01whole.pdf (2.78 MB)
An empirical investigation into asset allocation strategies: an Australian perspective
thesisposted on 2022-03-28, 02:34 authored by Abdel Karim Joubaili
Optimisation models developed since the development of modern portfolio theory diverge mainly in their approaches to the estimation mechanics of asset returns and risk parameters. The original structure of modern portfolio theory employs the estimates of the first two moments of returns, namely the mean and variance, in the optimisation algorithm. Critics of modern portfolio theory base their arguments on the fact that statistical estimates of the input parameters, namely return and risk, are subject to estimation error. As a result, the output of the modern portfolio theory optimisation model is usually unstable with counterintuitive portfolio weights. This has provoked ample research fixated on addressing those limitations by using various econometric and mathematical techniques. This research addresses two issues that constitute the crux of deliberations on portfolio choice problems. The first is the moments estimation mechanics and its effect on portfolio optimisation in a static framework. The other is the asset returns predictability and its employment in dynamic asset allocation strategies. In both cases, the performance of the portfolio optimisation rules is compared with the performance of the naïve allocation as a benchmark. This is because recent empirical evidence has highlighted the supremacy of naïve allocation over statistically driven optimisation strategies. This thesis forms an empirical investigation of the asset allocation models that appeared in the financial literature throughout the past few decades. This thesis first delves into the weaknesses and strengths of each of the optimisation theories suggestedin the financial literature, and then extends some of the optimisation theories into novel approaches. In essence, the core of this research is an investigation of the effcient market hypothesis and the effectiveness of asset returns predictability. This research endeavor is significant, first because of its empirical scrutiny of asset allocation strategies, vigorous analysis and robustness checks of those asset allocation strategies under several risk and returns configurations, and second because of its consideration of static optimisation models versus dynamic allocation techniques.