Schroders. The goal of including diversifi ers. What are alternative risk premia?

Transcription

1 Schroders For Financial Intermediary, Institutional and Consultant use only. Not for redistribution under any circumstances. Alternative risk premia Are they really alternative? A key question when considering an investment in alternatives is whether they are really alternative? That is to say what diversifi cation properties do the alternatives bring to a portfolio? In this note, we look at some common properties of Alternative Risk Premia (ARP), how they can provide diversifi cation and how the transparent exposures of ARP may allow for effective portfolio construction, a key component to delivering truly alternative sources of return. The goal of including diversifi ers A driving force behind the use of alternative investment strategies has been a desire to improve the diversifi cation of portfolios. We can broadly consider two categories of diversifying allocations. The fi rst of these is to diversify by adding exposure that is negatively correlated to the main risk(s) in a portfolio. This is normally added as a defensive or crisis risk allocation to protect the portfolio during bad times and is intended to have a high payoff over a short time horizon when needed. The second form of diversifi cation is an uncorrelated exposure which is designed to be effective over a longer time horizon. This form of allocation seeks to provide a more consistent source of positive returns, but may not provide large returns during bad times. ARP fall into the second of these diversifi cation categories (Figure 1), a space traditionally thought to be occupied by hedge funds What are alternative risk premia? The foundation of our ARP is a belief that human behavior is a key driver of asset prices beyond the traditional risk premia associated with growth, infl ation and slowdown. We believe that certain behaviors create biases in asset prices that lead to systematically exploitable opportunities; opportunities that are typically unrelated to broader macro fundamentals. In our research we discuss six cognitive biases (prospect theory, extrapolation, hindsight, herding, ambiguity and availability) and show how these can lead to ARP such as momentum, value and carry within a range of asset classes. The theoretical foundation of investor behavior provides a basis for ARP to add diversifi cation; however we must recognize that some behavioral biases are pro-cyclical and that ultimately these premia have to be accessed using traditional asset classes. As a result, there is a risk of correlation with traditional portfolio exposures in a similar way to that experienced with hedge funds. To better understand how ARP interact with a portfolio of traditional equity and bond exposures we start by examining the more familiar world of hedge fund exposure.

2 Figure 1: Negative, Positive and Uncorrelated exposures with the S&P 500 over the last 5 and 10 years (to December 2014) Correlation with Equities (S&P500) US 10 Yr S&P GSCI Equal Weighted ARP Schroders ARP 5yr 10Yr Source: Bloomberg, Schroders. Data as of December 31, Please see Appendix 2 for equally weighted ARP constituents. Schroders ARP includes our proprietary portfolio construction and beta hedging techniques. Hedge funds: the traditional diversifi ers and their challenges Hedge funds can provide returns that are lowly correlated with traditional equity exposures. However their risk and return drivers are opaque and when allocating to a portfolio of hedge funds it is quite common for aggregate exposures to drift towards higher correlation with equities, particularly when return opportunities become harder to fi nd. We have seen this in recent years as the correlation of a portfolio of investable hedge funds, represented by the HFRX Global Hedge Fund index, has a signifi cant and positive correlation to equities (represented as the S&P500 - Figure 2). We also see that exposures across the hedge fund strategy universe can become concentrated, illustrated in Figure 3, which shows the absorption ratio 1 for the same universe of hedge fund strategies. This suggests one risk factor can explain up to 9 of variation of returns in the hedge fund portfolio during times of stress. Figure 2: Correlation between hedge funds and equities Figure 3: Absorption ratio (AR) of variance and correlation amongst hedge fund strategies in HFRX Dec-2003 Dec-2004 Dec-2005 Dec-2006 Dec-2007 Dec-2008 Dec-2009 Dec-2010 Dec-2011 Dec-2012 Dec-2013 Dec Dec-2003 Dec-2004 Dec-2005 Dec-2006 Dec-2007 Dec-2008 AR of HF Correlation Dec-2009 Dec-2010 Dec-2011 Dec-2012 AR of HF Variance Dec-2013 Dec-2014 Source: Bloomberg, Schroders. Data as of December 31, Data in Figures 2 and 3 are based on a universe of HFRX index returns composed of the following sub indices: HFRX Equity Hedge Index, HFRX Event Driven Index, HFRX Macro/CTA Index, HFRX Relative Value Arbitrage Index, HFRX EH - Equity Market Neutral Index, HFRX ED - Merger Arbitrage Index, HFRX RV - FI Convertible Arbitrage Index, HFRX ED - Distressed Restructuring Index. Data simulated and not actual performance. Equities represented by the S&P500 Total Return Index. 1 Absorption ratio is the percent of risk that can be explained by the fi rst risk factor in a principal component analysis of strategy returns. We quote ARs for both variance and correlation, the latter adjusting for any bias created by more volatile strategies. 2

