Mean variance portfolio optimisation for long term investors

Speaker: Dr Duy-Minh Dang (The University of Queensland)

According to the Willis Towers Watson's Global Pension Assets Study 2016, about 85% of the pension plan assets under management in Australia are of the Defined Contribution type. This implies that retail investors are faced with managing investments over a truly long term period, about 50 years in total, consisting of roughly a 30 year accumulation  period followed by a 20 year decumulation phase. We discuss dynamic mean-variance asset allocation strategies for long term investors. We consider a portfolio consisting of a risk free bond and an equity index which follows a jump diffusion process. We derive the embedding result which converts the mean-variance objective into a form suitable for dynamic programming. We then discuss a numerical Hamilton-Jacobi-Bellman partial differential equation approach for solving the embedded mean-variance optimization problem under realistic constraints on the strategy. Extensive backtests show superiority of the multi-period mean variance strategy over the popular "glide path" strategy used by pension target date funds.