What s in a Name? IN THE CASE OF SMART BETA, IT S HARD TO TELL

Transcription

1 What s in a Name? IN THE CASE OF SMART BETA, IT S HARD TO TELL Sebastian Ceria, PhD, CEO Melissa Brown, CFA, Senior Director, Applied Research Ian Webster, COO Robert Stubbs, PhD, VP of Strategic Innovation Axioma, Inc. August 2016 Introduction Demand for passive investment strategies has exploded, and none has attracted more attention than smart beta. According to The Wall Street Journal, data from Morningstar shows that there has been a steady inflow into passive strategies since at least 2008, with more money flowing to passive versus active strategies since In 2015, more than $207 billion left active strategies, while almost $414 billion went into the passive category. And this is not just driven by private clients: according to FTSE Russell, the percentage of asset owners currently evaluating smart beta products has doubled since The attraction is plain to see. Passive investment via ETFs and other such products is easily accessible, has lower fees than traditional active management, and in the case of smart beta products, has the potential to outperform a cap-weighted passive index. So it is hardly surprising that these products are being aggressively marketed.

2 But do ETF buyers, especially those seeking smart beta strategies, really know what they are getting? Is it alpha? The term smart beta has been applied to ETFs that attempt to capture well-known investment strategies such as high dividend yield, value, and low volatility with transparent, rules-based approaches that come with lower fees than traditional investments, better tradeability, and other attractive characteristics. One provider defined smart beta as an index-based investment strategy that is not capitalization weighted (i.e., fundamental weighted, equal weighted, factor weighted, optimized, etc.) defining it by what it is not, not by what it is. We believe investors should know what the fund is. To address the question of whether buyers are getting what they think they are getting, let us first go back to the beginning and review Grinold and Kahn s 1995 seminal work that defined the sources of return. Exhibit 1. Grinold and Kahn: Sources of Equity Returns While CAPM would assert that all returns should fall into the passive category, Grinold and Kahn (and we) believe there are sources of alpha. Alpha can be systematic also known as factor investing or pure, i.e., a source of return that is uncorrelated with markets, elusive, and much more difficult to systematize. In this taxonomy, systematic returns have been taken over by the idea of smart beta. To further examine the concept of systematic alpha, we look to Ross s generalization of CAPM, Arbitrage Pricing Theory (APT). APT states that returns of a security comprise those from the beta to a market factor, plus returns that result from betas of other systematic factors, which are then divided into style factors, such as momentum, value, and size, along with other drivers, such as industry, country, and currency. We could include macroeconomic factors as well. Of course, there is a residual return that is uncorrelated to any of the other factors in the model. 1 Grinold and Kahn, Active Portfolio Management, 1995 What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 2

3 Exhibit 2. Ross s Arbitrage Pricing Theory as a Generalization of CAPM r = ß Market ß Style ß Industry ß Country r = ß R f R + ɛ f Market f Style f Industry f Country + ɛ, All of this theory leads us to the task at hand, which is to discuss smart beta strategies in more detail. These factors ultimately represent the systematic return that investors want to capture, and smart beta ETF providers need to construct portfolios that attempt to capture the factors. This can be done in a number of ways. For example, one could create decile portfolios, or do as we at Axioma do, create factor-mimicking portfolios ( FMPs ) 2. One must also consider how to build the portfolios, including choices such as which stocks to include or exclude and how to weight them. Portfolio construction options include optimization to maximize exposure to one or more factors while possibly minimizing exposures to other factors, choosing the top stocks and weighting them by capitalization, equally-weighting them, or perhaps risk-weighting them. Some products choose a benchmark, exclude the lowest-ranked stocks, and then reweight what is left. And then there is the choice of rebalance schedule, and efforts to minimize trading costs. In sum, the details matter. Capturing risk premia associated with systematic alpha is not just about identification of the factors, but also about how one creates a portfolio to capture them most effectively. 3 For many ETFs, the rules used to create these funds initial universe, weighting schemes, and other constraints vary widely, despite the similar sounding names. Some of these vehicles have done a much better job of capturing their marquee feature than others, while many that sound the same actually look quite different. In this paper, we focus on a few types of smart beta portfolios in order to highlight similarities and differences driven by methodology. But first, a short review of the basics. What is smart beta? The idea of capturing systematic returns has been around since the beginning of quantitative ( quant ) management. In fact, packaging those returns into transparent, replicable, and rulesbased products is what quants have been striving for from the early days. Smart beta portfolios attempt to target an index that has been created with specific rules in mind, with the added benefit of having a catchy name to help in the marketing of the funds. Just as quant manager is an umbrella term that ignores many of 2 FMPs are long-short portfolios with unit exposure to the factor in question and no exposure to any other factor. These portfolios are highly levered and generally unimplementable, but they do represent the purest representation of a factor return. 3 Note that it is not obvious how to define a factor. For example, Value can mean a number of different things to different investors. It is not the intention here to provide factor definitions, but rather to illustrate that the details matter in terms of how those factors are reflected in a portfolio. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 3

