Second, mutual fund managers are generally prohibited from taking short positions in the stocks in the stocks in which they invest. In contrast, hedge fund managers are allowed to take short positions. Practically, this "long only" constraint means that mutual fund managers can only make money from investing in assets that they believe to be undervalued, while hedge fund managers can make money from both undervalued and overvalued situations.
Third, mutual fund managers are generally limited in the amount of leverage (be it in the form of debt or derivatives) they can use to magnify their returns. From a regulatory point of view, there is a good reason for this: using leverage is a risky strategy, that magnifies not only gains, but losses as well (remember Long Term Capital Management?). Presumably, sophisticated "accredited investors" understand this risk, and are willing to take it when they invest in hedge funds, which can and do use leverage.
Now that we know, in general terms, how hedge funds make money, let’s look in somewhat more detail at the different strategies they employ.
The first major group of strategies used by hedge funds are known as "event-based" investing. "Event Driven" funds try to make their money by taking long or short positions based on their forecast about the outcome of an expected event. For example, some of these funds invest in the securities of companies involved in merger and acquisition transactions, while others invest in the debt and equity of firms facing serious financial problems. At the end of 2003, Event Drive hedge funds accounted for 17% of the assets in the CSFB/Tremont Investable Hedge Fund Index.
The second major investing strategy used by hedge funds is arbitrage. In the traditional meaning of the term, arbitrage was a low risk strategy, in which one simultaneously bought an asset in a market in which it appeared underpriced, while selling the same or a very similar asset in another market in which it appeared overpriced. As practiced by hedge funds, however, this strategy is considerably higher risk, and often involves holding open long and short (and highly leveraged) positions in assets whose alleged similarity occasionally turns out not to be the case (just ask the people who ran Long Term Capital Management).
Within the overall arbitrage strategy group, "Convertible Arbitrage" funds try to make money by taking advantage of pricing differences between a company’s convertible bonds (that is, bonds that have the option of being converted into equity shares at a later date) and its outstanding shares. For example, a hedge fund might buy a company’s convertibles while selling short its stock, assuming the latter was perceived to be overvalued. The profit on the strategy would come from both the interest earned on the bond, plus the profit earned on the short sale of the stock (when you sell a stock short, you receive a price for the shares today, but promise to deliver them at a later date. If the shares have declined in price by that date, you can buy the shares you need to deliver for a price that is lower than what you have received for them). However, because the profit margins on these convertible arbitrage trades are usually small, hedge funds in this category generally use substantial amounts of leverage to magnify their returns. At the end of 2003, approximately 11% of the total amount invested in hedge funds tracked by the CSFB/Tremont Index was invested in funds in this category.
"Fixed Income Arbitrage" funds try to profit by taking advantage of pricing differences between similar fixed income securities (buying the undervalued one, and shorting the overvalued one). Again, because the profit margins on individual transactions are small, these funds typically use large amounts of leverage. Long Term Capital Management was in this category, and provides a vivid example of how high leverage can quickly lead to a hedge fund’s demise if its view of the market proves incorrect. Fixed Income Arbitrage funds accounted for 10% of the total amount invested in the hedge funds tracked by CSFB/Tremont.
A third type of hedge fund is often included in the arbitrage category, but its fit there is awkward at best. The managers of "Equity Market Neutral" funds essentially hunt for pure alpha. That is, they take long and short positions in different companies depending on their view of those companies expected future performance relative to the overall market. However, they do not take any overall equity market (beta) risk, as they hedge it away by using derivative contracts, or by taking offsetting long and short positions. At the end of 2003, Equity Market Neutral funds accounted for 10% of the total capital of the CSFB/Tremont Hedge Fund Index.
The majority of money invested in hedge funds, however, is not in any of the strategies we have already discussed, but rather in what are broadly called "directional strategies." These funds try to earn high returns by taking large directional bets, in the expectation that overvalued assets they are short will fall in price and undervalued assets they are long will rise in price. Because directional trading typically generates higher profit margins per transaction, these funds generally use less leverage than the arbitrage funds.
