The Path to Achieving True Diversification In Managed Futures Strategies

Transcription

1 Investment Intelligence The Path to Achieving True Diversification In Managed Futures Strategies January 2018 Rahul Malhotra, Ph.D. Director, Portfolio Manager Multi Asset US Claudio Marsala Director, Portfolio Manager Multi Asset US Heather Bergman, Ph.D. Vice President, Portfolio Manager Multi Asset US Introduction With valuations in equities and other asset classes at lofty levels after a decade-long bull market, investors are understandably concerned about the prospects of a potential spike in volatility and low expected returns from traditional exposures to equities and bonds. To manage some of those risks, growing numbers of investors are allocating assets to alternative strategies. One of the most common ways to gain exposure to such strategies is through managed futures portfolios which are driven primarily by trend-following or momentum investing. In particular, managed futures strategies seek to identify price momentum in futures markets, including equities, fixed income, commodities and currencies (FX). While managed futures strategies have become more accessible to investors, the industry might be overlooking significant risks. Such risks could undermine performance over a long-term horizon. Most investors including large managed-futures managers use traditional notions of diversification as a core portfolio construction practice. For example, many managed futures managers tout the diversification advantage gained from trading a large number of instruments. Also, many managers place faith in investing across instruments that have historically appeared to have low correlation to each other over the long term. However, strategies relying on such standard approaches fail to account for important features of asset markets particularly commodities, though no asset class is exempt. Key Takeaways A misunderstanding of the true nature of risk-taking in managed futures leads to inefficient diversification and may expose managers to unanticipated risks that undermine long-term performance. Most investors including large managed futures managers build portfolios using traditional diversification strategies based on investing across a large number of instruments that have historically exhibited a low correlation to each other over time. We suggest a more effective approach based on the concept of risk clusters, or groups of assets with common risk characteristics that make it likely that correlations among them could increase over time and/or during times of market stress. Structuring a managed futures portfolio with risk clusters may protect against unexpected increases in correlations, thereby producing the type of smoother return profile sought after by absolute return investors, and potentially increasing portfolio returns over time. us.allianzgi.com/institutional For fund distributors and professional investors use only FOR INSTITUTIONAL USE ONLY

2 An alternative approach that properly accounts for both high and increasing correlations among seemingly different commodity markets, and various latent risks associated with commonalities between futures markets in all asset classes, could produce stronger and more persistent performance in a managed futures portfolio. Moreover, an approach that distributes portfolio risk across groups of assets into risk clusters, as opposed to balancing risks across individual instruments, may be advantageous from an absolute-return perspective. Misunderstanding the Nature of Risk In Managed Futures Managed futures can offer an attractive tool for institutions to access alternative strategies that, overall, offer low correlation to traditional asset classes. But managing the composition and distribution of risk in managed futures portfolios is paramount to long-term success. In this light, we believe that the managed futures industry might be overlooking significant risks that could undermine performance over a long-term horizon, especially with regard to potentially shifting correlations and positioning for latent risk. We address those risks below. Correlations are larger than commonly believed, and shifting There are about two dozen tradeable commodity markets, ranging from oil and gas in the energy sector to grains, such as corn and soybeans. The diverse nature of these markets leads to the belief that they would be highly uncorrelated to each other. Indeed, as Exhibit 1 shows, for most of history, the monthly correlation between these markets was quite small. However, if we dig a bit deeper, we see that over lower frequencies, such as quarterly or semi-annually, the correlations have been higher. Not only that, as we move to the right of the graph we see that in recent years and especially since the 2008 financial crisis that correlation has moved even higher. Exhibit 1: When seen over less-frequent intervals, commodities have shown a much stronger correlation to each other Correlation As of June 30, 2017 Based on average