4 the nuances in individual strategies, so is a term like high dividend yield smart beta portfolio. Quants typically embrace the idea of smart beta, because it confirms what they have been saying for decades. Still, we should expect the same level of rigor in the creation and analysis of a smart beta product as we would for any type of quantitatively driven strategy (or indeed for any strategy expected to produce consistent performance). And in many cases, these products do not get the scrutiny they should. So, how do we assess a smart beta strategy? Key questions include: Is the return that the product captures systematic and persistent? Is the systematic return that the product claims responsible for the returns (e.g., performance attribution)? Is the systematic return that the product claims the biggest contributor to risk? Are the rules used to rebalance and weigh the assets sensible? Are these rules consistent with the criteria above? How are transaction costs assessed and minimized, and how much turnover is needed to reach the goals? What is a suitable benchmark to monitor the quality of the systematic return? So smart beta is by definition quant, and we should therefore judge smart beta strategies with the tools that quantitative managers and their customers have been using to judge systematic strategies for decades. We propose that smart beta analytics should be an essential part of the marketing of the product. Case Study: High Dividend Yield ETFs In order to answer some of the questions listed above, we have examined a few types of smart beta ETFs and grouped them by their tagline descriptions. While the detailed descriptions of these funds highlight their features and thus their differences, we also know that many consumers of these products may not read the fine print, and therefore may not get what they expect. Let s look at some examples. For our first analysis, we chose four ETFs whose names contained high yield or dividend 4. SDY: SPDR S&P Dividend ETF VYM: Vanguard High Dividend Yield ETF FVD: First Trust Value Line Dividend Index Fund HDV: ishares Core High Dividend Performance of these ETFs since 2012 is shown in Figure 1. There are some fairly significant differences. For example, in 2013, SDY gained 31.2%, while HDV was up only 24.9%. In 2014, HDV recovered some ground by outperforming, gaining 17.1% versus 13-14% for the others. But investors who sold the underperforming fund, perhaps because they didn t understand the sources of the different returns, may not have benefitted from the rebound. 4 Of course, there are many more ETFs that fall into this category. We chose this subset of ETFs because they are relatively large, have a few years of return history, and have similar-sounding names. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 4

5 Figure 1. High Dividend Yield ETF Returns Cumulative Return Cumulative Return vs. Russell 3000 Source: State Street Global Advisors, Vanguard, First Trust, ishares, Axioma, Inc. What drives the performance differences? A number of characteristics stand out. On March 31, 2016, the average and median market capitalizations varied widely, although Price/Earnings, Price/Book, and Debt/Equity ratios were not substantially different from one another (Table 1). Interestingly, actual dividend yields showed a spread of about 67 basis points between the highest and lowest about a 20% difference versus the average yield. Sector weights varied dramatically. SDY was almost a quarter financials, VYM had about 13% of its weight there, and HDV had less than 8% (Figure 2). HDV s Energy weight was over 20%, while SDY s was just over 3%. FVD was about 24% utilities, whereas HDV was only about 7% in that sector. Note that sector weights vary over time, but in general not by very much. The differences across funds are more striking than the differences in one fund through time. One would think US Large Cap High Yield is US Large Cap High Yield, correct? In fact, the biggest names in each ETF are different. Exxon Mobil makes up almost 10% of HDV, 4% of VYM, and is not in the top five (and comprises less than 1%) of the others. HDV contains only 76 names and the top five names account for 35% of the ETF s weight, whereas VYM contains 426 names and the top five accounted for only 19% of the weight. The top 25 names in FVD make up only about 13% of its weight; the 195 holdings are more or less equally weighted. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 5

6 Table 1. Fundamental Characteristics of Dividend Yield ETFs Market Cap* Market Cap** P/E ratio*** P/B ratio*** Debt/ Equity** Dividend Yield* SDY $45,928,173,732 $15,060,490, % VYM $140,820,087,828 $114,700,672, % FVD $48,726,153,704 $22,547,056, % HDV $168,019,539,807 $163,596,206, % *Weighted average **Weighted median ***Weighted harmonic mean Data as of 3/31/2016 Source: State Street Global Advisors, Vanguard, First Trust, ishares, Axioma, Inc. Figure 2. High Dividend/High Yield ETF Sector Weights as of March 31, 2016 Source: State Street Global Advisors, Vanguard, First Trust, ishares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 6