"Long/Short Equity" funds are different from market neutral funds in that the long and short positions they take may be of different sizes. Long/Short funds may either invest in a broad range of asset classes, or be more narrowly focused (e.g., a biotechnology hedge fund). At the end of 2003, they accounted for 13% of the CSFB/Tremont Index.
"Global Macro" funds hunt for alpha using a market timing approach. They take long and short positions across a very broad range of asset classes and markets around the world, depending on their view of their respective future returns. These funds may also use substantial amounts of leverage on a tactical basis to increase the potential payoffs from some of their directional bets. Famous hedge funds, such as George Soros’ Quantum Fund or Julian Robertson’s Tiger Fund are in this class. They accounted for 13% of the CSFB/Tremont Index at the end of 2003.
"Managed Futures" funds invest in listed financial and currency futures, and their managers are usually called commodity trading advisors, or CTAs. They often employ momentum strategies. These funds accounted for 10% of the hedge fund assets tracked by CSFB/Tremont at the end of 2003.
"Emerging Markets" funds try to make money through superior market timing and security selection in markets that are often less liquid than those of developed countries. They accounted for 3% of the total hedge fund assets tracked by CSFB/Tremont at the end of 2003.
"Dedicated Short Bias" funds have greater than fifty percent of their assets invested in short equity market positions. Because of the difficulty of making money over the long term taking this approach (given that the economy grows, and markets rise, in far more years than they fall), dedicated short funds accounted for only 2% percent of total hedge fund assets at the end of 2003.
Finally, hedge funds which employed multiple investing strategies accounted for 11% of hedge fund assets at the end of 2003.
Now let’s move on to our analysis of how hedge funds fit into an investor’s portfolio.
In conducting this research, our first problem was the quality of the available data on hedge fund returns. To put it mildly, it is questionable at best. Because this data underlies much of the current enthusiasm for investing in hedge funds, it is critical that people understand its limitations.
To begin with, there are at least ten different indexes that claim to track the performance of the hedge fund universe. However, many of these indexes are constructed using different methodologies (e.g., how they classify different hedge fund strategies, whether they use equal or market capitalization based weighting, and whether they require audited results from the funds they include). Just to make things more interesting, hedge fund managers themselves decide whether or not to report their results to an index provider. For example, a fund with a poor performance record may choose not to report its results. At the other extreme, a fund with an outstanding performance record, which is closed to new investors, also may choose not to report its results. This is called "self-selection bias."
Moreover, reporting funds provide their results to different index providers. As a result, no index comes close to covering the entire hedge fund universe. But the problems don’t end there. When a hedge fund initially decides to report its results to an index provider, it delivers not only its current and future returns, but also a history of its past returns as well. Unfortunately, in the case of an indexed hedge fund product, you can only invest in a fund after it has been added to the index. In other words, what counts from an index investor’s point of view is performance after a fund has been added to an index, not before it. The extent to which fund returns are lower after they join an index than they were before this point is called "backfill bias."
Finally, the treatment of funds that leave an index can also create bias in the reported index returns. If either the returns of these funds are removed from the index database after they stop reporting, or if (in the case of failing funds) their final returns are not obtained, then the reported index returns can be biased upwards. This is known as "survivorship bias."
While a number of authors have examined the potential impact of these different biases, one of the best papers we’ve read on the subject is "A Reality Check on Hedge Fund Returns" by Posthuma and Van der Sluis. They directly examined the backfill bias in the TASS database (which contains over 3,000 hedge funds) over the period 1996 to 2002 (previous studies had only estimated the size of the problem). The authors found that more than half the reported returns in the database were backfilled. They went on to create a proxy for a truly investable index by (a) using only non-backfilled returns, (b) including the returns from funds which left the index; and (c) using two different approaches to estimate the final returns from failing funds (one assumed that investors received all their money back, while they second assumed a 50% loss of capital). When constructing their index, the authors equally weighted each hedge fund’s return. This is the practice used by almost all the major hedge fund index vendors, except CSFB/Tremont, which weights funds’ returns by their assets under management (in line with the way most other asset class indexes are constructed).
