Blog

KE1068 May 31, 2018 2018 by the Kellogg School of Management at Northwestern University. !is case was prepared by Professor Phillip A. Braun. Cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of e”ective or ine”ective management. Some details may have been #ctionalized for pedagogical purposes. To order copies or request permission to reproduce materials, call 800-545-7685 (or 617-783-7600 outside the United States or Canada) or e-mail custserv@hbsp.harvard.edu. No part of this publication may be reproduced, stored in a retrieval system, used in a spreadsheet, or transmitted in any form or by any means— electronic, mechanical, photocopying, recording, or otherwise—without the permission of Kellogg Case Publishing. P H I L L I P A . B R A U N Smart Beta Exchange-Traded Funds and Factor Investing It was early 2015 and executives in the iShares Factor Strategies Group were considering the launch of some new exchange-traded funds (ETFs). !e new ETFs were in a class of ETFs called smart beta funds. Most traditional ETFs’ portfolio weights were based on the market capitalization of a stock (stock price times the number of outstanding shares), but smart beta ETFs’ weighting schemes were based on #rms’ #nancial characteristics or properties of their stock returns. iShares was a division of BlackRock, Inc., an international investment management company based in New York. In 2014, BlackRock was the world’s leading asset manager, with over $4.5 trillion in assets under management.1 In 2014, iShares globally o”ered over 700 ETFs with almost $800 billion in net asset value in the US and $1 trillion globally—a 39% market share, making iShares the largest issuer of ETFs in the world.2 !e new smart beta multifactor ETFs being considered by iShares would provide investors with simultaneous exposure to four fundamental factors that had shown themselves historically to be signi#cant in driving stock returns: the stock market value of a #rm, the relative value of a #rm’s #nancial position, the quality of a #rm’s #nancial position, and the momentum a #rm’s stock price has had. While each of these factors existed in di”erent combinations and di”erent forms in ETFs already in the marketplace, no #rm was currently o”ering these four as a combination in a multifactor ETF. !e executives at iShares were unsure whether there would be demand in the marketplace for such multifactor ETFs, since their value added from an investor’s portfolio perspective was unknown. This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 2S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N G KE1068 K E L L O G G S C H O O L O F M A N A G E M E N T Factors Smart beta portfolios were driven by academic research that shows there is a common set of driving forces, or factors, that consistently explains stocks’ average returns and systematic risks. Professors Eugene Fama and Kenneth French have been at the forefront of developing factor models, having written a series of papers examining di”erent #nancially based factor models of stock returns. As the culmination of their decades of research, in 2013 Fama and French introduced a model that showed a systematic relationship between the average returns of stocks and #ve underlying factors.3 Based on the capital asset pricing model (CAPM) theory, the #rst factor Fama and French considered was the relative covariance of a #rm’s stock returns with a market portfolio. !e CAPM theory states that stock return movements can be broken down into two components: movements due to the returns of an underlying market portfolio (a market factor) and #rm-speci#c movements, with a stock’s average returns determined by its co-movements with the market portfolio. !is can be seen via the CAPM’s factor model: ( ), , ,i t f i i m t f i tr r r rD E H = + + (1) where ri,t is the return at time t for some security i, rf is the return to a risk-free asset, αi and βi are regression coe$cients, rm,t is the return to a market portfolio (the market factor) at time t, and εi,t is the regression’s residual. A stock’s co-movement with the market portfolio is measured by its CAPM beta (βi), the regression coe$cient above. Figure 1 shows the average returns on #ve portfolios ranked by the size of their betas using annual data from 1964 through 2014. !ese portfolios were created by sorting all stocks listed on the NYSE, AMEX, and NASDAQ by their betas and then splitting the stocks into quintile portfolios based on the level of their beta, with those stocks with betas in the bottom 20% of the distribution put into a smallest beta quintile portfolio, then the next 20% into the next highest quintile portfolio, and so on. Figure 1 shows no evidence of a strong upward sloping relationship between a portfolio’s beta and its average return, contrary to the prediction of the CAPM theory. !us, Fama and French concluded that the CAPM theory does not explain average returns very well.*4 !e second factor Fama and French explored was the size of a #rm as measured by its market capitalization, what is termed a size factor. Fama and French (and others) found a signi#cant negative relationship between the size of a #rm and its average return—that is, #rms with small market capitalizations earn higher average returns across time than #rms with large market capitalizations. Figure 2 plots the average returns of portfolios grouped into quintiles by their market capitalizations. What we see in Figure 2 is the small #rm e”ect; small #rms have higher average returns than large #rms, with a consistent rise in average returns as we go from the smallest to the largest #rm’s portfolio. * Note that Fama and French used a longer time frame, from 1927–1990, to construct the portfolios in this study than were used to construct the portfolios in Figure 1 (1964–2014). This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 3S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N GKE1068 K E L L O G G S C H O O L O F M A N A G E M E N T Figure 1: Relationship Between Average Returns and the CAPM Beta 0% 2% 4% 6% 8% 10% 12% 14% Smallest Betas Quintile 2 Quintile 3 Quintile 4 Largest Betas Av er ag e R et ur n Source: Author’s calculations using data from Ken French’s data library, accessed August 2016, http://mba.tuck.dartmouth.edu/ pages/faculty/ken.french/data_library.html.