7 Table 2. Top Holdings, Number of Names, and Concentration, March 31, 2016 SPDR Dividend ETF (SDY) Weight Dividend Yield Top 5 Stocks AT&T INC 2.04% 4.80% CATERPILLAR 1.87% 3.84% QUESTAR CORP 1.80% 3.39% CHEVRON CORP 1.78% 4.49% REALTY INCOM 1.75% 3.65% Vanguard High Dividend (VYM) Weight Dividend Yield Top 5 Stocks MICROSOFT CO 5.00% 2.34% EXXON MOBIL 4.11% 3.45% GENERAL ELEC 3.51% 2.89% JOHNSON & JO 3.51% 2.73% WELLS FARGO 2.91% 3.05% Total # names 107 Percent in Top 5 9.2% Top % Top % Total # names 426 Percent in Top % Top % Top % First Trust Value Line Dividend (FVD) Weight Dividend Yield Top 5 Stocks CLECO CORP N 0.59% 2.90% TRANSCANADA 0.54% 4.07% PUBLIC STORA 0.54% 2.36% HANOVER INS 0.53% 1.87% AVALONBAY CM 0.53% 2.63% ishares High Dividend Index (HDV) Weight Dividend Yield Top 5 Stocks EXXON MOBIL 9.46% 3.45% VERIZON COMM 7.32% 4.10% JOHNSON & JO 6.45% 2.73% CHEVRON CORP 6.38% 4.49% PFIZER INC 5.57% 3.78% Total # names 195 Percent in Top 5 2.7% Top % Top % Total # names 76 Percent in Top % Top % Top % Source: State Street Global Advisors, Vanguard, First Trust, ishares, Axioma, Inc. Risk characteristics also vary widely. A point-in-time risk analysis reveals some major differences in how risk is allocated in each fund (Table 3). In general, Axioma s dividend yield factor accounts for very little of the risk allocation, and its allocation differs by fund as well. Instead, the largest proportion of risk, especially for FVD, came from the negative exposure to Market Sensitivity. In other words, these funds tend to be low beta. VYM also gets a reasonable proportion of its risk from its low volatility bet. Also interesting is the 5% to 10% of variance that is the result of the portfolios Size beta. HDV and VYM actually have positive bets on size (stocks in the ETF are on average larger than the Russell 3000, while SDY and FVD have a negative Size exposure, i.e., small-cap bias). Finally, industry bets in total account for a high proportion of risk, but the biggest bets differ fund by fund. For example, for HDV, electric utilities don t crack the top five in terms of contribution to variance, but that industry makes up almost 15% of the risk in FVD. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 7

8 Table 3. Decomposition of Active Risk* Active % of Variance (Covariance Distributed) SDY VYM FVD HDV Total Active Risk % % % % Specific Active Risk 13.20% 16.15% 4.85% 22.90% Factor Active Risk 86.80% 83.85% 95.15% 77.10% Style 44.05% 51.14% 54.05% 40.56% Dividend Yield 2.30% 5.33% 1.48% 2.44% Earnings Yield -0.20% 1.23% 0.31% 0.34% Exch Rate Sens 3.02% 1.02% 0.87% 0.72% Growth 1.56% 4.96% 1.30% 2.33% Leverage 0.03% -0.86% 0.11% -0.68% Liquidity 0.57% 3.99% 6.02% 2.28% Mkt Sens 27.97% 22.73% 42.89% 20.87% MT Momentum -0.57% 0.27% 2.14% 0.55% MidCap -1.79% 0.44% -1.15% 0.12% Profitability -0.05% -0.58% -1.13% 0.33% Size 10.08% 5.12% 5.23% 5.96% Value 0.17% 0.03% -0.22% -0.01% Volatility 0.95% 7.44% -3.80% 5.30% Industry Total 42.62% 32.64% 41.01% 36.25% SDY Top 5 Industries VYM Industry Active % of Variance Industry Active % of Variance Gas Utilities 10.25% Electric Utilities 5.57% Multi Utils 6.72% Oil, Gas 4.30% Software 3.11% Internet Sw&Sv 3.73% Internet Sw&Sv 3.05% Multi Utilities 3.72% Machinery 2.58% Tobacco 3.53% FVD HDV Industry Active % of Variance Industry Active % of Variance Electric Utilities 14.69% Oil, Gas 13.84% Multi Utils 9.82% Tobacco 5.49% Gas Utilities 6.16% Multi Utils 2.47% Water Utilities 1.82% Pharma 2.23% Internet Sw&Sv 1.67% Household Prods 2.14% *Active risk is calculated versus the Russell 3000 and is based on Axioma s US4 medium-horizon model. Covariance is distributed across the factors. Data as of March 31, Source: State Street Global Advisors, Vanguard, First Trust, ishares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 8

9 Factor exposures can also vary significantly over time 5. While the exposure to Axioma s Dividend Yield factor has, of course, been positive, it has actually dipped recently for each of the ETFs (Figure 3). Exposure to both Market Sensitivity and Growth has been negative, but the magnitude has fluctuated, most notably for HDV s Market Sensitivity tilt. While FVD has had positive exposure to Value for most of the time since 2013, HDV s has largely been negative. VYM s has fluctuated between positive and negative. Figure 3. High Dividend/High Yield ETF Value 5 Exposures Through Time Dividend Yield Market Sensitivity Value Growth 5 Exposures here are absolute, not relative to a benchmark. Source: State Street Global Advisors, Vanguard, First Trust, ishares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 9