Posthuma and Van der Sluis found that due to backfill bias, average annual hedge fund index returns were overstated by 4.35% during the 1996 – 2002 period (10.73% before the bias was removed, versus 6.34% after). By strategy, the impact of backfill bias ranged from a high of 6.34% for Long/Short Equity to 3.13% for Global Macro, 2.60% for Equity Market Neutral, and 2.45% for Event Driven. However, these returns assumed that investors suffered no loss after a fund left the investable index. When the authors assumed that such funds incurred a 50% loss of capital, the overall return on the index declined to 7.43%, and the backfill bias rose to 7.24% -- essentially leaving the overall investable index return equal to zero. Moreover, the authors also found that the net impact of these biases also distorted (unfavorably) various measures of risk. Let’s look at these. Standard deviation (also known as volatility) measures the extent to which returns are disbursed around their average. In general, investors who are risk averse prefer lower levels of standard deviation, and seek to maximize the amount of return per unit of volatility they take on. Posthuma and Van der Sluis found that backfill and survivorship biases artificially lowered their hedge fund index’s reported standard deviation. Skewness measures the extent to which positive or negative returns are more probable. In a positively skewed distribution (which risk-averse investors prefer), positive returns are more probably than negative ones. A normal (bell curve) distribution has skewness equal to zero, because positive and negative returns are equally probable. In this case, the authors found that removing the survivorship and backfill biases made hedge fund returns’ skewness more negative. Finally, kurtosis measures the "peakedness" of the distribution of returns, relative to what would be found in a normal distribution. Positive kurtosis means the distribution has "fatter tails" than a normal distribution. Practically, positive kurtosis means that extreme returns – both positive and negative (skewness tells you which is more likely) – are more likely than they would be if returns were normally distributed. Investors’ kurtosis preference depends on the skewness of the distribution. If it is negative, risk-averse investors dislike positive kurtosis, because it means that big negative returns are more likely than big positive ones. On the other hand, if a distribution is positively skewed, then a risk averse investor may prefer somewhat higher than normal kurtosis, which would raise the probability of realizing big positive returns.
In their study of hedge fund returns, Posthuma and Van der Sluis found that the backfill and survivorship bias tended to depress reported kurtosis. Similar findings on the impact of survivorship and backfill bias on reported average returns, standard deviation, skweness and kurtosis were also reported by Professor Ross Barry of Macquarie University in his paper "Hedge Funds: A Walk Through the Graveyard." Last but not least, we should also mention that Posthuma and Van der Sluis found that there was "no persistence between the returns of the backfilled hedge fund returns and the non-backfilled returns." In other words, what is true of mutual funds also seems to be true in the hedge fund world: you can’t use past performance to pick future winners.
Impressive as it was, Posthuma and Van der Sluis’ study left out another important bias. The returns that hedge fund managers report each month to various index providers are based, in part, on changes in the market value of the assets in which they have invested. However, if those assets are sufficiently illiquid (as would be the case for example, with some distressed debt, exotic derivative instruments, or privately placed equity), it can be very difficult to obtain accurate market prices for them each month. As a result, estimated values are frequently used, in an approach that is not dissimilar to the way residential real estate is often re-valued by appraisers during the long period in between market transactions. In both cases, the appraisal approach leads to a higher degree of correlation between asset prices in succeeding periods than is normally found in liquid markets. Statistically, this is called "autocorrelation." Practically, the autocorrelation bias in some hedge fund returns causes reported standard deviations to be lower than what their "true" value probably is.
Perhaps the greatest limitation of most studies of hedge fund performance is the relatively short periods they cover. More than anything else, this is a function of the length of the available hedge fund index data series, which generally only go back as far as 1994. A number of researchers have tried to overcome this limitation by using regression modeling to artificially create a longer series of hedge fund performance data. In practice, this involves regressing the performance of a hedge fund (or hedge fund index) against the values of a number of other independent variables that have longer data series. If the model produces a reasonably good fit between the actual and predicted hedge fund performance, then the historical values of the independent variables can be used to project back into the past a longer series of estimated hedge fund returns. Of course, developing these regression models is not without its challenges, including which independent variables to use, and the actual form of the model itself (e.g., should it be a simple linear model or a more complex polynomial one?).
