Nowcasting causality in mixed frequency vector autoregressive models

This paper introduces the notion of nowcasting causality for mixed-frequency VARs as the mixed-frequency version of instantaneous causality. We analyze the relationship between nowcasting and Granger causality in the mixed-frequency VAR setting of Ghysels (2012) and illustrate that nowcasting causality can have a crucial impact on the significance of contemporaneous or lagged high-frequency variables in standard MIDAS regression models