10 Stress Tests So we can see that the funds have very different exposures, both from each other and through time. We next set out to determine what might happen to the ETFs, and how that might differ, under different stress scenarios. We applied distinct stresses 6 to Volatility (factor down 1%), Value (up 1%), Market Sensitivity (down 1%), Earnings Yield (up 1%), and Dividend Yield (up 1%). As anticipated, based on the different factor exposures that were largely in the same direction but of different magnitudes, we saw that stress test results were similar in direction but different in magnitude. The biggest difference was the stress on Value, which had quite a negative impact on FVD. This difference in return illustrates two things: 1) how all factor exposures, not just the one the ETF tilts on, can have a major impact on performance; and, 2) the importance of correlations in identifying risk. In the case of FVD, when Value was stressed, FVD s positive exposure to Value implied a positive return. However, FVD had the most negative exposure to Market Sensitivity of our four ETFs, and the highly positive correlation between Market Sensitivity and Value implied FVD should fall when Value does well, offsetting any benefit from its Value exposure. Figure 4. Stress Test Results, Dividend Yield ETFs Source: State Street Global Advisors, Vanguard, First Trust, ishares, Axioma, Inc. Factor Attribution Next, we subjected our set of high dividend yield ETFs to a factor-attribution analysis to identify the specific sources of performance for each of the funds. Attribution covered the period January 2012 through March 2016, and the Russell 3000 was used as the benchmark 7. The results were somewhat eye-opening, especially in that Dividend Yield was actually a negative contributor to the active return of all four of our portfolios (Table 4). The small-cap bias in SDY and FVD helped returns, while VYM s and HDV s larger-cap bias hurt. We colorcoded the industries to highlight similarities and differences across the funds. Energy Equipment and Services was among the top contributors to returns in all four ETFs, and Computers and Peripherals (with a big negative 6 Axioma s Portfolio Analytics tool was used to apply stress tests to the various ETF portfolios. 7 One could argue that the Russell 3000 is not the appropriate benchmark here, but its use allowed us to have a consistent comparison across all funds. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 10

11 weight) also helped in three of the four. Pharmaceuticals, on the other hand, was a top contributor to VYM and HDV, but one of the biggest detractors in SDY. Industry bets clearly led to very different payoffs. Most notably, the low beta (Market Sensitivity) and low Volatility bets in the portfolios were the most positive contributors (although Market Sensitivity s contribution in FVD was twice that in VYM and SDY s return from Volatility was more than twice that in FVD). This suggests that the appeal of high dividend yield may not actually be in the yield itself, but in its ability to capture some of the low volatility phenomenon. Table 4. Factor Performance Attribution Contribution Source of Return SDY VYM FVD HDV Portfolio 15.18% 14.14% 15.85% 13.41% Benchmark 14.45% 14.45% 14.45% 14.45% Active 0.73% -0.31% 1.40% -1.03% Specific Return -0.97% -0.41% -0.72% -2.54% Factor Contr. 1.71% 0.10% 2.12% 1.51% Style 1.65% 0.01% 1.10% -0.19% Dividend Yield -0.83% -0.91% -1.00% -1.49% Earnings Yield -0.04% 0.40% 0.24% 0.25% Exch Rate Sens 0.07% -0.07% -0.02% -0.08% Growth -0.01% 0.06% -0.01% 0.00% Leverage 0.02% -0.03% 0.01% -0.10% Liquidity -0.04% -0.12% -0.19% -0.10% Mkt Sens 1.13% 0.76% 1.58% 1.48% MT Momentum -0.38% -0.35% -0.51% -0.56% MidCap -0.28% 0.10% -0.18% 0.14% Profitability -0.06% 0.04% -0.33% 0.24% Size 0.94% -0.92% 0.90% -1.20% Value -0.22% -0.07% -0.03% -0.22% Volatility 1.35% 1.12% 0.64% 1.44% Industry 0.09% 0.10% 1.05% 1.36% SDY VYM FVD HDV Top 4 Industries Oil, Gas 0.59% Pharmaceuticals 0.54% Oil, Gas 0.30% Pharmaceuticals 1.23% Energy Eq & Svc 0.31% Energy Eq & Svc 0.25% Energy Eq & Svc 0.29% Tobacco 0.58% Comp & Periph 0.19% Tobacco 0.14% Multi Utilities 0.20% Energy Eq & Svc 0.30% Hshld Durables 0.16% Beverages 0.12% Comp & Periph 0.20% Comp & Periph 0.21% Multi Utilities 0.11% Comm Equip 0.07% Metals & Mining 0.11% Beverages 0.16% Bottom 4 Industries Pharmaceuticals -0.15% Media -0.08% Airlines -0.08% Media -0.13% Machinery -0.16% Hlth Care Svcs -0.12% Software -0.08% Electric Utilities -0.25% Chemicals -0.21% Biotechnology -0.37% Div Fin Svc -0.10% Biotechnology -0.34% Biotechnology -0.37% Oil, Gas -0.61% Biotechnology -0.32% Oil, Gas -0.58% Software Airlines -0.07% Life Sci Svcs -0.06& Hlth Care Svcs -0.13% Source: State Street Global Advisors, Vanguard, First Trust, ishares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 11