One of the most interesting approaches of this type is described in the paper "Risks and Portfolio Decisions Involving Hedge Funds" by Agarwal and Naik. They started with the observation that "a large number of equity oriented hedge fund strategies exhibit payoffs [return distributions] resembling a short position in a put option on the market index, and therefore bear significant left tail risk." As we shall see, this is a very important point to keep in mind.
For those readers who are a little unclear about what it means to be short a put option, it means that you are, in essence, an insurance company. In selling (or, as it is known, "writing") a put option you have promised the other party that, for a specified period of time (say the next 360 days), you stand ready to purchase a specified quantity of an asset (say, 1,000 shares of the exchange traded index fund that tracks the Russell 3000 Index) at a specified "strike price" (say, $65 per share). In exchange for making this promise (or, to be more accurate, taking on this risk), the purchaser pays you a premium. Now let’s think about what happens next. Suppose that over the next year, the R3000 ETF never trades below $65/share, and the holder of the option you have sold therefore chooses not to exercise it. At the end of the year, you add the option premium you earned to your other gains and losses to calculate the total return on your investment portfolio. You notice that said return is higher because of the option premium, so you decide to do the same thing again next year. And again, the ETF never trades below $65/share. So you do it again for a third year, and, again, nothing happens. Now think about how what you have done has affected the three-year portfolio returns you have reported to your key investors (say, your spouse).
First, the option premiums you earned raised your reported average return. Second, because those premiums were constant over three years, they reduced the reported standard deviation of your reported returns. Third, if in any year the option premium turned what otherwise would have been a negative return into a positive one, they made the skewness of your returns look more positive. Finally, because you never had to pay out under the insurance contract (er, the put option), you had no big negative returns – so selling the put option had no affect on the reported kurtosis of your returns.
By now, I rather suspect you’re getting the larger picture: Agarwal and Naik’s finding that "a large number of equity oriented hedge fund strategies exhibit payoffs [return distributions] resembling a short position in a put option on the market index, and therefore bear significant left tail risk" is potentially a ticking time bomb for all those people (think back to whichever self-style "hedge fund guru" you most recently encountered) who believe that, in essence, hedge funds are a free lunch that can magically improve a portfolio’s risk/return tradeoff.
This really isn’t news, however. The hints have been there for quite a while. It wasn’t just Long Term Capital Management that was brought down by the Russian Debt Crisis in 1998; a lot of other hedge funds went out of business then too. And, as we will demonstrate below, despite their acknowledged shortcomings, the existing hedge fund return indexes still give an indication that this is a very different animal from other asset classes.
But let’s for the moment go back to our investor who had discovered the joys of selling put options. Say he does it again in year four, but this time the market tanks, and the R3000 ETF drops to $25/share. At this price, the holder of the put option exercises it, which forces our intrepid investor to pay $65,000 for 1,000 ETFs that are only worth $25,000. Given his track record over the previous three years, that big a loss seems like it will take quite a bit of explaining to his investors (ouch!). On the other hand, a statistical analysis of his reported returns will finally reflect the large risk (in the form of greatly increased standard deviation and kurtosis, as well as negative skewness) that was lurking all along in his investment strategy. The moral of the story is simple. Investing is like the rest of life: if something seems too good to be true, it probably is too good to be true.
But back to Agarwal and Naik’s paper. When they used their regression model (which included option payoffs as some of its factors) to extend their estimated hedge fund index return data series back to 1927, they made another disturbing discovery. Hedge funds’ recent performance seems to be significantly better than their long-term performance. More specifically, they found that their projected average historical hedge fund returns were significantly lower, and their standard deviation higher than those estimated using just their more recent performance. Seems like another caution flag to us.
