| Summary: | Thesis (PhD)--Stellenbosch University, 2013. === A persistent problem for hedge fund researchers presents itself in the form of
inconsistent and diverse style classifications within and across database providers. For
this paper, single-manager hedge funds from the Hedge Fund Research (HFR) and
Hedgefund.Net (HFN) databases were classified on the basis of a common factor,
extracted using the factor axis methodology. It was assumed that the returns of all
sample hedge funds are attributable to a common factor that is shared across hedge
funds within one classification, and a specific factor that is unique to a particular hedge
fund. In contrast to earlier research and the application of principal component analysis,
factor axis has sought to determine how much of the covariance in the dataset is due to
common factors (communality). Factor axis largely ignores the diagonal elements of the
covariance matrix and orthogonal factor rotation maximises the covariance between
hedge fund return series.
In an iterative framework, common factors were extracted until all return series were
described by one common and one specific factor. Prior to factor extraction, the series
was tested for autoregressive moving-average processes and the residuals of such
models were used in further analysis to improve upon squared correlations as initial
factor estimates. The methodology was applied to 120 ten-year rolling estimation
windows in the July 1990 to June 2010 timeframe. The results indicate that the number
of distinct style classifications is reduced in comparison to the arbitrary self-selected
classifications of the databases. Single manager hedge funds were grouped in portfolios
on the basis of the common factor they share. In contrast to other classification
methodologies, these common factor portfolios (CFPs) assume that some unspecified
individual component of the hedge fund constituents’ returns is diversified away and that
single manager hedge funds should be classified according to their common return
components. From the CFPs of single manager hedge funds, pure style indices were
created to be entered in a multivariate autoregressive framework.
For each style index, a Vector Error Correction model (VECM) was estimated to
determine the short-term as well as co-integrating relationship of the hedge fund series with the index level series of a stock, bond and commodity proxy. It was postulated that
a) in a well-diversified portfolio, the current level of the hedge fund index is independent
of the lagged observations from the other asset indices; and b) if the assumptions of the
Efficient Market Hypothesis (EMH) hold, it is expected that the predictive power of the
model will be low. The analysis was conducted for the July 2000 - June 2010 period.
Impulse response tests and variance decomposition revealed that changes in hedge
fund index levels are partially induced by changes in the stock, bond and currency
markets. Investors are therefore cautioned not to overemphasise the diversification
benefits of hedge fund investments. Commodity trading advisors (CTAs) / managed
futures, on the other hand, deliver diversification benefits when integrated with an
existing portfolio.
The results indicated that single manager hedge funds can be reliably classified using
the principal factor axis methodology. Continuously re-balanced pure style index
representations of these classifications could be used in further analysis. Extensive
multivariate analysis revealed that CTAs and macro hedge funds offer superior
diversification benefits in the context of existing portfolios. The empirical results are of
interest not only to academic researchers, but also practitioners seeking to replicate the
methodologies presented.
|
|---|
