Hedge fund performance during the Internet bubble
Bachelor thesis financeColby Harmon 6325661 /10070168 Thesis supervisor: V. Malinova
Table of content
1. Introduction p. 3
2. Literature reviews of studies on mutual funds and hedge funds p. 5
2.1 Evolution of performance measures p. 5
2.2 Studies on performance of mutual and hedge funds p. 6
3. Deficiencies in peer group averages p. 9
3.1 Data bias when measuring the performance of hedge funds p. 9
3.2 Short history of hedge fund data p. 10
3.3 Choice of weight index p. 10
4. Hedge fund strategies p. 12
4.1 Equity hedge strategies p. 12
4.1.1 Market neutral strategy p. 12
4.2 Relative value strategies p. 12
4.2.1 Fixed income arbitrage p. 12
4.2.2 Convertible arbitrage p. 13
4.3 Event driven strategies p. 13
4.3.1 Distressed securities p. 13 4.3.2 Merger arbitrage p. 14 4.4 Opportunistic strategies p. 14 4.4.1 Global macro p. 14 4.5 Managed futures p. 14 4.5.1 Trend followers p. 15
4.6 Recent performance of different strategies p. 15
5. Data description p. 16
6. Methodology p. 18
6.1 Seven factor model description p. 18
6.2 Hypothesis p. 20 7. Results p. 21 7.1 Results period 1997-2000 p. 21 7.2 Results period 2000-2003 p. 22 8. Discussion of results p. 23 9. Conclusion p. 24
1. Introduction
A day without Internet today would be cruel and unthinkable. But 20 years ago this was nothing out of the ordinary. People back then got their news from the newspaper as well as from news channels. Internet started to have an impact on the world in the mid 90’s, with the rise of electronic mail, instant messaging and the World Wide Web (Wikipedia, history of the Internet). This development was also visible in the financial markets. In the late years of the 1990’s there was a large growth of the commercial internet sector. Due to this growth the ‘dot-com’ industries experienced huge rises in their stock prices. In the period from 1998 through February 2000 the internet industry earned over 1000 percent returns on its public equity. The internet sector covered roughly 6 percent of the market capitalization of public companies in the U.S. as well as 20 percent of all publicly traded equity volume (Ofek and Richardson, 2003).
A bubble in the financial market is characterized by a self perpetuating increase in the stock prices of a particular industry. This happens when speculators notice the fast increase in value and decide to buy expecting further rises (Wikipedia). They buy the share because of the quick rise in price and not because the share is undervalued. This generally causes companies to be overvalued. When a bubble is overinflated and bursts the prices of the companies’ shares drop fiercely. This causes many companies to go bankrupt or out of business.
Finally the internet bubble burst happened on March 10, 2000. On this date the NASDAQ composite peaked at 5,048.7 (day-end) (Nasdaq). By March 20, 2000, NASDAQ had lost over 10% of its peak. Nearing 2001 the bubble was collapsing at full speed. Many internet companies ran out of capital and taken over or liquidated. The stock market crash due to the internet bubble bursting added up to a loss of 5 trillion dollars in the market value of companies from March 2000 until November of 2002. The stock market crashed affected all kinds of investment vehicles, such as mutual funds and hedge funds. For hedge funds this was the third hit in a time period of three years. First the Asian currency crisis of 1997, then the Russian debt default of 1998 and now the dotcom crash of 2000.
Hedge funds are characterized as private investment vehicles for wealthy people and institutional investors. They usually take on the form of limited partnerships, where the partners and managers invest a large part of their own wealth to align the managers’ incentives with the funds performance (Fung and Hsieh, 1999).
Caldwell (1995) says that the first ever hedge fund was introduced by Albert W. Jones in 1949 using a strategy based on long and short positions in equity as well as leverage. He
used leverage to buy shares and went short on the other side to avoid market risk. Jones thus referred to his fund being ‘hedged’.
Hedge funds managers enjoy a huge flexibility when it comes to making investment decisions. This is all thanks to the limited regulatory oversight on hedge funds. Hedge funds, unlike mutual funds, are not required to register with the SEC and disclose their holdings. The regulation on limited partnerships makes this possible (Liang, 1999).
The fee structure of hedge funds is specially designed to motivate managers. The fee is based on factors, such as asset size. There is also a separate incentive fee to align the
managers’ performance with that of the fund. In general, the incentive fees are only paid after a ‘hurdle rate’ has been met. Most of the hedge funds also make use of a ‘high watermark’ provision. This entails that managers have to make up for previous losses in order to get paid the incentive fee. All these previously named features make sure that manager’s act in the invertors’ best interest (Liang, 1999)
Mutual funds generally speaking use relative benchmarks, such as S&P 500 for equity funds and the Lehman Brothers Aggregate index for bond funds, for their returns. This means that the funds returns are compared to a benchmark. The relative return is the difference between the absolute return of an asset and the return of the benchmark. As compared to hedge funds, which use absolute returns to measure their performance and often take highly speculative positions (HedgeCo.com)
As a result of benefits hedge funds have gained a lot of popularity. Since the
conception of the first hedge fund in 1949, there has been a huge growth. In the 1980’s there were around 100 funds. In the early years of 1990 the number of funds grew greatly and now there are over 10.000 hedge funds worldwide available to investors (Liang, 1999).
This paper expands existing performance of hedge fund literature by taking a closer look at particular hedge fund strategies in times of financial turmoil. By using asset-based style factors in a model of hedge fund risk (Fung and Hsieh, 2004), I compare different hedge fund strategies to the market index in the same period. I find that the hedge fund returns consistently underperform the market index in the period leading up to the bursting of the internet bubble. Also, I show that in times of financial turmoil the market return is more volatile than the returns of the three hedge funds.
In chapter two of this paper there will be a literature review of studies on mutual funds and hedge funds. In chapter three will be about deficiencies in peer group averages followed by different types of hedge fund strategies in chapter four. In chapter five I will discuss how I obtained my data for this research. The methodology section in chapter six will follow this. In
chapter seven I will show my results of my tests and this will be followed by a discussion about the results in chapter eight. Finally in chapter nine will be an overall conclusion.
2. Literature review of studies of mutual funds and hedge funds
In this chapter there will be a discussion on previous literature of performance studies and performance measures of hedge and mutual funds. First I will give a short historical insight in the evolution of the performance measures, followed by a discussion on previous literature in which they use these performance measures.