These portfolios are from French’s univariate portfolios, sorted by
beta, using annual value-weighted data from 1964–2014. Figure 2: Relationship Between Average Returns and Firm Size 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% 20% Smallest Firms Quintile 2 Quintile 3 Quintile 4 Biggest Firms Av er ag e R et ur n Source: Author’s calculations using data from Ken French’s data library, accessed August 2016, http://mba.tuck.dartmouth.edu/ pages/faculty/ken.french/data_library.html.
These portfolios are from French’s univariate portfolios, sorted by
firm size (market capitalization), using annual equal-weighted data from 1964–2014. This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 4S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N G KE1068 K E L L O G G S C H O O L O F M A N A G E M E N T At #rst glance it is not obvious why smaller #rms should have higher average returns. Fama and French showed that, controlling for other factors, size is related to a #rm’s pro#tability; small #rms have lower earnings on assets than big #rms.5 Firm size is also thought to proxy for speci#c risk factors associated with smaller #rms; researchers have explored the underlying sources of such risks, but the results are not conclusive. Some have argued that because small #rms’ stocks do not trade as often as larger #rms and are thus less liquid, investors in smaller #rms require higher returns for accepting this liquidity risk.6 Others suggest that size is correlated with information uncertainty— that is, smaller #rms are not often followed by investment banks while simultaneously having more volatile fundamentals.7 !e third factor that Fama and French considered was a value factor, which is the ratio of a #rm’s book value (as given by a #rm’s balance sheet, its assets minus its liabilities) to its market capitalization—its book-to-market (B/M) ratio.8 Figure 3 plots the average returns of #ve portfolios that were created by ranking stocks by their B/M ratio and then sorting them into quintile portfolios from low B/M ratios to high. What we see in Figure 3 is that the portfolios with higher B/M ratios have higher average returns, which is termed a value e”ect. Figure 3: Relationship Between Average Returns and the Book-to-Market Ratio 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% Lowest B/M Quintile 2 Quintile 3 Quintile 4 Highest B/M Av er ag e R et ur n Source: Author’s calculations using data from Ken French’s data library, accessed August 2016, http://mba.tuck.dartmouth.edu/ pages/faculty/ken.french/data_library.html.
These portfolios are from French’s univariate portfolios, sorted by the
B/M ratio, using annual value-weighted data from 1964–2014. Fama and French argue that this positive relationship between average returns and B/M ratios is because high B/M stocks are less pro#table and are relatively distressed, so these #rms are riskier and have higher average returns to re%ect that. Conversely, low B/M #rms have high returns on capital with sustained pro#tability (hence they are termed growth stocks); therefore, they are less risky and have lower average returns. !e B/M e”ect is called a value e”ect because “When a #rm’s This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 5S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N GKE1068 K E L L O G G S C H O O L O F M A N A G E M E N T market value is low relative to its book value, then a stock purchaser acquires a relatively large quantity of book assets for each dollar spent on the #rm. When a #rm’s market price is high relative to its book value the opposite is true.”9 !e fourth factor that Fama and French proposed was a pro#tability factor, the ratio of a #rm’s operating pro#t to its book value.* In Figure 4, the average returns of #ve portfolios are presented, sorted from low to high pro#tability. To create these portfolios, #rms were ranked by the Fama and French pro#tability measure and then sorted into quintile portfolios. As can be seen in Figure 4, more pro#table #rms earn signi#cantly higher average returns than less pro#table #rms. !e argument is simply that #rms with productive assets should have higher returns than #rms with unproductive assets. !is pro#tability factor is sometimes referred to as a quality factor— #rms with higher pro#tability are higher quality #rms. “While traditional value strategies #nance the acquisition of inexpensive assets by selling expensive assets, [a quality strategy] exploits a di”erent dimension of value, #nancing the acquisition of [quality] productive assets by selling unproductive assets.”10 Figure 4: Relationship Between Average Returns and Profitability 0% 2% 4% 6% 8% 10% 12% 14% Lowest Profitability Quintile 2 Quintile 3 Quintile 4 Highest Profitability Av er ag e R et ur n Source: Author’s calculations using data from Ken French’s data library, accessed August 2016, http://mba.tuck.dartmouth.edu/ pages/faculty/ken.french/data_library.html.
