Into a New Dimension. An Alternative View of Smart Beta

Transcription

1 Into a New Dimension An Alternative View of Smart Beta

2

3 Into a New Dimension An Alternative View of Smart Beta Table Of Contents Introduction 4 The alpha/beta debate has a long and evolving history 4 Some pure alpha is captured by beta 5 A broader framework for beta Into a new dimension 7 All very interesting, but why bother? 8 Some implementable solutions 9 Alpha and skill The role of the investor 11 Implications for alpha 11 In summary 12 Appendix 13

4 The investment landscape is changing. Smart beta, * a concept gaining significant investor interest, is a term we use to cover a broad spectrum of ideas that challenge the traditional blackand-white split between alpha and beta. We believe this is good news for investors. While many of the ideas have been around for years, there has been little interest in using a broader beta set to construct portfolios, in part because good implementation options were not available. In this article, we: Show that some of what was once called alpha can, in fact, be captured as beta Develop a broader framework for thinking about returns, adding an implementation strategy dimension to the traditional asset class definition Examine the diversification properties of some of the new or so-called alternative betas that come from this broader framework Consider implications for alpha and beta While many of the ideas have been around for years, there has been little interest in using a broader beta set to construct portfolios, in part because good implementation options were not available. We believe this is changing. Therefore, we think that investors should: Consider allocating directly to these new betas because they have low correlation to equity and credit markets Consider the beta exposures embedded in active mandates and the appropriateness of the fees they are paying for these exposures The alpha/beta debate has a long and evolving history Prior to modern portfolio theory, there was no easy way to understand performance investors could ascribe returns only to manager skill or perhaps luck. However, investors realized that their portfolio returns were linked to a large degree to a common driver, namely, the performance of the stock market as a whole. The development of the Capital Asset Pricing Model (CAPM) and indexation led to a new way to understand performance: The market (as described by a capitalizationweighted index) was beta, and excess return was active management, or alpha. The idea of market capitalization indexing in other major markets such as government and corporate bonds also took hold. Even in the early days of the CAPM, academics had identified groups of equity securities with certain characteristics such as value, low volatility and momentum that offered returns not easily explained by the simple market model. These groups of securities could be described and captured systematically. 1, 2 * See Appendix for the bulk beta/smart beta/alpha continuum. 4 towerswatson.com

5 Some pure alpha is captured by beta The most recent chapter in the debate relates to research on active managers, and in particular, hedge fund/returns once considered the ultimate expression of pure alpha investing. In fact, many studies show that a significant proportion of aggregate hedge fund returns can be explained by beta. To illustrate this point, in Figure 1, we show cumulative returns for hedge funds (as represented by the Hedge Fund Return Index (HFRI) Composite Index) 3 and a combination of beta investments. The thesis is simple: While individual funds may exhibit pure alpha, hedge funds, as a whole, pick up some common sources of return. This is particularly true for a broad collection of hedge funds, such as those represented by an index. As can be seen, the majority of hedge fund returns have been captured over this period by betas. Hedge funds did outperform modestly, despite the zero-sum game of active management and high fees. On a risk-adjusted basis, however, returns for the hedge fund index and the beta combination are similar. Figure 1. Illustrative cumulative return Dec 96 Dec 97 Dec 98 Dec 99 Dec 00 Dec 01 Dec 02 Dec 03 Dec 04 Dec 05 Dec 06 Dec 07 Dec 08 Dec 09 Dec 10 Dec 11 Dec 12 Dec 13 Hedge Fund Composite Index Beta combination Hedge Fund Composite Index Return over cash % per annum Risk (standard deviation) % Risk-adjusted return Beta combination This analysis and performance figures are backward-looking and intended to illustrate the portion of net hedge fund returns that can be captured by various market betas. The beta combination figures do not represent an actual portfolio return. Results of an actual portfolio invested in this way would vary depending on the timing of transactions, fees and other factors. Products designed to access the strategies utilized may not have been available to investors for all or part of the time span covered here. The hypothetical beta combination being used for comparison is shown gross of fees. Into a New Dimension An Alternative View of Smart Beta 5

