Hedge fund performance

Submission note: Submitted in total fulfilment of the requirements of the degree of Doctor of Philosophy to the Department of Economics and Finance, La Trobe Business School, College of Arts, Social Sciences and Commerce, La Trobe University, Victoria.

This thesis revisits the debate on hedge-fund performance by examining a number of scenarios. It employs novel quantitative methods and a large data sample consisting of 4,803 unique hedge funds covering a 20-year period from 1 January 1995 to 31 December 2014 to re-investigate three major interlinked aspects of hedge-fund performance. Sensitivity of ranking to different performance measures. Previous studies, based on a static approach, suggest that choice of performance measure does not affect hedge-fund ranking. Adding to the current literature, the thesis employs a dynamic approach through Fung and Hsieh’s (2004) eight-factor benchmark. It compares fund rankings based on the method of measurement, that is, absolute versus relative, and different benchmarks across1 various time horizons (36, 60, 120 and 180 months). The thesis finds that fund rankings differ across absolute and relative measures and more substantially in short-term horizons, which is a new result. Moreover, it also finds that the choice of benchmark does influence the ranking of funds. These results are robust in all investigation periods. Performance persistence. For the first time, the thesis assesses hedge-fund performance persistence using a variance ratio test approach. The analysis reveals that hedge-fund observed (smoothed) returns persist over periods of 1, 3, 6 and 12 months. However, the smoothing return bias is the main source of persistence, as the analysis finds very weak persistence in true (de-smoothed) returns. It also examines the persistence of the true alpha and finds similar results; however, the level of persistence in the true alpha is found to be higher than the level of persistence in the true return. The impact of fund characteristics on performance. Instead of restricting the analysis to a linear relationship between fee characteristics and performance at a single time horizon, the thesis also examines non-linear relationship across various time horizons (36, 60 and 120 months) using Friedman’s (1991) Multivariate Adaptive Regression Splines (MARS) for the first time. It finds evidence of a non-linear relationship between fees and performance. Significantly, the breakpoints, representing the minimum thresholds for a fund to commit to delivering superior results are endogenously identified by MARS and not exogenously imposed. The results reveal that management fees significantly and positively associate with hedge fund performance in the short term whereas incentive fees positively associate with hedge-fund performance in the long term.