Why Alt Beta?

Overview

Passive management has transformed the asset management industry. The first index fund was started in 1972. Since then, the U.S. index mutual fund industry has grown to more than $3 trillion. This represents about 20% of the total assets under management for the U.S. mutual fund industry.¹ This growth was driven by skepticism toward the value of active management as well as a relentless search for lower fees.

The next wave of innovation was “smart beta” funds, which modify passive strategies with factor tilts, such as favoring cheaper stocks to generate better returns than traditional market capitalization-based indices. This industry just passed $1 trillion in assets.²

Now “alt beta” funds are primed to be the next step. Alt beta aims to replicate some of the factor returns of hedge funds using automated trading rules implemented at low costs. The market is estimated at around $200 billion and is growing rapidly.³

This research note provides an overview of alt beta investments. It first illustrates how alt beta products can replicate returns for some categories of hedge funds, using foreign currency trading as an example. It then discusses the rationale for the existence of risk premia on alt beta products. Finally, it shows how alt beta products can be harnessed to add value to investor portfolios, through proper selection and structuring of the alt beta portfolio.

What is Alt Beta?

Alt beta strategies manufacture returns from systematic trading strategies that are expected to capture positive compensations for risk, or “risk premia”:

  • •  With low sensitivity to traditional asset classes (i.e., low beta, duration, spread duration)
  • •  Across various asset classes, such as equities, rates, currencies, and commodities
  • •  Across various styles, such as value, carry, trend, and volatility
  • •  At relatively low costs

Thus, alt beta strategies can be used to construct pure “zero correlation” portfolios.

Alt beta relies on a wide variety of risk premia. The rationale for this is analogous to the “equity premium.” Equity markets are very volatile. Over the long run, however, equities do tend to go up by an average of, say, 4–8% per annum. Most investors believe in the existence of a positive long-term equity premium—even though they may not agree on the exact number—because this represents a compensation for bearing non-diversifiable equity risk.

From the viewpoint of a long-term investor, there would be much interest in a product that also returns a premium around 4–8% but with low correlations to traditional asset classes and reasonable volatility. A properly structured alt beta product should be able to achieve this by collecting multiple risk premia across sources largely uncorrelated with each other. As we shall see, implementing such a strategy through total return swaps, for example, allows for leveraging up the portfolio to achieve a desired risk premium.

Example: Risk Premia for FX

As an illustration, consider an investment in a hedge fund that trades foreign currencies (FX). The hedge fund manager should add value by actively trading the portfolio, after trading costs and fees. Some of this positioning, however, represents exposures to risk factors that have been shown in academic research to provide positive returns over the long run. Such factors include:

  • •  “FX carry,” due to the fact that some currencies pay high interest rates, which provide higher returns that are not fully offset by a currency depreciation
  • •  “FX trend,” due to the fact that short-term currency movements tend to persist and are thus somewhat predictable
  • •  “FX value,” due to the fact that undervalued currencies using a metric such as Purchasing Power Parity (PPP) tend to revert back to their long-term, fair value equilibrium

These factors are well rooted in academic research, which has demonstrated that they provide positive returns, when averaged over multiple years. The averages, or expectations, of these factor returns constitute what we call “risk premia.”

Take the Brazilian currency, the real (BRL), for example. The short-term interest rate on BRL is around 7%, as opposed to 1.5% currently for the dollar in the U.S. So, investing in BRL-denominated cash should provide an excess return of 5.5% if the currency does not move. On the other hand, if the currency depreciates by 5.5%, the investment will be a wash. If the BRL depreciates by more, this will create a loss. Empirically, we observe wide swings in exchange rates, but, on average, higher interest rates are not fully offset by currency depreciations. This is what creates the FX carry risk premium.

Ideally, such premia should be traced to a source of inefficiency. For FX carry, this is probably due to market segmentations. In practice, interest rates are set unilaterally and independently by central banks. The Brazilian money supply amounts to approximately $100 billion, which reflects very large currency reserves. There would have to be very large flows of speculative capital to eliminate systematically this FX carry effect.
Likewise, there should be a good explanation for the “FX trend” effect. Academic research has tied this effect to participants in the currency markets that are not profit-oriented. This includes central banks, which intervene in that market to stabilize currencies as opposed to make a profit, as well as exporters and importers, who are hedging instead of speculating. For instance, any of these participants could be continuously buying the currency, creating a slight appreciation that would persist.

