index) and the other is a long/short portfolio that bets on the style underlying the particular Smart Beta product (Smart Beta – Cap Weight Index).11 In a long-only Smart Beta product you don’t get to choose how much of the long/short stuff you get. It just falls out of the particular way the product was constructed. There might be some legitimate reasons to buy these together, but why not consider breaking this tie-in sale? Isn’t that the next logical step? By the way, if you’re laughing a bit as we extend the “Smart Beta” concept by removing the actual “Beta” part, you are paying attention! We call this long/short portfolio style premia but it is really just Smart Beta where you’re going long the smart part and short (and we apologize for this...) the dumb part! Not only can a long/short strategy allow investors to explicitly pursue “both sides” of an investment style, it also creates a portfolio that is naturally hedged to the underlying market. That is, while long-only Smart Beta will strongly tend to rise and fall with equity markets, long/short style premia may offer a more diversifying, truly alternative return source.
Other Asset Classes Besides Equities
Since we’re going unconventional, why stop at equities? Style premia, such as value, momentum, low risk and others, can be harvested in other asset classes, including bonds, currencies and commodities.14 Importantly, these are not new, original styles tailored to these other asset classes, but the same ones being used in long-only equity Smart Beta. If you believe cheap stocks have higher expected returns than expensive stocks, why would you believe that it ends at stocks? Is valuation a phenomenon that is unique to stocks? Shouldn’t cheap bond markets outperform expensive bond markets? Shouldn’t cheap currencies outperform expensive currencies? The data suggest that the answer to each of these questions is yes. It turns out that many of the styles that work for predicting stock returns also work in other asset classes. And why shouldn’t they? Given the economic intuition and the empirical evidence, it’s hard not to make the small logical jump to implementing these styles in these other asset classes, too. However, the long-only smart beta construct makes benefitting from ALL the places Smart Beta works quite difficult.
Using simulations from 1990–2013, we find a significant benefit from combining style premia and implementing them across multiple asset classes. While simulations no doubt paint a rosier picture than what the future may bring, we believe that the results are directionally accurate. Namely, with style premia, diversification across both styles and asset classes can make a meaningful difference (this applies to both single and multi-factor versions). Again, we challenge someone to believe in a Smart Beta factor for a long-only equity tilt, and not believe in that same factor, with similar evidence, in other places.
Conclusion: Smart Beta in Perspective
Smart Beta and style premia investing bring decades of academic research to investors, and can help investors better understand — and potentially more reliably capture — sources of value added returns. They may be “less new” than sometimes advertised, but still often bring important innovation in the areas of transparency and clarity (though the clarity sometimes comes only after fighting through the marketing).
However, confusion exists on how best to compare and decide among different products. While there’s still plenty of room to argue about which are the best factors and best ways to implement and combine them, our effort here is to help clarify the true variety and objectives of these strategies, and the true range of possibilities.
Finally, many investors liken Smart Beta to an index strategy, which often results in manager selection based on fees alone. We caution against this (though we are happy to compete on this basis). While they share some elements with index strategies, such as transparency and rules-based decisions, Smart Beta and style premia are active strategies in the sense of trying to outperform. Though the concepts underlying them may appear straightforward, as Exhibit 1 shows, the amount and quality of active sources of returns varies, and this can determine how well or poorly the portfolio performs. We don’t eschew analysis based on fees. Rather, we encourage and advocate for it. Just keep in mind that, as with any strategy, we believe it should be done considering what is purchased, not just what is spent.
AQR is a global investment management firm that employs a systematic, research-driven approach to manage alternative and traditional strategies. As of June 30, 2014, we managed approximately $113 billion for institutional investors and investment professionals.