Performance in different market environments

Get Complete Project Material File(s) Now! »

Theoretical framework

Efficient market hypothesis

The efficient market hypothesis has since its origin been the leading theory explaining how price adjustments occur in markets. Financial markets are efficient and investors use information provided by companies, often in the form of quarterly- and annual reports, but also daily news and other analysis provided by different investors. Further, financial markets are also competitive and quick to respond to information. Therefore, in a situation with perfect information available, the market price would always represent the true price of a security (Fama , 1970).
The efficient market hypothesis also implies that the market does not have any memory. Hence, as new information becomes available, the security’s value can move up or down towards its fundamental value. A stock that reacts on daily information does not have a correlation with yesterday’s price level, which indicates a random walk7 and thereby a lack of memory in the market (Clarke, Jandik, & Mandelker, 2001). There are however, certain conditions that need to be fulfilled for the efficient market hypothesis to be valid8. Further, the theory assumes that arbitrary gains are random and that the efficient market would be impossible to out-perform without purchasing riskier assets (Fama & French, 2012).

Modern Portfolio Theory

Harry Markowitz was awarded with the Nobel Prize in Economics in 1990 for his theory called The Modern Portfolio Theory (1952) that has become one of the most central economic theories in the field of finance and investment (Elton & Gruber, 1997). Markowitz’s theory assumes that investors are risk-averse and that they can build their portfolio to maximize expected return at any given level of risk in the market. Furthermore, the theory explains that it is inadequate to focus on the expected return and risk of one specific security. Investors should rather invest in more than one security to reap the benefits from diversification and reduce risk in the portfolio (Elton & Gruber, 1997; Markowitz, 1952).

Capital asset pricing model and the efficient market portfolio

The Capital asset pricing model (CAPM) was founded and developed by William Sharpe (1964) and is based on Harry Markowitz portfolio selection theory (1952). Still today, the CAPM is used for various purposes such as evaluation of performance among managed portfolios and to calculate the cost of capital (Fama & French, 2004).
The CAPM is a single factor model that describes the relationship between an assets systematic risk and its expected return (Lintner, 1965; Perold, 2004). Within the scope of the model, a market proxy9 is an estimation of the true market portfolio where the true market portfolio is explained as a measurement for comparing different funds to the overall market or industry (Roll, 1977). The return of the market and such a portfolio is almost impossible to accomplish. Additionally, Richard Roll (1977) showed that the true market portfolio should consist of everything in the world that has a value. Based on this, you can only make qualified attempts to create the true market portfolio (Ferson & Harvey, 1991). The market proxy generally serves as a benchmark for systematic risk (Perold, 2004).
The following equation is the estimation of CAPM:
In the equation above, represents the return of fund , in month t; is the risk-free return in month t; denotes the return of the market portfolio in month t; is the error term, and are the intercept and slope of the regression respectively. The equations intercept is the so-called Jensen’s alpha (1968). Alpha measures the performance on a risk-adjusted basis and it is interpreted as a measure of over- and under-performance compared to the market proxy used (Capocci & Hübner, 2004). Jensen’s alpha will be discussed further in section 4.3.

Empirical measurements

Beta

Beta is a measure of a portfolios systematic risk or volatility in comparison to the market. The coefficient is calculated with a regression analysis and is based on historical data. A security’s beta measures how sensitive the security’s return is in comparison to the market. In the efficient portfolio, the value of beta is assumed to be one (Berk & DeMarzo, 2014). Therefore, the average asset return moves 1% for each 1% movement in the market. To clarify, the beta of a security is the expected percentage return change in its return, given a one-percentage change in the return of the market portfolio (Capocci, 2013).
To calculate an asset’s beta, divide the covariance of the assets return and the benchmark’s return by the volatility (variance) of the benchmarks return.
Beta, , determines the systematic risk for an investment and [   − ], which is the risk premium in the market. By doing so, CAPM can help investors combine a portfolio to generate the desired expected return (Hamberg, 2001).