Finally, the “FX value” effect represents slow, long-term movement back to equilibrium exchange rates that evolve according to relative inflation rates. For instance, a country with a currency that is abnormally cheap in real terms relative to others would be expected to have increasing exports because of its cheap manufactured goods. This would create a long-term increase in the demand for that currency, pushing it back to equilibrium.

Alt beta products attempt to replicate each of these “risk premia” by a systematic trading rule following these principles. For example, Binny (2005) uses:

  • •  For FX carry, buying 3 currencies with high interest rates, and selling 3 with low rates
  • •  For FX trend, buying (selling) a currency when its 5-day moving average is above (below) the 40-day average
  • •  For FX value, buying (selling) currencies where the current exchange rate is below (above) the long-term fair value derived from Consumer Price Indices

In practice, there are many different ways to implement these algorithms by changing the sample of currencies, the number of currencies bought or sold, the parameters of the trading rules, the construction of the portfolio, and so on. Evaluating various algorithms requires practical experience in building and assessing such models, in particular their soundness and robustness.

So, what is the empirical evidence in terms of the expected return and risk for these FX risk premia? Table 1 displays estimates over a long, 30-year period, from 1987 to 2016.4 Like the equity premium, all risk premia are defined in excess of the cash rate.

These three FX trading strategies all provide positive average returns, ranging from 3% to 5% per annum, with an annual volatility in the range of 9%. This confirms the existence of FX risk premia over this period, and is in line with the abundant academic research on the topic.

Example: Replicating FX Traders

Armed with these three FX risk premia, we now have tools to evaluate and try to replicate the performance of any FX hedge fund. As a reference, let us use the performance of the Barclay Currency Traders (BCT) Index, which is an equal-weighted composite of manager programs that trade currency futures and forward contracts with a long history.

Over the period from 1987 to 2016, this index has grown by an average of 6.8% annually, net of all management and incentive fees.

Let us now decompose the performance of this index and check whether it can be partially replicated by our FX risk premia. Over this period, cash returned 3.7% pa, so the BCT Index had an excess return of 6.8 – 3.7 = 3.1% pa. How much of this can be explained by the three FX premia?

The most simplistic approach is to assume that these FX traders have fixed exposures to these three risk factors. These exposures can be estimated with a multivariate regression of the BCT Index returns on the three FX factors. Results are shown in Table 2. Two of the slope coefficients are strongly significant, with positive values and t-statistics above 3. This means that FX traders act as trend followers and high carry currencies investors. There is no evidence that they follow value strategies, however.

The final step reports how much of the BCT Index performance can be attributed to FX risk premia, by combining the traders’ exposures with FX risk premia. This is shown in Figure 1.

The figure shows that, out of the 3.1% excess return provided by FX traders, 2.2% can be accounted for by FX risk premia. The remainder is pure alpha, 0.9%. Thus, simple mechanical trading rules can account for a good fraction of the active manager excess returns, in this case more than two-thirds of the total (2.2 out of 3.1%).

So, why pay high fees to hedge fund managers if a large fraction of their returns can be replicated at low cost? Essentially, this is the rationale behind alt beta.

These results do not imply, however, that all hedge funds can be so easily replicated. FX trading takes very few security-specific bets and has a relatively narrow universe of currencies. Likewise, Global Macro funds trade in broad indices, where strategies can be partially replicated by alt beta algorithms. At the other end of the spectrum are hedge fund strategies such as Relative Value that rely heavily on individual security selection. Indeed alt beta factors explain the highest fraction of alpha for Global Macro funds, and the lowest for Relative Value funds.7 Thus alt beta is only a substitute for some categories of hedge funds.

What are These Risk Premia?

Alt beta risk premia represent payoffs on long-short strategies typically used by hedge funds. For classification purposes, they can be usefully sorted into two broad categories, as described in Table 3. The first is by type, e.g., value, carry, trend/momentum, and volatility. These apply across asset classes, e.g., equities, fixed income/rates, currencies, and commodities.

The risk premium in each cell of this 4 x 4 table has been generally supported by extensive academic research and well documented by Ilmanen (2011).8 This research has attempted to explain these premia in terms of (1) rational risk factors, (2) behavioral factors, or (3) market segmentations.

