The Five Faces of Alpha (II): “True” alpha reveals itself

Today we bring you part 2 of Erik Einertson’s special to AllAboutAlpha.com “The Five face of Alpha”  (first installment here)…

Alpha #2: Insurance Beta

The second type of “Alpha” often found in the market can be referred to as insurance beta or Informationless Investing (Weisman 2002). This type of exposure has persistent tilts towards strategies that have an insurance like payout. The payout for this type of strategy will experience long periods of modest positive returns (and high IR) followed by short periods of large negative performance. While the selling of insurance can be part of a diversified portfolio, it is the tactical use of these strategies that creates real alpha, not the systematic tilts toward these exposures. Unfortunately, many investors in these types of strategies are unaware their portfolios have this type of risk exposure.

Insurance beta strategies are often used to improve short-term performance and can be identified by their characteristic negative skew distribution over time. Alpha that is derived from systematic tilts to Insurance Beta premiums should, at best, not be considered true alpha, and at worst, are a dangerous addition to a portfolio if the end investor is unaware of the risks. In many cases, the negative tail risk of these portfolios are at the exact time an investor is relying on their alternative investments to be uncorrelated with their other assets. Examples of this type of strategy include: explicit short volatility positions (selling puts), implicit short volatility positions like the currency carry trade (Frankel, 2007) and St. Petersburg Strategies (which involve starting with a simple bet, say 5$. If one wins, they bet $5 again the next time. If they lose, they double their bet after each losing period to get back to break-even. Profit is taken whenever one wins consecutively.) Other strategies may not immediately appear to be unrelated, but in reality exhibit strong negative skew and qualify as insurance beta type of strategies based on their skew distribution, such as the exploitation of credit and term spreads (the common strategy of many of the Structured Investment Vehicles–SIVs–used), market making (providing liquidity to an illiquid market) and even mortgage securities.

For purposes of categorization, we will say that strategies that have a long-term negative skew distribution perform significantly over equities and Insurance Beta strategies. (For this analysis I have defined this range to be a negative skew distribution larger than the Russell 3000 over the past 20 years ending June 2009, around -0.77.) This type of alpha is the least valuable form for investors’ portfolios, but because of the likelihood of short-term high IRs there is a bias towards investing in them. Even worse, because they show little volatility for long periods of time, they encourage leverage that exacerbates losses when the negative skew event occurs.

Risks – in most cases, this type of alpha exposes the portfolio to extreme risks that are not being fully valued by portfolio managers. Looking at Table 1 (to the right), which shows 20 years of data, the overall HFRI indices of Convertible Arbitrage, Distressed/ Restructuring, Emerging Market, Event Driven, Multi Strategy and Merger Arbitrage all seem to have some type of Insurance Beta exposures in their strategy. Interestingly enough, if you take the data from 1990 to December 2006 (prior to the recent economic turmoil), only the HFRI Multi-Strategy Index had less negative skew than the Russell 3000 (-0.26 vs -0.59) showing that many of these characteristics should have been well known by the marketplace prior to the most recent economic calamity.

Portfolio Effect – insurance Beta does not necessarily have a negative effect on portfolio diversification. It is the fact that so many of these strategies have their negative tail events simultaneously that makes them so devastating on portfolios.

Alpha #3: Exotic Beta

A third type of “alpha” is the use of exotic betas. These betas tend to be systematic tilts towards market premia that have persisted for long periods of time. An example of this would be equity managers tilting towards the traditional Fama and French factors of value and small-cap, as well as momentum and illiquidity premiums, emerging markets and commodities. Exotic betas will have many of the same characteristics of the betas referred to above with a few exceptions:

  1. They are not as easy to replicate with futures as general stock and bond market exposures (larger bid/ask spreads, less liquidity and higher tracking error).
  2. The exposure to this beta may be a byproduct of another instrument and may be intentional or unintentional. For example, quantitative firms focus on valuation in equities has caused a long-term valuation bias in their portfolios. For technical firms, a tilt towards momentum will be common.

For exotic beta, these persistent tilts are measurable by using a variety of tools: for example, a three factor model that measures for any Exotic Beta.

Recent published research has focused on categorizing a majority of hedge fund returns into these exotic betas, including important work from Robert Litterman describing Exotic Beta (Litterman 2007) and William Fung and David Hsieh’s development of a seven factor model that captures a large percentage of hedge fund returns (Fung, William and David Hsieh 2004), particularly for indices that include significant directional exposures. In fact, their seven factor model had an R2 of 87.6% with the HFRI Fund Weighted Composite from December 2002 to December 2004.

Using the framework I have described, Exotic Betas should not exhibit dramatically different risk profiles than standard market indices. While these betas tend to not be as complicated as insurance betas, they should not carry excessive negative skew or kurtosis. To qualify as an Exotic Beta, the skew distribution should be similar to that of equities and bonds. For a point of reference, a reasonable assumption is that the skew distribution will lie between the Barclay’s US Aggregate and the US Small Cap Equities (between -0.29 to -0.77 over the past 20 years ending December 31, 2009).

Risks – the biggest risk to this type of alpha is believing it is uncorrelated with the rest of one’s portfolio and expecting diversification and/or paying alpha level fees for something that should be much cheaper. If only 22% of Funds of Funds actually create alpha, should there really be a 2% and 20% fee structure for any beta?

