Title: Return prediction models and asset allocation
Authors: Ekaterini Panopoulou - University of Essex (United Kingdom) [presenting]
Iason Kynigakis - University of Kent (United Kingdom)
Abstract: The aim is to examine the out-of-sample return predictability for stock, bond and commodity indices using a large set of economic indicators. The methods used to construct the return forecasts are from the shrinkage, latent factor and forecast combination literature. The economic and statistical value of the forecasts is examined during recessions and expansions and for periods with negative and positive returns. The benefits of return predictability and of including commodities in a portfolio are further investigated through an out-of-sample multivariate asset allocation exercise based on the mean-variance optimization framework. The effects of return predictability are assessed around business cycles, different levels of risk aversion, investment constraints, transaction costs, alternative risk measures and different ways of estimating the covariance matrix.