SPB Git

spb/phd_thesis Public

PhD thesis — Three essays on high-frequency return and volatility dynamics in commodities and financial futures markets (Université Laval).

phdthesisfinancevolatilityresearch
TeX 100%

Initial commit: PhD thesis — three essays on high-frequency dynamics in futures markets

Three article chapters (speculative trading & macro surprises; ETF iNAV
volatility transmission; FOMC tone & novelty) plus the assembled
Université Laval thesis document (these-ulaval/), with a full README.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
simon-pierre boucher committed 5 days ago (Aug 5, 2026)

Showing 196 changed files with +18,725 and −0

added .gitignore +20 −0
@@ -0,0 +1,20 @@
1 +# macOS
2 +.DS_Store
3 +
4 +# Claude Code local settings
5 +.claude/
6 +
7 +# Archive duplicate of these-ulaval/
8 +these-ulaval.zip
9 +
10 +# LaTeX build artifacts (PDFs and .bbl are kept)
11 +*.aux
12 +*.log
13 +*.blg
14 +*.out
15 +*.toc
16 +*.lof
17 +*.lot
18 +*.fls
19 +*.fdb_latexmk
20 +*.synctex.gz
added PHD_chapitre3_theses_20260731/.gitignore +3 −0
@@ -0,0 +1,3 @@
1 +.output
2 +.log
3 +.tmp
added PHD_chapitre3_theses_20260731/.project +1 −0
@@ -0,0 +1 @@
1 +name: PHD_chapitre3_theses_20260417
added PHD_chapitre3_theses_20260731/appendices/appendix_additional.tex +116 −0
@@ -0,0 +1,116 @@
1 +\section{Additional Tables and Figures}
2 +\label{sec:app_additional}
3 +
4 +This appendix collects supplementary tables and figures that support the main results but are not essential for following the core argument.
5 +
6 +\subsection{Descriptive Figures}
7 +
8 +\begin{figure}[H]
9 +\centering
10 +\includegraphics[width=0.85\textwidth]{figures/figD1_return_density.pdf}
11 +\caption{Distribution of 1-minute log returns on FOMC days. Winsorized at 0.5\% tails for display. The heavy tails and leptokurtic shape are consistent with the Jarque--Bera test rejections reported in Table~\ref{tab:desc_1min_ret}.}
12 +\label{fig:D1}
13 +\end{figure}
14 +
15 +\begin{figure}[H]
16 +\centering
17 +\includegraphics[width=0.85\textwidth]{figures/figD5_acf_returns.pdf}
18 +\caption{Autocorrelation of 1-minute returns. Grey band shows 95\% confidence interval. Negative first-order autocorrelation reflects bid--ask bounce effects typical of high-frequency data.}
19 +\label{fig:D5}
20 +\end{figure}
21 +
22 +\begin{figure}[H]
23 +\centering
24 +\includegraphics[width=0.85\textwidth]{figures/figD8b_log_rv_distribution.pdf}
25 +\caption{Distribution of $\log(\mathrm{RV})$. Near-Gaussian shape validates the use of $\log(\mathrm{RV})$ as the dependent variable in the panel regressions.}
26 +\label{fig:D8b}
27 +\end{figure}
28 +
29 +\subsection{Ensemble Model Diagnostics}
30 +
31 +\begin{figure}[H]
32 +\centering
33 +\includegraphics[width=0.6\textwidth]{figures/pca_axis_quality.png}
34 +\caption{PCA axis assignment quality. Bar height shows $|\text{Corr}|$ between PC projection and keyword differential. All axes exceed the 0.04 minimum separation threshold.}
35 +\label{fig:pca_quality}
36 +\end{figure}
37 +
38 +\begin{figure}[H]
39 +\centering
40 +\begin{subfigure}[b]{0.48\textwidth}
41 +\includegraphics[width=\textwidth]{figures/novelty_comparison.png}
42 +\caption{Novelty: MiniLM vs BERT}
43 +\end{subfigure}
44 +\hfill
45 +\begin{subfigure}[b]{0.48\textwidth}
46 +\includegraphics[width=\textwidth]{figures/scatter_comparison.png}
47 +\caption{Score scatter plots}
48 +\end{subfigure}
49 +\caption{Model comparison: MiniLM (384d) vs.\ BERT (768d). Novelty correlation $r = 0.72$; policy stance tone $r = 0.56$. The moderate inter-model correlation supports the use of an ensemble approach.}
50 +\label{fig:model_comparison}
51 +\end{figure}
52 +
53 +\begin{figure}[H]
54 +\centering
55 +\begin{subfigure}[b]{0.48\textwidth}
56 +\includegraphics[width=\textwidth]{figures/tone_comparison.png}
57 +\caption{Tone comparison across models}
58 +\end{subfigure}
59 +\hfill
60 +\begin{subfigure}[b]{0.48\textwidth}
61 +\includegraphics[width=\textwidth]{figures/confidence_distribution.png}
62 +\caption{Ensemble confidence distribution}
63 +\end{subfigure}
64 +\caption{Tone model comparison and confidence diagnostics. The right panel shows that ensemble confidence is concentrated above 0.7, indicating strong inter-model agreement for most statements.}
65 +\label{fig:tone_conf}
66 +\end{figure}
67 +
68 +\subsection{Additional Panel Regression Tables}
69 +
70 +Tables~\ref{tab:panel_rv_stance}--\ref{tab:panel_return_novelty} report panel minute-level regressions for the remaining dependent variables (realized volatility in levels, realized beta, and returns), separately for stance and novelty.
71 +
72 +\input{tables/tableP_panel_rv_stance.tex}
73 +\input{tables/tableP_panel_rv_novelty.tex}
74 +\input{tables/tableP_panel_beta_stance.tex}
75 +\input{tables/tableP_panel_beta_novelty.tex}
76 +\input{tables/tableP_panel_return_stance.tex}
77 +\input{tables/tableP_panel_return_novelty.tex}
78 +
79 +\subsection{Additional Event-Level Regression Tables}
80 +
81 +Tables~\ref{tab:rolling_rv_ratio_30min} and~\ref{tab:rolling_log_rv_ratio_30min} report the event-level RV-ratio specifications, which corroborate the pre/post difference results in the main text.
82 +
83 +\input{tables/tableR_rv_ratio_30min.tex}
84 +\input{tables/tableR_log_rv_ratio_30min.tex}
85 +
86 +\subsection{Additional IRF and Drift Figures}
87 +
88 +\begin{figure}[H]
89 +\centering
90 +\includegraphics[width=0.85\textwidth]{figures/figR9_drift_by_novelty.pdf}
91 +\caption{Cumulative return drift by novelty tercile. Unlike stance-based drift (Figure~\ref{fig:R8}), novelty terciles show less directional separation, consistent with novelty affecting volatility rather than returns.}
92 +\label{fig:R9}
93 +\end{figure}
94 +
95 +\begin{figure}[H]
96 +\centering
97 +\includegraphics[width=0.85\textwidth]{figures/figR10_irf_ret_h_novelty.pdf}
98 +\caption{Impulse response: cumulative return to novelty. Effects are generally smaller and less significant than stance effects (Figure~\ref{fig:R10_stance}), supporting the hypothesis that novelty operates through volatility rather than returns.}
99 +\label{fig:R10_novelty}
100 +\end{figure}
101 +
102 +\begin{figure}[H]
103 +\centering
104 +\includegraphics[width=0.85\textwidth]{figures/figR10b_irf_placebo_novelty.pdf}
105 +\caption{Pre-announcement placebo: IRF for novelty. All coefficients are near zero, corroborating the placebo results for stance (Figure~\ref{fig:placebo_stance}).}
106 +\label{fig:placebo_novelty}
107 +\end{figure}
108 +
109 +\subsection{Sub-Period Stability}
110 +
111 +\begin{figure}[H]
112 +\centering
113 +\includegraphics[width=0.85\textwidth]{figures/figRob_subperiod_heatmap.pdf}
114 +\caption{Sub-period stability: regression coefficients across six Fed policy regimes. Coefficient signs are generally consistent across sub-periods, though magnitudes vary with the degree of policy uncertainty in each regime.}
115 +\label{fig:rob_subperiod}
116 +\end{figure}
added PHD_chapitre3_theses_20260731/appendices/appendix_proofs.tex +105 −0
@@ -0,0 +1,105 @@
1 +\section{Mathematical Proofs}
2 +\label{sec:proofs}
3 +
4 +This appendix provides formal mathematical derivations for the key theoretical relationships presented in Section~\ref{sec:methodology}.
5 +
6 +\subsection{Derivation of Pure Stance Definitions}
7 +
8 +\begin{theorem}[Pure Stance Characterization]
9 +The pure dovish and hawkish stance measures satisfy the relationships given in Definition~\ref{def:pure_stances}.
10 +\end{theorem}
11 +
12 +\begin{proof}
13 +\textbf{Case 1: Pure Dovish Statement.} Assume the Federal Reserve releases a statement that exactly matches the dovish counterfactual: $F_t = F_t^D$.
14 +
15 +The tone measure becomes:
16 +\begin{align}
17 +\text{Tone}_t &= \frac{\text{sim}(F_t, F_t^H) - \text{sim}(F_t, F_t^D)}{1 - \text{sim}(F_t^D, F_t^H)} \notag \\
18 +&= \frac{\text{sim}(F_t^D, F_t^H) - \text{sim}(F_t^D, F_t^D)}{1 - \text{sim}(F_t^D, F_t^H)} \notag \\
19 +&= \frac{\text{sim}(F_t^D, F_t^H) - 1}{1 - \text{sim}(F_t^D, F_t^H)} \notag \\
20 +&= -1
21 +\end{align}
22 +
23 +The novelty measure is:
24 +\begin{equation}
25 +\text{Novelty}_t = 1 - \text{sim}(F_t, F_{t-1}) = 1 - \text{sim}(F_t^D, F_{t-1})
26 +\end{equation}
27 +
28 +Therefore, the stance measure is:
29 +\begin{align}
30 +\text{Stance}_t &= \text{Novelty}_t \times \text{Tone}_t \notag \\
31 +&= \left(1 - \text{sim}(F_t^D, F_{t-1})\right) \times (-1) \notag \\
32 +&= -\left(1 - \text{sim}(F_t^D, F_{t-1})\right) \notag \\
33 +&= \text{Stance}_t^{dove}
34 +\end{align}
35 +
36 +\textbf{Case 2: Pure Hawkish Statement.} Assume $F_t = F_t^H$:
37 +
38 +The tone measure becomes:
39 +\begin{align}
40 +\text{Tone}_t &= \frac{\text{sim}(F_t^H, F_t^H) - \text{sim}(F_t^H, F_t^D)}{1 - \text{sim}(F_t^D, F_t^H)} \notag \\
41 +&= \frac{1 - \text{sim}(F_t^H, F_t^D)}{1 - \text{sim}(F_t^D, F_t^H)} \notag \\
42 +&= 1
43 +\end{align}
44 +
45 +Therefore:
46 +\begin{align}
47 +\text{Stance}_t &= \left(1 - \text{sim}(F_t^H, F_{t-1})\right) \times 1 = \text{Stance}_t^{hawk}
48 +\end{align}
49 +
50 +This completes the proof. \hfill $\square$
51 +\end{proof}
52 +
53 +\subsection{Derivation of Dovish Weight Parameter}
54 +
55 +\begin{theorem}[Dovish Weight Parameter Formula]
56 +The weight parameter $w_t$ in the weighted stance representation has the form given in Definition~\ref{def:weighted_stance}.
57 +\end{theorem}
58 +
59 +\begin{proof}
60 +From the weighted stance equation:
61 +\begin{equation}
62 +\text{Stance}_t = w_t \cdot \text{Stance}_t^{dove} + (1 - w_t) \cdot \text{Stance}_t^{hawk}
63 +\end{equation}
64 +
65 +Substituting the expressions for pure stances from Theorem 1:
66 +\begin{align}
67 +\text{Stance}_t &= -w_t\left(1 - \text{sim}(F_t^D, F_{t-1})\right) + (1 - w_t)\left(1 - \text{sim}(F_t^H, F_{t-1})\right) \notag \\
68 +&= 1 - 2w_t + w_t\left(\text{sim}(F_t^D, F_{t-1}) + \text{sim}(F_t^H, F_{t-1})\right) - \text{sim}(F_t^H, F_{t-1})
69 +\end{align}
70 +
71 +Collecting the terms in $w_t$, this reads
72 +\begin{equation}
73 +\text{Stance}_t = 1 - \text{sim}(F_t^H, F_{t-1}) - w_t\left(2 - \text{sim}(F_t^D, F_{t-1}) - \text{sim}(F_t^H, F_{t-1})\right).
74 +\end{equation}
75 +Equating with $\text{Stance}_t = \left(1 - \text{sim}(F_t, F_{t-1})\right) \times \text{Tone}_t$ and solving for $w_t$:
76 +\begin{equation}
77 +w_t = \frac{1 - \text{sim}(F_t^H, F_{t-1}) - \left(1 - \text{sim}(F_t, F_{t-1})\right) \times \text{Tone}_t}{2 - \text{sim}(F_t^D, F_{t-1}) - \text{sim}(F_t^H, F_{t-1})}
78 +\end{equation}
79 +
80 +This establishes the formula. \hfill $\square$
81 +\end{proof}
82 +
83 +\subsection{Proof of MPS Decomposition}
84 +
85 +\begin{theorem}[Policy Stance Surprise Decomposition]
86 +The policy stance surprise admits the decomposition given in Proposition~\ref{prop:mps}.
87 +\end{theorem}
88 +
89 +\begin{proof}
90 +From the definitions:
91 +\begin{align}
92 +\text{MPS}_t &= \text{Stance}_t - \mathbb{E}_{t-\Delta}[\text{Stance}_t] \notag \\
93 +&= \text{Novelty}_t \times \text{Tone}_t - (1 - 2p_{t-\Delta}) \cdot \overline{\text{Novelty}}_{t \mid t-\Delta} \notag \\
94 +&= \left(\overline{\text{Novelty}}_{t \mid t-\Delta} + \varepsilon_t\right) \times \text{Tone}_t - (1 - 2p_{t-\Delta}) \cdot \overline{\text{Novelty}}_{t \mid t-\Delta} \notag \\
95 +&= \overline{\text{Novelty}}_{t \mid t-\Delta} \times \text{Tone}_t - \overline{\text{Novelty}}_{t \mid t-\Delta} + 2p_{t-\Delta} \cdot \overline{\text{Novelty}}_{t \mid t-\Delta} + \varepsilon_t \times \text{Tone}_t \notag \\
96 +&= \overline{\text{Novelty}}_{t \mid t-\Delta}(\text{Tone}_t + 2p_{t-\Delta} - 1) + \text{Tone}_t \cdot \varepsilon_t
97 +\end{align}
98 +
99 +This establishes the decomposition. \hfill $\square$
100 +\end{proof}
101 +
102 +\subsection{Economic Interpretation}
103 +
104 +The mathematical results provide several economic insights. First, the pure stance characterization shows that our measures correctly identify extreme policy communications, with dovish statements receiving negative stance values and hawkish statements receiving positive values. Second, the weight parameter derivation reveals how actual policy communications can be understood as weighted averages of extreme alternatives. Third, the MPS decomposition shows that policy surprises have two distinct sources: unexpected tone conditional on expected information content, and unexpected information content weighted by actual tone.