  1. Rational Risk Factors: For example, the extra return on value equities relative to growth stocks could be due to some omitted risk factor, such as default risk, that is not properly measured in the traditional market beta. Exposures to such risk factors may be worthwhile if they carry a high enough compensation.
  2. Behavioral Factors: Momentum in prices could be due to systematic behavioral biases on the part of investors, who tend to buy assets that have gone up recently in price, thereby further pushing prices up. Or, some market participants could act for non-economic reasons.
  3. Market Segmentations: These can justify why prices are not fully arbitraged away between different markets, for a number of reasons including regulations, insufficient arbitrage capital, or disequilibrium between demand and supply. One example is equity carry, where dividends futures trade at an implied rate that can be systematically lower than expected dividend rates. This is due to the issuance of structured products by banks, which are promising clients a return tied to the capital appreciation of a stock index. Banks hedge this risk using total return indices, which leave a residual risk tied to dividend payments. Banks can hedge the remaining dividend risk by selling dividend futures, thus depressing implied dividend rates.

Such systematic explanations are quite useful, because they suggest why these risk premia should persist. The counterbalancing effect is always that arbitrage capital could move to take advantage of these risk premia, potentially reducing their size over the long run.

In practice, this reduction will depend on the size of the capital arbitrage flows. Long/short arbitrage capital typically originates from hedge funds and proprietary bank trading desks. Regulatory restrictions imposed on bank trading since the Global Financial Crisis, however, have decreased this second source of arbitrage capital.

Are These Risk Premia Reliable?

Like all sources of risk premia, the reliability of these positive average returns over time depends on the signal-to-noise ratio, i.e., the ratio of the expected return to its annual volatility. This is typically low, leading to wide swings in returns. So, an investor could experience a couple of years with negative returns even if the risk premium is positive over the long term.

This is not unique to alt beta risk premia, however. Reliability is also an issue with the equity premium. Even though most investors are thoroughly convinced that the equity premium is positive, the size of this premium is relatively small relative to the volatility of equity returns. Assume a 5% equity premium, for example. Say that over the long run, stock return volatility is around 20%, leading to a “Sharpe ratio” of 5%/20% = 0.25. This is rather low, and can lead to many years with negative returns on the stock market. This is illustrated in Table 4.

The second panel asks the question of how many years would be required to ascertain the statistical significance of the equity returns averaged over some period. Put differently, how many years would we need to be reasonably confident that this estimated mean reflects a positive premium?

This question can be answered by computing the usual t-statistic, which is the ratio of the estimated mean to its standard error. Say the estimated mean is indeed 5%. Its standard error depends on the number of observations. With 30 years of returns, this is σ/√Τ = 20%/√(30) = 4.6%. Taking the ratio of 5% to 4.6% gives a t-statistic of 1.10. We usually require t-statistics in excess of 2 to establish significance at the 95% confidence level. Because this t-statistic is lower, we cannot conclude that this mean return of 5% is statistically “significantly” positive.

As the table shows, to get a t-statistic above 2, we would need 100 years of data to prove with high confidence that the equity premium is positive. In other words, it is hard to prove statistically that there is a positive equity premium even though most investors would believe in its existence.

The same issue affects individual alt beta risk premia. Even with positive risk premia, there could be strings of years with negative returns.

This is why statistics based on history must be combined with judgment. Understanding the rationale for the alt beta risk premia is critical to assessing its continued existence. In other words, we need to understand whether the drivers of the risk premium are likely to have disappeared or not. As discussed previously, this includes evaluating (1) rational risk factors, (2) behavioral factors, or (3) market segmentations.

In addition, current market conditions may give useful clues. As an example, take the FX carry risk premium. We showed that this is largely driven by differences in nominal interest rates between different currencies, e.g. 7% on the Brazilian real vs. 1.5% on the U.S. dollar. If the exchange rate does not move, we would expect to reap 5.5% annually. On the other hand, if this difference were to shrink from 5.5% to 0%, we would not expect any FX carry risk premium on the Brazilian currency. Indeed, in the aftermath of the financial crisis, most G-10 currencies had low interest rates, within 1% of each other. So, the FX carry risk premium must have largely gone away for G-10 currencies. Emerging market currencies, however, still enjoy a wide spread in nominal interest rates, so should still benefit from the FX carry. Therefore, making a judgement as to the continued existence of some alt beta risk premia requires a good knowledge of their drivers and current market conditions.