2.1 Evolution in performance measures
About thirty years ago a commonly used performance measure based on CAPM was Jensen’s alpha (1968), just as the Sharpe’s (1966) reward-to-variability ratio. These were often used in the performance evaluation. Due to the more recent literature on cross-sectional variations in stock return there has been increasingly more interest in multi-factor models. Studies by Fama and French (1998) and Chan et al. (1996) show that the cross sectional variations of average returns on U.S. stock show almost no relation to Sharpe’s (1964) beta or Litners’ capital asset pricing model. Rather they identify other factors that have reliable powers to explain the cross section of average returns (Capocci and Hübner, 2004). These factors include the company size, leverage, price/earnings, book-to-market, dividend yield and the momentum effect (Elton et al., 1996). There are also other multi-factor models introduced and they include the three-factor model by Fama and French (1993), the Carhart (1997) four-factor model, and the international Fama and French model (1998).
However, studies in recent years have allowed for some doubt on the usefulness of the new models. The Fama-French three-factor model gets better results than the classical CAPM by adding variables as company size and book-to-market equity to the equation. But
according to Kothari and Warner (2001) it also detects significantly abnormal results (like timing) when none really exist. Additionally, the development of Carharts (1997) four-factor model proves to be better than the traditional CAPM and Fama-French three-factor model.
Carhart (1997) constructs a 4 factor model based on fama frenches 3 factor model plus an additional factor that captures Jagadeesh and Titman’s (1993) one year momentum
anomaly. This model is like a model of market equilibrium but then looks at four risk factors. By adding these variables they have a higher reliable power to explain the cross-section of average returns (Capocci and Hübner, 2004).
Also in literature on hedge funds have there been different models used to evaluate performance. Fung and Hsieh (1997) extend the asset class factor model by Sharpe (1992) and also find five dominant investment styles in hedge funds. Schneeweis and Spurgin (1998) follow in Fung and Hsieh’s footsteps by also using a style-based analysis on a multifactor approach. On the other side Ackermann et al. (1999) uses a single factor model that focuses only on risk. Extensions on the Fung and Hsieh (1997) model used by Liang (1999) are regressions that are based on funds’ characteristics that make use of the Sharpe ratio. A general asset-class factor consisting of excess returns on passive option-based strategies is suggested by Agarwal and Naik (2002), as well as on buy-and-hold strategies to benchmark the performance of hedge funds.
All the above suggests that it is necessary to use multi-factor models, rather than the CAPM model, to do performance studies. The only problem is that there is no unanimously accepted model. In this case it is advisable to use a number of models to compare results.
2.2 Studies on performance of mutual and hedge funds
Recently there has been a growing interest in hedge funds. This interest comes from their increasing prominence in the financial markets (Liang, 1999). But despite this growing interest there have been only a few performance studies that compare hedge funds to, for example, the market index or mutual funds. The main reason why this is the case is due to the private characteristics of hedge funds and there are difficulties to gain access to the funds data. It is interesting to first look at some studies that compare the two funds in order to find performance characteristics of hedge funds.
Generally speaking, performance studies comparing mutual funds to hedge funds can be divided into two categories, either they conclude or deny that hedge funds have
significantly higher realized returns than those that use a passive strategy (Capocci and Hübner,2004). In these studies they find that with this outperformance there is also an
increase in the volatility for hedge funds compared to mutual funds (Ackermann et al., 1999). But just as Liang (1999) they also find that although the hedge funds outperform mutual funds, that they perform worse than the market equity index.
Hedge fund managers often transact in the same markets as traditional fund managers. However, there is evidence that hedge fund returns have different characteristics than those of traditional fund managers. Fung and Hsieh (1997) show that hedge fund returns have a much lower correlation with standard asset returns than mutual funds do. This can be interpreted as that hedge fund managers are more skilled than traditional fund manager. But this is not
always the case. For there is evidence that hedge funds perform poorly when asset markets perform poorly. A different way to interpret this is that hedge funds are exposed to risk, just as the mutual funds are, but that the risk is different to the risk of mutual funds. Fung and Hsieh (2004) use asset based style factors to create hedge fund benchmarks that seize the common risk factors in hedge funds.
Fung and Hsieh (2001) show that hedge fund strategies often generate option-like returns and that linear-factor models have difficulty explaining them. Here they model these hedge fund returns by focusing on the ‘trend-following’ strategy. Trend followers bet on big moves. Just like option buyers, they profit when the market is volatile. By using look-back1 straddles they can explain trend-following funds’ returns better than standard asset indices.
Furthermore, Agarwal and Naik (2004) find similar results. They look at the systematic risk exposure of hedge funds using a buy-and-hold strategy and option-based strategies. Here they also find that the payoffs of the equity oriented hedge funds strategies exhibit non-linear option like behavior. Besides the non-linear exposure to the equity market risk, they find that hedge funds exhibit significant risk exposures to the Fama and French (1993) size and value factors, as well as the momentum factor by Carhart (1997).
Just as Fung and Hsieh (2004), Mitchel and Pulvino tried to identify common return components by using observable market risk factors. By looking at merger arbitrage funds Mitchel and Pulvino (2001) created an asset based style factor for these funds. Merger arbitrage, also known as risk arbitrage, is a strategy that buys the stock of the target firm and goes short on the stock of the acquirer. This strategy presumes that the announced merger transaction is completed. By simulating the return of a rule based merger arbitrage strategy, they show that this asset based style factor for merger arbitrage has the return characteristics close to those of merger arbitrage funds. Using all the stock and cash merger transactions from 1968 to 1998 they show that that the returns of merger arbitrage funds had low correlations with the S&P 500 index returns. Except for when the S&P 500 had a large decline. This decline went hand in hand with the worst performance of merger arbitrage funds. They concluded that the systematic risk in merger arbitrage is that many mergers will be canceled at the same time as the market sharply declines. Thus creating a systematic loss for merger arbitrage funds.