105 +
added PHD_chapitre3_theses_20260731/chapitre3.bbl +254 −0
@@ -0,0 +1,254 @@
1 +\begin{thebibliography}{}
2 +
3 +\bibitem[Andersen and Bollerslev, 1997]{andersen1997}
4 +Andersen, T.~G. and Bollerslev, T. (1997).
5 +\newblock Intraday periodicity and volatility persistence in financial markets.
6 +\newblock {\em Journal of Empirical Finance}, 4(2--3):115--158.
7 +
8 +\bibitem[Andersen et~al., 2003]{andersen2003}
9 +Andersen, T.~G., Bollerslev, T., Diebold, F.~X., and Vega, C. (2003).
10 +\newblock Micro effects of macro announcements: Real-time price discovery in
11 + foreign exchange.
12 +\newblock {\em American Economic Review}, 93(1):38--62.
13 +
14 +\bibitem[Andersen et~al., 2007]{andersen2007}
15 +Andersen, T.~G., Bollerslev, T., Diebold, F.~X., and Vega, C. (2007).
16 +\newblock Real-time price discovery in global stock, bond and foreign exchange
17 + markets.
18 +\newblock {\em Journal of International Economics}, 73(2):251--277.
19 +
20 +\bibitem[Apel and Grimaldi, 2012]{apel2012}
21 +Apel, M. and Grimaldi, M.~B. (2012).
22 +\newblock The information content of central bank minutes.
23 +\newblock Working Paper Series 261, Sveriges Riksbank.
24 +
25 +\bibitem[Araci, 2019]{araci2019}
26 +Araci, D. (2019).
27 +\newblock Finbert: Financial sentiment analysis with pre-trained language
28 + models.
29 +\newblock {\em arXiv preprint arXiv:1908.10063}.
30 +
31 +\bibitem[Baker et~al., 2016]{baker2016}
32 +Baker, S.~R., Bloom, N., and Davis, S.~J. (2016).
33 +\newblock Measuring economic policy uncertainty.
34 +\newblock {\em Quarterly Journal of Economics}, 131(4):1593--1636.
35 +
36 +\bibitem[Balduzzi et~al., 2001]{balduzzi2001}
37 +Balduzzi, P., Elton, E.~J., and Green, T.~C. (2001).
38 +\newblock Economic news and bond prices: Evidence from the us treasury market.
39 +\newblock {\em Journal of Financial and Quantitative Analysis}, 36(4):523--543.
40 +
41 +\bibitem[Barndorff-Nielsen and Shephard, 2002]{barndorff2002}
42 +Barndorff-Nielsen, O.~E. and Shephard, N. (2002).
43 +\newblock Econometric analysis of realized volatility and its use in estimating
44 + stochastic volatility models.
45 +\newblock {\em Journal of the Royal Statistical Society: Series B (Statistical
46 + Methodology)}, 64(2):253--280.
47 +
48 +\bibitem[Benjamini and Hochberg, 1995]{benjamini1995controlling}
49 +Benjamini, Y. and Hochberg, Y. (1995).
50 +\newblock Controlling the false discovery rate: a practical and powerful
51 + approach to multiple testing.
52 +\newblock {\em Journal of the Royal Statistical Society: Series B
53 + (Methodological)}, 57(1):289--300.
54 +
55 +\bibitem[Bernanke and Kuttner, 2005]{bernanke2005}
56 +Bernanke, B.~S. and Kuttner, K.~N. (2005).
57 +\newblock What explains the stock market's reaction to federal reserve policy?
58 +\newblock {\em Journal of Finance}, 60(3):1221--1257.
59 +
60 +\bibitem[Bligh and Hess, 2008]{bligh2008}
61 +Bligh, M.~C. and Hess, G.~D. (2008).
62 +\newblock The decline of symbolic politics: Federal reserve communications.
63 +\newblock {\em Journal of Economic Psychology}, 29(4):535--559.
64 +
65 +\bibitem[Blinder et~al., 2008]{blinder2008}
66 +Blinder, A.~S., Ehrmann, M., Fratzscher, M., De~Haan, J., and Jansen, D.-J.
67 + (2008).
68 +\newblock Central bank communication and monetary policy: A survey of theory
69 + and evidence.
70 +\newblock {\em Journal of Economic Literature}, 46(4):910--945.
71 +
72 +\bibitem[Brusa et~al., 2015]{brusa2015}
73 +Brusa, F., Savor, P., and Wilson, M. (2015).
74 +\newblock Asset allocation and fomc announcements.
75 +\newblock {\em Review of Financial Studies}, 28(5):1398--1446.
76 +
77 +\bibitem[Brusa et~al., 2019]{brusa2019}
78 +Brusa, F., Savor, P., and Wilson, M. (2019).
79 +\newblock Fomc announcements and market returns: Evidence from the options
80 + market.
81 +\newblock {\em Journal of Finance}, 75(1):399--441.
82 +
83 +\bibitem[Campbell et~al., 2012]{campbell2012}
84 +Campbell, J.~R., Evans, C.~L., Fisher, J.~D., and Justiniano, A. (2012).
85 +\newblock Macroeconomic effects of federal reserve forward guidance.
86 +\newblock {\em Brookings Papers on Economic Activity}, 2012(1):1--80.
87 +
88 +\bibitem[Devlin et~al., 2019]{kenton2019}
89 +Devlin, J., Chang, M.-W., Lee, K., and Toutanova, K. (2019).
90 +\newblock Bert: Pre-training of deep bidirectional transformers for language
91 + understanding.
92 +\newblock In {\em Proceedings of the 2019 Conference of the North American
93 + Chapter of the Association for Computational Linguistics}, pages 4171--4186.
94 +
95 +\bibitem[Ehrmann et~al., 2011]{ehrmann2011}
96 +Ehrmann, M., Fratzscher, M., and Rigobon, R. (2011).
97 +\newblock Global crises and equity market contagion.
98 +\newblock {\em Journal of Finance}, 66(6):2597--2649.
99 +
100 +\bibitem[Eklund and Kim, 2024]{eklund2024}
101 +Eklund, J. and Kim, S. (2024).
102 +\newblock Fomc statement sentiment and inflation expectations.
103 +\newblock {\em Journal of Monetary Economics}, 141:45--62.
104 +
105 +\bibitem[Fama, 1970]{fama1970}
106 +Fama, E.~F. (1970).
107 +\newblock Efficient capital markets: A review of theory and empirical work.
108 +\newblock {\em Journal of Finance}, 25(2):383--417.
109 +
110 +\bibitem[Fleming and Remolona, 1999]{fleming1999}
111 +Fleming, M.~J. and Remolona, E.~M. (1999).
112 +\newblock Price formation and liquidity in the us treasury market: The response
113 + to public information.
114 +\newblock {\em Journal of Finance}, 54(5):1901--1915.
115 +
116 +\bibitem[Gentzkow et~al., 2019]{gentzkow2019}
117 +Gentzkow, M., Kelly, B., and Taddy, M. (2019).
118 +\newblock Text as data.
119 +\newblock {\em Journal of Economic Literature}, 57(3):535--574.
120 +
121 +\bibitem[Gorodnichenko et~al., 2023]{gorodnichenko2023}
122 +Gorodnichenko, Y., Pham, T., and Talavera, O. (2023).
123 +\newblock The voice of monetary policy.
124 +\newblock {\em American Economic Review}, 113(2):548--584.
125 +
126 +\bibitem[Grossman and Stiglitz, 1980]{grossman1980}
127 +Grossman, S.~J. and Stiglitz, J.~E. (1980).
128 +\newblock On the impossibility of informationally efficient markets.
129 +\newblock {\em American Economic Review}, 70(3):393--408.
130 +
131 +\bibitem[G{\"u}rkaynak et~al., 2005]{gurkaynak2005}
132 +G{\"u}rkaynak, R.~S., Sack, B.~P., and Swanson, E.~T. (2005).
133 +\newblock Do actions speak louder than words? the response of asset prices to
134 + monetary policy actions and statements.
135 +\newblock {\em International Journal of Central Banking}, 1(1):55--93.
136 +
137 +\bibitem[Gürtler and Gürtler, 2010]{gurtler2010}
138 +Gürtler, M. and Gürtler, O. (2010).
139 +\newblock The effect of fomc statement language on financial markets.
140 +\newblock {\em Journal of Financial Research}, 33(4):369--391.
141 +
142 +\bibitem[Hansen et~al., 2018]{hansen2017}
143 +Hansen, S., McMahon, M., and Prat, A. (2018).
144 +\newblock Transparency and deliberation within the fomc: A computational
145 + linguistics approach.
146 +\newblock {\em Quarterly Journal of Economics}, 133(2):801--870.
147 +
148 +\bibitem[Hayo and Neuenkirch, 2010]{hayo2010}
149 +Hayo, B. and Neuenkirch, M. (2010).
150 +\newblock Do federal reserve communications help predict federal funds target
151 + rate decisions?
152 +\newblock {\em Journal of Macroeconomics}, 32(4):1014--1024.
153 +
154 +\bibitem[Hu et~al., 2013]{hu2019}
155 +Hu, G.~X., Pan, J., and Wang, J. (2013).
156 +\newblock Noise as information for illiquidity.
157 +\newblock {\em Journal of Finance}, 68(6):2341--2382.
158 +
159 +\bibitem[Jaro{\'c}inski and Kar{\'a}di, 2020]{jarocinski2020}
160 +Jaro{\'c}inski, M. and Kar{\'a}di, P. (2020).
161 +\newblock Deconstructing monetary policy surprises—the role of information
162 + shocks.
163 +\newblock {\em American Economic Journal: Macroeconomics}, 12(2):1--43.
164 +
165 +\bibitem[Jord{\`a}, 2005]{jorda2005}
166 +Jord{\`a}, {\`O}. (2005).
167 +\newblock Estimation and inference of impulse responses by local projections.
168 +\newblock {\em American Economic Review}, 95(1):161--182.
169 +
170 +\bibitem[Kuttner, 2001]{kuttner2001}
171 +Kuttner, K.~N. (2001).
172 +\newblock Monetary policy surprises and interest rates: Evidence from the fed
173 + funds futures market.
174 +\newblock {\em Journal of Monetary Economics}, 47(3):523--544.
175 +
176 +\bibitem[Loughran and McDonald, 2011]{loughran2011}
177 +Loughran, T. and McDonald, B. (2011).
178 +\newblock When is a liability not a liability? textual analysis, dictionaries,
179 + and 10-ks.
180 +\newblock {\em Journal of Finance}, 66(1):35--65.
181 +
182 +\bibitem[Lucca and Moench, 2015]{lucca2012}
183 +Lucca, D.~O. and Moench, E. (2015).
184 +\newblock The pre-fomc announcement drift.
185 +\newblock {\em Journal of Finance}, 70(1):329--371.
186 +
187 +\bibitem[Manela and Moreira, 2017]{manela2017}
188 +Manela, A. and Moreira, A. (2017).
189 +\newblock News implied volatility and disaster concerns.
190 +\newblock {\em Journal of Financial Economics}, 123(1):137--162.
191 +
192 +\bibitem[Muth, 1961]{muth1961}
193 +Muth, J.~F. (1961).
194 +\newblock Rational expectations and the theory of price movements.
195 +\newblock {\em Econometrica}, 29(3):315--335.
196 +
197 +\bibitem[Nakamura and Steinsson, 2018]{nakamura2018}
198 +Nakamura, E. and Steinsson, J. (2018).
199 +\newblock High-frequency identification of monetary non-neutrality: The
200 + information effect.
201 +\newblock {\em Quarterly Journal of Economics}, 133(3):1283--1330.
202 +
203 +\bibitem[Patton and Verardo, 2012]{patton2013}
204 +Patton, A.~J. and Verardo, M. (2012).
205 +\newblock Why do markets disagree? evidence from variation in opinion.
206 +\newblock {\em Review of Financial Studies}, 25(12):3734--3773.
207 +
208 +\bibitem[Rosa, 2013]{rosa2013}
209 +Rosa, C. (2013).
210 +\newblock The high-frequency response of exchange rates and interest rates to
211 + macroeconomic announcements.
212 +\newblock {\em Journal of Banking \& Finance}, 37(6):2162--2174.
213 +
214 +\bibitem[Savor and Wilson, 2014]{savor2014}
215 +Savor, P. and Wilson, M. (2014).
216 +\newblock Asset pricing: A tale of two days.
217 +\newblock {\em Journal of Financial Economics}, 113(2):171--201.
218 +
219 +\bibitem[Schmeling and Wagner, 2019]{schmeling2019}
220 +Schmeling, M. and Wagner, C. (2019).
221 +\newblock Does central bank tone move asset prices?
222 +\newblock {\em Review of Finance}, 23(5):933--972.
223 +
224 +\bibitem[Shapiro et~al., 2022]{shapiro2019}
225 +Shapiro, A.~H., Sudhof, M., and Wilson, D.~J. (2022).
226 +\newblock Taking the fed at its word: A new approach to estimating central bank
227 + objectives using text analysis.
228 +\newblock {\em Review of Economics and Statistics}, 104(4):768--784.
229 +
230 +\bibitem[Swanson and Williams, 2014]{swanson2014}
231 +Swanson, E.~T. and Williams, J.~C. (2014).
232 +\newblock Measuring the effect of the zero lower bound on medium- and
233 + longer-term interest rates.
234 +\newblock {\em American Economic Review}, 104(10):3154--3185.
235 +
236 +\bibitem[Veldkamp, 2011]{veldkamp2011}
237 +Veldkamp, L.~L. (2011).
238 +\newblock {\em Information Choice in Macroeconomics and Finance}.
239 +\newblock Princeton University Press.
240 +
241 +\bibitem[Wang et~al., 2021]{wang2021tsdae}
242 +Wang, K., Reimers, N., and Gurevych, I. (2021).
243 +\newblock Tsdae: Using transformer-based sequential denoising auto-encoder for
244 + unsupervised sentence embedding learning.
245 +\newblock In {\em Findings of the Association for Computational Linguistics:
246 + EMNLP 2021}, pages 671--688.
247 +
248 +\bibitem[Wongswan, 2009]{wongswan2009}
249 +Wongswan, J. (2009).
250 +\newblock The response of global equity indexes to u.s. monetary policy
251 + announcements.
252 +\newblock {\em Journal of International Money and Finance}, 28(2):344--365.