3 Figure 4: Factor attribution of hedge fund indices to common ARP HFRX Global Hedge Fund Average HFRX Style Factor Value Style Factor Momentum Style Factor Carry Style Factor Size Unexplained Source: Schroders, Bloomberg, monthly data from January 31, 1998 to December 31, Average HFRX is an equal cap weight of the selected constituents. Data simulated and not actual performance. Construction of the HFRX Average Index: As well as showing analysis using the HFRX Global Hedge Fund Index, we also include an equally cap weighted equivalent. This is based on the following sub indices: HFRX Equity Hedge Index, HFRX Event Driven Index, HFRX Macro/CTA Index, HFRX Relative Value Arbitrage Index, HFRX EH - Equity Market Neutral Index, HFRX ED - Merger Arbitrage Index, HFRX RV - FI Convertible Arbitrage Index, HFRX ED - Distressed Restructuring Index. Although hedge fund returns are opaque, we can use ARP to de-construct the return drivers of the HFRX Index and we fi nd the majority of returns can be explained by common ARP (Figure 4). This holds true even if we re-weight constituents to have equal allocation to each HF strategy within HRFX so as not to have a biased index. The idea that hedge funds, in aggregate, represent exposure to a combination of ARP is gaining traction across the industry, for example see Mladina This should not be confused with the suggestion that hedge funds do not provide real alpha, they can, but the important point is that when allocating to a portfolio of hedge funds the aggregate risk and return profi le can be well explained by ARP and that the correlation and concentration dynamics of hedge funds pose challenges for wider portfolio construction. The key element to understand is if the exposures in hedge funds and ARP are similar, how do ARP help us create a diversifying alternative portfolio? How do alternative risk premia help? ARP provides transparent exposure to systematic risk premia which may help in several ways: Each individual ARP has a consistent exposure to its systematic risk premia, leading to less concentration / style drift within the ARP portfolio. Each individual ARP has well understood core drivers, allowing effective portfolio construction. Each individual ARP has pricing transparency, allowing for effective management of transient fundamental premia exposures. Consistent exposure leading to less concentration In our earlier analysis we saw how the absorption ratio for a collection of hedge fund strategies within HRFX varied through time. Figure 5 shows the equivalent analysis for our current collection of ARP. Here we see that not only was the absorption ratio lower, but also it was more stable through time, thus helping to support ARP s lower correlation with equities in Figure 6, even for an equally weighted portfolio of ARP. 2 Mladina, P., Illuminating hedge Fund Returns to Improve Portfolio Construction. The Journal of Portfolio Management, Spring 2015, pp