6 Figure 2. Beta allocations matched to hedge fund strategies 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Relative value Equity l/s Event Macro Alternative beta Carry Vol premium Momentum Equity value Equity size Bulk beta Credit spread Gov bonds EM equity spread DM equity spread Composite index Figure 2 shows the beta allocations broken down for each of the main HFRI categories. As well as bulk betas such as equities and credit, we add alternative strategies such as value or momentum. The allocations are intuitive. For example, long/ short equity strategies have a greater exposure to equity market beta, whereas fixed income relative value strategies pick up bond and credit beta. Similarly, macro hedge funds include an exposure to momentum (trend) following strategies. The exposure to the volatility premium across most categories is likely to come from the typical lefttailed pattern of hedge fund returns. We can also assess the importance of beta by looking at return patterns. Figure 3 shows the percentage of monthly hedge fund returns that can be explained by beta a highly significant 84% for the broad HFRI index. Differences by strategy are again intuitive higher for equity long/short (more equity exposure) and lower for macro (more idiosyncratic strategies). Figure 4 shows that the finding is also remarkably consistent over time. Figure 3. Percentage of hedge fund returns explained by beta combination* HFRI strategy % Relative value 64 Equity I/s 84 Event 67 Macro 48 Composite index 84 Visually, the drawdown chart of Figure 5 also shows the closeness of fit: The dips appear at the same time and, broadly, to the same extent. Finally, we offer a note of caution. Our intention is not to replicate hedge fund returns per se, or to demonstrate that alpha has disappeared. The analysis is backward-looking and the betas fitted with hindsight. * R^2 measure from monthly regression of returns See Appendix Figure 4. Percentage of hedge fund returns (HFRI composite) explained by beta combination Rolling three years 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Dec 99 Dec 00 Dec 01 Dec 02 Dec 03 Dec 04 Dec 05 Dec 06 Dec 07 Dec 08 Dec 09 Dec 10 Dec 11 Dec 12 Dec 13 6 towerswatson.com

7 It should also be acknowledged that active managers have been responsible for identifying many of the strategies that are now being commoditized as beta. Nevertheless, the analysis does present some interesting issues for active management going forward (as discussed later). A broader framework for beta Into a new dimension To date, beta has been synonymous with asset classes and indices. An asset class groups together securities with similar characteristics, for example, equities or bonds. Once defined for inclusion in an index, the securities are generally buy-and-hold, changing little over time. As can be seen from Figure 2, we include examples of betas that do not fit into a traditional definition. At this stage, it is worth setting out a broader perspective on beta, although we shy away from a precise definition: Defined processes are applied to select a group of securities with common characteristics. Simplicity and transparency are preferable. The securities capture a premium for exposure to common sources of risk or return. The process has low fees and costs. Manager skill is primarily used for access (sourcing or selecting securities) and good implementation. The majority of passive assets globally are in equities and bonds. Because these are simple, cheap and liquid, we refer to them as bulk beta. However, under our broader description, we can extend bulk beta in two main ways: Implementation strategies. Here we refer to a strategy of owning or tilting toward securities with specific characteristics, and not holding (or shorting) those without. 4, 5 Examples are carry (own higher yielding securities) or momentum (buy securities that have recently appreciated). The idea also includes thematic investing owning assets with characteristics that should benefit from a long-term investment horizon. Extending asset classes to nontraditional areas such as currency or commodities. We also include strategies that are derived from an asset class. For example, the volatility strategy captures a premium for unanticipated volatility in an asset class, rather than returns from the asset class. Figure 5. Drawdown chart 0% -5% -10% -15% -20% -25% Dec 96 Hedge Fund Composite Index Dec 97 Dec 98 Dec 99 Dec 00 Dec 01 Dec 02 Dec 03 Dec 04 Beta combination Dec 05 Dec 06 Dec 07 Dec 08 Dec 09 Dec 10 Dec 11 Dec 12 Dec 13 Into a New Dimension An Alternative View of Smart Beta 7

