Moving Along Together

This study examines co-movement of three assets, i.e. stocks, bonds and commodities, in combination with macroeconomic and financial variables. I use forecast combinations, with a pre-selection based on Least Angle Regression, two factor based models, principal component analysis and partial least squares, and regimes switching models to investigate the predictability of co-movements. This is done on the Fisher and a custom transformation of the realised correlation. The main finding are that interest rates, spreads and bond market factors are important for Stock-Bond co-movement, macroeconomic factors are important for Commodity related co-movements and forecasts made with a custom transformation are more accurate than if the fisher transformation was used.