The purpose of this paper is to analyze the content of the statements that are released by the Federal Open Market Committee (FOMC) after its meetings, identify the main textual associative patterns in the statements and examine their impact on the US treasury market.
Latent semantic analysis (LSA), a language processing technique that allows recognition of the textual associative patterns in documents, is applied to all the statements released by the FOMC between 2003 and 2014, so as to identify the main textual “themes” used by the Committee in its communication to the public. The importance of the main identified “themes” is tracked over time, before examining their (collective and individual) effect on treasury market yield volatility via time-series regression analysis.
We find that FOMC statements incorporate multiple, multifaceted and recurring textual themes, with six of them being able to characterize most of the communicated monetary policy in the authors’ sample period. The themes are statistically significant in explaining the variation in three-month, two-year, five-year and ten-year treasury yields, even after controlling for monetary policy uncertainty and the concurrent economic outlook.
The main research implication of the authors’ study is that the LSA can successfully identify the most economically significant themes underlying the Fed’s communication, as the latter is expressed in monetary policy statements. The authors feel that the findings of the study would be strengthened if the analysis was repeated using intra-day (tick-by-tick or five-minute) data on treasury yields.
The authors’ findings are consistent with the notion that the move to “increased transparency” by the Fed is important and meaningful for financial and capital markets, as suggested by the significant effect that the most important identified textual themes have on treasury yield volatility.
This paper makes a timely contribution to a fairly recent stream of research that combines specific textual and statistical techniques so as to conduct content analysis. To the best of their knowledge, the authors’ study is the first that applies the LSA to the statements released by the FOMC.
The authors are grateful to the Editor-in-Chief, Professor Janis Zaima, for her help and support during the reviewing process, and to two anonymous reviewers for helpful comments and suggestions which greatly improved the paper. The authors have benefited from comments and suggestions by participants at the 13th annual conference of the Hellenic Finance and Accounting Association (H.F.A.A.) held in Volos, Greece; the 5th national conference of the Financial Engineering and Banking Society (F.E.B.S.) held in Athens, Greece; and the 22nd annual Conference of the Multinational Finance Society (M.F.S.) held in Halkidiki, Greece, where previous versions of the paper were presented. Special thanks are due to Yu-Chen Wei that acted as a discussant of the paper at the M.F.S. conference. The first author wishes to thank his colleagues at Alpha Bank A.E. for their support. The second author gratefully acknowledges the financial support provided by the Basic Research Funding Program of the A.U.E.B. research centre (project EP–1990–01) and by project EP–1688–01 of the A.U.E.B. research center. The views and suggestions in this paper are solely the responsibility of the authors and should not be interpreted as reflecting the views of the Athens University of Economics and Business (A.U.E.B.) or Alpha Bank A.E. or of any other person associated with the aforementioned.
Emerald Publishing Limited
Copyright © 2017, Emerald Publishing Limited