With these shortcomings in mind, we set out to explore the potential impact of using hedge funds in model portfolios with different compound annual real return objectives.
Our first step was to choose an index to use. We decided on the CSFB/Tremont Index, because it is (a) has a history dating back to 1994; (b) is relatively free of survivorship bias, and (c) is the only major index that is asset weighted. The latter factor makes it more comparable with the other returns series we use in our analysis.
Our next step was to develop inputs to use in our simulation optimization model. We began by looking at both the overall index, and a number of strategy sub-indexes, including Equity Market Neutral, Global Macro, and Event Driven. We chose the first two because our past analyses had found them to be potentially valuable additions to a portfolio, in terms of their impact not only on returns, but also on standard deviation, correlation, skewness and kurtosis (for a paper which also reaches this conclusion, see "Fund of Funds Portfolio Selection" by Davies, Kat, and Lu). We chose Event Driven because it provides a good contrast with the first two strategies.
The following table shows summary real return data covering the 1994 –2003 period.
| A$ | Aggregate Index | Equity Mkt. Neutral | Global Macro | Event Driven |
| Average Real Return | 9.58% | 4.78% | 14.46% | 8.38% |
| Std. Deviation | 12.38% | 10.04% | 15.86% | 9.76% |
| Average/Std. Deviation | 0.77 | 0.48 | 0.91 | 0.86 |
| Skewness | (0.06) | 0.50 | 0.11 | (0.02) |
| Kurtosis | (-.27) | (0.82) | (0.06%) | (0.86) |
| C$ | Aggregate Index | Equity Mkt. Neutral | Global Macro | Event Driven |
| Average Real Return | 9.36% | 8.57% | 13.10% | 9.46% |
| Std. Deviation | 8.38% | 6.02% | 12.51% | 6.53% |
| Average/Std. Deviation | 1.12 | 1.42 | 1.05 | 1.45 |
| Skewness | 0.30 | (0.21) | 0.01 | (1.18) |
| Kurtosis | (0.07) | (0.08) | 0.98 | 3.49% |
| Euro | Aggregate Index | Equity Mkt. Neutral | Global Macro | Event Driven |
| Average Real Return | 8.18% | 7.40% | 11.89% | 8.28% |
| Std. Deviation | 13.71% | 9.94% | 16.79% | 11.67% |
| Average/Std. Deviation | 0.60 | 0.74 | 0.71 | 0.71 |
| Skewness | 0.33 | 0.16 | 0.40 | (0.73) |
| Kurtosis | 0.33 | (0.29) | 1.08 | 2.00 |
| Yen | Aggregate Index | Equity Mkt. Neutral | Global Macro | Event Driven |
| Average Real Return | 11.37% | 10.56% | 15.17% | 11.46% |
| Std. Deviation | 16.87% | 12.25% | 20.74% | 14.34% |
| Average/Std. Deviation | 0.67 | 0.86 | 0.73 | 0.80 |
| Skewness | (0.46) | (0.34) | (0.53) | (0.89) |
| Kurtosis | 2.83 | 1.61 | 4.24 | 2.57 |
| UK Pound Sterling | Aggregate Index | Equity Mkt. Neutral | Global Macro | Event Driven |
| Average Real Return | 7.19% | 6.41% | 10.86% | 7.28% |
| Std. Deviation | 11.81% | 8.34% | 14.94% | 10.16% |
| Average/Std. Deviation | 0.61 | 0.77 | 0.73 | 0.72 |
| Skewness | 0.46 | 0.47 | 0.56 | (1.23) |
| Kurtosis | 1.74 | 1.14 | 1.12 | 6.31 |
| US$ | Aggregate Index | Equity Mkt. Neutral | Global Macro | Event Driven |
| Average Real Return | 8.86% | 8.07% | 12.58% | 8.95% |
| Std. Deviation | 8.52% | 3.08% | 12.18% | 6.04% |
| Average/Std. Deviation | 1.04 | 2.62 | 1.03 | 1.48 |
| Skewness | 0.06 | 0.10 | (0.60) | (3.36) |
| Kurtosis | 1.61 | 0.41 | 1.89 | 22.18 |
These tables are interesting for a number of reasons. First, most studies done to date have used hedge fund returns in U.S. dollars (the currency in which over 80% of hedge funds report their returns). As you can see, the U.S. dollar table confirms the findings from many of these studies that, at the level of the aggregate index, hedge funds’ impressive ratio of average return/standard deviation also requires the acceptance of quite a high level of kurtosis (i.e., a greater probability of experiencing extreme returns). The U.S. dollar table also show the relative attractiveness of the Equity Market Neutral strategy, and the unattractive skewness and kurtosis characteristics of the Event Driven strategy.