8 The terms alpha and skill are often used synonymously, and investors have often outsourced alpha generation to specialist fund managers. An interesting perspective can be gained by considering this broader beta set in two dimensions (Figure 6). Going down the table, we extend our asset class universe to diversifying areas such as currency, commodities and volatility. (We omit others for brevity.) Across the table, we extend beta to implementation strategies, separated into systematic and thematic ideas. The broader set of betas is not theoretical. The ticks in the table refer to betas where we have found good solutions to implement and continue to investigate new ones. We can also fit traditional alpha neatly into this framework. We would characterize alpha as discretionary processes to select markets/ securities to generate positive returns over time. We discuss alpha a little later within a broader context of skill. All very interesting, but why bother? As with most things in investment, risk and return are key considerations. A striking feature of these new betas is that they have great diversification properties, both between themselves and perhaps more importantly to traditional asset classes. 6 To illustrate this, Figure 7 shows correlations for pairs of betas. As can be seen, correlations are highest in traditional markets such as equities and credit. Figure 6. Broader opportunity set Implementation strategies examples Systematic Thematic Alpha Return drivers Asset classes Carry Momentum Value EM Deleveraging Stock selection Market (beta) selection Equity Equities Term/inflation Bonds Credit Bonds Currency Currency Insurance Bulk beta Volatility Commodities Alternative beta Alpha Figure 7. More diversification from alternative betas Correlation Equities vs. credit Equities vs. value Carry vs. momentum Equities vs. volatility premium Best = lowest Worst = highest Average Rolling three-year correlations from December 1996 to December towerswatson.com

9 Some implementable solutions Not all things that can be called beta will necessarily generate positive returns. Our approach is to consider the following questions: 1. Rationale What economic risks or behavioral effects are we capturing, and why should they exist? 2. Evidence What academic or other evidence is there? Do we have out-of-sample tests? Can we independently substantiate the findings? 3. Beliefs What do we believe about the world going forward? Will it be different from the past? We have been researching smart beta strategies for a number of years, often working with the investment community to develop new products from scratch. Figure 8 summarizes the performance characteristics of an implementable alternative beta portfolio shown in Figure 9, and compares it to hedge funds (HFRI index). The portfolio features: Strategies use long and short positions, where relevant, which result in low exposures to traditional markets (and therefore low correlation by design). Strategies that can be accessed cost-effectively elsewhere are avoided. For example, equity value can be accessed cheaply in long-only mandates with equity smart betas. The allocation uses portfolio construction principles, scaling for risk (including tail risk) and diversity. We do not try to replicate hedge fund returns, per se, with either individual strategies or a combination. Management fees are low relative to existing methods of access (say, 50bps per annum vs. a 2+20 hedge fund fee scale). As Figure 8 shows, this alternative beta mix would have performed well, especially after adjusting for risk. Sensitivity (measured by beta) to traditional asset classes such as equities has been low. Note also that the alternative betas do in fact explain some of the hedge fund returns, although not by design, confirming the presence of alternative beta in hedge fund returns. Figure 8. Illustrative performance summary Figure 9. Example alternative beta portfolio HFRI Index Average return % per annum Standard deviation % Max drawdown % Equity beta % of returns explained by alternative beta portfolio 44 December 1996 to December % Emerging market currency 7.5% FX carry 12.5% Multi-asset carry 17.5% Volatility premium 25% Reinsurance 17.5% Momentum 7.5% Commodities Alternative beta portfolio This hypothetical alternative beta portfolio analysis is shown net of estimated fees. Please see Appendix for fee assumptions and additional information. This analysis and performance summary are backward-looking and intended to illustrate the portion of hedge fund returns that can be attributed to various market betas. The figures do not represent an actual portfolio return. Results of an actual portfolio invested in this way would vary depending on the timing of transactions, fees and other factors. Products designed to access the strategies utilized may not have been available to investors for all or part of the time span covered here. Into a New Dimension An Alternative View of Smart Beta 9