absolute pair-wise correlations over rolling five-year periods between the following Bloomberg Commodity Total Return Sub Indexes: Energy, Industrial Metals, Precious Metals, Grains, Softs and Livestock. Source: Bloomberg Semi-Annual Quarterly Monthly Another example of this phenomenon is the correlation between oil and some commonly traded agricultural commodities. As depicted in Exhibit 2, these correlation levels have been on the rise. Similar patterns are seen among other commodity groups as well, including metals and oil. Most investors are aware that traditional diversification strategies failed to live up to expectations during the 2008 crisis. The simplest explanation of why standard portfolio diversification methods fell short is that, at the height of the crisis, long-established correlation patterns among asset classes and geographic regions broke down. What many investors don t fully appreciate is that factors other than one of history s greatest financial crises influence correlations, and that correlation patterns can shift over time for much more prosaic reasons reducing the effectiveness of diversification strategies built upon fixed correlation assumptions. Exhibit 2: Correlation between agricultural commodities and oil has been on the rise Correlation to Oil Corn As of June 30, 2017 Based on correlations using semi-annual returns between the Bloomberg WTI Crude Oil Sub-Index and the Bloomberg Sub-Indices for Corn, Soybean, Wheat, Soybean Meal, Soybean Oil, Sugar, Coffee, Cocoa and Cotton. Source: Bloomberg Wheat Soybean Oil Soybean Soybean Meal Sugar Coffee Cotton Cocoa Commodity markets once again offer a prime example. Just 10 to 15 years ago, the commodities market was made up primarily of suppliers/buyers purchasing and hedging exposures. Since that time, a new generation of commodity trading advisors (CTAs) has emerged. These firms now make up a large share of commodities trading volume, and they are trading based on similar algorithms. Exhibit 3 (next page) shows the close relationship between growth of the managed futures industry and the commodities market. Along with fundamental changes in global macroeconomics since the 2008 crisis, the increasing influence of CTAs trading with similar strategies could also be contributing to higher correlations between commodity sectors. Given high CTA trading volumes, it s likely that price movements in one direction across commodity sectors would be exacerbated due to concerted buying (or selling) by the trendfollowing CTA industry, thereby compounding the original correlation. While such shifts cannot be predicted, recognizing that they could potentially occur is key to providing consistent performance in managed futures strategies. 2

3 Exhibit 3: Growth of CTAs has gone hand in hand with an increase in commodity trading volumes Managed Futures Industry AUM vs Average Monthly Trading Volumes Commodities (US$ Billions) $400 $350 $300 $250 $200 $150 $100 $50 $ YTD 2017 As of September 30, 2017 Bars (LHS) represent managed futures industry assets under management in US$ billions as of December 31 of each year, except for year-to-date Line (RHS) represents average monthly trading volume for a basket of the 29 most actively traded commodity contracts in US$ notional. Source: Bloomberg, Morgan Stanley and BarclayHedge $8,000 $7,000 $6,000 $5,000 $4,000 $3,000 $2,000 $1,000 $0 Latent risk is a wild card In some ways, shifting correlation patterns within commodities caused by a change in the composition of the investor base can be viewed as a latent risk for the asset class. We define latent risk as any variable that has the potential to have a meaningful impact on the operation and performance of an asset class or market, but has until this point been dormant. In the case of commodities or any other asset class, for that matter a dramatic change in the makeup of the investor base has the potential to alter market dynamics. Prior to the last five to 10 years, this was a latent risk: It always existed in potential, but had not yet come into play. Our analysis suggests that such latent risks even if not fully realized have a meaningful impact on managed futures strategies. For another example of how these risks can materialize, one need look no further than the recent Sharpe ratios of momentum strategies in different asset classes. While momentum has underperformed in commodities possibly due to the reduced diversification, as mentioned above it has also underperformed in FX. Why? The financial crisis resulted in unprecedented coordination of action by central banks across the world. Interest rates hit the zero bound and central banks balance sheets ballooned. As a result, the interest rate carry across jurisdictions has decreased, which, in turn, has likely impacted the momentum behavior of global currencies relative to each other. These shifts may well have resulted in the underperformance of FX momentum strategies during this period (and provided a boost to bond and equity momentum). To return to the