What is equally interesting, however, is the way exchange rate changes affect the perception of these strategies’ results when they are expressed in different currencies. In general, the relationship between the Equity Market Neutral and Event Driven strategies remains the same. An exception to this, however, is the table showing real hedge fund returns expressed in Australian dollars, where the Global Macro approach comes out best.
Despite the attractiveness of Equity Market Neutral relative to the aggregate hedge fund index, we chose to use the latter in our asset allocation analysis because the only hedge fund index products available thus far are based on this measure. While we performed a sensitivity check to get a rough idea of the impact of moving away from the aggregate index, we did not, in this analysis, include Equity Market Neutral and Global Macro as separate asset classes.
In our analysis, we first used our simulation optimization model to develop optimal target return portfolios using inputs based on historical data. For hedge funds, we used returns from 1994 to 2003; for the other asset classes we used returns from 1973 to 2002. The correlation matrix we used, however, covers only the 1994 to 2003 period. Our second step was to repeat the portfolio construction process using our estimated future returns for each asset class as inputs. We then combined the historical and forward looking portfolios, weighting the former 67% and the latter 33%.
For hedge funds, our examination of comparable historical data showed a rather close relationship between the return on the aggregate hedge fund index and the return on the Wilshire 5000 U.S. equity index, albeit with a substantially lower standard deviation. The following table shows this data:
| Currency |
Average Hedge Fund Index Real Return (1994-2003)
|
Average Wilshire 5000 Real Return
(1994 - 2003) |
| A$ | 9.58% | 8.43% |
| C$ | 9.36% | 9.57% |
| Euro | 8.18% | 8.39% |
| Yen | 11.37% | 11.58% |
| UK Pound Sterling | 7.19% | 7.39% |
| US $ | 8.86% | 9.06% |
Given this historical data, we set our future hedge fund real return assumptions equal to the expected local currency real return on the Wilshire 5000 Index.
For the sake of brevity, the rest of this section will focus on our analysis of the impact of including hedge fund index products in U.S. Dollar portfolios whose objective is a minimum compound annual real rate of return. The inputs for this analysis are summarized in the following table:
| US$ |
Hist Ret
|
Fut Ret
|
Std Dev
|
RB
|
DB
|
FB
|
CP
|
C
|
DE
|
RE
|
EE
|
HF
|
| Real Bonds | 2.30% | 2.50% | 2.50% | 1.00 | 0.43 | 0.35 | -0.06 | 0.26 | -0.16 | -0.13 | -0.14 | -0.04 |
| Dom Bonds | 3.80% | 4.00% | 5.40% | 1.00 | 0.30 | 0.01 | 0.04 | -0.06 | -0.11 | -0.16 | 0.16 | |
| For Bonds | 9.50% | 3.61% | 11.20% | 1.00 | 0.02 | 0.17 | -0.02 | 0.23 | -0.05 | -0.18 | ||
| Comm Prop | 7.90% | 3.70% | 9.80% | 1.00 | 0.10 | 0.32 | 0.29 | 0.33 | 0.23 | |||
| Commodities | 8.10% | 8.10% | 18.30% | 1.00 | 0.07 | 0.14 | 0.12 | 0.17 | ||||
| Dom Equity | 7.30% | 6.20% | 16.30% | 1.00 | 0.79 | 0.73 | 0.54 | |||||
| For Equity | 7.00% | 5.57% | 17.20% | 1.00 | 0.71 | 0.42 | ||||||
| EM Equity | 9.60% | 7.50% | 24.00% | 1.00 | 0.52 | |||||||
| Hedge Funds | 8.86% | 6.20% | 8.52% | 1.00 |
Given the relatively questionable quality of the hedge funds return data we used, we capped the maximum allowable allocation to hedge funds at twenty percent of the target return portfolios.
