When Words Are Not Enough: Emotional Spillovers from Fed Chair Communication to Global Financial Markets

Yiru Chen, Muhammad Ali Nasir and Xingjie Wei

CAMA Working Papers from Centre for Applied Macroeconomic Analysis, Crawford School of Public Policy, The Australian National University

Abstract: Given the important role of the Federal Reserve Chair, their communications may generate substantial cross-border effects on international financial markets. Building on this perspective, this article investigates whether emotional signals conveyed by the Fed Chair during FOMC post-meeting press conferences generate emotional spillovers to global financial markets. Using press conferences held between 2011 and 2025, we construct multimodal emotional indices from textual, vocal, and facial cues using machine learning techniques, and examine their high-frequency effects on exchange rates and equity markets across six major economies. Using local projection methods with minute-level event windows, the results show that Fed Chair emotional signals have immediate and economically significant impacts on global financial markets. Among different modalities, facial and textual emotions exhibit stronger and more robust effects than vocal signals. In particular, positive emotional cues are generally associated with increases in equity ETF returns, while exchange rate responses display substantial cross-country heterogeneity, with emerging market currencies tending to appreciate. We also provide evidence that the effects of emotional cues are non-additive and depend on cross-modal interactions, with vocal signals conditioning the impact of textual and facial emotions. Moreover, the findings show that the transmission of emotional spillovers is shaped by macro-financial vulnerability and global risk conditions. On the one hand, economies with higher vulnerability exhibit stronger responses to emotional shocks, indicating that macro-financial vulnerability acts as a moderating mechanism. However, the effects of emotional signals, especially facial and textual emotion, are amplified during periods of higher global uncertainty. Our main findings remain robust across several robustness checks. In particular, for vocal and facial emotions, coefficient equality is not rejected in 85.8% and 78.3% of the market horizon comparisons, respectively, indicating that the main findings are not systematically driven by the choice of emotion extraction algorithms.

Keywords: central bank communication; multimodal signals; emotional spillover; high-frequency identification; machine learning; facial emotion analysis (search for similar items in EconPapers)
JEL-codes: E44 E52 E58 (search for similar items in EconPapers)
Pages: 44 pages
Date: 2026-10
References: Add references at CitEc
Citations:

Downloads: (external link)
https://openresearch-repository.anu.edu.au/server/ ... 4bee870ec46c/content (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:een:camaaa:2026-86

Access Statistics for this paper

More papers in CAMA Working Papers from Centre for Applied Macroeconomic Analysis, Crawford School of Public Policy, The Australian National University Contact information at EDIRC.
Bibliographic data for series maintained by Cama Admin ().