These portfolios are from French’s univariate portfolios, sorted by
profitability, using annual value-weighted data from 1964–2014. !e #fth Fama and French factor was an investment factor, which they measured as the percentage change in the value of a #rm’s assets over the course of a year.11 In Figure 5, #rms are ranked by this investment measure and then sorted into quintile portfolios, from conservative (low * Fama and French’s measure of operating pro#t is a #rm’s revenues minus costs of goods sold, minus selling, general, and administrative expenses, minus interest expense. This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 6S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N G KE1068 K E L L O G G S C H O O L O F M A N A G E M E N T levels of investment) to aggressive (high levels of investment). As shown in Figure 5, portfolios with conservative levels of investment have higher average returns than aggressive #rms. Fama and French’s rationale for this investment e”ect is that, holding a #rm’s expected revenues constant, a rise in a #rm’s investment implies lower future expected earnings and thus lower expected returns, while lower investment levels yield higher expected earnings and thus higher expected returns. Like the pro#tability factor, the investment factor is also termed a quality factor because #rms with lower levels of investment are higher quality #rms because they are not depressing their earnings via excessive investment and are putting less stress on their balance sheets. Figure 5: Relationship Between Average Returns and Investment 0% 2% 4% 6% 8% 10% 12% 14% 16% Lowest Investment Quintile 2 Quintile 3 Quintile 4 Highest Investment Av er ag e R et ur n Source: Author’s calculations using data from Ken French’s data library, accessed August 2016, http://mba.tuck.dartmouth.edu/ pages/faculty/ken.french/data_library.html.
These portfolios are from French’s univariate portfolios, sorted by
investment, using annual value-weighted data from 1964–2014. Separately from Fama and French, a series of authors have examined what is termed a momentum factor.12 !e momentum e”ect documents that stocks that have large price appreciation in one year continue to have high price appreciation the following year and stocks with negative or low price appreciation continue to do so the following year. In Figure 6, stocks are sorted into quintile portfolios based on their stock price appreciation in the previous year, from those stocks that performed in the bottom 20% of all stocks in the previous year, what is termed a loser portfolio, to stocks that performed in the top 20% over the last year, a portfolio of winners. As Figure 6 shows, the greater the price appreciation for a portfolio over the last year, the higher that portfolio’s average return. This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 7S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N GKE1068 K E L L O G G S C H O O L O F M A N A G E M E N T Figure 6: Relationship Between Average Returns and Momentum 0% 5% 10% 15% 20% 25% Lowest Momentum Quintile 2 Quintile 3 Quintile 4 Highest Momentum Av er ag e R et ur n Source: Author’s calculations using data from Ken French’s data library, accessed August 2016, http://mba.tuck.dartmouth.edu/ pages/faculty/ken.french/data_library.html.