Jensen’s alpha

Jensen’s alpha (1968) is one of the most frequently used measurements when evaluating hedge fund performance (Jensen, 1968). It is calculated by running equation 1, where Jensen’s alpha is calculated as:
When measuring performance of a hedge fund, or other asset classes, one needs to consider how much risk a fund contains (Ghysels, Santa-Clara, & Valkanov, 2005). A risk-adjusted measure is the excess return, calculated as the funds return subtracted with a risk-free investment, , of a fund in relation to a selected benchmark. Here, the benchmark is simply what the fund is supposed to perform under the assumptions of the CAPM (Jensen, 1968; Alexander & Dimitriu, 2005; Sharpe, 1964). The expected alpha of the efficient market portfolio is therefore 0, (E( )=0) (Merton, 1973). Hence, if the fund manager succeeds to perform a positive alpha, this is known as out-performing the market. To illustrate this, a common approach is to consider the security market line, SML. The SML is a chart resulting from running the CAPM formula that demonstrates how a fund performs in contrast to the market. is a measure of the overall performance of the fund and the managers’ ability to generate risk-adjusted return (Dybvig & Stephen, 1985). To simplify, a funds alpha value can only be determined in two different situations, either through an increase in the expected return given the same risk level, or through a decrease in risk given that the fund succeeds to maintain the same expected return (Ross, Westerfield, Jaffe, & Jordan, 2008).
Figure 1: The Security Market Line, SML
Note: The Security market line illustrates the relationship between the expected return and the beta. If a fund is plotted above the SML, as the square, it is considered under-valued and the fund has out-performed the benchmark/market. If a fund is plotted under the SML, as the triangle, it is considered over-valued and the fund has under-performed in comparison to the benchmark/market. Constructed by the authors.

Skewness and kurtosis

As a part of the moments10 in statistical analysis, skewness and/or kurtosis is present if the return falls outside of the normal distribution. Skewness shows the degree of the data’s variation of direction, the asymmetry of the distribution. If the skewness is positive, the asymmetric tail of the distribution tends to spread out towards the right, more positive values. The tail of a distribution with negative skewness tends to spread out towards the left, more negative values (left) (Scott & Horvath, 1980). Kurtosis portrays the degree of how peaked the distribution is. Distributions with a positive kurtosis indicate a peaked distribution, while negative kurtosis tends to have a flatter distribution (Greene, 2003).
In a normal distribution, the kurtosis equals three. An investment with a high kurtosis will have a distribution with “fat tails”, both in the negative and positive ends. In addition, high kurtosis tends to overestimate the likelihood of realizing the mean returns (Connoly & Hutchinson, 2012).
In investment situations, few returns have normal distributions. Expected returns are often predicted based on volatility, which are assumed to be normally distributed. By using the measures of skewness and kurtosis in investment decisions, investors may find more efficient indicators of the reliability of the investments volatility (Groenveld & Meeden, 1984; Sheikh & Qiao, 2009)

READ  The development and application of not-for-profit marketing

Method and data

Method

The research follows a time-series and comparative research design, that is in accordance with previous studies (Amenc et al., 2002; Capocci & Hübner, 2004; Eling & Faust, 2010; Naik et al., 2007). In order to compare hedge fund strategies with a selected benchmark, which is the measure of what the strategies try to out-perform, the paper uses a quantitative method. Hedge fund returns are collected as secondary data and statistically analyzed. The selected time interval is January 2002 to December 2016, which offers a large sample of returns over a long timespan.

Sample selection and data collection

The objective of this paper is to examine returns of hedge funds that operate in emerging markets. All hedge fund data is collected from the Eurekahedge database11, which comprises a large sample. To be considered an emerging market fund, at least 90% of the funds’ assets must be allocated in the region. In this research, emerging markets is defined in accordance with Eurekahedge as the regions: developing Europe (Central and Eastern), Latin America, Russia, Asia (excluding Japan, Australia and New Zealand), the Middle East, Africa and the Caribbean (Eurekahedge, 2017).
The hedge fund data regarding emerging markets on Eurekahegde consists of 349 equally weighted funds, categorized into different strategies. This study includes five strategies and the selection is described in section 5.3.1 below. Further, the selected benchmark is the Eurekahedge Hedge Fund index, from now on called the GHF index, which comprises 2797 global hedge funds.
All fund results included in this study are assembled monthly by Eurekahedge and measured in terms of profit/loss of the whole portfolio worth and are net of fees. There are no “twin” funds in the benchmark index, e.g. no onshore and offshore type of the same fund. New funds, with at least a three-month history, are rebalanced into the index while closed funds historical performance remains permanently.
To calculate the hedge funds risk-adjusted returns, data was collected for the one-month US Treasury Bill from the Federal Reserve’s database (Federal Reserve, 2017). Lastly, to enable a comparison between the hedge fund findings and more traditional investment alternatives, data was collected for the S&P 500 index from Yahoo Finance (Yahoo Finance, 2017).

Variables

This study includes five different hedge fund strategies to represent the strategy variable. The five strategies; arbitrage, managed futures, event driven, long/short and fixed income are selected for their variation in terms of investment characteristics. Some of which are based on reducing risk and others enhance return by taking higher risks. Arbitrage, long/short and fixed income are categorized as risk-reducing strategies while managed futures and event driven enhance risk more frequently (Connor & Woo, 2004). Further, long/short have a style of taking different positions in the market whereas fixed income locates their investments into various asset classes, such as currencies and equity (Fung & Hsieh, 1997). The event driven strategy is based on the managers’ speculation of events that will affect the market. Lastly, managed futures use sophisticated computer-programs to beat the market (Connor & Lasarte, 2004).
The benchmark variable is included as a measure to out-perform. As mentioned above it is the GHF index, which is useful since it provides performance of all underlying funds in the database, irrespective of region (Eurekahedge, 2017).

