US Monetary Policy Since the 1950s and the Changing Content of FOMC Minutes

Abstract

Content analysis is used to analyze 60 years of FOMC minutes. Since there is no unique algorithm to quantify content two different algorithms are applied. Wordscores compares content relative to a chosen benchmark while DICTION is an alternative algorithm that is specifically designed to capture various elements that capture the sentiment or tone conveyed in a text. The resulting indicators are then incorporated into a VAR. The content of FOMC minutes is found to be significantly related to the state of the economy, notably real GDP growth and changes in the fed funds rate. However, the relationship between content and macroeconomic conditions changes after 1993 when minutes are made public with a lag. Both content indicators also suggest substantive changes in the content of FOMC minutes since the 1950s in terms of the FOMC’s dovishness or hawkishness.