10

11 Overdiversified hedge fund strategies risk moving to industryaverage returns, and therefore, closer to the returns that can be captured with beta. Alpha and skill The role of the investor This broader beta framework adds a space between bulk beta and alpha. The terms alpha and skill are often used synonymously, and investors have often outsourced alpha generation to specialist fund managers. However, we think there is an opportunity for investors to capture new sources of returns as beta by applying governance, or skill in a broader sense. It is fair to say that these ideas require more time and expertise to understand and manage than traditional asset classes. Some of them are, in part, based on behavioral/structural explanations and therefore require investor beliefs about the rationale and sustainability of returns. We might also argue for a complexity premium, available to investors that have or can develop appropriate governance. The additional governance may be a barrier for some investors, particularly those with predominantly simple passive portfolios. Some of the ideas have limited capacity or cyclical return patterns, and their effectiveness may diminish over time. Similarly, over time, new strategies will emerge that can be commoditized as beta. So monitoring and selection of betas is very important a form of alpha in itself. Compared to alpha, the governance emphasis is shifted toward understanding, constructing and managing beta allocations rather than manager monitoring and evaluation. From a clear sheet of paper, we would argue that this broader beta framework entails more up-front governance, but less ongoing governance than alpha. Investors can, of course, have both beta and alpha, and indeed, this is what we would advocate, where governance allows. Implications for alpha In this paper, we make a case that some of what has been labeled alpha can be captured as beta. This is good news, as beta is (or should be) cheaper than alpha. We are not suggesting that alpha is not worth pursuing; genuine alpha is a source of uncorrelated returns and therefore much valued for portfolios. However, there are some implications for active management, particularly for hedge funds: Investors should consider the fees being paid for alpha, bearing in mind that some could in fact be beta. There is nothing wrong with including beta in an active mandate, providing that fees are appropriate. To help with this, we consider fees as a share of alpha. Manager selection is key. While this seems obvious, there may be a presumption that hedge fund managers are an above-average bunch, and therefore, the usual zero-sum game argument does not apply. While this may be true before fees, it is harder to justify net of fees. As shown earlier, we saw some additional return from hedge funds, but the argument is finely balanced. Overdiversified hedge fund strategies risk moving to industry-average returns, and therefore closer to the returns that can be captured with beta. This is exacerbated when fund of hedge funds are used. Into a New Dimension An Alternative View of Smart Beta 11

12 In summary In Figure 10, we draw together the key messages from this paper. The development of smart beta challenges the traditional split between alpha and beta, building a space between them. This creates an opportunity for investors that have or want to develop the skills to capture these sources of return. While many of the ideas have been around for years, there has been little interest in using a broader beta set to construct portfolios, in part because good implementation options were not available. We believe this is changing. Figure 10. The traditional split between alpha and beta Old definition New definition So what? Beta Returns from passive market capitalization exposure in traditional markets Returns from broader set of asset classes and strategies that are exposed to common risk (or other) factors Beta can be considered in a broader framework Investor skill needed Alpha Returns from active management Returns that cannot be explained by exposure to common risk factors Some of what was alpha is beta Portfolio implications Understand beta exposures Beliefs in new betas: Risk and return properties Consider appropriate exposure to betas (portfolio construction) and method of access Pay appropriate fees for alpha Consider overall contribution to return from beta and alpha sources Focus on true alpha generators 12 towerswatson.com

13 Appendix Bulk beta, smart beta and alpha Bulk beta traditional market cap passive investment in core asset classes such as equities and bonds. Bulk beta should be simple and liquid. Smart beta strategies that move away from market cap indexation in traditional asset classes. We therefore include nontraditional asset classes and systematic processes to capture common sources of risk or return. We divide smart beta strategies into three areas: Diversifying: Alternative asset classes that have low correlation with traditional markets; Systematic: Strategies that capture new risk premiums or exploit inefficiencies in market cap investing; Thematic: Capturing mispricing opportunities by being a long-term investor. As far as is practical, strategies should be beta-like in nature simple, low cost, transparent and so on. Alpha discretionary process to select securities and/or markets. Alpha cannot be explained by either bulk or smart beta. Beta and hedge fund regression results Data from December 1996 to December 2013 Sources: HFRI, FTSE, MSCI, Merrill Lynch, AQR, Ken French, Bloomberg, Towers Watson Beta strategies used: Developed-market equities: U.S. equities Emerging-market equity spread: Emergingmarket equities less global equities Government bonds: U.S. government bonds Credit spread: U.S. high-yield bonds less U.S. government bonds Equity size: Fama-French U.S. size factor Equity value: Fama-French U.S. value factor Momentum: Multi-asset trend following strategy based on prior 12 months performance Volatility premium: Short volatility strategy on the S&P 500 index Carry: FX carry strategy represented by 1.8*FTSE FRB10 index R^2 is a statistical measure giving the percentage of monthly return variation that can be explained by a given strategy. In this paper, we show the percentage of the variation in hedge fund returns that can be explained by a combination of beta strategies, with the allocations shown in Figure 2. The beta allocations depend somewhat on the way in which strategies are defined, as well as the time period used. Nevertheless, the broad conclusions in this paper are valid. Illustrative performance summary fee assumptions Beta strategy returns being analyzed are a combination of manager back-tests, indices and live data. A 3.5% per annum penalty has been applied to manager back-tests and index returns as a broad deduction to account for fees, expenses, trading costs and survivorship bias. Where manager live data were used, the data are net of fees. An additional 40 bps per annum was deducted across all strategies to simulate an investment advisory fee. Into a New Dimension An Alternative View of Smart Beta 13