The following tables show the impact of including a hedge fund index as a possible investment alternative in portfolios with target compound annual real returns (in US Dollars) of 7%, 5%, and 3%.
| US$ | 7% Historical | 7% Future |
7% Weighted |
| Real Return Bonds | 5% | 0% | 3% |
| Domestic Bonds | 0% | 0% | 0% |
| Foreign Bonds | 40% | 0% | 27% |
| Commercial Property | 20% | 0% | 13% |
| Commodities | 5% | 20% | 10% |
| Domestic Equity | 5% | 50% | 20% |
| Foreign Equity | 0% | 0% | 0%q |
| Emerging Equity | 10% | 15% | 12% |
| Hedge Funds | 15% | 15% | 15% |
| Total | 100% | 100% | 100% |
| Expected Annual Return | 8.3% | 6.8% | N/A |
| Expected Std. Deviation | 6.9% | 12.9% | N/A |
| Probability of Achieving Target | 76.0% | 38.0% | N/A |
| US$ | 5% Historical | 5% Future |
5% Weighted |
| Real Return Bonds | 5% | 5% | 5% |
| Domestic Bonds | 30% | 0% | 20% |
| Foreign Bonds | 30% | 5% | 22% |
| Commercial Property | 10% | 0% | 7% |
| Commodities | 5% | 20% | 10% |
| Domestic Equity | 5% | 50% | 20% |
| Foreign Equity | 0% | 0% | 0% |
| Emerging Equity | 5% | 10% | 7% |
| Hedge Funds | 10% | 10% | 10% |
| Total | 100% | 100% | 100% |
| Expected Annual Return | 6.9% | 6.4% | N/A |
| Expected Std. Deviation | 5.3% | 11.7% | N/A |
| Probability of Achieving Target | 93.0% | 62.0% | N/A |
| US$ | 3% Historical | 3% Future |
3% Weighted |
| Real Return Bonds | 55% | 15% | 42% |
| Domestic Bonds | 15% | 20% | 17% |
| Foreign Bonds | 10% | 15% | 12% |
| Commercial Property | 10% | 10% | 10% |
| Commodities | 5% | 10% | 7% |
| Domestic Equity | 5% | 10% | 7% |
| Foreign Equity | 0% | 5% | 2% |
| Emerging Equity | 0% | 5% | 2% |
| Hedge Funds | 0% | 10% | 3% |
| Total | 100% | 100% | 100% |
| Expected Annual Return | 4.3% | 4.8% | N/A |
| Expected Std. Deviation | 3.2% | 5.7% | N/A |
| Probability of Achieving Target | 97.0% | 91.0% | N/A |
Finally, as another way of assessing the potential impact of adding hedge funds to these target return portfolios, we backtested them using data from 1994 – 2003. When we were doing this, we also ran a simple sensitivity analysis to test the potential impact of using index funds tied to something other than the aggregate hedge fund index. Specifically, while we kept our hedge fund allocations unchanged, we substituted a simple mix of 50% Equity Market Neutral Index return and 50% Global Macro Index return for the Aggregate Hedge Fund Index Return. In the following table, these results are labeled "w/EG". The results were encouraging, and indicate that a superior asset allocation solution probably could be achieved by using a mix of hedge fund style indexes, rather than the aggregate index. We will undertake a more comprehensive analysis when and if hedge fund style-based index products are introduced.
