| Most of the commonly used performance measures of hedge funds, such as the Sharpe ratio and the Jensen alpha, assume an a priori frequency distribution of returns, which, under certain conditions, may result in erroneous inferences. Meanwhile, a non-parametric method allows data to determine the shape of the functional form rather then imposing the parametric straightjacket of rigid distributional assumptions. Distribution of the error term, for example, is not viewed as taking a specific functional form, and the relationship between the dependent and independent variables is not forced into a constraining parametric structure. The parametric method might especially become problematic when the distribution has fat tails and a sharp peak around zero (i.e. there might be more downside risk than upside potential). With a non-parametric approach, on the other hand, the approximation of the distribution of the returns and thus the subsequent inference can be carried out without any guidance/constraints from the theory. The second chapter measures performance of hedge funds following a non-parametric approach and compares its findings with the parametric performance measurement methods.; Furthermore, it is a challenging task to identify the systematic risk factors of hedge funds because of the voluntary disclosure of information by hedge funds and the limited availability of data. The previous literature has typically identified risk factors using stepwise regression methods, with the single model selected by the procedure used for subsequent statistical analysis. A serious shortcoming of any such procedure is that the reported uncertainty for the values such as future predictions or parameter estimates reflects only within model uncertainty, ignoring between model uncertainty—the uncertainty associated with the model selection procedure itself. Hence, stepwise methods can result in the underestimation of the uncertainty about the parameters, overestimation of the confidence in a particular model being “correct” and riskier decisions and poorer predictive ability. Third chapter employs Bayesian Model Averaging (BMA) to account for model uncertainty in the selection of factors that explains the hedge fund returns using a multi-factor model. It is shown that BMA leads to a better evaluation of factors for hedge funds, as well as improved predictive ability. |