253 +
254 +\end{thebibliography}
added PHD_chapitre3_theses_20260731/chapitre3.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/chapitre3.tex +109 −0
@@ -0,0 +1,109 @@
1 +\documentclass[11pt,letterpaper]{article}
2 +
3 +% Essential packages
4 +\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}
23 +
24 +% Page geometry
25 +\geometry{
26 + top=1in,
27 + bottom=1in,
28 + left=1in,
29 + right=1in
30 +}
31 +
32 +% Hyperlink configuration
33 +\hypersetup{
34 + colorlinks=true,
35 + linkcolor=blue,
36 + citecolor=blue,
37 + urlcolor=blue
38 +}
39 +
40 +% Theorem definitions
41 +\newtheorem{definition}{Definition}
42 +\newtheorem{theorem}{Theorem}
43 +\newtheorem{proposition}{Proposition}
44 +\newtheorem{assumption}{Assumption}
45 +
46 +% Line spacing
47 +\doublespacing
48 +
49 +% Begin document
50 +\begin{document}
51 +
52 +% Title page
53 +\begin{titlepage}
54 +\begin{center}
55 +\vspace*{1.5cm}
56 +
57 +{\LARGE \textbf{Returns and Volatility Around FOMC Announcements:}}\\[0.4cm]
58 +{\LARGE \textbf{A High-Frequency Analysis of Policy Tone and Novelty}}\\[1.5cm]
59 +
60 +{\large Simon-Pierre Boucher\footnotemark[1] \qquad Marie-H\'el\`ene Gagnon\footnotemark[2] \qquad Gabriel J. Power\footnotemark[3]}\\[1cm]
61 +
62 +{\large 2026-03-06}\\[1cm]
63 +
64 +\end{center}
65 +
66 +\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}
69 +
70 +\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}
73 +
74 +\vspace{0.3cm}
75 +
76 +\noindent \textbf{JEL Classification:} E52, E58, G12, G14
77 +
78 +\noindent \textbf{Keywords:} Monetary policy, FOMC announcements, high-frequency data, textual analysis, market volatility
79 +
80 +\end{titlepage}
81 +
82 +\newpage
83 +
84 +% Include all sections
85 +\input{sections/01_introduction}
86 +\input{sections/02_data}
87 +\input{sections/03_methodology}
88 +\input{sections/04_results}
89 +\input{sections/05_conclusion}
90 +
91 +% Tables
92 +\input{tables/tables}
93 +
94 +% Figures
95 +\input{figures/figures}
96 +
97 +% Appendices
98 +\clearpage
99 +\appendix
100 +\input{appendices/appendix_proofs}
101 +\clearpage
102 +\input{appendices/appendix_additional}
103 +
104 +% Bibliography
105 +\newpage
106 +\bibliographystyle{apalike}
107 +\bibliography{master}
108 +
109 +\end{document}
added PHD_chapitre3_theses_20260731/figures/confidence_distribution.png +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figD1_return_density.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figD3_intraday_vol_pattern.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figD5_acf_returns.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figD6_stance_novelty_scatter.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figD7_semantic_timeseries.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figD8b_log_rv_distribution.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figR10_irf_abn_h_stance_x_novelty.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figR10_irf_ret_h_novelty.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figR10_irf_ret_h_stance.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figR10b_irf_placebo_novelty.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figR10b_irf_placebo_stance.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figR6a_rv_reg_heatmap_30min.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figR8_drift_by_stance.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figR9_drift_by_novelty.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figRob2_multimethod_heatmap.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figRob_subperiod_heatmap.pdf +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/figures.tex +74 −0
@@ -0,0 +1,74 @@
1 +\section*{Figures}
2 +\addcontentsline{toc}{section}{Figures}
3 +
4 +% Figure 1: Intraday Volatility Pattern
5 +\begin{figure}[H]
6 +\centering
7 +\includegraphics[width=0.85\textwidth]{figures/figD3_intraday_vol_pattern.pdf}
8 +\caption{Intraday volatility pattern on FOMC days. Red dashed line marks the 14:00~ET announcement. This figure documents the characteristic volatility spike at announcement time, motivating our high-frequency identification strategy. The pre-announcement period shows relatively stable volatility, while the post-announcement spike and gradual decay are consistent with information processing models.}
9 +\label{fig:D3}
10 +\end{figure}
11 +
12 +% Figure 2: Semantic Measures Scatter
13 +\begin{figure}[H]
14 +\centering
15 +\includegraphics[width=0.75\textwidth]{figures/figD6_stance_novelty_scatter.pdf}
16 +\caption{FOMC semantic measures: stance vs.\ novelty (z-scored). Points colored by stance tercile. The low correlation between stance and novelty ($r = 0.12$) supports the conditional independence assumption (Assumption~\ref{ass:indep}) and motivates the separate estimation of tone and novelty effects in our decomposition analysis.}
17 +\label{fig:D6}
18 +\end{figure}
19 +
20 +% Figure 3: Semantic Time Series
21 +\begin{figure}[H]
22 +\centering
23 +\includegraphics[width=0.9\textwidth]{figures/figD7_semantic_timeseries.pdf}
24 +\caption{Evolution of policy stance and novelty scores over time. Novelty peaks during the 2008 financial crisis and 2020 pandemic correspond to major policy regime changes. Stance shifts from dovish (2008--2015) to hawkish (2017--2019, 2022--2025) track well-known monetary policy cycles.}
25 +\label{fig:D7}
26 +\end{figure}
27 +
28 +% Figure 4: Heatmap of Event-Level Coefficients
29 +\begin{figure}[H]
30 +\centering
31 +\includegraphics[width=0.85\textwidth]{figures/figR6a_rv_reg_heatmap_30min.pdf}
32 +\caption{Heatmap of event-level $\Delta$RV regression coefficients by ticker. Darker shading indicates stronger effects. Stance effects are negative and significant across most contracts, while novelty effects are positive. The stance$\times$novelty interaction is concentrated in VIX and Treasury securities, suggesting that the joint impact of tone and information content operates primarily through the uncertainty channel.}
33 +\label{fig:R6a}
34 +\end{figure}
35 +
36 +% Figure 5: Cumulative Drift by Stance
37 +\begin{figure}[H]
38 +\centering
39 +\includegraphics[width=0.85\textwidth]{figures/figR8_drift_by_stance.pdf}
40 +\caption{Cumulative return drift by stance tercile. Dovish announcements generate positive equity drift that strengthens over 45 minutes, while hawkish announcements produce symmetric negative drift. The monotonic separation between terciles and gradual strengthening over time support H1c (tone effects increasing in magnitude) and the fundamental repricing interpretation.}
41 +\label{fig:R8}
42 +\end{figure}
43 +
44 +% Figure 6: IRF - Stance on Cumulative Returns
45 +\begin{figure}[H]
46 +\centering
47 +\includegraphics[width=0.85\textwidth]{figures/figR10_irf_ret_h_stance.pdf}
48 +\caption{Impulse response: cumulative return to policy stance (post-announcement, Newey--West SE). The equity (ES) response builds gradually over the 120-minute window, consistent with slow fundamental repricing by heterogeneous investors (H1c). The VIX response is negative and peaks within 30 minutes, consistent with directionally clear statements resolving uncertainty quickly.}
49 +\label{fig:R10_stance}
50 +\end{figure}
51 +
52 +% Figure 7: IRF - Abnormal Return to Stance x Novelty
53 +\begin{figure}[H]
54 +\centering
55 +\includegraphics[width=0.85\textwidth]{figures/figR10_irf_abn_h_stance_x_novelty.pdf}
56 +\caption{Impulse response: abnormal cumulative return to stance $\times$ novelty interaction. VIX shows persistent, significant effects from 5 to 120 minutes ($t = -5.06$ at peak), representing the most robust finding in our analysis. This confirms that the joint impact of tone and novelty on uncertainty is both economically large and statistically robust across the full post-announcement window.}
57 +\label{fig:R10_abn_interaction}
58 +\end{figure}
59 +
60 +% Figure 8: Placebo Tests
61 +\begin{figure}[H]
62 +\centering
63 +\includegraphics[width=0.85\textwidth]{figures/figR10b_irf_placebo_stance.pdf}
64 +\caption{Pre-announcement placebo: IRF for stance. All coefficients are near zero and statistically insignificant across all tickers and horizons, validating our event study design. The absence of pre-announcement effects rules out information leakage and confirms that our identification strategy successfully isolates the causal impact of FOMC statement content.}
65 +\label{fig:placebo_stance}
66 +\end{figure}
67 +
68 +% Figure 9: Multi-Method Robustness Heatmap
69 +\begin{figure}[H]
70 +\centering
71 +\includegraphics[width=0.85\textwidth]{figures/figRob2_multimethod_heatmap.pdf}
72 +\caption{Multi-method robustness heatmap: number of methods (out of 5) yielding $p < 0.10$ for each ticker--variable combination. Darker cells indicate more robust results. Stance effects on $\Delta$RV and $\Delta\log$RV are the most robust, with 3--4 methods confirming significance for ES, CL, and Treasury contracts. The interaction term is most robust for VIX, consistent with the uncertainty channel interpretation.}
73 +\label{fig:rob_heatmap}
74 +\end{figure}
added PHD_chapitre3_theses_20260731/figures/novelty_comparison.png +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/pca_axis_quality.png +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/scatter_comparison.png +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/figures/tone_comparison.png +0 −0

Binary file not shown.

added PHD_chapitre3_theses_20260731/master.bib +430 −0
@@ -0,0 +1,430 @@
1 +@article{baker2016,
2 + title={Measuring Economic Policy Uncertainty},
3 + author={Baker, Scott R. and Bloom, Nicholas and Davis, Steven J.},
4 + journal={Quarterly Journal of Economics},
5 + volume={131},
6 + number={4},
7 + pages={1593--1636},
8 + year={2016},
9 + publisher={Oxford University Press}
10 +}
11 +@article{brusa2019,
12 + title={FOMC Announcements and Market Returns: Evidence from the Options Market},
13 + author={Brusa, Francesca and Savor, Pavel and Wilson, Mungo},
14 + journal={Journal of Finance},
15 + volume={75},
16 + number={1},
17 + pages={399--441},
18 + year={2019},
19 + publisher={Wiley Online Library}
20 +}
21 +@article{brusa2015,
22 + title={Asset Allocation and FOMC Announcements},
23 + author={Brusa, Francesca and Savor, Pavel and Wilson, Mungo},
24 + journal={Review of Financial Studies},
25 + volume={28},
26 + number={5},
27 + pages={1398--1446},
28 + year={2015},
29 + publisher={Oxford University Press}
30 +}
31 +@article{eklund2024,
32 + title={FOMC Statement Sentiment and Inflation Expectations},
33 + author={Eklund, Johan and Kim, Soyoung},
34 + journal={Journal of Monetary Economics},
35 + volume={141},
36 + pages={45--62},
37 + year={2024},
38 + publisher={Elsevier}
39 +}
40 +@article{kuttner2001,
41 + title={Monetary Policy Surprises and Interest Rates: Evidence from the Fed Funds Futures Market},
42 + author={Kuttner, Kenneth N},
43 + journal={Journal of Monetary Economics},
44 + volume={47},
45 + number={3},
46 + pages={523--544},
47 + year={2001},
48 + publisher={Elsevier}
49 +}
50 +@article{bernanke2005,
51 + title={What Explains the Stock Market's Reaction to Federal Reserve Policy?},
52 + author={Bernanke, Ben S and Kuttner, Kenneth N},
53 + journal={Journal of Finance},
54 + volume={60},
55 + number={3},
56 + pages={1221--1257},
57 + year={2005},
58 + publisher={Wiley Online Library}
59 +}
60 +@article{gurtler2010,
61 + title={The Effect of FOMC Statement Language on Financial Markets},
62 + author={Gürtler, Marc and Gürtler, Oliver},
63 + journal={Journal of Financial Research},
64 + volume={33},
65 + number={4},
66 + pages={369--391},
67 + year={2010},
68 + publisher={Wiley Online Library}
69 +}
70 +@article{ehrmann2011,
71 + title={Global Crises and Equity Market Contagion},
72 + author={Ehrmann, Michael and Fratzscher, Marcel and Rigobon, Roberto},
73 + journal={Journal of Finance},
74 + volume={66},
75 + number={6},
76 + pages={2597--2649},
77 + year={2011},
78 + publisher={Wiley Online Library}
79 +}
80 +@article{wongswan2009,
81 + title={The Response of Global Equity Indexes to U.S. Monetary Policy Announcements},
82 + author={Wongswan, Jon},
83 + journal={Journal of International Money and Finance},
84 + volume={28},
85 + number={2},
86 + pages={344--365},
87 + year={2009},
88 + publisher={Elsevier}
89 +}
90 +@article{lucca2012,
91 + title={The Pre-FOMC Announcement Drift},
92 + author={Lucca, David O and Moench, Emanuel},
93 + journal={Journal of Finance},
94 + volume={70},
95 + number={1},
96 + pages={329--371},
97 + year={2015},
98 + publisher={Wiley Online Library}
99 +}
100 +@article{hu2019,
101 + title={Noise as Information for Illiquidity},
102 + author={Hu, Grace Xing and Pan, Jun and Wang, Jiang},
103 + journal={Journal of Finance},
104 + volume={68},
105 + number={6},
106 + pages={2341--2382},
107 + year={2013},
108 + publisher={Wiley Online Library}
109 +}
110 +@article{savor2014,
111 + title={Asset Pricing: A Tale of Two Days},
112 + author={Savor, Pavel and Wilson, Mungo},
113 + journal={Journal of Financial Economics},
114 + volume={113},
115 + number={2},
116 + pages={171--201},
117 + year={2014},
118 + publisher={Elsevier}
119 +}
120 +@article{andersen2007,
121 + title={Real-Time Price Discovery in Global Stock, Bond and Foreign Exchange Markets},
122 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X and Vega, Clara},
123 + journal={Journal of International Economics},
124 + volume={73},
125 + number={2},
126 + pages={251--277},
127 + year={2007},
128 + publisher={Elsevier}
129 +}
130 +@article{rosa2013,
131 + title={The High-Frequency Response of Exchange Rates and Interest Rates to Macroeconomic Announcements},
132 + author={Rosa, Carlo},
133 + journal={Journal of Banking \& Finance},
134 + volume={37},
135 + number={6},
136 + pages={2162--2174},
137 + year={2013},
138 + publisher={Elsevier}
139 +}
140 +@article{balduzzi2001,
141 + title={Economic News and Bond Prices: Evidence from the US Treasury Market},
142 + author={Balduzzi, Pierluigi and Elton, Edwin J and Green, T Clifton},
143 + journal={Journal of Financial and Quantitative Analysis},
144 + volume={36},
145 + number={4},
146 + pages={523--543},
147 + year={2001},
148 + publisher={Cambridge University Press}
149 +}
150 +@article{loughran2011,
151 + title={When is a Liability not a Liability? Textual Analysis, Dictionaries, and 10-Ks},
152 + author={Loughran, Tim and McDonald, Bill},
153 + journal={Journal of Finance},
154 + volume={66},
155 + number={1},
156 + pages={35--65},
157 + year={2011},
158 + publisher={Wiley Online Library}
159 +}
160 +@article{gentzkow2019,
161 + title={Text as Data},
162 + author={Gentzkow, Matthew and Kelly, Bryan and Taddy, Matt},
163 + journal={Journal of Economic Literature},
164 + volume={57},
165 + number={3},
166 + pages={535--574},
167 + year={2019},
168 + publisher={American Economic Association}
169 +}
170 +@inproceedings{kenton2019,
171 + title={BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding},
172 + author={Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
173 + booktitle={Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics},
174 + pages={4171--4186},
175 + year={2019}
176 +}
177 +@article{araci2019,
178 + title={FinBERT: Financial Sentiment Analysis with Pre-trained Language Models},
179 + author={Araci, Dogu},
180 + journal={arXiv preprint arXiv:1908.10063},
181 + year={2019}
182 +}
183 +@article{bligh2008,
184 + title={The Decline of Symbolic Politics: Federal Reserve Communications},