Portfolio Effect – this type of alpha should be considered more valuable than Insurance beta at similar risk adjusted return expectations. They can serve a useful diversification benefit, especially if they are able to exploit some market inefficiency but still should not be mistaken for true alpha.

Alpha #4 – True skill alpha

True skill alpha can be defined as alpha that can be attributed to a manager’s tactical decisions and skill. This type of strategy is what most investors believe they are getting when they buy active management strategies. This type of strategy is more valuable than any of the three proceeding alpha strategies at similar risk adjusted returns because its returns should be uncorrelated with virtually all market indices over the long run, and the strategy should have virtually no skew in its distribution (between -0.3 to 0.3).

There are really two separate types of “True skill alpha” that need to be considered: tactical beta and uncorrelated alpha.

Tactical Beta – This beta uses tactical weightings of betas, insurance betas and exotic betas to create alpha. Its opportunity is best when markets have tendencies to either extreme positive (where going long on a particular beta versus another produces a positive payout) or negative (where going short a particular beta versus another produces a positive return.) From an investor’s point of view, a manager that is truly able to time the market would provide very strong negative correlations to traditional asset classes when they are going down, but high correlations when they are going up.

While this type of alpha will use general market betas, exotic betas andinsurance betas to achieve this performance, it is the tactical use of these exposures that cause it to be true alpha as opposed to those previous and inferior classifications of alpha.

Uncorrelated Alpha – this type of strategy should have very little correlation with the market at any point in time. The change in general market conditions should have very little effect on this type of strategy’s ability to earn alpha. This is the truly “uncorrelated” alpha source, although unlike tactical beta, there is little increase in opportunity when markets are decreasing. Carefully crafted Equity Market Neutral and currency strategies can be in this category.

Alpha #5 – Protection Alpha or The Ultimate Investment

The final, and most valuable, type of strategy is any strategy that can expect a positive return and still exhibit positive skew. This type of strategy acts like one is buying insurance and is the most elusive of alpha strategies.

An example of this type of strategy is risk-free arbitrage. Risk-free arbitrage should have a positive skew, as the alpha opportunity being exploited will not always be available. By taking advantage of mispricings when they exist, they will have periodic large positive payouts surrounded by frequent flat or–including fees–negative performance.

While short-term IRs of “The Ultimate Investment” may not be overly impressive, the ability of the portfolio to help moderate down-side risk is positive for overall portfolio risk. Some investors will shy away from these investments because the returns are episodic. What is discouraging is that more plans should be looking for these strategies. Unfortunately, reliance on inappropriate statistics used to evaluate alpha strategies, and a reliance on short-term performance, invariably leads participants towards other investments that are not as appropriate for their portfolio.  For this evaluation, these strategies should display a long-term skew distribution of over 0.4.

Conclusion

It has been well documented that the use of IR and raw performance have created inappropriate biases towards investments that are often inappropriate for end clients. With recent market turmoil, we have seen that many strategies that were advertised as “pure alpha” were really just exotic or insurance beta. A deeper examination of the sources of active management returns is needed to find what the true sources of out- performance are, especially where there is a lack of transparency. Often times, hedge funds use simple risk premia to create returns as opposed to true skill. The use of skew distribution is one such tool that can be used to find the methods a manager is using to generate returns.

Be Sociable, Share!

3 Comments

  1. Sami Al-Bashir
    April 8, 2010 at 11:54 am

    Thanks for the article, this is a great summary of the problems faced by investors when attempting to identify alpha. One question – in the table of HFRI indices showing a variety of metrics including skew why was equity long/short (equity hedged) not included? Seems odd considering that is where the majority of investor assets in the hedge fund space reside.


  2. David Harper
    April 10, 2010 at 2:39 pm

    Thank you for a truly good 2-part article. Question: insurance beta vexes me because I cannot grasp the beta factor in the same way that i can grasp (eg) the small cap premium in FF. After all, i don’t think it’s a volatility factor per se … is it possible what you are calling “insurance beta” is not so much common factor but rather merely that the typical regression-based alpha/beta methodology is not formally incorporating risk?


  3. Erik Einertson (Author)
    April 12, 2010 at 2:37 pm

    I will try and reply to both of your questions is one email.

    Sami – I used equity market neutral in the analysis. There is certainly is good reason to do the analysis on equity long/short equity hedged as well. I just used an abbreviated list and went from there. Looking back, it would have been good to add that to the list.

    David – To answer your question directly, I am calling beta something that doesn’t require real skill and is static. You and I can create this exposure (I am certainly not implying that you don’t have skill) without really accounting for what is happening in the market. A strategy that wrote deep out of the money puts, but uses market data and, for example, and stopped writing them early in 2008 would exhibit less skew and, as a result, I believe would be more alpha like. I am trying to move away from the definition of beta as just a regression based idea.

    The idea of insurance beta isn’t that it has “beta” in terms of exposure to normal market movements. A good example would be the CMO insurance sold in huge quantities by AIG. They were able to book profits from the day they wrote contracts on the CMOs and hadn’t really accounted for the possibility of a big downturn in housing when accounting for possible losses. In the end, the size of the insurance payment overwhelmed the strategy.

    I am very interested in doing further research to see if there is a long-term risk premium associated with skew. My instinct is that it varies considerably and at most times is negative because investors underestimate the risk of these extraordinary events and are attracted by the short-term high IRs.


Leave A Reply

← Did Equity Long/Short Hedge Funds Fail Investors in 2008 & 2009? The Five Faces of Alpha →