14 Footnotes 1. Many of the alternatively weighted equity strategies capture common betas such as value and small cap. See for example, The Surprising Alpha from Malkiel s Monkey and Upside-Down Strategies Arnott, Hsu, Kalesnik and Tindall, Journal of Portfolio Management, summer Within equity stock selection, these risk premiums are often referred to as style premia. The ideas come from work by Fama-French (value and small cap), Jegadeesh-Titman and later, Carhart (momentum), and are not new. The work on style premia has largely been used to understand and attribute active equity manager performance, rather than to consider the merits of the premium on a stand-alone basis. In part, this reflects controversy over the sources of return and dominance of thinking around the efficient market hypothesis. 3. The HFRI Fund Weighted Composite Index is a global, equal-weighted index of over 2,000 singlemanager funds that report to the HFR database. Constituent funds report monthly, net of all fees, performance in U.S. dollars and have a minimum of $50 million under management or a 12-month track record of active performance. The HFRI Fund Weighted Composite Index does not include funds of hedge funds. 4. The selection of desirable assets could be at the country or sector level, rather than individual security level (e.g., weighting equities toward lowercost countries or sectors.) In other asset classes, this makes more sense (country selection for government bonds) or is the only choice (there are no securities in FX). 5. It is possible to isolate the pure risk premium by constructing long and short portfolios holding securities with desirable characteristics and shorting those with undesirable characteristics, balancing holdings in such a way as to remove or significantly reduce the underlying market risk. This approach is often used in academia, for example, with the Fama-French value and small capitalization factors for equities. In practical terms, the cost of doing this needs to be carefully considered. Implementation with liquid futures markets makes the approach more cost effective. For stockselection strategies, long/short strategies are less cost effective. Access to premiums can be achieved in long-only portfolios by tilting toward desirable characteristics, depending on the exposures to the premium needed. 6. This does not guarantee future diversity. We are aware of the potential for correlation and downside risks, depending on the specific strategy being considered. 14 towerswatson.com

15 Further information For further information, please contact your Towers Watson consultant or Author: Phil Tindall Americas Contact: Douglas Smith, CFA Into a New Dimension An Alternative View of Smart Beta 15

16 Please note This document was prepared for general information purposes only and should not be considered a substitute for specific professional advice. In particular, its contents are not intended by Towers Watson to be construed as the provision of investment, legal, accounting, tax or other professional advice or recommendations of any kind, or to form the basis of any decision to do or to refrain from doing anything. As such, this document should not be relied upon for investment or other financial decisions, and no such decisions should be made on the basis of its contents without seeking specific advice. This document is based on information available to Towers Watson at the date of issue and takes no account of subsequent developments after that date. In addition, past performance is not indicative of future results. In producing this document, Towers Watson has relied upon the accuracy and completeness of certain data and information obtained from third parties. This document may not be reproduced or distributed to any other party, whether in whole or in part, without Towers Watson s prior written permission, except as may be required by law. In the absence of its express written permission to the contrary, Towers Watson and its affiliates, and their respective directors, officers and employees, accept no responsibility and will not be liable for any consequences howsoever arising from any use of or reliance on the contents of this document, including any opinions expressed herein. About Towers Watson Towers Watson is a leading global professional services company that helps organizations improve performance through effective people, risk and financial management. With more than 14,000 associates around the world, we offer consulting, technology and solutions in the areas of benefits, talent management, rewards, and risk and capital management. Copyright 2014 Towers Watson. All rights reserved. TW-NA towerswatson.com /towerswatson