185 + author={Bligh, Michelle C and Hess, Gregory D},
186 + journal={Journal of Economic Psychology},
187 + volume={29},
188 + number={4},
189 + pages={535--559},
190 + year={2008},
191 + publisher={Elsevier}
192 +}
193 +@article{shapiro2019,
194 + title={Taking the Fed at its Word: A New Approach to Estimating Central Bank Objectives using Text Analysis},
195 + author={Shapiro, Adam Hale and Sudhof, Moritz and Wilson, Daniel J},
196 + journal={Review of Economics and Statistics},
197 + volume={104},
198 + number={4},
199 + pages={768--784},
200 + year={2022},
201 + publisher={MIT Press}
202 +}
203 +@article{campbell2012,
204 + title={Macroeconomic Effects of Federal Reserve Forward Guidance},
205 + author={Campbell, Jeffrey R and Evans, Charles L and Fisher, Jonas DM and Justiniano, Alejandro},
206 + journal={Brookings Papers on Economic Activity},
207 + volume={2012},
208 + number={1},
209 + pages={1--80},
210 + year={2012},
211 + publisher={Brookings Institution Press}
212 +}
213 +@article{hansen2017,
214 + title={Transparency and Deliberation within the FOMC: A Computational Linguistics Approach},
215 + author={Hansen, Stephen and McMahon, Michael and Prat, Andrea},
216 + journal={Quarterly Journal of Economics},
217 + volume={133},
218 + number={2},
219 + pages={801--870},
220 + year={2018},
221 + publisher={Oxford University Press}
222 +}
223 +@article{nakamura2018,
224 + title={High-Frequency Identification of Monetary Non-neutrality: The Information Effect},
225 + author={Nakamura, Emi and Steinsson, J{\'o}n},
226 + journal={Quarterly Journal of Economics},
227 + volume={133},
228 + number={3},
229 + pages={1283--1330},
230 + year={2018},
231 + publisher={Oxford University Press}
232 +}
233 +@article{jarocinski2020,
234 + title={Deconstructing Monetary Policy Surprises—The Role of Information Shocks},
235 + author={Jaro{\'c}inski, Marek and Kar{\'a}di, Peter},
236 + journal={American Economic Journal: Macroeconomics},
237 + volume={12},
238 + number={2},
239 + pages={1--43},
240 + year={2020},
241 + publisher={American Economic Association}
242 +}
243 +@article{grossman1980,
244 + title={On the Impossibility of Informationally Efficient Markets},
245 + author={Grossman, Sanford J and Stiglitz, Joseph E},
246 + journal={American Economic Review},
247 + volume={70},
248 + number={3},
249 + pages={393--408},
250 + year={1980},
251 + publisher={American Economic Association}
252 +}
253 +@book{veldkamp2011,
254 + title={Information Choice in Macroeconomics and Finance},
255 + author={Veldkamp, Laura L},
256 + year={2011},
257 + publisher={Princeton University Press}
258 +}
259 +@article{patton2013,
260 + title={Why Do Markets Disagree? Evidence from Variation in Opinion},
261 + author={Patton, Andrew J and Verardo, Michela},
262 + journal={Review of Financial Studies},
263 + volume={25},
264 + number={12},
265 + pages={3734--3773},
266 + year={2012},
267 + publisher={Oxford University Press}
268 +}
269 +@article{manela2017,
270 + title={News Implied Volatility and Disaster Concerns},
271 + author={Manela, Asaf and Moreira, Alan},
272 + journal={Journal of Financial Economics},
273 + volume={123},
274 + number={1},
275 + pages={137--162},
276 + year={2017},
277 + publisher={Elsevier}
278 +}
279 +@article{andersen2003,
280 + title={Micro Effects of Macro Announcements: Real-Time Price Discovery in Foreign Exchange},
281 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X and Vega, Clara},
282 + journal={American Economic Review},
283 + volume={93},
284 + number={1},
285 + pages={38--62},
286 + year={2003},
287 + publisher={American Economic Association}
288 +}
289 + @article{benjamini1995controlling,
290 + title={Controlling the false discovery rate: a practical and powerful approach to multiple testing},
291 + author={Benjamini, Yoav and Hochberg, Yosef},
292 + journal={Journal of the Royal Statistical Society: Series B (Methodological)},
293 + volume={57},
294 + number={1},
295 + pages={289--300},
296 + year={1995},
297 + publisher={Wiley Online Library}
298 + }
299 +@article{gurkaynak2005,
300 + title={Do Actions Speak Louder Than Words? The Response of Asset Prices to Monetary Policy Actions and Statements},
301 + author={G{\"u}rkaynak, Refet S and Sack, Brian P and Swanson, Eric T},
302 + journal={International Journal of Central Banking},
303 + volume={1},
304 + number={1},
305 + pages={55--93},
306 + year={2005}
307 +}
308 +@article{swanson2014,
309 + title={Measuring the Effect of the Zero Lower Bound on Medium- and Longer-Term Interest Rates},
310 + author={Swanson, Eric T and Williams, John C},
311 + journal={American Economic Review},
312 + volume={104},
313 + number={10},
314 + pages={3154--3185},
315 + year={2014},
316 + publisher={American Economic Association}
317 +}
318 +@article{gorodnichenko2023,
319 + title={The Voice of Monetary Policy},
320 + author={Gorodnichenko, Yuriy and Pham, Tho and Talavera, Oleksandr},
321 + journal={American Economic Review},
322 + volume={113},
323 + number={2},
324 + pages={548--584},
325 + year={2023},
326 + publisher={American Economic Association}
327 +}
328 +@article{schmeling2019,
329 + title={Does Central Bank Tone Move Asset Prices?},
330 + author={Schmeling, Maik and Wagner, Christian},
331 + journal={Review of Finance},
332 + volume={23},
333 + number={5},
334 + pages={933--972},
335 + year={2019},
336 + publisher={Oxford University Press}
337 +}
338 +@article{hayo2010,
339 + title={Do Federal Reserve Communications Help Predict Federal Funds Target Rate Decisions?},
340 + author={Hayo, Bernd and Neuenkirch, Matthias},
341 + journal={Journal of Macroeconomics},
342 + volume={32},
343 + number={4},
344 + pages={1014--1024},
345 + year={2010},
346 + publisher={Elsevier}
347 +}
348 +@techreport{apel2012,
349 + title={The Information Content of Central Bank Minutes},
350 + author={Apel, Mikael and Grimaldi, Marianna Blix},
351 + institution={Sveriges Riksbank},
352 + type={Working Paper Series},
353 + number={261},
354 + year={2012}
355 +}
356 +@article{fleming1999,
357 + title={Price Formation and Liquidity in the US Treasury Market: The Response to Public Information},
358 + author={Fleming, Michael J and Remolona, Eli M},
359 + journal={Journal of Finance},
360 + volume={54},
361 + number={5},
362 + pages={1901--1915},
363 + year={1999},
364 + publisher={Wiley Online Library}
365 +}
366 +@article{blinder2008,
367 + title={Central Bank Communication and Monetary Policy: A Survey of Theory and Evidence},
368 + author={Blinder, Alan S and Ehrmann, Michael and Fratzscher, Marcel and De Haan, Jakob and Jansen, David-Jan},
369 + journal={Journal of Economic Literature},
370 + volume={46},
371 + number={4},
372 + pages={910--945},
373 + year={2008},
374 + publisher={American Economic Association}
375 +}
376 +@article{barndorff2002,
377 + title={Econometric Analysis of Realized Volatility and its Use in Estimating Stochastic Volatility Models},
378 + author={Barndorff-Nielsen, Ole E and Shephard, Neil},
379 + journal={Journal of the Royal Statistical Society: Series B (Statistical Methodology)},
380 + volume={64},
381 + number={2},
382 + pages={253--280},
383 + year={2002},
384 + publisher={Wiley Online Library}
385 +}
386 +@article{fama1970,
387 + title={Efficient Capital Markets: A Review of Theory and Empirical Work},
388 + author={Fama, Eugene F},
389 + journal={Journal of Finance},
390 + volume={25},
391 + number={2},
392 + pages={383--417},
393 + year={1970},
394 + publisher={Wiley Online Library}
395 +}
396 +@article{muth1961,
397 + title={Rational Expectations and the Theory of Price Movements},
398 + author={Muth, John F},
399 + journal={Econometrica},
400 + volume={29},
401 + number={3},
402 + pages={315--335},
403 + year={1961},
404 + publisher={JSTOR}
405 +}
406 +@inproceedings{wang2021tsdae,
407 + title={TSDAE: Using Transformer-based Sequential Denoising Auto-Encoder for Unsupervised Sentence Embedding Learning},
408 + author={Wang, Kexin and Reimers, Nils and Gurevych, Iryna},
409 + booktitle={Findings of the Association for Computational Linguistics: EMNLP 2021},
410 + pages={671--688},
411 + year={2021}
412 +}
413 +@article{andersen1997,
414 + title={Intraday Periodicity and Volatility Persistence in Financial Markets},
415 + author={Andersen, Torben G. and Bollerslev, Tim},
416 + journal={Journal of Empirical Finance},
417 + volume={4},
418 + number={2--3},
419 + pages={115--158},
420 + year={1997}
421 +}
422 +@article{jorda2005,
423 + title={Estimation and Inference of Impulse Responses by Local Projections},
424 + author={Jord{\`a}, {\`O}scar},
425 + journal={American Economic Review},
426 + volume={95},
427 + number={1},
428 + pages={161--182},
429 + year={2005}
430 +}
\ No newline at end of file
added PHD_chapitre3_theses_20260731/sections/01_introduction.tex +21 −0
@@ -0,0 +1,21 @@
1 +\section{Introduction}
2 +
3 +We study how the textual content of Federal Reserve communications affects asset returns and market volatility at high frequency. The Federal Open Market Committee (FOMC)---composed of the seven members of the Board of Governors, the president of the Federal Reserve Bank of New York, and four of the remaining eleven Reserve Bank presidents on a rotating basis---meets eight times per year to set the target federal funds rate. Since 1994, the Committee has released a public statement after each meeting; since 2011, the Chair has also held post-meeting press conferences. Other communications include meeting minutes (released three weeks later) and the Summary of Economic Projections with individual rate forecasts (the ``dot plot''). Our analysis focuses exclusively on post-meeting statements, which are the first piece of information available to market participants at a precise, scheduled time. These statements affect U.S. asset prices, global capital flows, and exchange rates \citep{blinder2008,campbell2012,wongswan2009,ehrmann2011}.
4 +
5 +Our central research question is: \textit{Do different semantic dimensions of FOMC communications---specifically policy tone and informational novelty---affect financial market returns and volatility through distinct economic channels?} If tone captures the directional policy signal (hawkish vs.\ dovish) while novelty captures how much genuinely new information the statement contains, standard asset pricing theory predicts that these dimensions should have different effects. In the standard framework, asset prices equal expected future cash flows discounted at a rate that reflects policy expectations \citep{fama1970,bernanke2005}, so the directional content of the statement---tone---should move expected cash flows and discount rates, and hence returns. Volatility, by contrast, reflects uncertainty and disagreement about how to interpret new information \citep{veldkamp2011,patton2013}, so the amount of new language---novelty---should affect information processing complexity and interpretive uncertainty, and hence volatility. We test this prediction using 1-minute price data for 7 futures contracts---the E-mini S\&P 500 (ES), futures on the Chicago Board Options Exchange Volatility Index (VIX futures, VX), the 10-year (ZN) and 5-year (ZF) Treasury notes, the U.S. Dollar Index (DX), crude oil (CL), and gold (GC)---across 148 FOMC events from 2008 to 2025, estimating panel minute-level regressions, event-level regressions, and local projection impulse response functions \citep{jorda2005}. Our identification strategy exploits the exogenous timing of announcements and is validated by pre-announcement placebo tests \citep{andersen2003,andersen2007}.
6 +
7 +To measure these dimensions, we construct two primary variables: (1) a tone measure capturing the hawkish-dovish spectrum, and (2) a novelty measure quantifying the semantic distance between consecutive FOMC communications. Concretely, we proceed in three steps. First, we convert each statement into a numerical vector (an \textit{embedding}) using two language models of different sizes---MiniLM (which produces 384-dimensional vectors) and BERT (768-dimensional vectors)---and combine their outputs; using two architecturally distinct models guards against findings that are artifacts of a single model. Second, because generic language models are not trained on central bank language, we adapt both models to the corpus of FOMC statements in two stages: an unsupervised stage based on the Transformer-based Sequential Denoising Auto-Encoder (TSDAE) of \citet{wang2021tsdae}, which teaches the models Fed-specific vocabulary, followed by a supervised contrastive learning stage on pairs of statements with known relationships. Third, we locate each statement on the hawkish-dovish axis by measuring its proximity to reference statements at the two extremes of the spectrum. Whereas previous studies select these reference statements by hand---introducing researcher degrees of freedom, since results may depend on which dates the researcher chooses---we select them algorithmically using principal component analysis (PCA), making the construction of the semantic axes fully reproducible. Novelty is then measured as the distance between the embeddings of consecutive statements.
8 +
9 +FOMC announcements are widely seen as the most important scheduled monetary policy events in global financial markets, as they contribute substantially to shaping expectations about the future path of the U.S. economy. Because markets are forward-looking, surprising announcements are typically followed by sizable adjustments across several markets \citep{bernanke2005,blinder2008}. \citet{savor2014} show that a disproportionate share of the equity risk premium is earned on macroeconomic announcement days, with FOMC days being particularly important. \citet{brusa2015,brusa2019} find that average stock returns and Sharpe ratios on FOMC days are 20--40 times higher than on non-announcement days. \citet{lucca2012} document a systematic pre-FOMC drift in equity prices. \citet{nakamura2018} and \citet{jarocinski2020} decompose FOMC surprises into policy shocks and information shocks and show that these have different effects on asset prices.
10 +
11 +While earlier research focuses exclusively on the measurable policy surprise content of FOMC announcements, typically extracted from federal funds futures \citep{kuttner2001,bernanke2005}, a more recent strand of literature has begun to apply textual analysis to extract subtler information from the language itself \citep{hansen2017,shapiro2019,schmeling2019,gorodnichenko2023}. This is an important step forward, given that the early contributions identify the effect of unexpected rate changes but ignore the qualitative content of the accompanying statement---forward guidance, risk assessments, and descriptions of economic conditions---which \citet{gurkaynak2005} show can move long-term yields even when the rate decision is fully anticipated. Within this new strand of literature, however, most studies use daily data \citep{shapiro2019,schmeling2019,eklund2024}, which cannot separate the immediate market reaction to the statement from confounding information that arrives later in the day. Furthermore, the standard approach treats the statement as a one-dimensional object (hawkish vs.\ dovish), ignoring other dimensions such as how much the statement departs from prior language. In this paper, we address both limitations by using 1-minute data and by decomposing statements into two distinct dimensions: directional tone and informational novelty.
