CMStatistics 2020: Start Registration
View Submission - CFE
Title: Estimating the effects of monetary policy: An automated narrative approach Authors:  Miguel Acosta - Columbia University (United States) [presenting]
Abstract: The aim is to investigate whether the macroeconomic effects of monetary policy can be separately identified from the effects of the information revealed by monetary policy actions and announcements. This task is complicated by the fact that both policies are emitted at the same time; thus, current estimates of monetary policy shocks are generally unable to distinguish between the two. An empirical strategy is presented for distinguishing between these effects. An information shock is identified via an ``automated narrative approach''. Specifically, Fed watchers describe what information was revealed in each policy announcement. By applying machine learning and natural language processing techniques, the surprise component of the Fed's discussion about different topics (e.g., inflation or output) in its post-meeting announcements can be measured. These measures can be used as detailed proxies for the information content of monetary policy in order to study and estimate the effects of monetary policy and communications policy on financial market and macroeconomic outcomes.