Stock-bond return correlations: Moving away from "one-frequency-fits-all" by extending the DCC-MIDAS approach  

This paper explores the determinants of U.S. stock-bond correlations estimated at various frequencies. For this purpose, the two-component DCC-MIDAS model of correlation (Colacito et al. 2011) is used and extended to incorporate a third correlation frequency component. Subsequently, macroeconomic and financial variables are studied as determinants of each component. We show that the daily correlation component is driven by financial market factors, while the monthly component is more influenced by macroeconomic factors. Finally, the yearly component is determinedbyfundingopportunitiesintheeconomy. Theseresultsareimportantasthey show that different correlation components and determinants should be considered for different investment horizons