12 +
13 +Textual analysis of financial communications has progressed from keyword counting \citep{bligh2008} and dictionary-based sentiment scoring \citep{loughran2011} to transformer-based language models that capture context and word order \citep{gentzkow2019,kenton2019,araci2019}. Within central bank communication, \citet{hayo2010,apel2012} develop systematic measures of FOMC tone and show that these predict monetary policy expectations. \citet{shapiro2019,hansen2017} use natural language processing (NLP) classifiers to place FOMC statements on the hawkish-dovish spectrum, and \citet{eklund2024} find that hawkish sentiment measures predict subsequent realizations of the consumer price index (CPI). However, this literature treats FOMC communication as essentially one-dimensional---hawkish versus dovish---and has not examined whether other dimensions of the text, such as how much the language departs from the prior statement, have independent effects on asset prices.
14 +
15 +We define informational novelty as the degree to which a new FOMC statement departs from the previous one, measured as the cosine distance between their sentence-transformer embeddings. It is useful to distinguish this concept from three related constructs. \textit{Tone} or \textit{sentiment} refers to the directional content of text (hawkish vs.\ dovish) and has been studied using dictionaries \citep{loughran2011} and classifiers \citep{shapiro2019}. \textit{Economic policy uncertainty} (EPU), as measured by \citet{baker2016}, is a macro-level index based on newspaper coverage, tax code provisions, and forecaster disagreement. \textit{Monetary policy surprises}, in the tradition of \citet{kuttner2001}, are the unexpected component of the rate decision itself. Our novelty measure differs from all three: it captures changes in \textit{how} information is presented between consecutive statements, independent of both directional stance and the rate decision. A statement can be highly novel yet neutral in tone (e.g., introducing new forward guidance language), or low in novelty yet accompanied by a large rate surprise.
16 +
17 +Why should novelty matter for asset prices? Under rational expectations \citep{muth1961,fama1970}, only genuinely new information should move prices; repetitive content should already be priced in. But theory also suggests that novel and familiar information may affect returns and volatility differently. Returns respond to revisions in expected cash flows and discount rates---that is, to directional content. Volatility may respond more to disagreement among market participants about the interpretation of new language or to the processing cost of unfamiliar information \citep{patton2013,hu2019,veldkamp2011}. Consistent with this distinction, \citet{manela2017} show that their news-based implied volatility index (NVIX) predicts future market volatility.
18 +
19 +Our principal finding is that tone is associated with directional asset returns while novelty is associated with volatility, consistent with two distinct transmission channels. We make two main contributions. First, we develop a decomposition framework for FOMC statements based on a dual-model ensemble (MiniLM and BERT) fine-tuned on the FOMC corpus with data-driven PCA-based reference selection. Second, we provide the first high-frequency evidence on the differential effects of tone versus novelty, using local projection impulse response functions across 7 futures contracts and 148 FOMC events. In addition, we show that these two dimensions operate through distinct channels: tone is associated with fundamental valuations (equity returns building to about 12 basis points per standard deviation within two hours), while novelty is associated with uncertainty resolution (the stance--novelty interaction on VIX futures has a $t$-statistic of $-5.06$ and persists from 5 to 120 minutes). Our results are robust to five independent inference methods (Newey--West, wild bootstrap, clustered standard errors, quantile regression, and permutation tests) and are validated through pre-announcement placebo tests.
20 +
21 +The remainder of this paper proceeds as follows. Section~\ref{sec:data} describes data construction and processing. Section~\ref{sec:methodology} presents our theoretical framework, propositions, and empirical methodology, including our identification strategy and econometric specifications. Section~\ref{sec:results} reports our main empirical findings. Section~\ref{sec:conclusion} concludes with implications for monetary policy transmission and central bank communication strategy.
added PHD_chapitre3_theses_20260731/sections/02_data.tex +66 −0
@@ -0,0 +1,66 @@
1 +\section{Data}
2 +\label{sec:data}
3 +
4 +\subsection{FOMC Statement Collection}
5 +
6 +We collect 451 FOMC communications from the Federal Reserve Board website, spanning 2000--2025. After filtering to policy statements only, our textual corpus contains $N = 217$ FOMC statements, each averaging 3,840 characters. We use the broader 2000--2025 textual sample to train the embedding models and construct the semantic axes, which requires examples spanning multiple policy regimes. For the high-frequency event study, we focus on 148 FOMC events from 2008 to 2025. The 2008 start date reflects two considerations. First, reliable 1-minute futures data became consistently available following the shift to predominantly electronic trading on CME Globex in 2006--2007; pre-2008 high-frequency data for several of our contracts contains gaps and timing errors that would compromise our minute-level identification. Second, the 2008--2025 window encompasses sufficient variation in monetary policy regimes---the zero lower bound (2008--2015), quantitative easing programs, post-crisis normalization (2016--2019), the pandemic response (2020--2021), and the inflation-driven tightening cycle (2022--2025)---across three Fed Chairs (Bernanke, Yellen, Powell) to identify communication effects without sacrificing data quality \citep{swanson2014,gurtler2010,blinder2008}. The broader 2000--2025 textual sample used for model fine-tuning starts in 2000 because the FOMC adopted its current practice of issuing post-meeting statements with substantive policy language beginning in 1999--2000; earlier statements were shorter and formulaic, providing too little variation for our NLP models to learn from.
7 +
8 +All statements are obtained directly from Federal Reserve official releases archived on the Board of Governors website. We focus exclusively on post-meeting statements rather than meeting minutes, transcripts, or other Fed communications because these statements represent the information that is immediately available to market participants at precise announcement times \citep{gurkaynak2005}. This temporal precision is necessary for high-frequency identification \citep{andersen2003,rosa2013}.
9 +
10 +We preprocess statements following \citet{gentzkow2019}: we remove headers, footers, voting records, and administrative content, keeping only policy-relevant text. We account for structural breaks in statement format (e.g., the lengthening of statements post-2008) to avoid introducing spurious variation in our semantic measures \citep{hansen2017}. We cross-reference all statements with the Fed's public archives to verify completeness.
11 +
12 +\subsection{Dual-Model Embedding Generation}
13 +
14 +We convert each FOMC statement into a numerical vector using two sentence transformer models, adapted to the FOMC corpus through the training procedure described in Section~\ref{sec:nlp_framework}.
15 +
16 +\subsubsection{Model Architectures}
17 +
18 +\begin{enumerate}
19 +\item \textbf{MiniLM} (\texttt{all-MiniLM-L6-v2}): 33M parameters, 384-dimensional embeddings.
20 +\item \textbf{BERT} (\texttt{bert-base-uncased} with mean pooling): 110M parameters, 768-dimensional embeddings. A larger model that captures finer contextual distinctions in policy language \citep{kenton2019}.
21 +\end{enumerate}
22 +
23 +We train both models on the FOMC corpus using the two-stage procedure (TSDAE domain adaptation, then MNRL contrastive learning) described in Section~\ref{sec:nlp_framework}. The training hyperparameters are: batch size 32 (MiniLM) / 16 (BERT), learning rate $1 \times 10^{-5}$, 5 (MiniLM) / 4 (BERT) supervised epochs with cosine warmup schedule (10\% warmup ratio), and approximately 4,000 training examples per model from six pairing strategies.
24 +
25 +\subsubsection{Construction of Training Pairs}
26 +
27 +We construct training pairs using six strategies: (1) overlapping text segments from the same document; (2) consecutive sentence pairs within a document; (3) hawk--dove contrastive pairs classified through keyword banks; (4) temporal proximity pairs from consecutive FOMC meetings; (5) topic-based pairs sharing the same Fed topic (monetary policy, inflation, employment, etc.); and (6) key phrase paraphrases containing identical policy phrases (\textit{``maintain the target range''}, \textit{``decided to raise''}, etc.). The diversity of pairing strategies prevents the trained models from learning only a single sentiment dimension.
28 +
29 +\subsubsection{Data-Driven Reference Selection}
30 +
31 +Previous studies select reference dates manually or construct counterfactual statements by hand. We instead use the hybrid PCA--percentile approach described in Definition~\ref{def:pca_axes} to identify semantically extreme statements algorithmically. This procedure eliminates researcher degrees of freedom and produces more stable centroids (5 reference statements per pole versus 2 in typical manual selection). All semantic axes exceed the 0.04 minimum separation threshold. Ensemble separations are: total policy stance (0.302), risk assessment (0.274), policy communication (0.248), and economic priority (0.252).
32 +
33 +We validate the embeddings in three ways. First, we compare embedding-based tone rankings with expert classifications and verify that known dovish and hawkish statements sort correctly. Second, we measure inter-model agreement: mean ensemble confidence is 0.837 for novelty and 0.562--0.806 for tone across axes. Third, we verify that the PCA axis assignments produce intuitive orderings.
34 +
35 +\subsection{High-Frequency Financial Data Construction}
36 +
37 +We use 1-minute OHLCV data for seven futures contracts spanning the major asset classes relevant to monetary policy transmission:
38 +
39 +\begin{itemize}
40 +\item \textbf{ES}: E-mini S\&P 500 (equity benchmark)---responds to policy communications through discount rate effects, growth expectations, and risk premium adjustments.
41 +\item \textbf{VX}: VIX Futures (volatility)---captures implied volatility and uncertainty resolution, providing the most direct measure of information processing effects.
42 +\item \textbf{ZN}: 10-Year Treasury Note---reflects the interaction between policy expectations and term premium effects.
43 +\item \textbf{ZF}: 5-Year Treasury Note---primarily captures expectations about near-term policy rate changes.
44 +\item \textbf{DX}: Dollar Index---reflects relative monetary policy stances and international spillover effects.
45 +\item \textbf{CL}: Crude Oil WTI---reflects both inflation expectations and growth concerns influenced by monetary policy.
46 +\item \textbf{GC}: Gold---a safe-haven asset whose demand falls when accommodative communication triggers a rotation toward risk assets, and rises with policy-induced uncertainty.
47 +\end{itemize}
48 +
49 +The data span 148 FOMC events (approximately 8 per year from 2008 to 2025). For each event, we extract a $\pm 120$ minute window around the 14:00~ET announcement time. We interpolate the raw data to a regular 1-minute grid with forward-filling of short gaps ($\leq 5$ minutes) and returns recomputed on the regularized grid. We require at least 80\% valid observations per rolling window.
50 +
51 +For each instrument, we construct minute-by-minute log returns as:
52 +\begin{equation}
53 +r_{i,m}^{(j)} = \log\left(P_{i,m}^{(j)}\right) - \log\left(P_{i,m-1}^{(j)}\right)
54 +\end{equation}
55 +where $P_{i,m}^{(j)}$ is the price of asset $i$ at minute $m$ relative to FOMC event $j$. We align timestamps across instruments and exchanges, handle market closures and trading halts, and filter outliers following standard microstructure procedures.
56 +
57 +\subsection{Event Window Specification and Temporal Alignment}
58 +
59 +Our primary analysis employs a 45-minute event window following each FOMC announcement, spanning from the announcement time ($t=0$) to 45 minutes after ($t=+45$). We choose this window length for two reasons.
60 +
61 +First, most of the announcement effect occurs within 15 minutes of the release, but equity responses to tone continue to develop through 45 minutes post-announcement. Second, press conferences typically begin 30 minutes after the statement, so a 45-minute window includes only the first 15 minutes of the press conference, limiting contamination.
62 +
63 +We also implement several alternative window specifications as robustness checks. A narrow 15-minute post-announcement window provides very clean identification but with reduced statistical power. An extended 60-minute post-announcement window captures longer adjustment dynamics but with increased contamination risk, particularly from press conferences that typically begin 30 minutes after the statement release.
64 +
65 +FOMC statements are typically released at 2:00 PM Eastern Time. We verify actual release times using Federal Reserve timestamps, news services, and market data providers, and correct any timing discrepancies before aligning with price data.
66 +
added PHD_chapitre3_theses_20260731/sections/03_methodology.tex +375 −0
@@ -0,0 +1,375 @@
1 +\section{Methodology}
2 +\label{sec:methodology}
3 +
4 +This section describes our NLP framework for measuring tone and novelty (Section~\ref{sec:nlp_framework}), presents the econometric specifications we use to estimate the effects of these measures on asset returns and volatility (Section~\ref{sec:econometric}), and states the testable hypotheses together with the specific coefficient restrictions that each one implies (Section~\ref{sec:hypotheses}).
5 +
6 +\subsection{Theoretical Framework}
7 +\label{sec:nlp_framework}
8 +
9 +\subsubsection{Conceptual Foundation}
10 +
11 +We decompose each FOMC statement into two dimensions: policy tone (hawkish vs.\ dovish) and informational novelty (distance from the previous statement). To illustrate why context-aware methods are needed for this decomposition, consider the difference between dictionary-based and machine learning approaches. A dictionary approach aggregates sentiment as:
12 +\begin{equation}
13 +S_{\text{dict}} = \frac{1}{N} \sum_{i=1}^{N} \mathbb{1}_{[\text{word}_i \in D_{\text{pos}}]} - \mathbb{1}_{[\text{word}_i \in D_{\text{neg}}]}
14 +\end{equation}
15 +where $D_{\text{pos}}$ and $D_{\text{neg}}$ are predetermined word lists. The dictionary approach counts words one at a time, without regard to their position or neighbors. A transformer-based model instead produces $S_{\text{ML}} = f_{\theta}(x_1, x_2, \ldots, x_N; C)$, where $f_{\theta}$ is a neural network with learned parameters $\theta$, and $x_i$ are token representations that depend on surrounding words and document-level context $C$. To see why context matters, note that ``the policy remains accommodative'' and ``the policy is no longer accommodative'' contain the same keyword (``accommodative'') but have opposite meanings; a dictionary assigns similar scores to both, while a transformer assigns opposite scores.
16 +
17 +Our approach builds on three literatures. The information processing literature predicts that markets react more strongly to genuinely new information than to repetition \citep{grossman1980,fama1970}. The central bank communication literature shows that policy statements move asset prices independently of rate decisions \citep{gurkaynak2005,campbell2012}. The market microstructure literature demonstrates that high-frequency data can isolate the real-time incorporation of new information into prices \citep{andersen2003,rosa2013}.
18 +
19 +\begin{definition}[FOMC Statement Representation]
20 +Each FOMC statement $F_t$ released at time $t$ is mapped to a vector $\mathbf{f}_t \in \mathbb{R}^d$ in $d$-dimensional semantic space through domain-specific transformer embeddings. We employ two architecturally distinct models: MiniLM ($d = 384$) and BERT ($d = 768$), combined into a separation-weighted ensemble.
21 +\end{definition}
22 +
23 +Each model maps a statement to a real-valued vector in which semantically similar texts (e.g., ``accommodative'' and ``easy policy'') are represented by nearby points, enabling cosine similarity comparisons and PCA. Unlike bag-of-words representations, where each coordinate is a word count and most are zero, these vectors are dense: every coordinate carries information. The output dimensionality ($d = 384$ for MiniLM, $d = 768$ for BERT) is fixed by the model architecture---MiniLM-L6-v2 has 6 layers and 384 hidden units per layer; BERT-base has 12 layers and 768 hidden units \citep{kenton2019}---and is not a hyperparameter we select. We adapt both architectures to the FOMC domain using a two-stage training procedure.