12 Lastly, we looked at liquidity in each of the funds to see if their assets might be so large that, in the event of a major liquidation of the fund, they may have to trade a large portion of their constituents daily volume. We calculated the dollar holdings of each based on the weight and the approximate dollar value of the fund as of March 31, 2016, and then compared that with the 60-day median daily trading volume 8. Some of the funds are quite large, and some are very concentrated. So, for example, Exxon Mobil made up about 9.5% of HDV as of March 31, 2016, and we estimate the size of the fund at that time to be about $5.6 billion. Therefore, the fund held about $530 million in XOM, which was about 41% of its $1.2 billion average trading volume at the time. We then averaged the percent of volume for the top five names as well as for the whole portfolio (Table 5). In addition, we calculated the number of days it would take to fully trade out of the portfolio. SDY s biggest holdings tended to be in somewhat less liquid names, and it therefore ends up holding a lot of the average daily volume. To be clear, this is only a major problem in the extremely unlikely scenario of having to liquidate the fund, but even a major rebalancing could potentially increase trading costs in the fund. The other funds were much more liquid, either because they were less concentrated (such as VYM) or had a lower level of assets under management (HDV). Table 5. Fund Liquidity Characteristics Average % of ADV* AUM* Top 5 Whole Portfolio Days to Trade SDY $14 bil 144% 223% 2.3 VYM $20 bil 59% 31% 0.4 FVD $2.1 bil 23% 32% 0.3 HDV $6 bil 39% 27% 0.4 *Approximate, as of July Source: State Street Global Advisors, Vanguard, First Trust, ishares, Yahoo Finance, Axioma, Inc. High Dividend Yield versus Low Volatility Given the exposures and sources of return in our dividend yield funds, to us the next logical step was to compare the high dividend yield ETFs with low volatility ETFs to determine if we are getting the same returns in a different package. For this exercise, we looked at three low or minimum volatility ETFs: SPLV: Powershares S&P 500 Low Vol LGLV: SPDR Russell 1000 Low Vol ETF USMV: ishares Edge MSCI Minimum Volatility At first glance, there are clear differences between our high dividend and low volatility funds. The tracking error between HDV and SPLV, for example, is 6.86%. Some are more similar, however, with the tracking error between FVD and SPLV at just 3.4% (Table 6). Correlations of active returns of these funds were also quite high (Table 7). And SDY and FVD both have higher correlations (on average) with the low volatility funds than with the other high yield funds. 8 For stocks where the longer-term average volume was unavailable, we used the 20-day average. There were only a handful of stocks for which we did not have the 60-day figure. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 12

13 Table 6. Tracking Errors Between High Yield and Low Volatility Funds SDY VYM FVD HDV LGLV 4.32% 4.77% 3.49% 6.20% SPLV 5.25% 6.09% 3.40% 6.86% USMV 4.98% 5.17% 3.76% 6.12% Source: State Street Global Advisors, Powershares, ishares, Axioma, Inc. Table 7. Correlation of Daily Active Returns (vs. Russell 3000), January 2012 March 2016 SDY VYM HDV FVD LGLV SPLV USMV VYM HDV FVD LGLV SPLV USMV Source: State Street Global Advisors, Vanguard, First Trust, ishares, Powershares, Axioma, Inc. So, what are the similarities and differences between the two types of funds? Many exposures are quite similar, although there are a few differences. When we compare exposures of our ETFs we see that the Dividend Yield ETFs not surprisingly have higher exposures to Axioma s Dividend Yield factor than do the Low Volatility ETFs. Both sets have very little exposure to Earnings Yield, and the Low Volatility ETFs have slightly more negative exposures to Value and Volatility. However, they all have quite negative exposures to Market Sensitivity and, as we noted before, that Market Sensitivity exposure can have a significant impact on performance. Overall, the exposures of both types of funds are quite similar, and almost all in the same direction. Figure 5. Factor Exposures of Dividend Yield ETFs vs. Low Volatility ETFs As of March 31, 2016 Source: State Street Global Advisors, Vanguard, First Trust, ishares, Powershares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 13

14 Stress Tests and Performance Attribution Redux The next step was to create the same stress tests for the Low Volatility funds as those we ran for the High Dividend Yield ETFs, in order to see if they exhibit the same reactions to a given stress. In Figure 6, we compare the results shown above with those for the low volatility ETFs. The stress reactions for the low volatility ETFs were largely in the same direction as those for the dividend yield ETFs, although the magnitudes were different. The one major difference was that stressing the dividend yield factor to go up 1% had a positive impact on the dividend yield portfolios (not surprisingly, of course), but had a negative impact on the low volatility portfolios. In this case, it was not only the result of the Market Sensitivity factor (a more negative exposure times a positive factor return), but also the result of a positive Medium-Term Momentum exposure and a negative correlation of that factor with dividend yield that drove the return. The Momentum exposure is likely to fluctuate between positive and negative over time, and therefore this chart is very dependent on the time period of the stress test. Figure 6. Stress Test Results, Dividend Yield vs. Low Volatility ETFs Source: State Street Global Advisors, Vanguard, First Trust, ishares, Powershares, Axioma, Inc Performance attribution comparing High Dividend Yield and Low Volatility Funds revealed that the sources of return for the two types of fund were quite similar (Table 8). Dividend Yield contributed negatively in both types of fund, as did Medium-Term Momentum. The low Market Sensitivity and Volatility bets, on the other hand, were major positive contributors. Size, which lost an average of about 3% per year over the period of the study, had a big impact on all except USMV, but that impact varied in size based on the fund s exposure. The color-coded industry table (Table 9) shows that many of the same industries were the biggest contributors to the returns of the different types of ETFs. Energy Equipment and Services was one of the top four What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 14

