Date of Award


Degree Type


Degree Name

Doctor of Philosophy


Faculty of Business and Public Management

First Advisor

Professor David Allen

Second Advisor

Professor Tim Brailsford

Third Advisor

Professor Robert Faff


When making conclusions about the performance of managed funds, it is critical that the framework in which such performance is measured provides an accurate and unbiased environment. In this thesis I search for true performance of the two major classes of funds- equity as well as fixed interest managed funds. Focusing, first on the former class, I examine five measurement models across three risk-free proxies, nine benchmarks proposed by the extant literature (covering conditional and unconditional as well as single and multi factor definitions) and over three independent periods in an effort to identity (in a consistent setting) the most accurate and least biast methodology. I also use the Australian dataset, which inherently mitigates any data biases that may potentially afflict US studies of these methodologies, since these were developed from the same dataset on which they were later tested. Not finding a pre-existing benchmark that is objective yet informative, I develop an independent model that satisfies these, sourcing from fifteen factor candidates across four categories. I find that teaming up a fund based market factor with well-defined proxies for size, value, momentum and conditional dividend yield provides the optimal benchmark. The latter class comprising fixed-interest managed funds is a segment left largely unexplored in the financial literature and neglected outright in the Australian context. I examine three risk-free proxies, six benchmark classes encompassing twenty-one potential factors, across five models and two independent time frames in an effort to establish the most informative and least biased setting. The task is complicated by two issues - an acute lack of Australian data (demanding additional bootstrap simulations and bridging tests with the US markets) and the need for a two-pass (time-series and cross-sectional) analysis, arising from the different information content benchmarks carry in these two dimensions. My results, consistent across time, show that a correct combination of a bond market variable, a mixture of interest rate factors and economic factors as well as the proxy for movements in the equity markets yield the optimal benchmark. Both fund classes point to Jensen's Alpha as the preferred model, but Treynor and Mazuy's definition of a quadratic measure is adequate if timing-selectivity separation is required. Neither class is significantly sensitive to the choice of risk free proxy featuring in the performance measures.