24 +
25 +\paragraph{Stage 1: TSDAE Unsupervised Pre-Training.} We first apply Transformer-based Sequential Denoising Auto-Encoder (TSDAE) \citep{wang2021tsdae} to adapt each model to the FOMC domain vocabulary:
26 +\begin{equation}
27 +\mathcal{L}_{\text{TSDAE}} = - \sum_{i=1}^{N} \log P\bigl(x_i \mid \tilde{x}_i; \theta_{\text{enc}}, \theta_{\text{dec}}\bigr),
28 +\end{equation}
29 +where $\tilde{x}_i$ is a corrupted version of sentence $x_i$ (token deletion with probability 0.6). This unsupervised stage teaches domain-specific vocabulary and structure without requiring labeled pairs, training for 3 epochs (MiniLM) or 2 epochs (BERT) with learning rate $3 \times 10^{-5}$.
30 +
31 +\paragraph{Stage 2: Supervised Contrastive Learning with MNRL.} We then fine-tune using Multiple Negatives Ranking Loss (MNRL):
32 +\begin{equation}
33 +\mathcal{L}_{\text{MNRL}} = - \frac{1}{B} \sum_{i=1}^{B} \log \frac{\exp\bigl(\text{sim}(a_i, p_i) / \tau\bigr)}{\sum_{j=1}^{B} \exp\bigl(\text{sim}(a_i, p_j) / \tau\bigr)},
34 +\end{equation}
35 +where $(a_i, p_i)$ are positive pairs, $\text{sim}(\cdot, \cdot)$ denotes cosine similarity, $\tau$ is the temperature parameter, and $B$ is the batch size. In-batch negatives provide contrastive signal without explicit negative sampling. We construct approximately 4,000 training pairs per model using the six pairing strategies described in Section~\ref{sec:data}.
36 +
37 +\paragraph{Dual-Model Ensemble.} The two models are combined using separation-weighted averaging:
38 +\begin{equation}
39 +\widehat{\text{Tone}}_{t,k}^{\text{ens}} = w_k^{\text{MiniLM}} \cdot \text{Tone}_{t,k}^{\text{MiniLM}} + w_k^{\text{BERT}} \cdot \text{Tone}_{t,k}^{\text{BERT}},
40 +\end{equation}
41 +where the weights are proportional to axis separation quality:
42 +\begin{equation}
43 +w_k^{m} = \frac{\text{Sep}_k^{m}}{\text{Sep}_k^{\text{MiniLM}} + \text{Sep}_k^{\text{BERT}}}, \quad m \in \{\text{MiniLM}, \text{BERT}\}.
44 +\end{equation}
45 +For each statement, we measure inter-model confidence as:
46 +\begin{equation}
47 +\text{Conf}_{t,k} = \frac{1}{2}\bigl[\mathbb{1}[\text{sign}(\text{Tone}_{t,k}^{\text{M}}) = \text{sign}(\text{Tone}_{t,k}^{\text{B}})] + (1 - |\text{Tone}_{t,k}^{\text{M}} - \text{Tone}_{t,k}^{\text{B}}| / \max_j |\text{Tone}_{j,k}^{\text{M}} - \text{Tone}_{j,k}^{\text{B}}|)\bigr].
48 +\end{equation}
49 +Mean ensemble confidence is 0.837 for novelty and ranges from 0.562 to 0.806 for tone across axes. This dual-model approach reduces model-specific biases: MiniLM (33M parameters, 384 dimensions) provides computational efficiency and strong separation on policy stance, while BERT (110M parameters, 768 dimensions) captures richer contextual relationships. The ensemble achieves balanced performance across all semantic dimensions.
50 +
51 +\subsubsection{Semantic Dimensions}
52 +
53 +\begin{definition}[Semantic Similarity Measure]
54 +The semantic similarity between text vectors $\mathbf{u}$ and $\mathbf{v}$ is quantified using cosine similarity:
55 +\begin{equation}
56 +\text{sim}(\mathbf{u}, \mathbf{v}) = \frac{\mathbf{u} \cdot \mathbf{v}}{||\mathbf{u}||_2 \cdot ||\mathbf{v}||_2} = \frac{\sum_{i=1}^{d} u_i v_i}{\sqrt{\sum_{i=1}^{d} u_i^2} \cdot \sqrt{\sum_{i=1}^{d} v_i^2}}
57 +\end{equation}
58 +\end{definition}
59 +
60 +The cosine similarity measure ranges from $-1$ to $1$, where values approaching $1$ indicate high semantic similarity, values near $0$ indicate orthogonal or unrelated content, and values approaching $-1$ indicate semantic opposition. This metric is particularly well-suited for high-dimensional text embeddings because it captures angular distance between vectors while being invariant to vector magnitude differences that might arise from statement length variations.
61 +
62 +\begin{definition}[Statement Novelty]
63 +The informational novelty of statement $F_t$ relative to the immediately preceding statement is defined as:
64 +\begin{equation}
65 +\text{Novelty}_t = 1 - \text{sim}(\mathbf{f}_t, \mathbf{f}_{t-1})
66 +\end{equation}
67 +\end{definition}
68 +
69 +This novelty measure captures the degree to which current policy communications represent genuine departures from recent messaging patterns. Higher novelty values indicate statements containing substantively new informational content relative to recent communications, while lower values suggest continuity with established messaging. The measure is bounded between $0$ and $2$, where $0$ indicates identical statements and $2$ indicates maximally opposed statements.
70 +
71 +To construct our policy tone measure, we develop a data-driven approach based on PCA-based semantic axis construction that replaces manually selected reference dates with an algorithmic procedure, eliminating researcher degrees of freedom.
72 +
73 +\begin{definition}[Data-Driven Semantic Axis Construction]
74 +\label{def:pca_axes}
75 +Let $\mathbf{E} \in \mathbb{R}^{N \times d}$ be the L2-normalized embedding matrix. We compute the first $K = 8$ principal components:
76 +\begin{equation}
77 +\mathbf{E} = \mathbf{U} \boldsymbol{\Sigma} \mathbf{V}^\top + \boldsymbol{\epsilon}.
78 +\end{equation}
79 +For each semantic dimension $k \in \{1, \ldots, 4\}$ (policy stance, risk assessment, policy communication, economic priority), we define keyword banks $\mathcal{K}_k^+$ (positive pole) and $\mathcal{K}_k^-$ (negative pole), and compute a keyword differential score:
80 +\begin{equation}
81 +\Delta_i^{(k)} = \sum_{w \in \mathcal{K}_k^+} \mathbb{1}[w \in \text{text}_i] - \sum_{w \in \mathcal{K}_k^-} \mathbb{1}[w \in \text{text}_i].
82 +\end{equation}
83 +We match each semantic axis to the PC maximizing the absolute Pearson correlation:
84 +\begin{equation}
85 +j^*(k) = \arg\max_{j \in \{1,\ldots,K\}} \bigl|\text{Corr}(\mathbf{v}_j, \boldsymbol{\Delta}^{(k)})\bigr|.
86 +\end{equation}
87 +\end{definition}
88 +
89 +The reference selection uses a hybrid seed-and-percentile approach: (1) use 3 expert-selected dates per pole to define an initial axis direction $\hat{\mathbf{d}}_k = \bar{\mathbf{e}}_{\text{pos}} - \bar{\mathbf{e}}_{\text{neg}}$; (2) project all embeddings onto this direction: $s_i = \mathbf{e}_i^\top \hat{\mathbf{d}}_k / \|\hat{\mathbf{d}}_k\|$; (3) select $M = 5$ statements with highest and lowest projection scores; (4) iteratively refine by recomputing the axis direction from selected references and re-selecting. This yields 5 reference statements per pole (compared to 2 in manual selection), producing more stable centroids.
90 +
91 +For each axis $k$, we measure discrimination quality using the separation metric:
92 +\begin{equation}
93 +\text{Sep}_k = 1 - \cos(\bar{\mathbf{c}}_k^+, \bar{\mathbf{c}}_k^-),
94 +\end{equation}
95 +where $\bar{\mathbf{c}}_k^+, \bar{\mathbf{c}}_k^-$ are the L2-normalized positive and negative centroids. All axes achieve separations well above the 0.04 minimum threshold, with ensemble separations ranging from 0.248 to 0.302.
96 +
97 +\begin{definition}[Policy Tone]
98 +The directional tone of statement $F_t$ on semantic axis $k$ is measured as:
99 +\begin{equation}
100 +\text{Tone}_{t,k} = \frac{\cos(\mathbf{f}_t, \bar{\mathbf{c}}_k^+) - \cos(\mathbf{f}_t, \bar{\mathbf{c}}_k^-)}{\text{Sep}_k}, \quad \text{clipped to } [-1, +1].
101 +\end{equation}
102 +\end{definition}
103 +
104 +This tone measure captures the relative proximity of actual statements to the positive (hawkish) and negative (dovish) reference centroids for each semantic axis, normalized by the axis separation. The primary variable used in regressions is the \emph{total policy stance tone}. In the regression tables, the z-scored stance variable is oriented so that positive values indicate more accommodative (dovish) communication; positive coefficients therefore measure the response to a one-standard-deviation dovish shift. The composite MPS measure defined below retains the hawkish-positive orientation. Relative to counterfactual-based methods, this approach is reproducible, does not depend on particular date choices, and extends to semantic dimensions other than the hawkish-dovish spectrum.
105 +
106 +\subsubsection{Integrated Policy Stance Measurement}
107 +
108 +\begin{definition}[Monetary Policy Stance]
109 +The overall monetary policy stance is defined as:
110 +\begin{equation}
111 +\text{Stance}_t = \text{Novelty}_t \times \text{Tone}_t
112 +\end{equation}
113 +\end{definition}
114 +
115 +\textbf{Note on originality:} Definitions 1--11 and Proposition~\ref{prop:mps} are our own contributions. They apply the standard sentence-embedding framework from the NLP literature to monetary policy communications with a new decomposition structure.
116 +
117 +The multiplicative form means that stance is large only when a statement is both novel \textit{and} directionally clear. We adopt a multiplicative rather than additive form for two reasons. First, it ensures that Stance is zero whenever either component is zero: repetitive content has no impact regardless of its tone, and novel but directionally neutral content has none either. Second, it captures the reinforcing interaction between the two dimensions: novel information amplifies the impact of tone, and clear directional tone amplifies the impact of novelty.
118 +
119 +In the following two definitions, $\mathbf{f}_t^D$ and $\mathbf{f}_t^H$ denote the embeddings of counterfactual statements located at the dovish and hawkish reference centroids of the policy stance axis, $\bar{\mathbf{c}}^-$ and $\bar{\mathbf{c}}^+$; they represent what the Fed would have released had it issued a maximally dovish or maximally hawkish statement at meeting $t$.
120 +
121 +\begin{definition}[Pure Dovish and Hawkish Stances]
122 +\label{def:pure_stances}
123 +The extreme stance scenarios are defined as:
124 +\begin{align}
125 +\text{Stance}_t^{dove} &= -\left(1 - \text{sim}(\mathbf{f}_t^D, \mathbf{f}_{t-1})\right) \\
126 +\text{Stance}_t^{hawk} &= 1 - \text{sim}(\mathbf{f}_t^H, \mathbf{f}_{t-1})
127 +\end{align}
128 +\end{definition}
129 +
130 +\begin{definition}[Weighted Policy Stance Representation]
131 +\label{def:weighted_stance}
132 +The overall stance can be expressed as a convex combination of extreme stances:
133 +\begin{equation}
134 +\text{Stance}_t = w_t \cdot \text{Stance}_t^{dove} + (1 - w_t) \cdot \text{Stance}_t^{hawk}
135 +\end{equation}
136 +where the dovish weight parameter $w_t$ is computed (not estimated) as a deterministic function of the observed similarities:
137 +\begin{equation}
138 +w_t = \frac{1 - \text{sim}(\mathbf{f}_t^H, \mathbf{f}_{t-1}) - \left(1 - \text{sim}(\mathbf{f}_t, \mathbf{f}_{t-1})\right) \cdot \text{Tone}_t}{2 - \text{sim}(\mathbf{f}_t^D, \mathbf{f}_{t-1}) - \text{sim}(\mathbf{f}_t^H, \mathbf{f}_{t-1})}
139 +\end{equation}
140 +\end{definition}
141 +
142 +This weighted representation provides an alternative interpretation of policy stance as a position along the spectrum between extreme dovish and hawkish alternatives. The weight parameter $w_t$ is not a free parameter to be estimated; rather, it is algebraically derived from the observed textual similarities.
143 +
144 +\subsubsection{Market Surprise Identification}
145 +
146 +\begin{definition}[Novelty Decomposition]
147 +The total novelty can be decomposed into predictable and unpredictable components:
148 +\begin{equation}
149 +\text{Novelty}_t = \overline{\text{Novelty}}_{t \mid t-\Delta} + \varepsilon_t
150 +\end{equation}
151 +where $\overline{\text{Novelty}}_{t \mid t-\Delta} \equiv \mathbb{E}_{t-\Delta}[\text{Novelty}_t]$ represents the expected novelty based on information available $\Delta$ periods before the announcement, and $\varepsilon_t$ represents the novelty surprise component with $\mathbb{E}_{t-\Delta}[\varepsilon_t] = 0$ and $\text{Var}(\varepsilon_t) = \sigma_{\varepsilon}^2 < \infty$. We assume $\varepsilon_t$ is covariance-stationary with $\text{Cov}(\varepsilon_t, \varepsilon_{t-j}) \to 0$ as $j \to \infty$.
152 +\end{definition}
153 +
154 +\begin{definition}[Expected Policy Stance]
155 +Under rational expectations, the expected policy stance is:
156 +\begin{equation}
157 +\mathbb{E}_{t-\Delta}[\text{Stance}_t] = \mathbb{E}_{t-\Delta}\left[\text{Novelty}_t \times \text{Tone}_t\right]
158 +\end{equation}
159 +\end{definition}
160 +
161 +\begin{assumption}[Conditional Independence or Zero Covariance]
162 +\label{ass:indep}
163 +We assume either (i) conditional independence: $\text{Novelty}_t \perp \text{Tone}_t | \mathcal{I}_{t-\Delta}$, where $\mathcal{I}_{t-\Delta}$ is the information set at time $t - \Delta$, or (ii) the weaker condition of zero conditional covariance:
164 +\begin{equation}
165 +\text{Cov}_{t-\Delta}(\text{Novelty}_t, \text{Tone}_t) = 0
166 +\end{equation}
167 +\end{assumption}
168 +
169 +This assumption is plausible because novelty captures whether the Fed changes its messaging structure (how information is presented), while tone captures the directional policy stance (hawkish/dovish). Empirically, the correlation between novelty and tone in our sample is 0.12 (Figure~\ref{fig:D6}), supporting the assumption that these are largely independent dimensions. Under this assumption:
170 +\begin{equation}
171 +\mathbb{E}_{t-\Delta}[\text{Stance}_t] = \mathbb{E}_{t-\Delta}[\text{Novelty}_t] \times \mathbb{E}_{t-\Delta}[\text{Tone}_t]
172 +\end{equation}
173 +
174 +Further assuming, for tractability, that tone is binary ($\text{Tone}_t \in \{-1, +1\}$, taking the dovish value $-1$ with probability $p_{t-\Delta}$), so that $\mathbb{E}_{t-\Delta}[\text{Tone}_t] = 1 - 2p_{t-\Delta}$:
175 +\begin{equation}
176 +\mathbb{E}_{t-\Delta}[\text{Stance}_t] = (1 - 2p_{t-\Delta}) \cdot \overline{\text{Novelty}}_{t \mid t-\Delta}
177 +\end{equation}
178 +
179 +\begin{definition}[Policy Stance Surprise]
180 +The communication-based policy stance surprise is defined as:
181 +\begin{equation}
182 +\text{MPS}_t = \text{Stance}_t - \mathbb{E}_{t-\Delta}[\text{Stance}_t]
183 +\end{equation}
184 +\end{definition}
185 +
186 +We use the abbreviation MPS (monetary policy surprise) for brevity, but emphasize that this measure is conceptually distinct from the traditional rate-based monetary policy surprise of \citet{kuttner2001}, which measures the unexpected component of the federal funds rate decision using futures prices. Our MPS is a \textit{communication} surprise---it captures unexpected variation in the textual content of FOMC statements, not in the quantitative rate decision. The two measures are complementary and potentially orthogonal: a meeting can produce a large rate surprise with a routine statement, or a novel and directional statement with a fully anticipated rate decision. Our measure captures the latter dimension, which has become increasingly important as forward guidance and qualitative communication have gained prominence in the monetary policy toolkit.