topic at hand, the possibility of an unprecedented, synchronized, global move to quantitative easing and historically low interest rates that alter traditional correlations among global currencies has always existed, but, until now, had never materialized. That s latent risk. Additional latent risks exist across and within asset classes due to geographic or other commonalities, which may be well recognized but not necessarily revealed in the data just yet. A good example of these potential latent risks may be seen in the divergence between northern and southern Europe in recent years. While it has always been known that the two regions differ significantly in economic structure and indebtedness, as well as in socio-demographic profile, these differences were not obvious in the asset-price data e.g., in their equity returns before the financial crisis. However, as debt-to-gdp ratios increased post-crisis and the fiscal integrity of individual countries was called into question, the diverse economic behaviors of the two blocks became apparent in their respective equity and debt markets. Exhibit 4 shows how equities in the most prominent southern European countries have reduced their linkage with northern European stocks as represented by German equities. France is the closest to northern Europe among the southern countries, with the spectrum extending to Greece, which is the least similar to the northern countries and is considered an emerging market today. A table of bond correlations shows a similar divergence. Together, the two factors of large, constantly shifting correlations and latent risk create a potentially damaging fault line in the traditional diversification strategies employed in most managed futures portfolios. Sprinkling assets across countries and industries according to traditional diversification approaches gives investors, in our view, a false sense of security. We believe that latent unbalanced risks that run throughout their seemingly diversified portfolios represent a real threat to long-term performance. Exhibit 4: Correlations among European equities have decreased since the 2008 financial crisis Correlation with German Equities Correlation analysis of MSCI Local Indexes for the respective countries using quarterly returns. Source: Bloomberg France Italy Spain Portugal Greece

4 The Virtues of Risk Balancing Most managed futures strategies are built on the basic assumption that momentum investing has similar Sharpe ratios across individual markets. This simple assumption can encourage managers to add greater overall momentum risk in commodity markets because, at least superficially, they seem highly differentiated from each other, as compared to global equities and bonds. However, as we saw earlier in Exhibits 1 and 2, correlations between commodity sectors were always higher than believed and are now even higher since the financial crisis. Such issues also arise within the four broad asset classes, namely equities, bonds, commodities and FX. Managers achieve diversification by allocating risk budgets equally across these various investments. For example, when investing in equities, managed futures portfolios often maintain equal risk allocations across each major country market. This approach seems to make sense: If equity momentum from these countries has historically demonstrated similar Sharpe ratios, spreading your risk budget broadly and evenly among them should produce an even higher Sharpe ratio and deliver attendant diversification benefits. As a result, managed futures portfolios may end up assigning equal risk budgets to individual momentum strategies based on a variety of geographically diverse equity markets, such as the United States, Canada, the United Kingdom, Germany, France, Italy, Japan and Australia, among others, without considering the commonalities between them. We believe that the diversification achieved by this approach could well prove illusory. Because this traditional portfolio construction fails to take into account issues like shifting correlations and latent risk, the very diversification benefits it produces could evaporate over time or in the face of a catalyzing event. This may result in an unbalanced portfolio of momentum risks. Prudent management demands that we address these risks. We do so by grouping potential investments into risk clusters, or groups of assets that could be affected in similar ways by future catalyzing events. For example, risk clusters in equity markets are easily visualized due to their strong geographic dependencies. In our methodology, we treat the US on par with Europe as a whole rather than treating Europe as an inherently more diverse set due to the many countries within it. This is because European countries are subject to similar risks and variables, including macroeconomic forces, strong trading relationships and demographic factors, among others. While it makes sense to treat Europe as a single entity at the top level of clustering, there is further segregation of