1 “A look-‐back straddle consists of a pair of look-‐back call and put options. A look-‐back option is a call (put) option giving the holder the retroactive right to buy (sell) the underlying asset at its minimum (maximum) within the look-‐back period.” (Fund and Hsieh, 2004)
In an analysis of fixed income funds Fung and Hsieh (2002) look at a different source of common risk factors and find that these funds are usually exposed to yield spreads. Fixed income funds usually buy bonds with low credit ratings or less liquidity. After this they hedge the interest rate risk by going short on U.S. Treasury bonds. Which in comparison have the highest credit rating and are more liquid. The yield spread is the difference between the yields of the two bonds.
Often studies apply conventional models for constructing asset class indexes, which assume that the underlying assets are reasonably homogeneous and the dominant strategy is to buy the asset and hold it. However, hedge funds diversify the makeup and performance of their underlying assets. They have dynamic investment styles and may take highly levered bets. Fung and Hsieh (2004) introduce a different approach to benchmark hedge funds returns by using asset-based style factors in a model of hedge fund risk. They propose a model that is similar to models based on arbitrage pricing theory with dynamic risk factor coefficients. With seven asset-based style factors they explain a large amount of the monthly return variations.
A different approach in analyzing hedge funds was taken by Capocci and Hübner (2004). They analyze hedge fund performance using a number of asset pricing models; including the Carhart (1997) model combined with the Fama and French (1998) and also the Agarwal and Naik (2002) models. Also Capocci and Hübner add a new factor that takes into account that some hedge funds invest in emerging bond markets. By doing so they show evidence of limited persistence for the middle decile funds, but not for the extreme performers.
Another aspect in the hedge fund world is the lock-up period2 and there have been a number of studies on this subject. For example, Field and Hanka (2001), Ofek and Richardson (2000) both report similar price and volume-effects around the end of a lock-up period around different time periods. Unfortunately, these papers do not give an explanation of these effects. In order to explain the rise and fall of stock prices in the Internet sector Ofek and Richardson (2003) use the effects of short sales restrictions and heterogeneity. They estimated that at the peak, the whole internet sector, hundreds of stocks, was priced as if the average future earnings growth rate all these firms would exceed the growth rates experienced by a few of the fastest growing individual firms in the past. Also the required rate of return would be 0%
2 “A lock-‐up period is a window of time in which investors of a hedge fund are not allowed to redeem or sell shares. The lock-‐up period helps portfolio managers avoid liquidity problems while capital is put to work in sometimes illiquid investments.” (investopedia)
for the next decade. These extreme levels of valuation would unequivocally lead you to believe there would be another asset price bubble.
A different view comes from Shiller (2000), he says that the stock price increase was induced by irrational euphoria amidst individual investors, fed by the media, who maximized television ratings by supplying pseudo news. There will always be rational and irrational market participants. However, there are two opposite views on whether the rational traders correct the price of behavioral traders. The supporters of Friedman (1953) and Fama (1965), who advocate the efficient market hypothesis, argue that rational speculation would not only get rid of riskless arbitrage opportunities, but also the exploitation of other forms of
mispricing that need imperfectly hedged and risky trades. This applies to the technology bubble, because there is not a close substitute that would be able to hedge a short position in the technology sector.
3. Deficiencies in peer group averages
The models mentioned above all use information obtained from commercial data vendors. Information on hedge fund returns, risk and fee structure is often not publicly available. This is because hedge funds are not obliged to disclose this information. There is no centrally organized group who gathers this information, rather there are only a few commercial data vendors who posses the information on hedge funds. Due to the fact that the funds are not obliged to disclose their information, this only happens on a voluntary basis. Thus making the information prone to biases (Liang, 2000).
A method frequently applied to model hedge fund risk is to use a broad-based index of hedge funds (Fung and Hsieh, 2004). For example, hedge fund indexes available from Hedge Fund Research (HFR), Zurich Capital Markets (ZCM) and Credit Suisse. These indexes are constructed by averaging individual hedge funds. By doing so these indexes are vulnerable to biases, as shown by Brown et al. (1999). Below we will discuss several biases.
3.1 Data biases when measuring the performance of hedge funds
Indexes created from averaging individual hedge funds can take over errors that were first in the hedge fund database. Brown et al. (1999) and Fung and Hsieh (2000) both write about the existence of these data biases. I will now introduce several biases.
- Self-selection bias
Hedge funds are not obligated, unlike mutual funds, to publicly disclose their activities. According to Fung and Hsieh (2004), the self-selection bias occurs when the sample of funds in the data is not a representative sample of the universe of hedge funds. Because the hedge funds decision is voluntary only the well performing funds will report and the poorly performing funds won’t.
- Survivorship bias
Due to the short lifetime of many hedge funds there are many new entrants and many departures each year. In the case that we would only examine funds that have survived until now, we would overestimate past returns. This is because many of the worst performing hedge funds would have not survived and would have disappeared. This type of bias is well known in mutual funds (Brown et al., 1992)
- Instant-history bias/ Backfill bias
When a fund enters a database, all or part of the its historical data is recorded ex-post in the database. It is likely that funds only publish their results when their performance is ‘good’. If the performance is ‘bad’, they don’t publish any results. Thus, when the
performance is backfilled, the average return in the database is upwardly biased (Fung and Hsieh, 2004).
3.2 Short history of hedge fund data
There is another drawback to using hedge funds indexes. It doesn’t matter if they are based on individual funds or on funds of hedge funds. The drawback is that reliable data on hedge funds only begins in the 1990’s (Fung and Hsieh, 2004). This period corresponds with one of the greatest bull markets in the U.S. This bullish period is only shows a few periods of market declines. Furthermore, it does not give a long enough history to show how hedge funds perform in different market environments (Fung and Hsieh, 2004)
4.3 Choice of index weights
There are different methods for weighting the shares of each security in an index. Each method has its advantages and its disadvantages. The three basic methods are price weighted, value weighted and equal weighted (Financial Education).
Price weighted indices
Price weighted indices are calculated by adding up the share price of each security, followed by dividing the sum of prices by the total number of securities in the index. One of the main advantages of using this method is that it is easy to calculate. And because of the ease of obtaining historical pricing information, it allows for back testing of the index. Also this method uses a constant number of shares, so it is not necessary to re-weight the index because of daily price fluctuations. One of the disadvantages here is that companies with higher share prices receive higher weights without any real reason. Thus, the performance of such a stock is not representative of the overall group. Also actions such as stock splits require re-weighting of the index, even though they do not alter the value of the securities. (Financial Education)
Value weighted indices
Value weighted indices are based on the market capitalization of each company in the index. This kind of indices tracks the returns of all publically traded shares in the index. When shares are not regularly traded (family ownership), weighting can be based on the float, or traded shares. By using this method there is an automatic adjustment for corporate actions and share price changes. Also, the economic changes in the stock market are reflected more accurately. A disadvantage here is when there are inefficiencies in pricing; this method gives a higher weight to overvalued stocks and a lower weight to undervalued ones (Financial Education).