These portfolios are from French’s univariate portfolios, sorted by
momentum, using annual equal-weighted data from 1964–2014. In a recent study, Fama and French found that, after controlling for the market, size, value, pro#tability, and investment factors, a momentum factor had no signi#cance in determining average returns.13 Others, however, #nd their evidence inconclusive, and continue to see a relevance for a momentum factor even after these other factors are controlled for.14 !ere are two schools of thought as to why we see a relationship between a stock or portfolio’s momentum and its average returns: the rational “markets are e$cient” school and the behaviorist “markets are ine$cient” school. A rational market perspective on the causes of the momentum e”ect would hold that #rms with high momentum face greater cash %ow risks and/or higher discount rates because of the nature of their investment sets.15 !e behaviorists view the momentum e”ect as resulting from investors’ cognitive biases. For example, investors who choose to invest in a high-return stock are simply extrapolating past performance into the future, exhibiting a kind of herding behavior.16 To understand how factors are a set of driving forces across #nancial securities’ returns, the next step is to consider an actual factor model. A factor model is a regression that relates securities’ returns to a set of factor portfolios. For example, the CAPM theory implies that there is only one factor driving security returns, the market portfolio. Given this, equation (1) is the CAPM’s factor model. For the six factors just discussed, the factor model would be a six-factor regression model of the form: This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 8S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N G KE1068 K E L L O G G S C H O O L O F M A N A G E M E N T ( ), , , / , , , , , ( ) ( ) ( ) ( ) ( ) i t f i i m t f i Size t i B M t i Profitabilty t i Investment t i Momentum t i t r r a b r r s F h F r F c F m F H = + + + + + + + (2) where ri,t is again the return on some stock, rm,t is the return on the market, rf is the return to a risk-free asset, FSize,t through FMomentum,t are the factor portfolios for each of the other #ve variables discussed above, ai is the regression intercept, bi to mi are the regression slope coe$cients, and εi,t is the regression’s error term. From this regression you can see that this six-factor model is the CAPM shown in equation (1) plus #ve additional factors. !e factor portfolios, FSize,t through FMomentum,t, can be created in di”erent ways. A simpli#ed example of one standard approach is to use the decile portfolios with the highest average return for a particular variable—for example, the decile with the smallest #rms would be used for the size factor portfolio, FSize,t; and the decile with the highest B/M ratios would be used for the B/M or value factor portfolio, FB/M,t, and so on. !ese factor portfolios are called long-only portfolios because they only include purchases of stocks in the portfolio. An example of a second standard approach, following Fama and French’s work, is to use a long-short factor portfolio. Simplistically, a long-short factor portfolio is constructed by taking the returns on the decile with the highest average return less the returns on the decile with the lowest average return. It is called long-short because the portfolio includes both purchases (long) as well as shorts—you short a security by borrowing the security from a broker and then selling it. For example, using this approach the size factor portfolio, FSize,t, would be the returns to the decile with the smallest #rms less the returns to the decile portfolio with the biggest #rms.* Fama and French call this the SMB portfolio, for small #rms minus big #rms. Fama and French’s other long- short portfolios for their #ve-factor model are: HML (high minus low): !e returns of the decile portfolio that includes the #rms with the highest B/M ratios less the returns to the decile portfolio with the lowest B/M ratios. RMW (robust minus weak): !e returns of the decile portfolio that includes the #rms with the most robust pro#tability less the returns to the decile portfolio with the weakest pro#tability. CMA (conservative minus aggressive): !e returns of the decile portfolio that includes the #rms with the lowest (most conservative) levels of investment spending less the returns to the decile portfolio with the highest (most aggressive) investment spending. Background on Smart Beta ETFs !e intent behind smart beta ETFs is to capture the high expected returns identi#ed for factors in the work of Fama and French and others.17 !e term “smart beta” is a fairly new marketing * !is simpli#es what Fama and French actually do, which is to take the bottom 30% of #rms in market size for the small #rm portfolio and the top 30% of #rms in market size for the big #rm portfolio. !is is also true for the other Fama-French long-short portfolios de#ned in the case. This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 9S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N GKE1068 K E L L O G G S C H O O L O F M A N A G E M E N T identi#er; some of the alternative names that have been used include strategic beta, active beta, enhanced index, alternative beta, and scienti#c beta. !ere is no one speci#c de#nition of what a smart beta portfolio is. However, a common theme across de#nitions is that smart beta portfolios are constructed such that the emphasis is weighting stocks in these portfolios not on the traditional measure of market capitalization, but by incorporating into their weighting scheme some aspect of a security’s fundamental value, such as a stock’s B/M ratio, pro#tability, or a characteristic of the security’s performance, such as a stock’s momentum.