Performance measurement

Together with excess returns, this study will observe the strategies alpha, which measures the strategies risk-adjusted (systematic risk) performance (Jensen, 1968). Jensen’s alpha is frequently used in previous research and can illustrate how well a fund performs in comparison to the benchmark (Capocci & Hübner, 2004; Dichev & Yu, 2011; Eling, 2009; Eling & Faust, 2010; Fung & Hsieh, 2004; Liang, 1999)

Data analysis

To evaluate the strategies performance, several regressions were run using the CAPM. All regressions were treated with a heteroscedasticity and autocorrelation consistent covariance matrix, which accounts for potential autocorrelation and heteroscedasticity in the data (Newey & West, 1986; Greene, 2003). In addition, an augmented Dickey Fuller test was conducted to determine if non-stationarity was present.
Besides running regressions on the dataset for each strategy, the time period was divided into sub-periods, which enabled rolling regressions and thereby increased number of results to analyze. The complete dataset for each strategy contains 180 months and every sub-period 48months, with a 24-month overlap for each period. The last sub-period contains 36 observations. The null hypothesis, H0, for all regressions is expressed as alpha equals zero. In accordance with previous studies, the regressions are performed with significance levels of 1%, 5% and 10% (Capocci & Hübner, 2004; Eling, 2009; Eling & Faust, 2010).

Quality of method

The entire dataset in this study is secondary and collected through publicly accessible databases. Therefore, there is limited control over the quality. Since hedge funds are not required to report historical performance, the collected data might not be unbiased and this is a disadvantage of this type of information that one should be aware of when analyzing the results (Fung & Hsieh , 2000). Further, the data is hand collected by the authors and thereby carries the risk of human error.
The strategies were selected because of their diversity. However, there are more strategies in emerging markets and arguments could be made to suggest another combination of strategies based on different selection criteria. The data available for hedge funds is efficient to interpret using the CAPM, which generates desired estimations. The model has however been criticized for being simplistic and academic studies regarding hedge funds often add other performance measurement models to further analyze the variety in fund returns (Agarwal & Naik, 2000; Capocci & Hübner, 2004; Eling & Faust, 2010; Fama & French, 2004; Liang, 1999). Since alpha is determined in proportion to a benchmark assessed as suitable for the fund, it is important that the chosen benchmark is representative, otherwise one can draw misleading conclusions (CFA Institute, 2013).

Potential biases

Eurekahedge is a well-established database that covers numerous hedge funds (Capocci, 2013). The data might suffer from three potential biases, namely; backfilling bias, survivorship bias and selection bias (Fung & Hsieh , 2000). Backfilling bias occurs when a new hedge fund is added to the database and is asked to provide historical performance. Hedge fund managers could for instance refuse to provide performance history if the result is poor. In contrast, managers also have strong incentives to overestimate their performance since this could bring more clients. This line of reasoning suggests that only managers with a good track record has the incentive to backfill their past performance which could lead to a bias (Eling & Faust, 2010). However, this problem could be solved by simply subtracting the past 12-24 monthly returns of the recently added funds (see Fung & Hsieh , 2000; Capocci & Hübner, 2004). Eling (2009) concluded that this technique would decrease the monthly return by 0.18 % on average while Fung and Hsieh (2000) reported that the monthly return would decrease by 0.12 % on average.

Table of Contents
1 Introduction 
1.1 Background
1.2 Problem statement
1.3 Purpose
2 Hedge funds 
2.1 Hedging
2.2 Emerging markets
2.3 What are hedge funds
2.4 Hedge funds vs. mutual funds
2.5 Hedge fund strategies
2.6 Classification of hedge fund strategies
2.7 Related literature
3 Theoretical framework 
3.1 Efficient market hypothesis
3.2 Modern Portfolio Theory
3.3 Capital asset pricing model and the efficient market portfolio
4 Empirical measurements 
4.1 Beta
4.2 Jensen’s alpha
4.3 Skewness and kurtosis
5 Method and data 
5.1 Method
5.2 Sample selection and data collection
5.3 Variables
5.4 Performance measurement
5.5 Data analysis
5.6 Quality of method
5.7 Potential biases
6 Empirical results
6.1 Descriptive statistics
6.2 Correlation
6.3 Performance measurement results
6.4 Performance in different market environments
7 Conclusion
7.1 Future research
References
GET THE COMPLETE PROJECT
The Performance of Hedge Fund Strategies in Emerging Markets

Related Posts