Equal weighted indices
Equal weighted indices invest the same amount in each stock. This method is quite easy and probably more like the reality of how investors actually weight their holdings. Due to the fact that price fluctuations can change the weight of each security, it is necessary to rebalance more often. This method gives a higher relative weight to small companies than to larger ones (Financial Education)
In my thesis I will use the equally weighted returns of the S&P500. This way of weighting stocks is also used in similar studies when looking at the performance of hedge funds, Ofek and Richardson (2003) and Liang (1999).
4. Hedge fund strategies
The huge number of hedge fund strategies and the wide array of instruments used within hedge funds make it hard to get an idea of the hedge fund industry. The different investments used and the strategies employed lead to a number of different representative strategies. All strategies can be accredited to six classes: Equity hedge strategy, Relative value strategy, Event driven strategy, opportunistic strategy, managed futures strategy and multi-strategies strategy (Hauser, 2005). Due to the fact that these multi-strategies invest in other parts of the market, their performance is not homogeneous. In order to give you an idea of their recent performances, see table 1 at the end of this chapter.
4.1 Equity Hedge Strategies
These strategies’ source of return is close to that of traditional investments. The goal is to profit from stock price increases by means of long positions within the portfolio.
Simultaneously, managers take short positions in an attempt to limit the risk exposure of the long positions. A frequently seen combination is that of long holdings of equity or derivatives with short sales of stocks (Hauser, 2005).
4.1.1 Market neutral strategy
Market neutral funds are characterized as funds that constantly look to avoid major risk factors. The main way to do this is by selling expected overvalued stocks and buying undervalued stocks. When this strategy is used in a certain sector to remove a particular industry risk from holdings, managers employ a ‘pairs trading’ strategy. This entails that the long and short positions originate from the same industry. The key to this strategy lies in the correct valuation of stocks (Hauser, 2005).
4.2 Relative value strategies
This type of hedge funds looks for gains by capitalizing on irregularities in the pricing of stocks, bonds or derivatives. They do this by taking positions on forward interest rates, the spread between different yields and also on price differences between related securities. These funds are commonly known as arbitrage funds (Financial policy forum).
4.2.1 Fixed income arbitrage
The bond arbitrage has similar characteristics to that of share arbitrage. A manager wishes to take advantage of price differences at different places. In fixed arbitrage they only
look at bonds. Compared to shares bonds have two major differences. The first one being that bonds have a predetermined maturity. The other difference is that bonds provide fixed
payments of interest until maturity (Hauser, 2005).
When applied to fixed-income arbitrage strategies the goal is to generate a profit due to the change of arbitrage spreads. The difference between the interest rates is called the spread. Managers generally make use of corporate and government bonds with the same characteristics. Usually, the instruments available are government bonds, corporate bonds, municipal bonds, bond options, bonds futures and other bond derivatives. By using a
combination of these instruments managers try to remove the market risk of the interest rates. A risk factor for fixed income arbitrage funds is that the spread moves in the other direction that anticipated. A different risk factor that is applicable is that the underlying company or country defaults (Hauser, 2005).
4.2.2 Convertible Arbitrage
This is the most complex strategy of all relative value strategies. A convertible bond is a security made up of partly a traditional bond and the other part is a stock. As a result the value of this convertible bond is made up of two parts as well. The first part is the fixed interest rate and the final value of the bond. This is the largest part of the convertible bonds price. The second part is the right of the owner to trade in the bond for a share at the end of the maturity. For this reason, the value of the stock impacts the price of the convertible bond, just as the default risk and the interest level (Hauser, 2005).
4.3 Event Driven Strategies
This type of strategy looks at special events and tries to identify the factors that might affect corporate valuations. They then try to trade in such a way that they can derive value when these events take place. The two main strategies are ‘distressed securities’ and ‘merger arbitrage’ (Hauser, 2005).
4.3.1 Distressed Securities
These are funds that take a position in a companies equity whose securities price is likely to be affected by situations such as: corporate bankruptcies, reorganizations, distressed sales and corporate restructurings. The usual way of acquiring these positions is through bank debt or high yield corporate bonds (Hauser, 2005).
4.3.2 Merger Arbitrage
Merger arbitrage, also known as risk arbitrage, makes use of corporate events that are publicly reported a long time prior to the actual event. They gain from situations like mergers and acquisitions, spin offs, recapitalizations, share buy backs, exchange offers and leveraged buyouts. The most frequent strategy is that of the merger and acquisitions. Funds invest in companies who are a part of the announced mergers and acquisition. The funds then assume that the deal will be completed. They generally profit by taking a long position in the equities of the target company and going short on the equities of the acquiring company. The profit here to be made is when the deal follows through and the acquiring company has to make an offer that is above the targets actual stock price. The offer has to be higher because otherwise the shareholders will not sell their shares. This effect of this is that the price of the targets stock will increase towards the price the acquirer is going to pay. The manager will be able to make arbitrage profits by financing long positions in the stocks of the target by means of selling short the acquirers’ stock (Hauser, 2005).
4.4 Opportunistic Strategies
Some managers have decided to change their approach so they can adapt to dynamic market conditions. By this way they want to take advantage of possible investment
opportunities (Hauser, 2005)
4.4.1 Global Macro
This type of fund relies on macro economic analysis so they can make bets on expected market movements. The profit to be earned here is from changes in the value of an entire asset class. Movements can be a consequence of shifts in the world economy.
The risk associated with this strategy is that the anticipated situation will not happen or that the situation will not have the effect on the position. To take advantage of the anticipated situation, managers take positions in stocks, bonds, currencies and commodities (Hauser, 2005).