1994 –2003 Backtesting Analysis
| US$ | 3% Real Tgt | 3% Tgt w/HF | 3% Tgt w/EG |
| Average Real Return | 4.96% | 5.15% | 5.20% |
| Std. Deviation | 3.96% | 3.96% | 3.92% |
| Average/Std. Deviation | 1.25 | 1.30 | 1.33 |
| Skewness | 0.22 | 0.26 | 0.29 |
| Kurtosis | 1.29 | 1.48 | 1.50 |
| CAGR 1994-2003* | 4.71% | 4.92% | 4.92% |
| US$ | 5% Real Tgt | 5% Tgt w/HF | 5% Tgt w/EG |
| Average Real Return | 6.04% | 6.16% | 6.30% |
| Std. Deviation | 8.09% | 6.25% | 6.01% |
| Average/Std. Deviation | 0.75 | .99 | 1.05 |
| Skewness | -0.65 | -0.41 | -0.35 |
| Kurtosis | 1.11 | 0.76 | 0.63 |
| CAGR 1994-2003* | 5.45% | 5.75% | 5.92% |
| US$ | 7% Real Tgt | 7% Tgt w/HF | 7% Tgt w/EG |
| Average Real Return | 5.81% | 6.53% | 6.74% |
| Std. Deviation | 9.03% | 7.96% | 7.59% |
| Average/Std. Deviation | 0.64 | 0.82 | 0.89 |
| Skewness | -0.67 | -0.73 | -0.66 |
| Kurtosis | 1.47 | 1.69 | 1.49 |
| CAGR 1994-2003* | 5.13% | 5.98% | 6.24% |
Looking at these results, (as well as results denominated in five other currencies, which are not shown here) made a number of points clear. First, the potential impact of hedge funds seems to depend on a portfolio’s target real return. For the 3% target real return portfolios, the impact was minimal, and the impact of improved return/standard deviation was usually offset by higher kurtosis (the impact on skewness was usually minimal). The inclusion of hedge funds seemed to provide the greatest benefits to the 5% target real return portfolios, and more often than not, this did not require taking on more skewness and/or kurtosis-related risk. On the other hand, the benefits of hedge funds to the 7% target real return portfolios, while usually significant in the area of return/standard deviation, typically required a worsening of those portfolios’ skewness and kurtosis. Finally, as previously noted, even a simple replacement of the aggregate hedge fund index returns with a 50/50 mix of Equity Market Neutral and Global Macro style index returns in many cases significantly reduced hedge funds’ negative impacts while preserving many of their portfolio return and standard deviation benefits.
However, we reiterate that when it comes to investing in this area, a healthy degree of skepticism and caution are still warranted. Not only is the quality of the underlying data suspect, but a larger question remains unanswered. Can it really be that almost 7,000 hedge fund managers possess either the superior information or the superior models needed to consistently deliver superior returns over a long period of time, even as more and more money is pursuing the same general investment strategy? Wouldn’t this also imply the equally sudden development of an opposite class of traders who are somehow doomed to be consistent losers?
| Market Condition |
Normal
|
Inflation
|
Deflation
|
| Reasons to Invest in Absolute Value (Hedge Fund) Strategies |
Equity Market Neutral seems likely to boost returns while lowering portfolio risk Global Macro can provide additional returns through tactical asset allocation for a small portion of your portfolio, minimal additional risk |
Since all equity is a claim on residual cash flow, and since companies can eventually adjust their prices when faced with inflation, equity returns should suffer less than fixed rate bond returns
Unsettled environment should favor Global Macro strategy |
Unsettled environment should favor global macro strategy. |
| Reasons Not to Invest in Absolute Value (Hedge Fund) Strategies |
Liquidity is low, so not appropriate for investors who make regular withdrawls from portfolio. With large amount of new money flowing into hedge funds, historical risk/return relationships will probably worsen in the future |
Other asset classes (e.g., real return bonds, commodities, and residential property) provide better protection against inflation
Hedge funds haven't really been tested under these conditions |
Other asset classes – such as investment grade bonds – provide better protection against deflation than equity (including equity market neutral funds). |