* Regardless what de#nition is used for smart beta ETFs, the bottom line is that they are proxies for the factor portfolios, the Fi,t’s, that we discussed in the previous section. Smart beta ETFs are considered a combination of passive and active investing. !e funds are passive because they passively mimic what are termed factor indexes and hence do not require any input from a portfolio manager. !ey are considered active because their weights deviate from standard market capitalization weights. As Cli” Asness, a founder, managing principal, and chief investment o$cer at AQR Capital Management, a global investment management #rm that has been at the forefront in creating momentum factor portfolios, has stated, “A portfolio that deviates from market weights . . . must be balanced by other investors who are willing to take the other side of those bets. For example, for every value investor, who tilts toward or selects cheap value stocks, there must be an investor on the other side who is underweighting value and overweighting expensive, growth stocks. Hence, as everything must add up to the market-weighted portfolio, everyone at once cannot hold or tilt toward value at the same time.”18 Smart beta portfolios are not active in the sense that fund managers are searching for mispriced securities. Although some justify smart beta portfolios from the behaviorist perspective that the factors are mispriced,19 it is not necessary to use this justi#cation to motivate smart beta strategies.20 Besides smart beta ETFs that capture the size, value, quality, and momentum factors we have discussed, there are also dividend, minimum-volatility, and other factor-based smart beta stock ETFs, as well as bond ETFs that capture factors speci#c to bonds. Dividend ETFs in general try to capture potentially higher average returns from investing in stocks paying high dividends. 21 !e rationale for minimum-volatility (also called low volatility) ETFs is the documented evidence that stocks with low volatility or risk have higher average returns than high volatility stocks.22 !ere is also a class of multifactor ETFs, whose goal is to provide exposure to a set of two or more factors simultaneously. In 2014, mutual funds that mimic underlying factors had existed for a while, but smart beta ETFs were newer to the marketplace. Figure 7 shows the level of assets under management in * !is is not a strict de#nition, however; some #rms classify traditional value, growth, dividend, and small #rm portfolios whose weighting scheme uses market capitalization weights into the smart beta category, while other #rms do not. Note that Fama and French consider the B/M ratio and their pro#tability measure to be better predictors of average returns than the dividend yield; see https://famafrench.dimensional.com/questions-answers/qa-dividends- is-bigger-better.aspx, accessed January 2018. !e market statistics in this section are the author’s calculations using data from the Center for Research in Security Prices and a partial list of ETFs from http://www.etf.com, accessed September 2016. Note that the group of smart beta ETFs included in this analysis excludes traditional value, growth, dividend, and small #rm portfolios whose weights are based on market capitalizations, which ETF.com classi#es as smart beta. This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 1 0 S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N G KE1068 K E L L O G G S C H O O L O F M A N A G E M E N T smart beta ETFs listed in the US from 2003, when the #rst smart beta ETFs were introduced, through 2014. !e smart beta market grew from around $1.5 billion in 2003 to just over $160 billion at the end of 2014, a cumulative annual growth rate of 47%.* Figure 7 also shows the breakdown of assets under management by smart beta category, with dividend ETFs having the most assets under management and momentum ETFs with the least. Figure 8 shows the number of smart beta ETFs and their share of the overall US ETF market. In 2003, there were only six smart beta ETFs; this had grown to 277 by 2014. In 2003, smart beta ETFs were 1% of the overall US ETF market and this grew to an 8% market share by 2014. Figure 7: Market Size and Breakdown for US Smart Beta Exchange-Traded Funds $0 $20 $40 $60 $80 $100 $120 $140 $160 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 N et A ss et V al ue (B illi on s $) Momentum Growth Quality Other Bonds Volatility Size Multifactor Dividend Value Sources:
Author’s calculations using data from the Center for Research in
Security Prices and a partial list of smart beta ETFs from http://www.etf.com.
