Multiple monetary policy shocks from daily data: A heteroskedasticity IV approach

We extend the heteroskedasticity IV estimator of Rigobon and Sack (2004) to multiple monetary policy shocks by imposing recursive zero restrictions on the impact matrix. Unlike high-frequency identification, the approach requires neither intraday tick data nor precise announcement timestamps, making it applicable to countries and historical periods where such data are unavailable. Applied to US FOMC announcements, we find causal effects similar to those of high-frequency identification. The heteroskedasticity-based instrument passes weakinstrument tests for the target shock, whereas results are mixed for high-frequency surprises. For the path shock, we obtain strong instruments for the majority of specifications and the underlying shock series is highly correlated with the high-frequency counterpart.