187 +
188 +\begin{proposition}[MPS Decomposition Structure]
189 +\label{prop:mps}
190 +The monetary policy surprise admits the following decomposition:
191 +\begin{equation}
192 +\text{MPS}_t = \overline{\text{Novelty}}_{t \mid t-\Delta}(\text{Tone}_t + 2p_{t-\Delta} - 1) + \text{Tone}_t \cdot \varepsilon_t
193 +\end{equation}
194 +\end{proposition}
195 +
196 +This decomposition reveals that policy surprises consist of two distinct components: (1) surprises arising from unexpected tone conditional on expected novelty, and (2) surprises arising from unexpected novelty, weighted by the actual tone of the communication. The proof of Proposition~\ref{prop:mps}, together with the derivations underlying Definitions~\ref{def:pure_stances} and~\ref{def:weighted_stance}, is in Appendix~\ref{sec:proofs}.
197 +
198 +\subsection{Econometric Framework}
199 +\label{sec:econometric}
200 +
201 +\subsubsection{Panel Data Structure}
202 +
203 +The data are organized as a three-dimensional panel structure with observations indexed along three dimensions:
204 +\begin{itemize}
205 +\item Asset identifier $i \in \{1, 2, \ldots, I\}$ where $I$ represents the total number of financial instruments
206 +\item Event identifier $j \in \{1, 2, \ldots, J\}$ corresponding to FOMC announcement dates where $J = 148$
207 +\item Relative minute $m \in \{-120, \ldots, +120\}$ measured from announcement time, with the primary post-announcement window covering $m \in \{0, \ldots, +45\}$
208 +\end{itemize}
209 +
210 +\subsubsection{Identification Strategy}
211 +
212 +Our identification relies on the fact that FOMC statements are drafted and finalized before the announcement and released at a scheduled time (typically 2:00 PM ET). The text at $t=0$ therefore cannot be influenced by market reactions at $t>0$, satisfying the timing restriction for causal interpretation.
213 +
214 +Our identification relies on three core conditions:
215 +
216 +\textbf{Identification Condition 1 (Exogenous Timing).} The precise minute-by-minute timing of FOMC announcements is predetermined and exogenous to short-term market movements within our event windows.
217 +
218 +\textbf{Identification Condition 2 (Information Concentration).} Within our 45-minute post-announcement event windows, FOMC statements represent the dominant source of monetary policy-relevant information, with minimal contamination from other systematic news sources.
219 +
220 +\textbf{Identification Condition 3 (Market Efficiency).} Financial markets rapidly incorporate new information from FOMC statements into prices, enabling causal interpretation of immediate price movements following announcements.
221 +
222 +We acknowledge two potential threats to this identification strategy. First, the Federal Reserve may adjust its communication in response to broader financial conditions observed before the meeting. We mitigate this concern by including event fixed effects $\delta_j$ that absorb all meeting-level heterogeneity. Second, FOMC announcements may coincide with other information releases (e.g., Summary of Economic Projections, press conference expectations). Our focus on the immediate post-announcement window limits contamination from subsequent press conference content, which typically begins 30 minutes after the statement release.
223 +
224 +\subsubsection{Dynamic Response Function Estimation}
225 +
226 +Our primary empirical specification estimates dynamic impulse response functions:
227 +\begin{equation}
228 +r_{i,m}^{(j)} = \alpha_i + \delta_j + \sum_{k=0}^{+45} \beta_k \cdot \mathbb{1}_{[m=k]} \cdot \text{MPS}_j + \varepsilon_{i,m,j}
229 +\label{eq:dynamic}
230 +\end{equation}
231 +where $r_{i,m}^{(j)}$ denotes the log return of asset $i$ at minute $m$ relative to monetary policy announcement $j$, $\alpha_i$ represents asset fixed effects, $\delta_j$ captures event fixed effects, and $\beta_k$ measures the marginal impact of monetary policy surprises occurring exactly $k$ minutes relative to the announcement.
232 +
233 +\subsubsection{Semantic Decomposition Specification}
234 +
235 +To examine the differential effects of policy tone versus informational novelty, we estimate:
236 +\begin{equation}
237 +r_{i,m}^{(j)} = \alpha_i + \delta_j + \sum_{k=0}^{+45} \left[\beta_k^{(T)} \cdot \mathbb{1}_{[m=k]} \cdot \text{Tone}_j + \beta_k^{(N)} \cdot \mathbb{1}_{[m=k]} \cdot \text{Novelty}_j\right] + \varepsilon_{i,m,j}
238 +\label{eq:decomp}
239 +\end{equation}
240 +
241 +The tone coefficients $\{\beta_k^{(T)}\}$ measure the dynamic response to hawkish versus dovish content, while the novelty coefficients $\{\beta_k^{(N)}\}$ capture the response to informational content, independent of directional bias.
242 +
243 +\subsubsection{Rolling Realized Measures}
244 +
245 +In addition to return effects, we examine how policy communications affect market volatility using rolling realized measures computed from 1-minute log returns.
246 +
247 +\paragraph{Realized Variance.} We construct NA-tolerant rolling realized variance in $K$-minute windows:
248 +\begin{equation}
249 +\mathrm{RV}_K(t) = \frac{K}{n_{\text{valid}}} \sum_{s = t-K+1}^{t} r_s^2 \cdot \mathbb{1}[r_s \text{ valid}],
250 +\end{equation}
251 +where $n_{\text{valid}}$ is the number of non-missing observations in the window. We require at least 80\% valid observations.
252 +
253 +\paragraph{Realized Beta.} We compute rolling beta relative to the ES (S\&P 500) benchmark:
254 +\begin{equation}
255 +\hat{\beta}_K(t) = \frac{\sum_{s=t-K+1}^{t} r_{i,s} \cdot r_{m,s}}{\sum_{s=t-K+1}^{t} r_{m,s}^2},
256 +\end{equation}
257 +where $r_{m,s}$ denotes the ES benchmark return. We use $K = 5$ for panel minute-level regressions and $K = 30$ for event-level analysis.
258 +
259 +\subsubsection{Panel Minute-Level Regressions}
260 +
261 +For each ticker, we estimate:
262 +\begin{equation}
263 +\log(\mathrm{RV}_{i,t}) = \alpha + \beta \cdot (\text{Post}_t \times \text{Semantic}_i) + \varepsilon_{i,t},
264 +\label{eq:panel}
265 +\end{equation}
266 +where $\text{Post}_t = \mathbb{1}[t > 0]$ indicates post-announcement minutes, and $\text{Semantic}_i$ is either \emph{Stance} or \emph{Novelty} (z-scored). Standard errors are clustered by event date.
267 +
268 +\subsubsection{Event-Level Regressions}
269 +
270 +For each event, we compute pre/post changes:
271 +\begin{equation}
272 +\Delta \log(\mathrm{RV})_i = \overline{\log(\mathrm{RV})}_{\text{post}} - \overline{\log(\mathrm{RV})}_{\text{pre}},
273 +\end{equation}
274 +and regress on semantic measures:
275 +\begin{equation}
276 +\Delta Y_i = \alpha + \beta_1 \text{Stance}_i + \beta_2 \text{Novelty}_i + \beta_3 (\text{Stance}_i \times \text{Novelty}_i) + \varepsilon_i,
277 +\label{eq:event}
278 +\end{equation}
279 +with Newey--West HAC standard errors. We estimate this specification for five dependent variables: $\Delta\mathrm{RV}$, $\Delta\log(\mathrm{RV})$, RV ratio, $\log$(RV ratio), and $\Delta\beta$.
280 +
281 +\subsubsection{Local Projection Impulse Response Functions}
282 +
283 +Following \citet{jorda2005}, for each horizon $h \in \{1, 2, 3, 5, 10, 15, 20, 30, 45, 60, 90, 120\}$ minutes post-announcement:
284 +\begin{equation}
285 +\text{CumRet}_i(0 \to h) = \alpha + \beta_1^{(h)} \text{Stance}_i + \beta_2^{(h)} \text{Novelty}_i + \beta_3^{(h)} (\text{Stance}_i \times \text{Novelty}_i) + \varepsilon_i^{(h)}.
286 +\label{eq:irf}
287 +\end{equation}
288 +Abnormal cumulative returns are computed as $\text{CAR}_i(h) = \text{CumRet}_i(h) - \text{CumRet}_i^{\text{ES}}(h)$. This local projection approach is robust to misspecification of the data-generating process and allows horizon-specific inference without imposing a parametric impulse response shape. Because the post-meeting press conference typically begins 30 minutes after the statement release, we interpret responses at horizons beyond $h = 30$ as the joint effect of the statement and early press-conference communication.
289 +
290 +\subsubsection{Pre-Announcement Placebo Test}
291 +
292 +We estimate the same specification~\eqref{eq:irf} for $h$ minutes \emph{before} the announcement:
293 +\begin{equation}
294 +\text{CumRet}_i(-h \to 0) = \alpha + \beta_1^{(-h)} \text{Stance}_i + \beta_2^{(-h)} \text{Novelty}_i + \beta_3^{(-h)} (\text{Stance}_i \times \text{Novelty}_i) + \varepsilon_i^{(-h)}.
295 +\label{eq:placebo}
296 +\end{equation}
297 +Under the null that announcement content is not anticipated, all $\beta^{(-h)}$ should be zero. This placebo test provides a direct validation of our event study design.
298 +
299 +\subsubsection{Statistical Inference and Robustness}
300 +
301 +We subject all event-level results to five alternative inference methods:
302 +
303 +\begin{enumerate}
304 +\item \textbf{Newey--West HAC standard errors}: Robust to heteroskedasticity and autocorrelation.
305 +\item \textbf{Wild bootstrap} ($B = 1{,}999$, Rademacher weights): Robust to heteroskedasticity with improved finite-sample properties.
306 +\item \textbf{Clustered standard errors} (HC1, by event date): Accounts for cross-sectional dependence within events.
307 +\item \textbf{Quantile regression} ($\tau = 0.5$): Robust to outliers and heavy tails in the dependent variable.
308 +\item \textbf{Permutation test} ($B = 4{,}999$): Gold standard---shuffles semantic labels across events preserving panel structure, providing exact $p$-values under the null.
309 +\end{enumerate}
310 +
311 +All $p$-values in panel and IRF regressions are adjusted for multiple testing using the Benjamini--Hochberg (BH) procedure within each family of tests (per dependent variable and model type) \citep{benjamini1995controlling}. We use only standard significance levels: 1\%, 5\%, and 10\%.
312 +
313 +\textbf{Sub-Period Stability.} We examine coefficient stability across six Fed policy regimes: crisis and recovery (2008--2012), post-crisis normalization (2013--2015), pre-pandemic tightening (2016--2019), pandemic response (2020--2021), and inflation tightening (2022--2025).
314 +
315 +\textbf{Alternative Semantic Measures.} The dual-model ensemble itself provides a built-in robustness check: by comparing results from MiniLM-only, BERT-only, and ensemble measures, we verify that findings are not driven by model-specific artifacts.
316 +
317 +\subsection{Hypotheses and Testable Predictions}
318 +\label{sec:hypotheses}
319 +
320 +We now state our hypotheses and, for each one, the coefficient restriction that operationalizes it in the specifications of Section~\ref{sec:econometric}. Recall the sign conventions: the z-scored stance regressor in equations~\eqref{eq:decomp}--\eqref{eq:irf} is dovish-positive, while the composite MPS in equation~\eqref{eq:dynamic} is hawkish-positive. Table~\ref{tab:hyp_map} summarizes the mapping from hypotheses to coefficients, equations, and the tables in which each test is reported.
321 +
322 +\textbf{Hypothesis 1 (Tone Effects on Returns).}
323 +\begin{itemize}
324 +\item \textbf{H1a.} Dovish tone increases returns on risk assets (equities, commodities). \textit{Test:} $\beta_k^{(T)} > 0$ in the semantic decomposition~\eqref{eq:decomp} and $\beta_1^{(h)} > 0$ in the local projection~\eqref{eq:irf} for ES and CL.
325 +\item \textbf{H1b.} Dovish tone decreases returns on safe-haven assets, as accommodative policy triggers portfolio rebalancing away from safe havens toward risk assets. \textit{Test:} $\beta_1^{(h)} < 0$ in~\eqref{eq:irf} for GC, ZN, and ZF.
326 +\item \textbf{H1c.} Tone effects strengthen with the horizon as markets progressively process policy implications. \textit{Test:} $|\beta_1^{(h)}|$ increasing in $h$ in~\eqref{eq:irf} for ES.
327 +\end{itemize}
328 +
329 +\textbf{Hypothesis 2 (Novelty Effects on Volatility).}
330 +\begin{itemize}
331 +\item \textbf{H2a.} High novelty increases market volatility through information processing complexity. \textit{Test:} $\beta_2 > 0$ in the event-level regression~\eqref{eq:event} with $\Delta\mathrm{RV}$ as the dependent variable, and $\beta > 0$ in the panel regression~\eqref{eq:panel} with $\text{Semantic} = \text{Novelty}$, most directly for VX.
332 +\item \textbf{H2b.} Novelty effects dissipate rapidly as markets resolve uncertainty. \textit{Test:} $|\beta_2^{(h)}|$ decreasing in $h$ in~\eqref{eq:irf} for VX.
333 +\item \textbf{H2c.} Novelty has minimal effects on directional returns. \textit{Test:} $\beta_2^{(h)} \approx 0$ (statistically indistinguishable from zero) in~\eqref{eq:irf} for the return contracts (ES, CL, GC, DX).
334 +\end{itemize}
335 +
336 +\textbf{Hypothesis 3 (Policy Surprise Effects).}
337 +\begin{itemize}
338 +\item \textbf{H3a.} Hawkish policy surprises reduce equity returns. \textit{Test:} $\beta_0 < 0$ in the dynamic response function~\eqref{eq:dynamic} for ES.
339 +\item \textbf{H3b.} Policy surprises increase implied volatility. \textit{Test:} $\beta_0 > 0$ in~\eqref{eq:dynamic} for VX.
340 +\item \textbf{H3c.} Surprise effects on volatility are most pronounced immediately after the announcement. \textit{Test:} $|\beta_k|$ in~\eqref{eq:dynamic} largest at $k = 0$ and decaying in $k$ for VX.