15 contributors to all six funds, while Computers and Peripherals boosted returns of five out of six. Biotechnology detracted from returns of all six. Pharmaceuticals was a top contributors in three two of the high yield and one of the low volatility and was a detractor in one of each type as well. Table 8. Style Attribution, Dividend Yield vs. Low Volatility Comparison 9 Source of Return SDY VYM FVD HDV SPLV USMV Portfolio 15.18% 14.14% 15.85% 13.41% 14.39% 14.86% Benchmark 14.45% 14.45% 14.45% 14.45% 14.45% 14.45% Active 0.73% -0.31% 1.40% -1.03% -0.05% 0.42% Specific Return -0.97% -0.41% -0.72% -2.54% -2.30% -2.76% Factor Contr. 1.71% 0.10% 2.12% 1.51% 2.25% 3.18% Style 1.65% 0.01% 1.10% -0.19% 2.82% 2.06% Dividend Yield -0.83% -0.91% -1.00% -1.49% -0.81% -0.41% Earnings Yield -0.04% 0.40% 0.24% 0.25% 0.00% -0.17% Exch Rate Sens 0.07% -0.07% -0.02% -0.08% 0.09% 0.09% Growth -0.01% 0.06% -0.01% 0.00% -0.06% -0.04% Leverage 0.02% -0.03% 0.01% -0.10% 0.09% 0.06% Liquidity -0.04% -0.12% -0.19% -0.10% -0.04% -0.07% Mkt Sens 1.13% 0.76% 1.58% 1.48% 1.77% 1.66% MT Momentum -0.38% -0.35% -0.51% -0.56% -0.26% -0.15% MidCap -0.28% 0.10% -0.18% 0.14% 0.07% 0.07% Profitability -0.06% 0.04% -0.33% 0.24% -0.19% 0.16% Size 0.94% -0.92% 0.90% -1.20% 0.44% 0.02% Value -0.22% -0.07% -0.03% -0.22% -0.15% -0.31% Volatility 1.35% 1.12% 0.64% 1.44% 1.88% 1.15% Industry 0.09% 0.10% 1.05% 1.36% -0.56% 1.16% Table 9. Top and Bottom Industry Attribution, Dividend Yield vs. Low Volatility Comparison SDY VYM FVD HDV SPLV USMV Top 4 Industries Oil, Gas 0.59% Pharmaceuticals 0.54% Oil, Gas 0.30% Pharmaceuticals 1.23% Oil, Gas 0.65% Pharmaceuticals 0.33% Energy Eq & Svc 0.31% Energy Eq & Svc 0.25% Energy Eq & Svc 0.29% Tobacco 0.58% Energy Eq & Svc 0.31% Energy Eq & Svc 0.31% Comp & Periph 0.19% Tobacco 0.14% Multi Utilities 0.20% Energy Eq & Svc 0.30% Comp & Periph 0.21% Oil, Gas 0.23% Hshld Durables 0.16% Beverages 0.12% Comp & Periph 0.20% Comp & Periph 0.21% Metals & Mining 0.11% Comp & Periph 0.20% Bottom 4 Industries Pharmaceuticals -0.15% Media -0.08% Airlines -0.08% Media -0.13% Pharmaceuticals -0.26% Div Fin Svc -0.09% Machinery -0.16% Hlth Care Svcs -0.12% Software -0.08% Electric Utilities -0.25% Biotechnology -0.28% Multiline Retail -0.10% Chemicals -0.21% Biotechnology -0.37% Div Fin Svc -0.10% Biotechnology -0.34% Multi Utilities -0.34% Media -0.12% Biotechnology -0.37% Oil, Gas -0.61% Biotechnology -0.32% Oil, Gas -0.58% Electric Utilities -0.50% Biotechnology -0.15% Source: State Street Global Advisors, Vanguard, First Trust, ishares, Powershares, Axioma, Inc. 9 This table excludes LGLV, which did not start until What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 15

16 Table 10. Low Volatility ETF Liquidity Characteristics Average % of ADV* AUM Top 5 Whole Portfolio Days to Trade SPLV $7.6 bil 56.8% 55.1% 0.55 LGLV $67 mil 0.67% 0.50% 0.01 USMV $14.6 bil 43.6% 51.0% 0.64 *Approximate, as of July Source: State Street Global Advisors, ishares, Powershares, Axioma, Inc. The vast differences in characteristics of similar-sounding portfolios is not limited to high yielding funds. In the Appendix, we show similar results for a series of US large-cap Value funds. Conclusion Our results suggest a number of conclusions about how investors should be thinking about the proliferation of smart beta portfolios. The goal of these portfolios is (or should be) to provide rules-based, transparent, repeatable, scalable strategies that provide systematic return to a single factor or series of factors. This mirrors and should sound very familiar to the strategies of most quantitative equity managers, and should therefore be subject to a similar degree of scrutiny. By hewing to the definition of the space as what it is not (a market-cap weighted index), providers are granted a lot of leeway. To be sure, some products in this field do exactly what they are supposed to do, but others may not. This is not to say that one fund is right and another is wrong. We believe that funds should come with an ingredients label, to make sure investors know what they are getting. The old saw that financial products are sold, not bought rings very true in the case of smart beta products. Clever marketing should not get in the way of a well-thought-out and well-executed investment process. As we have shown, many smart beta strategies are not providing pure factor exposures, despite claims (or implications) to the contrary. Risk analysis and performance attribution confirms that funds that sound the same can be quite different, and that funds that sound different can be very much the same. The bottom line is that we need to benchmark smart beta strategies, providing the same kind of detailed analysis that managers of other types of products are required to show. Smart beta products may not actually be smart or, for that matter, beta. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 16

