Composite hedge fund indices are typically used to track the relative performance of individual hedge fund managers in terms of generated return and sensitivities to market risk factors. Time and data dependency in these indices could affect research conducted by both practitioners and academics.
What’s Inside?
The authors examine whether hedge fund return–based style analysis is dependent on time and database choice by performing a series of empirical tests on major composite hedge fund indices. They also improve on past explanations of hedge fund return dynamics and risk characteristics and explain sensitivities to traditional asset classes and trading styles (e.g., momentum trading).
How Is This Research Useful to Practitioners?
Investors often face difficulties in assessing hedge fund performance because the funds are not required to track a predefined benchmark. Composite hedge fund indices, which aim to represent average returns from the hedge fund universe, are particularly useful in the process of appraising investments. Return-based style analysis can be used as a basis for hedge fund analysis and as a measure of return sensitivity to various market factors.
The authors indicate that when using a single time frame analysis of 1994–2009 monthly returns, annualized return, standard deviation, skewness, and kurtosis are at comparable levels across all three major composite hedge fund index providers—that is, the Center for International Securities and Derivatives Markets (CISDM), Hedge Fund Research (HFR), and Credit Suisse/Tremont (CSFB/Tremont). But the regression analysis of performance in multiple periods indicates significant differences in standard deviation, skewness, and kurtosis of particular indices, especially in the CSFB/Tremont index for periods before 2003.
All three composite hedge fund indices experience significant correlations with equities (S&P 500 Index and Russell 2000 Index) and credit risk factors, but they do not reflect the interest rate factor. Furthermore, all indices experienced higher correlation with the equity market around October 2008, which is related to large movements in equity valuation because of the financial crisis. The authors conclude that the CSFB/Tremont index has significantly lower correlation with the S&P 500 and Russell 2000 indices than do the other two hedge fund indices.
They also find that all three indices have similar time-series patterns in their correlations with various equity, bond, currency, and commodity trading and momentum factors. The authors conclude that the addition of variables beyond those that link performance to traditional asset classes (i.e., equity, interest rate, and credit risk) may have little impact on the explanatory power of hedge fund indices, except during limited periods of time.
How Did the Authors Conduct This Research?
The hedge fund universe has no single, commonly accepted database, and each database might have a different set of reporting managers. Therefore, data limitation can arise, particularly when fund managers—and their past performance data—are added to or removed from an index (i.e., backfill and survivorship bias). The authors use data reported at the time of publication, which are believed to overcome the biases and are superior to historical returns in a current database.
The return-based style analysis is based on monthly returns from 1994 to mid-2009 and includes data from three major composite hedge fund database providers (HFR, CISDM, and CSFB/Tremont), thus representing an average of industry returns.
To calculate average return and risk characteristics of each index, the authors perform a single time frame analysis. Subsequently, they use four-year rolling monthly returns to run a number of multifactor regression models to analyze the correlations of returns of hedge fund indices with traditional market factors (i.e., S&P 500, Russell 2000, Barclays Capital U.S. Corporate High Yield, and Barclays Capital U.S. Government), as well as correlations with various equity, bond, currency, and commodity trading and momentum factors.
The authors discuss the differences among the strengths of correlation with each of the risk factors and explain how these correlations could affect investors as well as academics. They confirm dependency on time and database choice by using t-statistics to measure the significance of identified differences in key risk characteristics and factor sensitivities among the three indices.
They do not explain, however, how the results of empirical tests might be influenced by different weighting structures in the three indices (e.g., the CSFB/Tremont index is asset weighted, whereas the HFR and CISDM are equally weighted).
Abstractor’s Viewpoint
The authors explain the sensitivities of major indices to asset classes, thus improving on past explanations of risk characteristics and return determinants in the hedge fund universe. The study is comprehensive, and a series of empirical tests supports the conclusions. But the use of data beyond the year 2009 would possibly show the impact of recent market volatility and make the research more relevant from the perspective of investors and academics. Furthermore, the study suggests that the major composite hedge fund index CSFB/Tremont reports lower correlations with market factors than do the other two indices, which may affect the performance appraisal process, empirical research, and perception of the overall attractiveness of the hedge fund industry.