Abstract
The monetary policy announcements of major central banks generate vast discussions on social media, providing new insights into how financial markets and the public interpret central bank communication. In this paper, we assemble a novel dataset of over 110 million Twitter messages surrounding policy decisions of the European Central Bank, the Federal Reserve, and the Bank of England. Using advanced AI-based text classification, we distinguish between non-relevant, broadly related, and strictly related tweets, and further classify strictly related tweets into hawkish and dovish categories. This framework enables us to capture social media users’ perceptions of the policy tilt of central banks in real time. Based on this classification, we construct a novel measure of disagreement among users about future monetary policy decisions. Our empirical analysis shows that higher social media disagreement is strongly associated with larger financial market reactions following monetary policy announcements. These findings highlight the value of social media as a complementary source of information for understanding how central bank communication is received and how expectations are formed.