17 Appendix: US Large-Cap Value Fund Comparison Another example of large differences in similar-sounding funds can be found in large-cap US Value portfolios, which can also look quite different from one another. Here is the list of Value ETFs we studied; as in the highyield funds, these were chosen because they are relatively large and have enough returns history to allow us to adequately evaluate them. PWV: Powershares Dynamic Large-Cap Value IVE: ishares S&P 500 Value JKF: ishares Morningstar Large-Cap Value VTV: Vanguard Value FTA: First Trust Large-Cap Value AlphaDEX Figure A 1. Value ETF Returns Cumulative Return Value ETFs Cumulative Return vs. Russell Source: Vanguard, First Trust, ishares, Powershares, Axioma, Inc. Table A 1. Fundamental Characteristics of Value ETFs Market Cap* Market Cap** P/E ratio*** P/B ratio*** Debt/ Equity** Dividend Yield* PWV $120,872,907,173 $91,313,046, % IVE $110,284,151,558 $64,897,856, % JKF $151,911,855,601 $153,435,628, % VTV $128,731,989,595 $82,156,455, % FTA $37,485,526,195 $17,321,285, % *Weighted average **Weighted median ***Weighted harmonic mean as of 3/31/2016 Source: Vanguard, First Trust, ishares, Powershares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 17

18 Figure A 2. Value ETF Sectors Weights as of March 31, 2016 Source: Vanguard, First Trust, ishares, Powershares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 18

19 Table A 2. Top Holdings, Number of Names, and Concentration, March 31, 2016 Powershares Dynamic Large Cap Value (PWV) Weight Top 5 Stocks IBM 3.59% CHEVRON CORP 3.54% ORACLE CORP 3.46% INTEL CORP 3.40% CISCO SYS 3.37% Total # names 50 Percent in Top % Top % Top % ishares S&P 500 Value (IVE) Weight Top 5 Stocks EXXON MOBIL 3.94% BERKSHIRE HATH 3.02% AT&T INC 2.74% WELLS FARGO 2.54% PROCTER & GAMBLE 2.48% Total # names 366 Percent in Top % Top % Top % ishares Morningstar Large Cap Value (JKF) Weight Top 5 Stocks EXXON MOBIL 6.59% GENERAL ELEC 5.63% AT&T INC 4.58% WELLS FARGO 4.24% VERIZON COMM 4.18% Total # names 86 Percent in Top % Top % Top % Vanguard Value (VTV) Weight Top 5 Stocks MICROSOFT CORP 4.26% EXXON MOBIL 3.56% GENERAL ELEC 3.08% J&J 3.07% BERKSHIRE HATH 2.79% First Trust Large Cap Value AlphaDEX (FTA) Weight Top 5 Stocks NEWMONT MINING 1.17% URBAN OUTFITTERS 1.15% MICHAEL KORS 1.13% CENTURYLINK INC 1.01% CUMMINS INC 0.99% Total # names 314 Total # names 204 Percent in Top % Percent in Top 5 5.5% Top % Top % Top % Top % Source: Vanguard, First Trust, ishares, Powershares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 19

20 Active % of Variance (Covariance Distributed) PWV IVE JKF VTV FTA Total Active Risk % % % % % Specific Active Risk 34.11% 12.88% 18.23% 18.60% 7.10% Factor Active Risk 65.89% 87.12% 81.77% 81.40% 92.90% Style 38.30% 33.97% 31.40% 37.83% 57.17% Dividend Yield 1.95% 3.38% 4.49% 2.47% 2.31% Earnings Yield 8.57% -0.84% -0.01% 1.00% -2.23% Exch Rate Sens 1.21% 1.77% 1.11% 0.73% 2.32% Growth 3.05% 7.44% 6.02% 8.40% 1.76% Leverage 0.25% -0.24% 0.29% -0.07% -1.61% Liquidity 1.49% -0.33% 0.87% 1.14% 4.30% Mkt Sens 13.81% 1.04% 5.36% 7.28% 7.21% MT Momentum 0.03% 9.57% 0.33% 0.63% 18.06% MidCap 0.62% 0.00% 0.06% 0.24% 0.79% Profitability 0.15% -3.84% -3.15% -4.25% -0.36% Size 4.71% 3.86% 8.20% 7.52% 2.76% Value 0.43% 11.52% 4.31% 7.27% 9.87% Volatility 2.02% 0.64% 3.53% 5.46% 12.00% Industry 27.59% 53.15% 50.21% 43.52% 35.76% Industry Top 5 Industries PWV IVE JKF Active % of Variance Table A 3. Decomposition of Active Risk* Industry Active % of Variance Industry Active % of Variance Electric Utilities 12.15% Oil, Gas & Consumable Fuels 16.90% Oil, Gas & Consumable Fuels 20.53% Oil, Gas & Consumable Fuels 5.03% Biotechnology 5.54% Biotechnology 5.53% Internet Software & Services 3.35% Internet Software & Services 5.15% Software 3.52% Multi Utilities 3.05% Commercial Banks 4.94% Electric Utilities 3.02% Internet & Catalog Retail 1.83% Software 4.69% Internet Software & Services 2.85% VTV FTA Industry Active % of Variance Industry Active % of Variance Oil, Gas & Consumable Fuels 8.55% Oil, Gas & Consumable Fuels 7.15% Biotechnology 7.51% Biotechnology 4.22% Internet Software & Services 5.38% Energy Equipment & Services 4.20% Commercial Banks 4.23% Pharmaceuticals 3.91% Internet & Catalog Retail 3.35% Metals & Mining 2.40% *Active risk is calculated versus the Russell 3000 and is based on Axioma s US4 medium-horizon fundamental model. Covariance is distributed across the factors. Data as of March 31, Source: Vanguard, First Trust, ishares, Powershares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 20

