Picking the best cherries: Analysing the use of macro-nance variables in predicting monthly realized volatility

This paper analyses the use of macroeconomic and finance variables in predicting monthly realized volatility in four different asset classes: Equities, commodities, foreign exchange rates, and bonds. The predictability is analysed with four different estimation technique classes: Penalized regressions, dynamic factor models, forecast combinations, and bootstrap aggregation. The results, evaluated both statistically and economically, reveal there is predictive content in the macro-finance variables. However, both the estimation technique and subset of variables which are most relevant appear to be asset class specific.