risks within Europe, with the UK, Northern Europe (e.g., Germany, Netherlands, et. al.) and South Europe (e.g., Italy, Spain, et. al.) forming three distinct risk clusters. We may expect that, due to structural economic and/or geopolitical factors, these clusters would show similar trajectories (i.e., momentum) in times of stress. There are similar considerations with Asian equities, given parallel economic structures among export-oriented economies, such as Taiwan and South Korea, while Japan has its own distinct trajectory. Meanwhile, Australia and Canada are bound together through commodity-linked economies. To better illustrate our point, Exhibit 5 shows an example of momentum risk allocation through our cluster-based approach. At the top level, we clustered momentum risk equally between the four major asset classes, and then formed smaller sub-clusters within each asset class. In practical terms, this translates into a more balanced distribution of exposures, in our view. Similar formulations are made in other asset classes. For example, in the case of commodities, it helps to cluster on the basis of usage and sources of production. Commodity sub-sectors such as energy, Exhibit 5: Grouping risk in clusters yields, in our view, a more balanced approach to managing momentum risk in managed futures Momentum Risk Equities Bonds Commodity Currencies US Europe Asia Developed Commoditylinked Cluster 1 Cluster 1 Cluster 1 UK Japan Canada Cluster 2 Cluster 2 Cluster 2 North Europe Asia ex- Japan Australia South Europe Source: Allianz Global Investors 4

5 metals and agriculture are natural starting points for the clustering approach, with sub-clusters such as oil-linked energy futures versus natural gas, soybean-based products, among others. Such an approach can add significantly to portfolio performance. To illustrate this, we first constructed simple hypothetical momentum strategies using some of the most liquid futures contracts in the world. 1 For each future contract, we evaluated the price return over the previous one-month, three-month and 12-month periods, assigning +1 if the return was positive and -1 if negative, and then calculated its total score as the weighted average of the three periods. 2 Finally, the position for each contract was sized as proportional to the total score divided by the prevailing volatility. 3 This allowed us to construct four hypothetical momentum portfolios corresponding to each of the four broad asset classes by equal-weighting contract positions within each asset class. Returns are evaluated on a monthly basis. Exhibit 6 juxtaposes those four hypothetical momentum portfolios with the SG Trends Index, 4 which measures the performance of a pool of CTA managers. The chart shows that, since the end of the global financial crisis, investment returns for trend-following managers have tracked very closely the returns on commodity momentum strategies (measured on an equal-weighted basis as well). Additionally, the SG Trends Index has had a very low Sharpe ratio, more similar to commodity and FX momentum than to equities and bonds (Exhibit 7). The result of this analysis strongly suggests an over-allocation to commodities. To further confirm this conclusion, we performed a simple regression analysis of the SG Trends Index against the four Exhibit 7: Momentum has underperformed in commodities and FX since 2009 Sharpe Ratio SG Trends Index 0.69 Sharpe Ratio for the four major asset class hypothetical momentum strategies and the SG Trends Index during the eight years from July 2009 to June Past performance is not indicative of future results. Hypothetical data results have certain inherent limitations. See hypothetical performance disclosure at the end of this document for additional information. Source: Bloomberg and Allianz Global Investors Commodities Bonds FX Equities hypothetical momentum strategies defined above. We found that, since the financial crisis, commodity momentum has contributed over 50% of the total variability explainable in our model. Aside from a spurious sense of diversification in commodities, it is possible that the strong performance of commodity momentum prior to 2008 motivated CTA managers to take undue risks in this asset class. Our balanced risk approach, on the other hand, results Exhibit 6: CTAs returns have, on average, tracked those of commodity momentum strategies Trend-Following by Asset Class in the Post-Crisis Periods Cumulative Return (base = 100) SG Trends Index Commodities Bonds Currencies Equities As of June 30, 2017 Return profiles of the four hypothetical momentum strategies corresponding to equities, bonds, commodities and FX, plotted against that of the SG trends Index. In order to provide a fair comparison, each of the four strategies is scaled to the volatility of the SG Trends Index during the period shown. Past performance is not indicative of future results. Hypothetical data results have certain inherent limitations. See hypothetical performance disclosure at the end of this document for additional information. Source: Bloomberg and Allianz Global Investors 1 See Appendix for a full list of futures contracts used to construct momentum portfolios. 2 The three-month score is weighted twice as much as the other scores to align with industry practices. 3 We use an exponentially weighted moving average (EWMA) volatility measure with a three-month half-life and 12-month window. 4 The SG/Newedge Trends Index is designed to track the 10 largest (by AUM) trend following CTAs and be representative of the trend followers in the managed futures space. The Index is equal weighted and calculates the daily rate of return for a pool of CTAs selected from the larger managers open to new investment. 5

6 Exhibit 8: A hypothetical risk-cluster approach enhanced the performance of managed future strategies Cumulative Return (base = 100) 220 SG Trends Index 200 Balanced Risk (across asset classes) 180 Balanced Risk (across and within asset classes) As of June 30, 2017 Past performance is not indicative of future results. Hypothetical data results have certain inherent limitations. See hypothetical performance disclosure at the end of this document for additional information. Source: Bloomberg and Allianz Global Investors in lower exposures to commodity momentum, as compared to the approximately 50% average industry exposure. This type of risk management approach has been helpful, particularly during the large negative performance contribution from commodity momentum in Relative to the rest of the industry, our cluster-based approach has also resulted in lower allocations to European equities and larger allocations to US equities, with similar benefits on the risk and potential portfolio outperformance. Furthering the analysis, Exhibit 8 shows how the performance of managed futures strategies would have changed had they utilized a risk cluster-based approach similar to the one we propose in this paper. The chart depicts the average performance of trend-following managers from 2009 to June 30, 2017, using the SG Trends Index as a benchmark, which is then compared to two strategies using our proposed approach: 1) Risk balanced across the four strategies based on equities, bonds, commodities and FX, and 2) risk balanced across and within asset classes. In the first case, we equal-weight the futures contracts within each of the four broad asset classes, while in the second case they are weighted as per the clustering scheme shown in Exhibit 5. In both cases, the four asset class momentum strategies are scaled to equal volatility, then combined in equal parts, and finally re-scaled to equal the volatility of the SG Trends Index during the period shown. As the chart illustrates, significant performance improvements were achieved over this period by the use of risk clustering. While the outperformance would be somewhat less than shown due to transaction costs, the futures contracts used in our analysis are the most liquid in the world, and we estimate transaction costs to be less than 1% or so (per annum). As these examples demonstrate, the use of risk clusters can have a potentially large impact on the performance of a managed futures portfolio. But how would that impact play out over a long-term horizon, and what specific benefits would it bring to investors? For many investors, the primary benefit of this approach will be that, due to the improved stability of risk exposures in a managed futures portfolio built with clusters, the strategy can potentially deliver a less volatile and smoother long-term return profile in which no single risk factor dominates the results. As a whole, a managed futures portfolio built using a careful risk balancing approach may stand a higher chance of delivering a consistent return profile over the long term a characteristic that is in keeping with the expectations and needs of investors allocating to an absolute return strategy. Over an extended horizon, this targeted consistency in the return profile should contribute to a potential outcome appreciated equally by investors of all types: Improved investment performance for the portfolio as a whole with the potential benefit of a lower volatility drag. Conclusion Most managed futures products employ strategies that claim to achieve diversification through trading a large number of instruments. However, this intrinsically fails to account for shifts in asset correlation and other latent risks in a portfolio. As a result, the diversification benefits such strategies promise could prove ephemeral and even disappear in the context of poor momentum performance across related assets due to market stresses or crowding behavior. Our risk-balancing approach, called risk clustering, seeks to address this shortcoming by allocating a risk budget across groups of assets with common risk characteristics that could be correlated over time or in periods of market stress. In this way, we seek diversification by balancing risks across a large number of risk clusters, rather than the number of instruments traded. This can reduce volatility and potentially yield a more consistent return profile for the strategy. 6

7 Appendix List of futures contracts utilized to construct the hypothetical momentum portfolios Equities Bonds Commodities FX AEX Australia 10 YR Aluminum AUD ASX SPI 200 Australia 3 YR Cocoa CAD CAC 40 Canada 10 YR Coffee CHF DAX Euro BTP Copper Euro FTSE 100 Euro Bund Corn GBP FTSE MIB Euro Buxl Cotton JPY Hang Seng Euro OAT Gasoil NOK IBEX Euro Schatz Gasoline NZD KOSPI 200 Japan 10 YR Gold SEK MSCI Taiwan UK Gilt Heating Oil OMXS30 US 10 YR Lean Hogs S&P 500 US 2 YR Live Cattle S&P/TSX 60 US Long Bond Natural Gas Topix Nickel Oil Brent Oil WTI Platinum Silver Soybean Soybean Meal Soybean Oil Sugar Wheat Zinc 7

8 Investing involves risk. The value of an investment and the income from it may fall as well as rise, and investors may not get back the full amount invested. Past performance is not indicative of future results. This document is being provided for informational purposes only and should not be considered investment advice or recommendations of any particular security, strategy or investment product. Statements concerning financial market trends are based on current market conditions, which will fluctuate. Forecasts and estimates have certain inherent limitations, and are not intended to be relied upon as advice or interpreted as a recommendation. The opinions expressed herein represent the current, good faith views of the author(s) at the time of publication and are provided for limited purposes, are not definitive investment advice, and should not be relied on as such. The information presented in this article has been developed internally and/or obtained from sources believed to be reliable; however, Allianz Global Investors does not guarantee the accuracy, adequacy or completeness of such information. Predictions, opinions, and other information contained in this article are subject to change continually and without notice of any kind and may no longer be true after the date indicated. Any forwardlooking statements speak only as of the date they are made, and Allianz Global Investors assumes no duty to and does not undertake to update forward-looking statements. Forward-looking statements are subject to numerous assumptions, risks and uncertainties, which change over time. The hypothetical performance information contained in this presentation is being provided for an institutional audience only and cannot be shared, reproduced or distributed to third parties or the public without the express written consent of AllianzGI US. Hypothetical performance is being presented for illustrative purposes only and does not represent actual performance of any client account. The results are based on hypothetical allocations which have certain inherent limitations. No representation is being made that any account will or is likely to achieve profits or losses similar to these being shown. The hypothetical performance results do not represent the results of actual trading and might not reflect the impact that material economic and market factors would have on the decision-making if AllianzGI US was actually managing a client s money. Fees may vary depending on, among other things, the applicable fee schedule and portfolio size. Investors should not assume they will have investment results that are similar to the hypothetical performance shown. There are frequently material differences between hypothetical performance results and actual results subsequently achieved by a particular investment strategy. Additional information regarding policies for calculating and reporting returns is available upon request. Hypothetical returns have many inherent limitations, only some of which are described here. The returns were developed with the benefit of hindsight and do not reflect the impact that material economic and market factors might have had on investment decisions if client funds were actually managed in the manner shown. The hypothetical returns have been generated based on certain assumptions that may not be reasonable or applicable for particular clients. Changes in the assumptions may have a material impact on the hypothetical data and returns presented. Performance is shown for a limited period of time. Performance over a different market cycle may not be as favorable as the performance shown and may result in losses. There can be no assurance that any client account will achieve profits similar to those shown or avoid incurring substantial losses. Allianz Global Investors is a global asset management business that operates under the marketing name Allianz Global Investors through affiliated entities world wide, including Allianz Global Investors U.S. LLC (AllianzGI US) a SEC registered investment adviser. No part of this material may be reproduced in any form, or referred to in any other publication, without express written permission. us.allianzgi.com 2018 Allianz Global Investors U.S. LLC 1633 Broadway, New York, NY 10019, us.allianzgi.com