21 Figure A 3. Value ETF Exposures Through Time Value Earnings Yield Growth Market Sensitivity Momentum Source: Vanguard, First Trust, ishares, Powershares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 21

22 Table A 4. Factor Performance Attribution Contribution Source of Return PWV IVE JKF VTV FTA Portfolio 14.01% 13.68% 12.05% 14.12% 13.37% Benchmark 14.45% 14.45% 14.45% 14.45% 14.45% Active -0.44% -0.77% -2.40% -0.33% -1.08% Specific Return -1.60% -0.46% -1.79% -0.71% 0.55% Factor Contr. 1.16% -0.31% -0.60% 0.38% -1.63% Style 1.52% 0.67% 0.37% 0.53% 0.37% Dividend Yield -0.62% -0.31% -0.72% -0.49% -0.17% Earnings Yield 1.10% 0.49% 0.99% 0.55% 0.83% Exch Rate Sens -0.06% -0.11% -0.25% -0.11% -0.10% Growth -0.01% 0.02% 0.03% 0.00% 0.00% Leverage -0.04% 0.07% -0.02% 0.04% 0.06% Liquidity 0.03% 0.02% 0.00% -0.02% 0.20% Mkt Sens 0.90% 0.36% 0.50% 0.35% 0.15% MT Momentum -0.14% -0.32% -0.47% -0.27% -0.67% MidCap 0.18% 0.09% 0.19% 0.06% -0.21% Profitability -0.13% -0.39% -0.35% -0.34% -0.36% Size -0.81% -0.54% -1.22% -0.66% 1.06% Value 0.21% 0.42% 0.47% 0.38% 0.64% Volatility 0.91% 0.86% 1.22% 1.05% -1.07% Industry -0.33% -0.94% -0.97% -0.15% -1.99% PWV IVE JKF VTV FTA Top 5 Industries Hlth Care Svcs 0.33% Comp&Periph 0.16% Pharmaceuticals 0.50% Pharmaceuticals 0.34% Multi Utilities 0.32% Pharmaceuticals 0.31% Div Fin Svcs 0.10% Energy Eq&Svc 0.31% Energy Eq&Svc 0.21% Electric Utilities 0.19% Energy Eq&Svc 0.24% Hlth Care Svcs 0.05% Div Fin Svcs 0.20% Comp&Periph 0.16% Comp&Periph 0.12% Comp&Periph 0.14% Food&Stap Retl 0.05% Div Tele Svcs 0.12% Div Fin Svcs 0.09% Specialty Retail 0.10% Aerospace&Def 0.14% Div Tele Svcs 0.04% Comp&Periph 0.11% Aerospace&Def 0.06% Hlth Care Svcs 0.10% Bottom 5 Industries Hlth Care Eq&Su -0.10% Airlines -0.06% Media -0.10% Comml Banks -0.05% Metals & Mining -0.13% Beverages -0.11% Comml Banks -0.08% IT Services -0.12% Media -0.08% Biotechnology -0.33% Biotechnology -0.15% Software -0.09% Software -0.12% IT Services -0.09% Pharmaceuticals -0.38% Comml Banks -0.16% Biotechnology -0.39% Biotechnology -0.33% Biotechnology -0.33% Energy Eq&Svc -0.71% Oil, Gas -1.04% Oil, Gas -0.65% Oil, Gas -1.76% Oil, Gas -0.58% Oil, Gas -0.80% Source: Vanguard, First Trust, ishares, Powershares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 22

23 Table A 5. Fund Liquidity Characteristics Average % of ADV* AUM Top 5 Whole Portfolio Days to Trade PWV $1 bil 4.5% 5.7% 0.05 IVE $10.7 bil 32.7% 10.1% 0.16 JKF $309 mil 1.6% 1.7% 0.02 VTV $43.8 bil 123.9% 50.0% 0.75 FTA $830 mil 6.8% 2.9% 0.04 *Approximate, as of July Source: Vanguard, First Trust, ishares, Powershares, Axioma, Inc. What s in a Name? In the Case of Smart Beta, It s Hard to Tell page 23

24 Contact us to learn more about how Axioma can bring more information and insights to your investment process. United States and Canada: Europe: Asia: Sales: Client Support: Careers: