Empirical investigations of hedge fund risk & returns: Dynamics of alpha & beta decompositions and risk profiling

The central theme of this thesis is to investigate whether or not hedge fund industry generate pure alpha or excess returns that justifies an increasing asset allocation to this industry from investors. Using sophisticated time series econometric techniques and data that cover all market cycles from 1991 to 2007, we provide overwhelming empirical evidence that hedge funds in general do not generate pure alpha or manager skill based excess returns. We test the robustness of this result overtime using rolling window estimation and moving statistical significance through time. The dwindling alpha or no statistically significant alpha is reported even for emerging market funds that attracted strong investors' attention recently. We provide strong evidence that hedge funds increasingly expose investors to systematic or beta returns which investors can easily & cheaply replicate using derivatives without incurring hefty hedge fund fees. The evidences reported in this dissertation are consistent with the prediction of efficient market hypothesis in that in an informationally efficient market, it is difficult to generate & sustained excess returns adjusted for systematic (beta) risk. We have also examined whether or not there is "strategy drift" in hedge funds by testing market neutrality of market neutral hedge fund strategies. We have documented that there no conclusive evidence on market neutrality of hedge funds defying the claim of market neutral hedge fund strategies. In an industry where transparency is a major issue, this is a significant finding for investors to pay attention to since strategy drift can lead to a drift in portfolio optimization.;We have also analyzed risk profiling of hedge fund indices using Vector Auto-Regression (VAR) techniques and Granger causality. Risk transparency is a major issue in hedge fund industry and the risk profiling using VAR and Granger causality address this central issue of hedge fund investing. It attempts to unveil the underlying causal factors for hedge fund return indices. It is based on the idea that the returns of hedge funds leave "foot prints" and using those foot prints the Granger causality tests profiles the "underlying routs" that the have taken during their trading strategies. Using Merrill Lynch multifactor index as a single factor model framework and the 18 factors in a multivariate vector auto regressions, we tested the Granger causality of hedge fund and fund of fund return series. We find overwhelming evidence that the Mrrill index "Granger cause" all hedge fund index return series" which implies that index components investigation could help profile the risk exposure of the indices which do not otherwise disclose the position level transparencies. The limitation of this finding is that the Merrill index is a bundle of other indices and therefore investors cannot pinpoint their exposure say to equity only factor using the results of Merrill index.;To address this and other limitation of the Merrill factor, we used 18 factors that cover almost all markets and conducted the multivariate VAR and Granger causality tests. We also matched the time period that the 18 factor cover and the Merrill factor for consistency. We found that the majority of the 18 factors individually "Granger cause the hedge fund returns" and collectively Granger cause all the hedge fund return indices at a very high confidence level. These findings have two important implications: (a) As an improvement on the finding of the Merrill index, now we know the individual risk factor's causal power, for instance, the equity factor or the bond factor, in Granger sense and the individual factor causal power can directly be used to profile the individual factor exposure of the hedge fund indices. (b) The finding that the 18 factors collectively "Granger cause" the hedge fund returns series is consistent with the findings using Merrill index which provides robustness of the finding using completely different data sources. (c) The high individual & collective "causal power" of the 18 factors can be used as a compliment to the high "expanalatory power" that we found in answering other research question of this dissertation. Although the estimations and the empirical methodologies are different, the evidence that they are pointing towards the same direction is an important empirical result.;This dissertation is presented as follows. The first chapter introduces the hedge fund industry dynamics & the research context, the research questions and hypothesis for the questions posed. The second chapter surveys the literature. The third presents the data, empirical strategy and the econometric models. The fourth, the main body of this dissertation, provides empirical findings and interpretations for all the research questions. The final chapter provides concluding remarks and draws implications from the dissertation's empirical results.