4 Figure 5: Absorption ratios (AR) for hedge funds and ARP variance Figure 6: Correlation of hedge funds and an equally weighted ARP portfolio with the S&P Dec-2003 Dec-2004 Dec-2005 Dec-2006 Dec-2007 Dec-2008 Dec-2009 Dec-2010 Dec-2011 Dec-2012 Dec-2013 Dec-2014 Dec-2003 Dec-2004 Dec-2005 Dec-2006 Dec-2007 Dec-2008 Dec-2009 Dec-2010 Dec-2011 Dec-2012 Dec-2013 Dec-2014 AR of HF Variance AR of ARP Variance HFRX Equal Weighted ARP Source: Bloomberg, Schroders. Data as of December 31, Data in Figures 5 and 6 is based on ARP diversifi cation note data, which is based on HFRX factor analysis. Average HFRX is an equal cap weight of the selected constituents. Data simulated and not actual performance. Well understood core drivers, allowing effective portfolio construction Just as with asset classes, ARP labels can be misleading. Our research into the drivers of different ARP allows us to make important distinctions. For example, Carry is not a homogenous factor, expressing different properties when implemented in different asset classes. When implemented long/short in equities through a dividend yield strategy, it can have relatively defensive or quality like properties, unlike the properties of FX or volatility carry which are more pro-cyclical. However during 2008/9 it seriously underperformed due to a relative concentration in the fi nancial sector. This illustrates why careful construction of exposures and an active approach to the risk management of these strategies is critical. In addition some ARP have complementary features, a product of their behavioral drivers. For example, value and momentum share a symbiotic relationship where herding behavior can lead to momentum in prices which, when unchecked, sets up tremendous value opportunities. When valuations become overstretched and correct, momentum strategies are capable of generating positive returns, causing correlation between value and momentum strategies to become negative during times of drawdown in the value strategy. We illustrate this in Figure 7 by showing Trend, a sub-set of Momentum. Figure 7: Trend provides defensive properties when Value premia are in draw down. Value max drawdowns and correlation between Value and Trend Value Max DD -2% -4% -6% -8% -1-12% -14% Dec-2006 Jun-2007 Dec-2007 Jun-2008 Dec-2008 Jun-2009 Dec-2009 Jun-2010 Dec-2010 Jun-2011 Dec-2011 Jun-2012 Dec-2012 Jun-2013 Dec-2013 Jun-2014 Dec Valeu and Trend Correlation Figure 8: Target risk allocations by risk premia type, Value, Carry, Size and Momentum. Illustrative of the Schroders ARP risk allocations. Carry Value Global Carry World Value Fixed Income RV EM Carry Target Risk Allocation FX Value Trend German Small Cap FX Growth UK Small Cap World Quality US Small Cap Momentum Size Value Max DD Value and Trend Correlation Source: Schroders, Bloomberg as of December This should not be viewed as a recommendation to buy/sell. Source: Schroders, For illustration only. Risk premia categories shown are for illustrative purposes only. 4

5 It is therefore important to look past the labels attached to ARP and to study their drivers and return profi les during portfolio construction. One way to approach this is to think about categories of ARP and to allocate risk to these groups. Figure 8 illustrates an approach to this. As well as grouping ARP into categories, in order to manage correlations over shorter time horizons, it is important to consider how the interactions of the premia vary across the return distribution. A portfolio construction based on the interactions between different premia during periods of negative return can improve the drawdown characteristics of an ARP portfolio as illustrated in Figure 9 and Figure 10. Figure 9: Improved left tail profi le when using portfolio construction techniques focused on downside risks % -3.8% -3.2% -2.6% Equal Capital Weight Schroders Alternative Risk Premia Figure 10: Improved drawdown profi le when using portfolio construction techniques focused on downside risks -2% -4% -6% -8% -1-12% -14% -16% Feb-08 Feb-09 Feb-10 Feb-11 Feb-12 Feb-13 Feb-14 Equal Capital Weight Schroders Alternative Risk Premia Source: Schroders. Bloomberg, data from January 31, 2007 to July 23, Equal capital weight represents a portfolio of ARP without applying the portfolio construction techniques included in Schroders Alternative Risk Premia portfolio. Effective management of transient fundamental premia exposures We have seen that hedge funds have time varying correlation with equities and the same is true for ARP strategies. In our research note on the interaction of behavioral ARP with fundamental premia, the premia that drive most traditional portfolios, we highlighted there are time variant and non-linear interactions across the return distribution. Here we show two examples from that note (Figure 11 and Figure 12), illustrating how the Size and Value ARP interact with growth, infl ation and slowdown premia. Figure 11: Factor attribution of Fama French Size strategy to fundamental premia May-2005 Jan-2006 Sep-2006 May-2007 Jan-2008 Sep-2008 May-2009 Jan-2010 Sep-2010 May-2011 Jan-2012 Sep-2012 May-2013 Jan-2014 Slowdown Growth Inflation Specific Figure 12: Factor attribution of Fama French Value strategy to fundamental premia May-2005 Jan-2006 Sep-2006 May-2007 Jan-2008 Sep-2008 May-2009 Jan-2010 Sep-2010 May-2011 Jan-2012 Sep-2012 May-2013 Jan-2014 Slowdown Growth Inflation Specific Source: Schroders, as of April Risk premia categories shown are for illustrative purposes only and should not be viewed as a recommendation to buy/sell. 5

