Exploring the Cyclical Predictability of Sector-Specific Premia

Abstract

This dissertation simplifies and substantially improves equity premium and sector return forecasts by combining two prediction models that perform well during different regime states as in Tsiakas et al. (2020) through Dynamic Model Averaging (DMA) of Raftery et al. (2010). This approach allows for statistically significant and economically valuable equity premium forecasts during both recessions and expansions, without needing to predict the regime state explicitly. Extending this approach to sector returns, the adaptability of DMA to sector-specific dynamics allows for return forecasts that outperform all considered alternatives and provide significant economic value in a modified Risk-to-Reward Timing strategy based on Kirby and Ostdiek (2012) over an equally weighted benchmark portfolio spanning all sectors.