4.5 Managed Futures
Professional money managers, known as commodity trading advisors, usually invest on the basis of mathematical models. One reason why this type of fund is interesting for investors is that it provides a diversified exposure to a large variety of markets. Managed
futures investment vehicles consist of: future contracts, forward contracts and options. All these investment vehicles are financial contracts for buying and selling of an index, stock, bond or commodity at some future date. These contracts are often used in the following way. Contracts for purchasing a particular asset pose as a long position. Managers may hedge a portfolio or take advantage of negative price movements by selling contracts for an asset in the future. By working in this fashion an advantage can be taken by trading advisors of any price trend (Hauser, 2005).
4.5.1 Trend Follower
The most used strategy used by commodity trading advisors is the systematic trend following strategy. The goal is to profit from the perpetuation of positions of long term trends in the market place. It doesn’t matter in what direction the price moves, managed future trading advisors are able to gain from price trends. If they expect a rising market they can buy futures positions and if they expect the market to decline they can sell futures positions Fung and Hsieh, 1997).
4.6 Recent performance of different strategies
As mentioned above there are many different kinds of hedge fund strategies and each fund performs different than others. In the table 1, you will see the recent performance of eight different hedge fund styles measured in their year to date returns over the past four years. Noticeable is that the year to date returns are all positive, except for the year 2011. Managers of hedge funds attribute the industry wide loss to things like Europe’s debt crisis, a slower than expected recovery of the U.S. economy and unforeseen events like Japan’s nuclear disaster. This all came together to create a tricky trading environment in which there were large and unpredictable stock price swing (Reuters).
Differences in performance for hedge fund strategies
Table 1 Year to
Date return
Year 2010 2011 2012 2013
Equity Long Short 7.27% -4.58% 6.35% 13.85%
Market Neutral 3.64% 0.16% 1.88% 8.61% Fixed income arbitrage 11.65% 4.48% 9.32% 8.69% Convertible arbitrage 12.15% 0.11% 8.74% 8.07% Event driven 10.01% -3.74% 8.46% 10.81% Distressed securities 14.01% -5.41% 12.22% 16.85% Merger arbitrage 6.05% 3.80% 3.82% 3.91% Global macro 6.74% -3.65% 2.59% 4.81% Average 9.94% -1.104% 6.67% 9.45%
Notes: Data obtained from www.barclayhedge.com. YTD describes the performance from the
beginning of the year, 1 January, until the year-end.
In this paper the focus is on three hedge fund strategies, namely equity long short, market neutral and global macro. Table 1 shows that in recent years the equity long short strategy has earned the highest returns compared to the other two strategies. But on the other side this fund also has the largest volatility of the three strategies in the past four years.
5. Data description
The data on the market equity returns returns came from CRSP. They cover the S&P500 equally weighted total returns index on a monthly basis from January 1997 until January 2004. Thus covering the time span of three years prior to the internet bubble and three years after the bubble.
Taking the Russel 2000 index total monthy return and subtracting the S&P500 total monthly return I calculated the size spread factor. This is the spread between the returns of large capitalization stocks and the returns on small capitalization stocks (Fung and Hsieh, 2004).
For the bond oriented risk factors the data was available at the Board of Governors of the Federal Reserve System. Here I obtained the monthly change in the 10-year treasury constant maturity yield. This is applicable to the bond market factor.
The credit risk factor was also available at the Board of Governors of the Federal Reserve System. To calculate the credit spread I first found the monthly change in Moody’s Baa yield and subtracted the 10-year constant maturity yield.
Fung and Hsieh provided the trend following data themselves. They used this data in their own research in 2001. This is data on the look-back straddles they used to identify evidence of trend following.
Data on the particular hedge fund strategies was obtained from
www.barclayshedge.com. This is data on their performance based on 464 funds for the long-short equity index, 86 funds for the market neutral index and 138 funds for the global macro index. Because disclosing data of hedge funds is on a voluntary basis, the sample is prone to biases. In this thesis my sample is prone to the self-selection bias. Barclays hedge database notes above the table of their returns: “only funds that provide us with net returns are included in the index calculation”.
Table 2 and 3 show the summary statistics of individual hedge fund returns of the three strategies prior to the crash and after.
Table 2 Summary statistics 1997-2000
Strategy Number of
funds
Obs Mean St. Dev. Min Max
Equity Long-Short 464 39 .0237359 .033276 -.0692 .1099 Global Macro 138 39 .0128923 .0240562 -.0281 .0753 Market Neutral 86 39 .0101308 .0114394 -.01 .0307
Notes: In this table you see the average monthly returns of three hedge fund styles from January 1997
to February 2000. Data obtained from www.barclayshedge.com.
Table 3 Summary statistics 2000-2003
Strategy Number of
funds
Obs Mean Std. Dev. Min Max
Equity Long-Short 464 44 .005287 .0162096 -.0287 .0473 Global Macro 138 44 .0087455 .0143275 -.0154 .0556 Market Neutral 86 44 .0052978 .0086224 -.0171 .0297
Notes: In this table you see the average monthly returns of three hedge fund styles from March 2000
6. Methodology
In this section will describe what model I will use to test my hypotheses, and will conclude with a description of my hypotheses.
6.1 Seven-factor model description
The seven-factor asset based style (ABS) model contains three main categories of risk factors, namely the equity oriented risk factors, bond oriented risk factors and trend following risk factors. Each category of the risk factor contains its own sub-factors.
The process of creating the ABS factors happens by first extracting the common sources of risk in hedge fund returns and then linking these common sources of risk to observable market prices (Fung and Hsieh, 1997).
Equity oriented risk factors
The sub risk factors here are the equity market factor and the size spread factor. The equity market factor shows the market risk and is measured by looking at the Standard & Poor’s 500 index for monthly total stock return less the risk free rate of the 10 year treasury bills.
Fung and Hsieh (2004) conclude that long short equity hedge funds have exposure to the stock market as well as to the spread between returns on large capitalization stocks and returns on small capitalization stocks. This is known as the size spread factor. It makes use of the Wilshire Small Cap 1750 less the Wilshire Large Cap 750 monthly returns.
Bond oriented risk factors
Fung and Hsieh (2002) analyzed fixed income hedge funds and found that these funds are usually exposed to yield spreads. This is because fixed income funds generally buy bonds with lower credit ratings and then hedge the interest rate risk by taking a short position in U.S. T bonds, which have the highest credit ratings. And the difference between the two types of bonds is called the yield spread.
