spb/phd_thesis Public

1\documentclass[11pt,letterpaper]{article}23% Essential packages4\usepackage[utf8]{inputenc}5\usepackage[T1]{fontenc}6\usepackage[english]{babel}7\usepackage{amsmath,amssymb,amsthm}8\usepackage{graphicx}9\usepackage{booktabs}10\usepackage{longtable}11\usepackage{setspace}12\usepackage{natbib}13\usepackage{hyperref}14\usepackage{geometry}15\usepackage{caption}16\usepackage{subcaption}17\usepackage{float}18\usepackage{pdflscape}19\usepackage{rotating}20\usepackage{adjustbox}21\usepackage{threeparttable}22\usepackage{enumitem}2324% Page geometry25\geometry{26    top=1in,27    bottom=1in,28    left=1in,29    right=1in30}3132% Hyperlink configuration33\hypersetup{34    colorlinks=true,35    linkcolor=blue,36    citecolor=blue,37    urlcolor=blue38}3940% Theorem definitions41\newtheorem{definition}{Definition}42\newtheorem{theorem}{Theorem}43\newtheorem{proposition}{Proposition}44\newtheorem{assumption}{Assumption}4546% Line spacing47\doublespacing4849% Begin document50\begin{document}5152% Title page53\begin{titlepage}54\begin{center}55\vspace*{1.5cm}5657{\LARGE \textbf{Returns and Volatility Around FOMC Announcements:}}\\[0.4cm]58{\LARGE \textbf{A High-Frequency Analysis of Policy Tone and Novelty}}\\[1.5cm]5960{\large Simon-Pierre Boucher\footnotemark[1] \qquad Marie-H\'el\`ene Gagnon\footnotemark[2] \qquad Gabriel J. Power\footnotemark[3]}\\[1cm]6162{\large 2026-03-06}\\[1cm]6364\end{center}6566\footnotetext[1]{PhD student in finance, Universit\'e Laval, Quebec City QC Canada G1V 0A6, email: simon-pierre.boucher.1@ulaval.ca}67\footnotetext[2]{Professor of Finance and Research Fellow, CRREP, Universit\'e Laval, email: marie-helene.gagnon@fsa.ulaval.ca}68\footnotetext[3]{IG Wealth Management Chairholder, Professor of Finance and Research Fellow, CRREP and CRIB, Universit\'e Laval, email: gabriel.power@fsa.ulaval.ca}6970\begin{abstract}71\noindent We decompose FOMC statements into policy tone (hawkish/dovish) and informational novelty (departure from previous messaging) and estimate their effects on high-frequency asset returns and volatility. Using 1-minute data for 7 futures contracts across 148 FOMC events (2008--2025), we find that tone predicts directional returns (a one-standard-deviation dovish shift is associated with equity gains that build to about 12 basis points within two hours) while novelty predicts volatility changes (the stance--novelty interaction on VIX persists from 5 to 120 minutes, $t = -5.06$). Policy stance is associated with realized volatility changes in 6 of 7 contracts ($p < 0.01$). Pre-announcement placebo tests and five independent inference methods validate these results. To construct our measures, we train a dual-model ensemble (MiniLM and BERT) on FOMC communications with data-driven PCA-based reference selection.72\end{abstract}7374\vspace{0.3cm}7576\noindent \textbf{JEL Classification:} E52, E58, G12, G147778\noindent \textbf{Keywords:} Monetary policy, FOMC announcements, high-frequency data, textual analysis, market volatility7980\end{titlepage}8182\newpage8384% Include all sections85\input{sections/01_introduction}86\input{sections/02_data}87\input{sections/03_methodology}88\input{sections/04_results}89\input{sections/05_conclusion}9091% Tables92\input{tables/tables}9394% Figures95\input{figures/figures}9697% Appendices98\clearpage99\appendix100\input{appendices/appendix_proofs}101\clearpage102\input{appendices/appendix_additional}103104% Bibliography105\newpage106\bibliographystyle{apalike}107\bibliography{master}108109\end{document}110