How Federal Reserve Discussions Respond to Increased Transparency

42 Pages Posted: 21 Oct 2015

Date Written: October 19, 2015


We analyze how the behavior of the Federal Open Market Committee changed after the statutory enforcement of transparency laws in 1993. To do this, we develop techniques to describe how language use changes over time. For a set of widely used vector space metrics, we demonstrate how to decompose aggregate changes into each individual dimension's contribution (such as a particular word's influence). Our approach can be generalized to account for associations between document dimensions (such as word definitions or meanings). Using various documents released by the Federal Reserve from 1976-2007, covering both years in which the FOMC knew its deliberations would eventually be made public, and years in which it believed no records were kept, we find that FOMC deliberations became more similar to the always-public press releases in the transparency regime. FOMC members shifted their comments towards popular economic subjects, such as "inflation" and "growth," and away from personal opinions, like "think." In this setting, the observed changes are not purely substitution across words with the same meaning, as the results are robust to accounting for semantic relations.

Keywords: Vector Space Models, Semantic Similarity, Growth Decompositions, Natural Experiments, Monetary Policy

JEL Classification: C19, C55, E02, E58

Suggested Citation

Egesdal, Michael and Gill, Michael and Rotemberg, Martin, How Federal Reserve Discussions Respond to Increased Transparency (October 19, 2015). Available at SSRN: or

Michael Egesdal

Airbnb ( email )

888 Brannan St
San Francisco, CA 94103
United States

Michael Gill (Contact Author)

Harvard University ( email )

1875 Cambridge Street
Cambridge, MA 02138
United States

Martin Rotemberg

New York University (NYU) ( email )

Bobst Library, E-resource Acquisitions
20 Cooper Square 3rd Floor
New York, NY 10003-711
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
PlumX Metrics