They also state that yield spreads have the tendency to move together, even more so in times of market distress. For this reason fixed income funds can be modeled as being exposed to credit risk.
The credit risk is measured by subtracting the 10 year constant maturity T bond off the yield on Moody’s Baa bonds.
Trend following risk factors
In 2001, Fung and Hsieh modeled a common return component from trend following funds by using a portfolio of look back straddles. They argued that trend followers bet on big moves. This is similar to option buyers, they make money when the market is volatile. They showed that these option portfolios have a high correlation with trend following funds as well as having the same kind of return characteristics.
In this model I will take into consideration three factors of trend followers, namely bond, currency and commodity factors. The data on this matter has been provided by Fung and Hsieh (2001).
The regression is as follows:
Rι-Rf= α + β1(Equity M) +β2(Size Spread)+β3(BondM)+β4(Credit spread) +β5(BondTF)+β6(CurrencyTF)+β7(CommodityTF)+ε
Ri-Rf stands for the excess returns of the individual hedge funds I will be looking at. I will regress on the monthly returns of the global-marco strategy, equity long-short strategy and the market neutral strategy. I subtract the risk-free rate from the hedge fund returns, and I do the same for the S&P500 returns. Because I do this, the returns that are left are the returns in excess of the market index. Therefor, the alpha will show the outperformance of the hedge funds relative to the market equity index when positive and the underperformance when negative.
Variables Proxy Measure
Equity Market Factor SPequalwreturn Standard & Poor's 500 index monthly total return
Size Spread Factor Size Spread Russell 2000 index monthly total return -/- Standard & Poors 500 monthly total return
Bond Market Factor Treasury10year Change in the 10-year treasury constant maturity yield
Credit Spread Factor Credit Spread Monthly change in Moody's Baa yield -/- 10-year treasury constant maturity yield Bond Trend Following
Factor
Bondlookbackstraddle Return of a portfolio of lookback straddles on bond futures
Currency Trend Following Factor
Currencylookbackstraddle Return of a portfolio of lookback straddles on currency futures
Commodity Trend Following
This model should have a higher explanatory power over the Carhart (1997) four factor model due to the additional variables. For this reason I will be using this model to empirically test if these three hedge fund strategies outperformed the market equity index before and after the bursting of the Internet bubble.
6.2 Hypothesis
In studies done by Ackermann et al. (1999) and Liang (1999) on the performance of hedge funds, the performance is compared to the performance of mutual funds, or the effect of incentive fees on hedge fund return is looked at. These studies have found that hedge funds consistently outperform mutual funds, but many times underperform compared to the market index. It is interesting how hedge funds perform in times of financial turmoil. For this reason I will be focusing my research on hedge fund performance in times of the Internet bubble compared to the market index. My first hypothesis is:
H1: The market outperforms the three hedge fund strategies in the period 1997-2000. I will be looking at three hedge fund strategies, namely two hedging strategies and one opportunistic strategy. The first two strategies, equity long-short and market neutral, are strategies that look to consistently minimize or remove market risk factors. The opportunistic strategy, global macro, seeks to capitalize by anticipating on expected market movements. These movements can be a consequence of the world economy (Hauser, 2005).
The time period of January 1997 – February 2000 is the period leading up to the bursting of the Internet bubble on 10 March 2000. The years 1997 and 1998 have been
turbulent for the financial market due to the Asian currency crisis (1997) and the Russian debt default (1998) (Liang, 1999). Because I think this was a tricky trading environment, I expect that the market will outperform the hedge fund strategies.
Other studies, Mitchell and Pulvino (2001) and Fung and Hsieh (1997), have found that hedge fund strategy returns have a low correlation with the market returns. Mitchell and Pulvino (2001) primarily looked at merger arbitrage funds and found that in times of poor performance of the market that the merger arbitrage funds did poorly as well. This was attributable to the fact that when the market doesn’t perform well, less mergers and
acquisitions, spin offs and recapitalizations occur, which could explain the poor performance. Brunnermeier and Nagel (2004) show that hedge funds were aware of the internet bubble and
argue that the hedge funds only captured the upside and avoided much of the downturn. Therefor, my second hypothesis is:
H2: In the period of 2000-2003 the three hedge fund strategies outperform the market. In contrast to Mitchell and Pulvino (2001), I will not test an event-driven strategy, but will look at how two hedging strategies and one opportunistic strategy performed in the period after the crash. I expect that the market neutral and equity long short hedge fund strategies will outperform the market, due to their dynamic hedging qualities. And also expect the global macro strategy to perform better, because its ability to bet on expected market movements. As previously mentioned, hedge funds have low correlations with the market index, so in contrast to Mitchel and Pulvino (2001) I don’t think the strategies will be influenced as much by the market decline.
7. Results
In this section I will discuss the results I obtained using the previously explained model. First the results of the period 1997-2000 will be discussed, followed by the results of the period 2000-1997. In the appendix you can find the results of the regressions.
7.1 Results period 1997-2000
In all three regressions the alphas are negative. This means that the three strategies performed worse on the return basis than the market equity index. The alphas of the global macro and market neutral were significant at the 5% level, while the alpha of equity long short was significant at the 1% level3. The explanation can partially be found in that the period from 1997-2000 was quite a volatile period for hedge funds. For example, in 1998 a lot of hedge funds ran into trouble because of the Russian debt crisis, which rapidly followed the Asian financial crisis of 1997(Liang, 2000). Also this period is the represents the end of one of the longest bull markets in U.S. stock history (Liang, 2000).
Although the S&P performed better when we look at the returns during the period of 1997-2000, its returns were also more volatile than any of the three hedge fund strategies. The market equity index shows a volatility of 5.004%, while the highest volatility measured is that of the equity long short strategy with 3.320%. The strategy with the lowest volatility is the
market neutral strategy with 1.154%. Effectively this means it is riskier to invest in the market than in one of the hedge fund strategies.
It is less risky to invest in hedge funds because they can effectively reduce their risk by doing dynamic hedging and diversifying their portfolios over different financial instruments.
Table 4 Volatility
1997 - 2000
2000 - 2003 Variable St. Dev. St. Dev. S&P 500 excess .0500441 .0592835 Long Short excess .0332079 .0188826 MarketNeutral excess .0115466 .0099462 GlobalMacro excess .0244937 .0165327
Notes: Volatility of three hedge fund strategies and S&P 500 index excess returns of the period
1997-2000 and 1997-2000-2003. Factor definitions can be found in table 5.