6 To directly tackle this time varying exposure, we beta hedge liquid representations of growth and duration risk premia though the S&P 500 and US Treasury futures, capping the portfolio beta exposure to 0.2. While over time this is expected to cause a drag on returns, since these are risk premia with positive return expectations, the addition of the beta hedge helps reduces portfolio volatility, but more importantly, limits the portfolio correlations to fundamental premia, maintaining the alternative nature of the portfolio. We see this in the correlation to equities, which for the fi nal portfolio averages near zero. Figure 13: Correlation of hedge funds, an equally weighted ARP portfolio and the Schroders ARP portfolio with the S&P Dec-2003 Dec-2004 Dec-2005 Dec-2006 Dec-2007 Dec-2008 Dec-2009 Dec-2010 Dec-2011 Dec-2012 Dec-2013 Dec-2014 HFRI Equal Weighted ARP Schroders ARP Source: Schroders. Data based on HFRX Factor Analysis with the average HFRX being an equal capital weight of the selected constituents Dataset is simulated and not actual performance. The end result ARP correlation characteristics across environments Having brought together these different stages of portfolio construction to build a strategy that shows low and fairly stable long term correlation with equities, it is important to check if this result or our analysis has sensitivity to economic regimes of growth and infl ation. To conduct this analysis we need more data history than is available for our own strategies. For ARP we have done extensive work extending datasets back in time using various academic data sources to aid test regime sensitivity, creating the MA Proxy (refer to Appendix 4 for further information). Absolute return levels of the extended data should be treated cautiously because these return streams are missing transaction costs and have some uncertainty about how fully they could be implemented. However, these return drags should not affect the shorter term dynamics of the ARP, allowing us to test regime dependency and relative performance in different regimes more fully. Figure 14: Simulated performance by growth regime 25% 2 15% 1 5% Figure 15: Simulated performance by infl ation regime 25% 2 15% 1 5% S&P 500 US10yr MA S&P 500 US10yr Proxy MA S&P 500 US10yr Proxy MA Proxy S&P 500 US10yr MA S&P 500 US10yr Proxy MA S&P 500 US10yr Proxy MA Proxy High Growth Medium Growth Low Growth High Inflation Medium Inflation Low Inflation Source: Schroders. Timeframe for economic classifi cation is from 1951 to December 2014 based on the OECD US Composite Leading Indicator for growth and US CPI for infl ation. The performance numbers are from 1982 to December There is no guarantee that volatility and return objectives can be met. Data simulated and not actual performance. Please see the appendix and Important Information for full details regarding the simulations and defi nitions. 6

7 We fi nd that ARP performance was stable during different growth regimes (Figure 14), showing much less dependency than equities. We also see that there was a slight preference for normal infl ation environments, although this result is harder to determine due to a relatively benign infl ation environment over much of the available sample period (Figure 15). If we perform the same analysis and calculate conditional correlations against equities we fi nd again a fairly stable relationship under different growth and infl ation regimes (Figures 16 and 17 respectively). We therefore fi nd that a carefully constructed ARP portfolio maintains its alternative nature though various different economic regimes. Figure 16: Conditional correlation of the MA proxy vs. S&P 500 under different growth regimes 5 Figure 17: Conditional correlation of the MA proxy vs. S&P 500 under different infl ation regimes 5 25% 25% -25% -25% -5 High Growth Regime Medium Growth Regime Low Growth Regime -5 High Inflation Regime Medium Inflation Regime Low Inflation Regime Correlation vs S&P 500 % of periods Correlation vs S&P 500 % of periods Source: Schroders. Timeframe for economic classifi cation is from 1951 to December 2014 based on the OECD US Composite Leading Indicator for growth and US CPI for infl ation. The performance numbers are from 1982 to December There is no guarantee that volatility and return objectives can be met. Data simulated and not actual performance. Please see the appendix and Important Information for full details regarding the simulations and defi nitions. Conclusions We have seen that ARP has the potential to provide exposure that is un-correlated to the fundamental premia that drive most tradition portfolios and provide return generators that are found in common hedge fund strategies. Therefore, such a risk/return profi le makes ARP a strong choice for an alternatives allocation. Key to achieving this is careful construction of pure ARP and robust portfolio construction to address the combined issue of risk concentration and the transient exposures to fundamental premia. Strategic Beta Research Series This paper draws on previous papers published under the Strategic Beta research series. For more detail on the topics discussed in this paper please refer to the following Strategic Beta research notes: Understanding and investing in behavioral risk premia Capturing behavioral biases: an empirical analysis How behavioral and fundamental premia interact 7