How to Harness Risk Premia?

We have shown that individual alt beta risk premia are somewhat uncertain, like all risk premia. Such risk premia, however, can be harnessed by the power of diversification. These premia should be constructed to have low correlation with general markets. They also tend to have very low correlations across each other, if suitably selected. This can help us build portfolios that take advantage of the “Fundamental Law of Active Management”, using what Grinold (1989) calls “breadth.”9

Indeed, pooling together many risk premia should help us improve the Sharpe ratio of the portfolio. Table 4 illustrates a hypothetical portfolio with ten risk premia, all with fairly low Sharpe ratios, conservatively estimated at 0.25, but with zero cross-correlation. For simplicity, we assume that all risk premia are scaled to the same volatility of 10%, and are all assigned the same weight of 10%. So, all risk premia have the same expected return of 2.5%, in excess of cash.

Since all expected returns are the same, the portfolio expected excess return is also 2.5%. Next, the variance of the portfolio can be calculated at V = ΣN (w × σ)2 = N × (w × σ)2 = 10 × (10%×10%)2 = 0.001. This gives a volatility of 3.2% only. As a result, the Sharpe ratio has now increased from 0.25 for individual risk premia to SR = 2.5%/3.2% = 0.79 for the portfolio.

This shows that cross-sectional diversification across risk premia factors should help lower portfolio volatility, ensuring more stable and reliable returns.

Investors, however, will in general require higher levels of returns than 2.5%. This can be easily achieved by leveraging up the portfolio. For instance, alt beta risk premia can be accessed by total return swaps, which do not need to be fully funded upfront and therefore allow leverage by increasing the notional amount.

Assume that the investor requires a target volatility of σ = 8%. As shown in Table 5, the portfolio can then be leveraged by a factor of 8%/3.2% = 2.5. This would increase the portfolio expected return, in excess of cash, to ER = SR × σ = 0.79 × 8% = 6.3%. Adding a risk-free rate currently around 1.5% gives an expected total portfolio return of 7.8%.

Higher levels of leverage and returns can be achieved as long as the portfolio holds enough cash to meet the initial margin requirements and as a buffer against potential margin calls.

Conclusions

Alt beta products are the next logical step in the progression toward automation of active management. Indeed, such trading algorithms aim to capture compensated risk factors, called “risk premia,” that are largely specific to the hedge fund industry.

This note has discussed the nature of alt beta risk factors and has illustrated some of their salient features. They should be largely uncorrelated with traditional asset classes, providing a “zero correlation” alternative product. They should be implementable at relatively low costs. Perhaps their most interesting feature is their “breadth,” due to the large variety of styles, asset classes, and implementation algorithms. This should lead to a selection of strategies that are largely uncorrelated with each other. As a result, a properly constructed portfolio of alt beta factors should be quite diversified, leading to a Sharpe ratio that is much higher than its components.

This breadth can be implemented through a portfolio of total return swaps, for example. Such swaps can easily allow changing leverage, thereby creating enough flexibility to match the risk/return profile of the alt beta portfolio to that desired by the investor.

 

1  The first index mutual fund was Qualidex, which aimed to replicate the Dow Index. Estimates of the size of the U.S. mutual fund and index mutual fund industries are from the Investment Company Institute’s Fact Book (2017).

2  Financial Times, December 28, 2017.

3  Albourne (Notional AUM Invested in Dynamic Beta Strategies–Survey Results, 2017) estimates the alt beta market (called long/short dynamic beta) at $216 billion.

4  Binny, James (2005), “Currency Management Style through the Ages,” Journal of Alternative Investments 8, 52–59.

5  Source: Binny (2005), or ABN AMRO Bank for data from 1987 to 1989. After that, the risk premia are taken from Deutsche Bank, and are available on Bloomberg. All average returns are taken as monthly averages annualized. This allows an exact decomposition of returns into risk premia components that add up to the total, unlike when using compound returns.

6  This approach has been used by academic research that examines the value added of foreign currency managers. See Levich, Richard and Momtchil Pojarliev (2008), “Do Professional Currency Managers Beat the Benchmark?” Financial Analysts Journal 64, 18–32.