7.2 Results period 2000-2003
Observable is that in the period after the bursting of the bubble the alphas of global macro and market neutral strategy are negative and the alpha of the equity long short strategy is positive4. This would insinuate that the equity long short strategy outperformed the market equity index and the other two underperformed. But when we look at the significance of the results we find that all three results are not significant. Because they are not significant we unfortunately cannot conclude that these strategies out- or underperformed the market equity index.
Again it is interesting to look at the standard deviation of the strategies. Here we find, just as the period before the bursting of the bubble, that the market index returns are more volatile than the returns of the hedge fund strategies. The equity market index has a volatility of 5.928% while the highest volatility of the hedge fund strategy is that of the long-short strategy and is 1.888%. The market neutral strategy has the lowest volatility with 0.994%.
After the bursting of the Internet bubble it would be a lot less risky to invest in these hedge fund strategies.
8. Discussion of results
The factor model that I used is designed to estimate the exposure of a diversified portfolio of hedge funds. But according to Fung and Hsieh (2004) there is a limitation for when you try to explain the performance of niche hedge fund styles. As a result you could run into some insignificant results. They say that if you want to look at more specific hedge fund styles it better to find risk factors that fit that purpose. Mitchell and Pulvino (2001) showed this for merger arbitrage funds as well as Fung and Hsieh (2001) showed this for trend following hedge funds. So in order to find more significant results for the hedge fund styles I used, the results would be better described by identifying more narrow benchmarks. For future research additional and more fitting risk factors should be identified that are more specific to these hedge fund styles.
A different aspect that might have affected my results is the bias that is present in the hedge fund database of Barclayhedge.com. Due to the lack of regulation for hedge funds to disclose their results, there will be biases when testing. Fung and Hsieh (2004) and Brown et al.(1999) have proven this in their research. Unfortunately I was not able to access the larger hedge fund databases like TASS, HFR and MAR, which are used in other researches and are more aware of the biases. This might partially explain why my results for the period 2000-2003 were insignificant.
My results show that the market index constantly outperforms the three hedge fund strategies. This is in line with the findings of Ackermann et al. (1999) and Liang (1999). As their results show that the hedge funds perform better than mutual funds, but the results are lower than the market index. But in contrast to their research, I find, just as Liang (2001), that the market index returns are more volatile than those of the hedge fund strategies. This may be attributable to the different dynamic hedging strategies employed by hedge funds to reduce risk.
9. Conclusion
Using a sample of publicly accessible hedge fund data I investigated three hedge fund strategies’ (global macro, market neutral, equity long-short) performance compared to the market index in the period 1997-2000 and 2000-2003 using a seven factor model (Fung and Hsieh, 1997) that takes into account several common risk factors. This period 1997-2000 was the period just after the Asian currency crisis (1997) and during the time of the Russian debt default (1998) but before the dotcom crash, also known as the Internet bubble. The period 2000-2003 is the period just after the dotcom crash.
In the period 1997-2000 leading up to the Internet bubble I found that the market index outperformed all three hedge fund strategies. This for me was not very surprising due to the prior crises that had just taken place. Although the market index returns were higher than those of the hedge fund strategies, the hedge funds were a lot less volatile. This I believe is mainly due to the hedge funds ability to dynamically hedging strategies and their ability to use different investment instruments to reduce their risk.
The period after the bursting of the bubble of March 10 2000 I unfortunately cannot say much about because the results of my empirical tests have insignificant values. But just as the period before, I noticed that the market index was more volatile than the returns of the hedge fund strategies.
Appendix
Regression of hedge fund strategies for period 1997-2003 (standard errors in parentheses)
Table 5 Period 1997-2000
Factor
Long-Short Global Macro Market Neutral
Intercept -.1582046 -.0429741 -.0429741 (.0475069)** (.0190454)* (.0190454)* SPexcess .5097095 .1467164 .1467164 (.0921375)** (.0369376)** (.0369376)**
Size Spread -.0000553 -8.32e-06 -8.32e-06
(.0000135)** (5.41e-06) (5.41e-06) Treasury 10y 1.144628 -.1766651 -.1766651 (.6656438) (.2668546) (.2668546) CreditSpread .0816781 -.7974734 .2025266 (.1540973) (.0617771)** (.0617771)** BondLBS -.0240229 .0047225 .0047225 (.0265511) (.0106442) (.0106442) CurrencyLBS .0328006 .0130065 .0130065 (.0269525) (.0108052) (.0108052) CommodityLBS -.0109388 .0100658 .0100658 (.0281665) (.0112918) (.0112918) R-squared 0.6189 0.8874 0.4933
Notes: The dependent variable is the average monthly return of the hedge fund strategy minus the risk
free rate over a period of 39 months. Intercept is the measure of outperformance when positive and underperformance when negative. SPexcess = S&P500 returns minus risk free rate; SizeSpread = Russel 2000 index – S&P500 monthly total return; Treasury 10y= month-end to month-end change in U.S. Federal Reserve 10-year constant maturity yield; CreditSpread = month-end to month-end change in the difference between Moody’s Baa yield and the Federal Reserve’s 10-year constant maturity yield; BondLBS = return of a portfolio of lookback straddles on bond futures; CurrencyLBS = return of a portfolio of lookback straddles on currency; CommodityLBS = return of a portfolio of lookback straddles on commodity futures.