8 Appendix 1: HFRX universe and the average HFRX index HFRX Equity Hedge Index, HFRX Event Driven Index, HFRX Macro/CTA Index, HFRX Relative Value Arbitrage Index, HFRX EH - Equity Market Neutral Index, HFRX ED - Merger Arbitrage Index, HFRX RV - FI Convertible Arbitrage Index, HFRX ED - Distressed Restructuring Index. Average HFRX refers to an equally weighted index of these sub indices. Appendix 2: Schroders ARP universe and the Equally Weighted ARP portfolio defi nition The below represents the investment universe for the Schroders ARP portfolio. The portfolio is an equally capital weighted combination of the below ARP. Please note at present Volatility Alpha is not invested with the ARP solution due to minimum size restrictions, it is invested within our Strategic ARP Volatility Alpha Equity Value Equity Quality UK Small Cap / Large Cap German Small Cap / Large Cap Trend US Small Cap / Large Cap FX Carry EM FX Carry FX Growth FX Value Fixed Income Relative Value Instrument Long Option collars on S&P, FTSE, Dax, Eurostoxx, Nikkei TRS Long customized global value basket v MSCI World TRS Long customized global quality basket v MSCI World TRS FTSE 250 Index v FTSE 100 future TRS Dax Mid-Cap Index v Dax future Long basket of futures & ETFs based on proprietary trend model Long Russell 2000 future v S&P500 future Long 7 high yield FFX v 7 low yield FFX Long 5 high yield EMFFX v 5 low yield EMFFX Long 7 high growth FFX v 7 low growth FFX Long 7 undervalued FFX v 7 overvalued FFX Long attractive G6 govt. bonds v short unattractive G6 govt. bonds Beta portfolios. Appendix 3: Regime identifi cation for fi gure and fi gure 1 Growth regime indicator: OECD US Composite Leading Indicator Data: Monthly from Classifi cation: The YoY change is sorted where Top 33% of the months are classifi ed as High Growth Regime Middle 33% of the months are classifi ed as Medium Growth Regime Bottom 33% of the months are classifi ed as Low Growth Regime Inflation regime indicator: US CPI Data: Monthly from Classifi cation: The YoY change is sorted where Top 33% of the months are classifi ed as High Infl ation Regime Middle 33% of the months are classifi ed as Medium Infl ation Regime Bottom 33% of the months are classifi ed as Low Infl ation Regime 8

9 Appendix 4: MA proxy defi nition In order to measure the dynamics of an Alternative Risk Premia portfolio under various growth and infl ation regime we have extended our ARP back through time using a combination of available data sources. This process requires some compromise between data availability and representativeness. The details of the composition of the MA Proxy through time are: Risk Group Strategies Schroders From Proxy From US Size 1979 Fama French 1976 Size UK Size 1986 Gregory 1980 German Size 1988 N/A Volatility Sterling 2006 Bloomberg 1986 Carry Value FX Carry 1999 Burnside 1976 EM FX Carry Equity Value Proprietary research N/A Fama French 1976 Fixed Income Relative Value 1993 N/A FX Value 2002 AMP Value 1979 FX Growth 2002 N/A Momentum Cross Asset Trend 1994 Barclays CTA Index 1987 Equity Quality 1996 AMP QMJ 1976 Details of the Proxy data sets used are: US Size and Equity Value (Fama-French): Size (SMB) and Value (HML) factors from tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html UK Size (Gregory): Size factor from / research/famafrench/fi les Gregory, Alan, Rajesh Tharyan, and Angela Christidis. Constructing and testing alternative versions of the Fama French and Carhart models in the UK. Journal of Business Finance & Accounting 40, no. 1-2 (2013): Volatility Selling (CBOE): Average of CBOE S&P 500 2% OTM BuyWrite Index (BXY) and CBOE S&P 500 PutWrite Index (PUT) FX Carry (Burnside): Currency data from Speculation_Data.html Burnside, Craig, Martin S. Eichenbaum, and Sergio Rebelo. Carry trade and momentum in currency markets. No. w National Bureau of Economic Research, FX Value and Equity Quality (AMP): Asness, Moskowitz and Pedersen Value and Quality factors from Asness, Cliff, Andrea Frazzini, and Lasse H. Pedersen. Quality minus junk. Available at SSRN (2013). Asness, Clifford S., Tobias J. Moskowitz, and Lasse Heje Pedersen. Value and momentum everywhere. The Journal of Finance 68, no. 3 (2013): Cross-Asset Trend (Barclays CTA Index): Barclays US Managed Futures Industry BTOP 50 Index (Bloomberg Ticker: BARCBTOP Index) OECD CLI: 9