Note that the group of smart beta ETFs in the analysis excludes
traditional value, growth, dividend, and small firm portfolios whose weights are based on market capitalizations. * If traditional market capitalization–based value, growth, and small #rm ETFs are included in the analysis, the smart beta market size in 2014 was $360 billion. This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 1 1 S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N GKE1068 K E L L O G G S C H O O L O F M A N A G E M E N T Figure 8: Number of US Smart Beta Exchange-Traded Funds and Market Share Relative to Total US ETF Market 0 50 100 150 200 250 300 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 N um be r o f F un ds Pe rc en ta ge M ar ke t S ha re Number of Smart Beta Funds Smart Beta Share of Total ETF Market Sources: Author’s calculations using data from the Center for Research in Security Prices and the list of ETFs from http://www.etf.com. See the source notes to Figure 7 for more detail. In 2014, iShares had a 38% overall US ETF market share and a 22% share of the smart beta market. Its closest competitor, State Street Corporation, had a 23% overall ETF market share, but only a 10% smart beta market share. !e other major player in the smart beta market was Invesco PowerShares Capital Management, which had a 15% smart beta market share, but only a 5% overall ETF market share. !e Vanguard Group had a 21% overall ETF market share but o”ered no smart beta ETFs.* At the end of 2014, iShares o”ered 19 di”erent smart beta ETFs in the US with $36 billion in assets under management. A list of these ETFs is presented in Table 1, along with their net asset values. iShares introduced its #rst dividend ETF in 2003, the iShares Select Dividend ETF (DVY), and by 2014, iShares o”ered a range of smart beta ETFs, both international and domestic US, for equities as well as bonds. In 2013, iShares introduced four smart beta ETFs directly related to the research of Fama and French and others: its size, value, quality, and momentum factor ETFs. * Vanguard did not sell smart beta funds because it questioned the value the approach added for investors. See, for example, C. B. Philips et al., “An Evaluation of Smart Beta and Other Rules-Based Active Strategies,” Vanguard Research (2015). Vanguard did, however, o”er a variety of capitalization-weighted ETFs similar to smart beta ETFs, such as its dividend appreciation ETF (VIG), its value ETF (VTV), and its small cap ETF (VB). This
document is authorized for use only in Prof. Fong wai ming’s FINA6112
FC & FD Term 1, 22 Aug 2021-22, Prof. Fong & Sam (2) at The
Chinese University of Hong Kong (CUHK) from Aug 2021 to Feb 2022. 1 2 S M A R T B E T A E T F S A N D F A C T O R I N V E S T I N G KE1068 K E L L O G G S C H O O L O F M A N A G E M E N T Table 1: List of Smart Beta ETFs Sold by iShares in 2014 Name Ticker Net Asset Value ($ in millions) iShares Asia/Pacific Dividend ETF DVYA 55 iShares Core Dividend Growth ETF DGRO 152 iShares Core High Dividend ETF HDV 5,197 iShares Edge MSCI Min Vol Asia ex Japan ETF AXJV 5 iShares Edge MSCI Min Vol EAFE ETF EFAV 1,356 iShares Edge MSCI Min Vol Emerging Markets ETF EEMV 1,907 iShares Edge MSCI Min Vol Europe ETF EUMV 4 iShares Edge MSCI Min Vol Global ETF ACWV 1,561 iShares Edge MSCI Min Vol Japan ETF JPMV 10 iShares Edge MSCI Min Vol USA ETF USMV 3,581 iShares Edge MSCI USA Momentum Factor ETF MTUM 482 iShares Edge MSCI USA Quality Factor ETF QUAL 721 iShares Edge MSCI USA Size Factor ETF SIZE 213 iShares Edge MSCI USA Value Factor ETF VLUE 517 iShares Emerging Markets Dividend ETF DVYE 213 iShares International Select Dividend ETF IDV 4,163 iShares MSCI USA Equal Weighted ETF EUSA 62 iShares Select Dividend ETF DVY 15,554 iShares Yield Optimized Bond ETF BYLD 9 Sources: Center for Research in Security Prices and http://www.etf.com. Note that traditional market capitalization–weighted portfolios are excluded from this analysis. !e multifactor ETFs that iShares was considering would be a combination of the size, value, quality, and momentum factors together, encompassing both international and domestic US stocks. As shown in Figure 7, multifactor ETFs only had $40 million in net asset value in 2003, which grew to $18 billion by 2014, a 76% cumulative annual growth rate. Prior to 2013, all the multifactor ETFs in the market captured just two e”ects, combining the size e”ect with one of the other factors. !e type of multifactor ETFs that iShares was considering, which capture three or four factors simultaneously, was very new to the market. At the end of 2014, only three #rms o”ered such multifactor products in the US market, with a combined net asset value of only about $150 million.*23 * A small competitor, WisdomTree Investments Inc., introduced a multifactor ETF based on the size, quality, and dividend factors in 2013, the WisdomTree US Small Cap Quality Dividend Growth Fund (DGRS), which at the end of 2014 only had a market value of $25 million. In mid-2014, State Street o”ered a set of 12 international multifactor ETFs, its SPDR MSCI Quality Mix ETF suite, and had announced the release of a US multifactor ETF for early 2015. !ese MSCI Quality Mix ETFs combined the value, low volatility, and quality factors into one ETF; at the end of 2014, this group of ETFs only had a total market value of $66 million. J. P. Morgan Asset Management introduced two multifactor ETFs in 2014: the JPMorgan Diversi#ed Return Global Equity ETF (JPGE) and the Diversi#ed Return International Equity ETF (JPIN), which brought together the size, value, momentum, and volatility factors and