* Significant at the 5% level ** Significant at the 1% level
Regression of hedge fund strategies for period 2000-2003 (standard errors in parentheses)
Table 6 Period 2000-2003
Factor Long-Short Global Macro Market Neutral
Intercept .0268054 -.0046194 -.0046194 (.0165253) (.0066437) (.0066437) SPexcess -.0586551 -.0176937 -.0176937 (.0466531) (.0187561) (.0187561)
Size Spread .0000227 -6.93e-06 -6.93e-06
(.0000245) (9.86e-06) (9.86e-06) Treasury 10y -.7427976 -1.020.305 -1.020.305 (.8826535) (.3548553)** (.3548553)** CreditSpread .2875544 -.7427893 .2572107 (.1344045)* (.0540349)** (.0540349)** BondLBS -.0193534 -.0104895 -.0104895 (.0151857) (.0061051) (.0061051) CurrencyLBS .0242141 .0123776 .0123776 (.015225) (.0061209) (.0061209) CommodityLBS -.0220933 .0070244 .0070244 (.0211888) (.0085186) (.0085186) R-squared 0.4478 0.8879 0.6605
Notes: The dependent variable is the average monthly return of the hedge fund strategy minus the risk
free rate over a period of 44 months. Intercept is the measure of outperformance when positive and underperformance when negative. SPexcess = S&P500 returns minus risk free rate; SizeSpread = Russel 2000 index – S&P500 monthly total return; Treasury 10y= month-end to month-end change in U.S. Federal Reserve 10-year constant maturity yield; CreditSpread = month-end to month-end change in the difference between Moody’s Baa yield and the Federal Reserve’s 10-year constant maturity yield; BondLBS = return of a portfolio of lookback straddles on bond futures; CurrencyLBS = return of a portfolio of lookback straddles on currency; CommodityLBS = return of a portfolio of lookback straddles on commodity futures.
* Significant at the 5% level ** Significant at the 1% level
References
Ackermann, C., Ravenscraft, D., 1998. The impact of regulatory restrictions on fund performance: a comparative study of hedge funds and mutual funds. Working paper, University of Notre-Dame Ackermann, C., McEnally, R., Ravenscraft, D., 1999. The performance of hedge funds: risk, return, and incentives. Journal of Finance 54, 833-874
Agarwal, V., Naik, N.Y., 2002. Characterizing systematic risk of hedge funds with buy-and-hold and option based strategies. Working paper, London Business School
Agarwal, V., Naik, N.Y. 2004. Risk and portfolio decisions involving hedge funds. Review of Financial Studies 17 (1), 63-98
Amin, G.S., Kat, H.M., 2003. Hedge fund performance 1990-2000: Do the “money machines” really add value? Journal of Financial and Quantitative Analysis 38, 251-263
Barclays Hedge Alternative Investment Database. http://www.barclayhedge.com/
Breeden, D.T., 1979. An intertemporal asset-pricing model with stochastic consumption and investment opportunities. Journal of Financial Economics 7, 265-276
Brown, S.J., Goetzman, W., Ibbotson, R.G., 1992. Survivorship bias in performance studies. Review of Financial Studies 5 (4), 553-580
Brown, S.J., Goetzman, W., Ibbotson, R.G., 1999. Offshore hedge funds: survival & performance 1989-1999. Journal of Business 72 (1), 91-117
Carhart, M.M., 1997. On persistence in mutual fund performance. Journal of Finance 52, 57-82 Chan, L.K., Jegadeedh, N., Lakonishok, J., 1996. Momentum strategies. Journal of Finance 51, 1681-1714
Elton, E.J., Gruber, M.J., Blake, C.M., 1996. Survivor bias and mutual funds performance, Review of financial studies 9 (4), 1097-1120
Fama, E.F., 1965. The behavior of stock market prices, Journal of Business 38, 34-105
Fama, E.F., French, K.R., 1993. Common risk factors in the returns on stocks and bonds. Journal of Financial Economics 33, 3-56
Fama, E.F., French, K.R., 1996. Multifactor explanations of asset pricing anomalies. Journal of Finance 51, 55-84
Fama, E.F., French, K.R., 1998. Value versus growth: the international evidence. Journal of Finance, 1975-1999
Field, L.C., Hanka, G., 2001. The expiration of IPO share lockups. Journal of Finance 56 (2), 471-500 Financial education: http://financial-education.com/2007/06/23/index-weighting-methods/
Financial policy forum: http://www.financialpolicy.org ‘Hedge Funds and Funds of Funds’ Friedman, M., 1953. The case for flexible exchange rates, Essays in Positive Economics. Fung, W., Hsieh, D.A., 1997. Empirical characteristics of dynamic trading strategies: the case of hedge funds. Review of Financial Studies 10, 275-302
Fung, W., Hsieh, D.A., 2002. The risk in fixed-income hedge fund styles. Financial Analyst Journal 12 (2), 16-27
Fung, W., Hsieh, D.A., 2004. Hedge fund benchmarks: a risk-based approach. Financial Analyst Journal 60 (5), 65-80
Hauser, F., 2005, Multi factor models in hedge fund and fund of fund management. University of Karlsruhe.
HedgeCo : www.hedgeco.net
Hendrikcks, D., Patel, J., Zeckerhauser, R., 1993. Hot hands in mutual funds: short-run persistence of performance. Journal of Finance 48, 93-130
Jensen, M.C., 1968. The performance of mutual funds in the period 1945-1964. Journal of Finance 23 (2)
Liang, B., 1999. On the performance of hedge funds. Financial Analyst Journal 55 (4), 72-85
Liang, B., 2001. Hedge funds performance: 1990-1999. Financial Analysts Journal, January/February, 11-18
Managed Account Reports (MAR). http://www.marhedge.com/
Mitchell, M., Pulvino, T., 2001. Characteristics of risk and return in risk arbitrage. Journal of Finance 56, 2135-2175
Nasdaq: http://futures.tradingcharts.com/historical/ND/1999/3/linewchart.html
Ofek, E., Richardson, M., 2000. The IPO lockup period: Implications for market efficiency and downward sloping demand curves. Working paper, New York University
Ofek, E., Richardson, M., 2003. Dotcom mania: The rise and fall of internet stock prices. Journal of Finance 58, 1113-1137
Reuters: http://www.reuters.com/article/2012/01/11/us-hedgefunds-idUSTRE80A1CV20120111 Schneeweis, T., Spurgin, R., 1998. Multifactor analysis of hedge funds, managed futures and mutual fund return and risk characteristics. Journal of Alternative Investments 1, 1-24
Sharpe, W.F., 1964. Capital asset prices: a theory of market equilibrium under conditions of risk. Journal of Finance 19, 425-442
Sharpe, W.F., 1966. Mutual fund performance. Journal of Business 39 (1), 119-138
Sharpe, W.F., 1992. Asset allocation: management style and performance measurement. Journal of Portfolio Management 18, 7-19
Wikipedia : http://en.wikipedia.org/wiki/History_of_the_Internet http://en.wikipedia.org/wiki/Dot-com_bubble