10 This page left purposely blank

11 This page left purposely blank

12 Important information: The returns are presented as gross returns, including cash, reinvestment of dividends, interest and other income earned in the period and are calculated on a trade date basis after transaction charges (brokerage commissions), but before taxes and management and custody fees. Performance would have been reduced by such fees and the effect of these fees on performance compounds over time. As an illustration see the chart below. The value of a $5,000,000 account would be reduced by the following amounts due to the compound effect of the management fees. (This has been calculated assuming an assumed constant return of 1 per annum* and a hypothetical management fee of 0.75% per annum, which has been applied on a simple average of opening and closing annual fund values). Gross value Net value Compound effect 1 year $5,500,000 $5,447,500 $52,500 3 years $6,655,000 $6,466,238 $188,762 5 years $8,052,550 $7,675,491 $377, years $12,968,712 $11,782,633 $1,186,079 *The assumed 1 return is hypothetical and should not be considered a representation of past or future returns. The actual effect of fees on the value of an account over time will vary with future returns, which cannot be predicted and may be more or less than the amount assumed in this illustration. Actual fees may differ from the assumed rate presented above. Please consult the Firm s Advisory Brochure (ADV Part 2) for a description of the fees. Past performance is not a guide to future returns. The value of investments can fall as well as rise as a result of market movements. Investments in smaller companies may be less liquid than in larger companies and price swings may therefore be greater than in larger company funds. Exchange rate changes may cause the value of overseas investments to rise or fall. Less developed markets are generally less well regulated than the US, they may be less liquid and may have less reliable custody arrangements. Investors should be aware that investments in emerging markets involve a high degree of risk and should be seen as long term in nature. The Strategy will invest in some higher-yielding bonds (non-investment grade). The risk of default is higher with non-investment grade bonds than with investment grade bonds. Higher yielding bonds may also have an increased potential to erode your capital sum than lower yielding bonds. The Strategy will invest in Property Funds and Property Investment Companies. It may be difficult to deal in these investments because the underlying properties may not be readily saleable which may affect liquidity. The views and forecasts contained herein are those of the Multi-Asset team and are subject to change. The information and opinions contained in this document have been obtained from sources we consider to be reliable. No responsibility can be accepted for errors of facts obtained from third parties. Reliance should not be placed on the views and information in the document when taking individual investment and/or strategic decisions. The opinions stated in this presentation include some forecasted views. We believe that we are basing our expectations and beliefs on reasonable assumptions within the bounds of what we currently know. However, there is no guarantee that any forecasts or opinions will be realized. This document is not an offer or a solicitation to acquire or dispose of an interest in securities or other investment instruments described herein. This document contains outline, indicative terms for discussion purposes only and is not intended to provide the sole basis for evaluation of the instruments described. Terms are purely indicative and may change in line with market conditions. Any recipient of this document agrees that the appropriateness of any described structure to its particular situation will be independently determined, including consideration related to the legal, tax and other related aspects of any transaction. Further information about Schroders can be found at Schroder Investment Management North America Inc. 875 Third Ave 22nd Floor, New York, NY (212) MA-ARP