7  See Coupe and Norman (2016), “Don’t Bet it Alt on Beta,” PAAMCO Viewpoint.

8  Ilmanen, Antti (2011), Expected Returns: An Investor’s Guide to Harvesting Market Rewards, Wiley.

9  Grinold, Richard (1989), “The Fundamental Law of Active Management”, Journal of Portfolio Management 15, 30—37.

 

 

Philippe Jorion - PAAMCO

Philippe Jorion, PhD is a Managing Director in the Risk Management Group. He oversees the development and implementation of risk measures for PAAMCO hedge funds, sectors, and client portfolios, all of which are based on full position-level transparency. He is charged with the active management of risk, which involves building tools for monitoring managers, customizing funds, and helping to manage overall portfolios. As chair of the firm’s Investment Oversight Committee, he is involved in all stages of the investment process. He also serves as a chaired Professor of Finance at the Paul Merage School of Business at the University of California at Irvine. He is a frequent speaker at academic and professional conferences and is on the editorial boards of a number of finance journals. Philippe has authored more than 100 publications on the topic of risk management and international finance. Some of his most notable work includes the Financial Risk Manager Handbook (Wiley 6th ed. 2010), which provides the core body of quantitative methods and tools for financial risk managers; Big Bets Gone Bad: Derivatives and Bankruptcy in Orange County (Academic Press 1995), the first account of the largest municipal failure in US history; and Value at Risk: The New Benchmark for Managing Financial Risk (McGraw-Hill 3rd ed. 2006), the first definitive book on VAR. In March 2013, the Financial Analysts Journal honored Philippe Jorion and Rajesh K. Aggarwal’s article titled “Is There a Cost to Transparency?” with the 2012 Graham and Dodd Scroll Award. From 2012 to 2014, Philippe served on the Federal Reserve’s Model Validation Council, an advisory council of economists that evaluates the models used in bank stress tests. Philippe holds an MBA and a PhD from the University of Chicago and a degree in engineering from the University of Brussels. Philippe has over thirty years of experience in investments and risk management.

 

Pacific Alternative Asset Management Company, LLC (“PAAMCO U.S.”) is the investment adviser to all client accounts and all performance of client accounts is that of PAAMCO U.S. Pacific Alternative Asset Management Company Asia Pte. Ltd. (“PAAMCO Asia”), Pacific Alternative Asset Management Company Europe LLP (“PAAMCO Europe”), PAAMCO Miren Portföy Yönetimi A.Ş. (“PAAMCO Turkey”), Pacific Alternative Asset Management Company Mexico, S.C. (“PAAMCO Mexico”), and PAAMCO Colombia S.A.S. (“PAAMCO Colombia”) are subsidiaries of PAAMCO U.S. “PAAMCO” refers to PAAMCO U.S., PAAMCO Asia, PAAMCO Europe, PAAMCO Turkey, PAAMCO Mexico, and PAAMCO Colombia, collectively.

This document contains the current, good faith opinions of the authors but not necessarily those of Pacific Alternative Asset Management Company, LLC and its subsidiaries (collectively, “PAAMCO”). The document is meant for educational purposes only and should not be considered as investment advice or a recommendation of any type. This document may contain forward-looking statements. These are based upon a number of assumptions concerning future conditions that ultimately may prove to be inaccurate. Such forward-looking statements are subject to risks and uncertainties and may be affected by various factors that may cause actual results to differ materially from those in the forward-looking statements. Any forward-looking statements speak only as of the date they are made and PAAMCO assumes no duty to and does not undertake to update forward-looking statements.

Pacific Alternative Asset Management Company is a registered trademark in the United States, Canada, Japan, Singapore Australia and Mexico. PAAMCO is a registered trademark in the United States, Canada, Europe, Japan, Australia and Mexico. Pacific Alternative Asset Management Company Europe and PAAMCO Europe are registered trademarks in Europe. Pacific Alternative Asset Management Company Asia and PAAMCO Asia are registered trademarks in Singapore. completeAlpha is a registered trademark in Singapore, Japan, the EU, the U.S. and Canada and it is a trademark of PAAMCO in Australia.

PAAMCO is a Core Supporter of the Standards Board for Alternative Investments.