Abstract: Financial market’s high-frequency responses to FOMC announcements are often used as measures of structural monetary policy shocks. Several recent papers question the exogeneity of these “monetary policy surprises”. For example, Jarocinski and Karadi (2020) show that one in three monetary policy surprises raise stock prices, against conventional wisdom. In this project, we employ a high-dimensional VAR technique to recover what is driving those unusual responses. By controlling for those predictors, we provide a new measure of monetary policy shocks free of such biases. According to our estimates, a one-standard-deviation contractionary shock induces an 11% retraction in GDP after one year, roughly 50% more than the benchmark.