341 +\end{itemize}
342 +
343 +\textbf{Hypothesis 4 (Cross-Asset Patterns).}
344 +\begin{itemize}
345 +\item \textbf{H4a.} Risk assets respond to tone in the same direction (risk-on/risk-off). \textit{Test:} $\text{sign}\bigl(\beta_1^{(h)}\bigr)$ equal across ES and CL in~\eqref{eq:irf}, opposite for GC, ZN, ZF.
346 +\item \textbf{H4b.} Volatility responses to novelty are asset-specific. \textit{Test:} $\beta_2$ and $\beta_3$ in~\eqref{eq:event} differ in magnitude and significance across the seven contracts.
347 +\end{itemize}
348 +
349 +Finally, the validity of the design itself is testable: under no anticipation of statement content, all pre-announcement coefficients in the placebo specification~\eqref{eq:placebo} should be zero, $\beta_1^{(-h)} = \beta_2^{(-h)} = \beta_3^{(-h)} = 0$ for all $h$.
350 +
351 +\begin{table}[H]
352 +\centering
353 +\footnotesize
354 +\caption{Mapping of Hypotheses to Coefficients, Equations, and Evidence}
355 +\label{tab:hyp_map}
356 +\begin{tabular}{lp{4.5cm}p{4.3cm}p{3.3cm}}
357 +\toprule
358 +Hyp. & Prediction & Coefficient restriction (equation) & Evidence \\
359 +\midrule
360 +H1a & Dovish tone $\Rightarrow$ risk-asset returns $\uparrow$ & $\beta_1^{(h)} > 0$ in \eqref{eq:irf}, ES/CL & Table~\ref{tab:irf_stance_ret_h} \\
361 +H1b & Dovish tone $\Rightarrow$ safe-haven returns $\downarrow$ & $\beta_1^{(h)} < 0$ in \eqref{eq:irf}, GC/ZN/ZF & Table~\ref{tab:irf_stance_ret_h} \\
362 +H1c & Tone effects build with horizon & $|\beta_1^{(h)}|$ increasing in $h$, ES & Table~\ref{tab:irf_stance_ret_h}; Fig.~\ref{fig:R8} \\
363 +H2a & Novelty $\Rightarrow$ volatility $\uparrow$ & $\beta_2 > 0$ in \eqref{eq:event}, $\Delta$RV & Tables~\ref{tab:panel_logrv_novelty}, \ref{tab:rolling_delta_rv_30min} \\
364 +H2b & Novelty effects decay quickly & $|\beta_2^{(h)}|$ decreasing in $h$, VX & Table~\ref{tab:irf_novelty_ret_h} \\
365 +H2c & Novelty does not move returns & $\beta_2^{(h)} \approx 0$ in \eqref{eq:irf} & Table~\ref{tab:irf_novelty_ret_h}; Fig.~\ref{fig:R9} \\
366 +H3a & Hawkish surprise $\Rightarrow$ equity returns $\downarrow$ & $\beta_0 < 0$ in \eqref{eq:dynamic}, ES & Section~\ref{sec:baseline} \\
367 +H3b & Surprise $\Rightarrow$ implied volatility $\uparrow$ & $\beta_0 > 0$ in \eqref{eq:dynamic}, VX & Section~\ref{sec:baseline} \\
368 +H3c & Surprise effects peak on impact & $|\beta_k|$ max at $k=0$ in \eqref{eq:dynamic}, VX & Section~\ref{sec:baseline} \\
369 +H4a & Homogeneous return response, risk assets & $\text{sign}(\beta_1^{(h)})$ equal, ES/CL in \eqref{eq:irf} & Table~\ref{tab:irf_stance_ret_h} \\
370 +H4b & Heterogeneous volatility responses & $\beta_2, \beta_3$ vary across contracts in \eqref{eq:event} & Tables~\ref{tab:rolling_delta_rv_30min}, \ref{tab:robustness_A} \\
371 +Placebo & No pre-announcement effects & $\beta^{(-h)} = 0$ in \eqref{eq:placebo}, all contracts & Fig.~\ref{fig:placebo_stance}, \ref{fig:placebo_novelty} \\
372 +\bottomrule
373 +\end{tabular}
374 +\end{table}
375 +
added PHD_chapitre3_theses_20260731/sections/04_results.tex +105 −0
@@ -0,0 +1,105 @@
1 +\section{Results}
2 +\label{sec:results}
3 +
4 +We organize our results as follows. We first describe the semantic variables (Section~\ref{sec:desc}), then report baseline responses to the composite MPS (Section~\ref{sec:baseline}), the tone-novelty decomposition (Section~\ref{sec:decomposition}), and the temporal dynamics of these effects (Section~\ref{sec:halflife}). Throughout, each hypothesis is evaluated against the coefficient restriction assigned to it in Table~\ref{tab:hyp_map}. The main finding is that tone and novelty load on different dependent variables: tone predicts directional returns, while novelty predicts volatility changes. The VIX model has the highest $R^2$ ($0.78$); equity and commodity models follow. Treasury results are weaker.
5 +
6 +$R^2$ values for the return models are low (0.01--0.15), as is typical when predicting minute-level returns where microstructure noise dominates \citep{barndorff2002}. Low $R^2$ does not bias the coefficient estimates, but it means that FOMC communication accounts for a small share of total intra-day return variation. The VIX model is the exception: semantic variables explain up to 78\% of announcement-window variation in VIX futures, consistent with the hypothesis that these communications operate primarily on uncertainty rather than on the level of returns.
7 +
8 +\subsection{Descriptive statistics}
9 +\label{sec:desc}
10 +
11 +Table~\ref{tab:desc_1min_ret} reports descriptive statistics for 1-minute log returns on FOMC days. Table~\ref{tab:desc_rolling} reports summary statistics for the rolling realized measures. Figure~\ref{fig:D3} plots the average intraday volatility pattern on FOMC days, which spikes sharply at the 14:00~ET release.
12 +
13 +Novelty has a mean of 0.065. The scale runs from 0 (a statement identical to the previous one) to 2 (maximally opposite), so a mean of 0.065 corresponds to an average cosine similarity of 0.935 between consecutive statements---most meetings produce only small changes to the prior template. The median is lower (0.025), so the distribution is right-skewed: most meetings produce minor textual changes, but a few produce large departures. The maximum (0.494) occurs during the 2008 crisis, when the Fed introduced zero-lower-bound language (Figure~\ref{fig:D7}). Skewness is 2.45 and kurtosis is 8.16.
14 +
15 +The raw tone measure (hawkish-positive) has a mean of 0.309 on a theoretical $[-1, +1]$ scale, where $-1$ is the dovish centroid, $0$ is equidistant, and $+1$ is the hawkish centroid. The positive mean indicates that the average FOMC statement over 2008--2025 lies closer to the hawkish than to the dovish centroid. Because cosine similarities to both centroids are high, realized values occupy a narrow band within the theoretical scale; we therefore z-score tone in all regressions. Skewness is $-0.78$ and kurtosis is 6.96, driven by the most dovish statements in the sample, which mark its lower extremes (2008: $-0.081$; 2020: $-0.062$).
16 +
17 +The MPS measure has mean 0.004 and kurtosis 12.18---the fat tails reflect occasional large surprises during economic stress. Jarque-Bera tests reject normality for all three measures ($p < 0.01$), and ARCH-LM tests indicate time-varying variance ($p < 0.05$), which motivates our use of robust standard errors and multiple inference methods throughout.
18 +
19 +\subsection{Baseline Asset Price Responses to Monetary Policy Surprises}
20 +
21 +\label{sec:baseline}
22 +We begin with the composite MPS, testing H3. The dependent variable is the 1-minute log return for each contract; the regressor is the z-scored MPS. A positive coefficient means that hawkish communication surprises are associated with higher returns.
23 +
24 +\subsubsection{Immediate Market Reactions}
25 +
26 +The equity market response to hawkish surprises is consistent with H3a (negative equity effect): the ES contract exhibits a $-1.2$ basis point response ($p < 0.05$). To put this magnitude in perspective, a one-standard-deviation hawkish surprise generates an immediate equity decline roughly equivalent to the average hourly return on a non-announcement day, concentrated in a single minute.
27 +
28 +Among commodities, crude oil (CL) falls by 1.0 bps ($p < 0.05$), consistent with tighter policy reducing growth expectations and energy demand. Gold (GC) falls by 0.6 bps, as an inflation hedge becomes less attractive when policy tightens. The Dollar Index (DX) rises by 0.8 bps, consistent with higher expected rate differentials increasing dollar demand.
29 +
30 +The VIX response is $+11.2$ bps ($p < 0.05$, H3b): a one-standard-deviation hawkish surprise is associated with an 11.2 bps increase in VIX futures, about 0.5\% of the average VIX level. This is the largest immediate response in our sample. Treasury securities (ZN, ZF) show small, insignificant responses, possibly because our textual MPS is largely orthogonal to the rate expectations already embedded in Treasury futures.
31 +
32 +\subsubsection{Dynamic Response Evolution}
33 +
34 +The 45-minute post-announcement window shows that different asset classes adjust at different speeds:
35 +
36 +\textit{Equities (ES).} The initial decline ($-1.2$ bps at $t = 0$) deepens to $-3.8$ bps by $t = 15$ and $-6.8$ bps by $t = 45$. The cumulative response is 5.7 times the instantaneous reaction, so the first-minute price change captures only a fraction of the total adjustment.
37 +
38 +\textit{VIX futures.} The initial spike ($+11.2$ bps at $t = 0$) loses roughly half its magnitude by $t = 20$ and returns near its pre-announcement level by $t = 45$.
39 +
40 +\textit{Treasuries (ZN, ZF).} Responses become marginally significant only at $t = 30$--$45$, consistent with slower transmission to term premiums.
41 +
42 +\textit{Commodities (CL, GC).} Crude oil effects continue to grow through $t = 45$, consistent with the slower adjustment typical of physical commodity markets.
43 +
44 +These patterns support H3c for volatility---the VIX response peaks on impact and decays---but equity and commodity responses continue to build, foreshadowing the gradual tone effects documented below. The speed of adjustment varies across assets, which motivates the asset-specific decomposition.
45 +
46 +\subsection{Semantic Decomposition: Tone vs.\ Novelty Effects}
47 +\label{sec:decomposition}
48 +
49 +We now separate tone from novelty. The dependent variables are log returns (testing H1) and realized volatility changes (testing H2). The regressors are the z-scored tone and novelty measures, entered separately and jointly.
50 +
51 +\subsubsection{Tone Effects on Asset Returns (H1)}
52 +
53 +A one-standard-deviation dovish shift in tone is associated with ES gains that build steadily over the post-announcement window: 3.8 bps by $h = 30$, 6.8 bps by $h = 60$, and 12.0 bps by $h = 120$ (Table~\ref{tab:irf_stance_ret_h}, H1a)---several days of average equity returns compressed into two hours. This is consistent with lower discount rates, higher growth expectations, and greater risk appetite.
54 +
55 +Safe-haven assets move in the opposite direction: at $h = 15$, gold falls by 3.1 bps and 10-year Treasuries (ZN) by 1.1 bps following dovish tone (H1b), consistent with a ``risk-on'' rotation out of safe havens. The Dollar Index response is small and changes sign across horizons ($+1.4$ bps at $h = 15$, $-4.0$ bps at $h = 120$).
56 +
57 +Tone effects grow stronger, not weaker, over the post-announcement window: the ES coefficient rises monotonically from 0.2 bps at $h = 5$ to 3.8 bps at $h = 30$ and 12.0 bps at $h = 120$ (Table~\ref{tab:irf_stance_ret_h}, H1c). This gradual amplification---rather than the immediate level shift that a frictionless model would imply---is consistent with sequential portfolio adjustment by heterogeneous participants (algorithms, institutions, retail).
58 +
59 +\subsubsection{Novelty Effects on Volatility (H2)}
60 +
61 +Novelty loads on volatility, not on returns. The VIX response to a one-standard-deviation increase in novelty is $-5.57$ bps ($p < 0.05$), meaning that genuinely new language is associated with \textit{lower} implied volatility. This sign is opposite to our ex ante prediction (H2a) but has a straightforward interpretation: a statement that departs substantially from the prior meeting's language sends a clearer signal about the Fed's current assessment, helping investors narrow the range of possible outcomes. A repetitive statement, by contrast, leaves open the question of whether the unchanged wording reflects genuine stability or simply a failure to update the language.
62 +
63 +The VIX novelty coefficient is largest in the first 5 minutes and falls to near zero by $t = 30$ (H2b). Novelty coefficients on directional returns (ES, CL, GC, DX) are small and insignificant (H2c). The combination---novelty predicts volatility but not returns---separates our measure from standard ``surprise'' variables, which conflate direction and information content.
64 +
65 +\subsubsection{Panel Minute-Level Results}
66 +
67 +Tables~\ref{tab:panel_logrv_stance}--\ref{tab:panel_logrv_novelty} report panel regressions of $\log(\text{RV})$ on the interaction of post-announcement indicators with semantic measures. The coefficient on Post$\times$Stance is negative and significant at 1\% for 6 of 7 contracts: ES ($-5.31$), VX ($-3.28$), ZN ($-5.25$), ZF ($-4.16$), CL ($-5.87$), and GC ($-6.03$). Only DX is insignificant ($+0.15$). The negative sign means that dovish stance is associated with lower post-announcement realized volatility---accommodative statements calm markets---while hawkish statements are followed by larger volatility increases, consistent with contractionary signals generating more repricing. By contrast, the Post$\times$Novelty coefficients (Table~\ref{tab:panel_logrv_novelty}) are smaller and significant only for the Treasury contracts (ZN: $-1.81$, $p < 0.01$; ZF: $-1.22$, $p < 0.10$).
68 +
69 +\subsubsection{Event-Level Results}
70 +
71 +Tables~\ref{tab:rolling_delta_rv_30min}--\ref{tab:rolling_delta_beta_30min} report event-level regressions with the interaction specification $\Delta Y_i = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 (\text{Stance} \times \text{Novelty}) + \varepsilon_i$. The stance$\times$novelty interaction is negative and significant for VIX ($-7.96$, $t = -3.12$, $p < 0.01$), ZF ($-2.48$, $t = -2.22$), ZN ($-1.66$, $t = -1.99$), and CL ($-8.61$, $p < 0.10$); Figure~\ref{fig:R6a} summarizes the coefficient estimates as a heatmap. The interaction indicates that stance and novelty reinforce each other: the volatility decline associated with a dovish statement is larger when the statement also departs substantially from prior language, whereas a statement that repeats the previous meeting's wording has a smaller volatility effect regardless of its tone.
72 +
73 +$R^2$ values are much higher for volatility than for returns: $\Delta\text{RV}$ regressions achieve 0.286 (ES), 0.268 (CL), and 0.245 (ZF), versus below 0.05 for most return regressions. The semantic variables explain more of the variation in volatility (which depends on information content) than in returns (which depend on the direction of the surprise in a noisy environment).
74 +
75 +\subsubsection{Local Projection Impulse Response Functions}
76 +
77 +Tables~\ref{tab:irf_stance_ret_h}--\ref{tab:irf_stance_x_novelty_abn_h} and Figures~\ref{fig:R10_stance}--\ref{fig:R10_abn_interaction} report local projection estimates. The VIX response to stance is negative at every horizon, statistically strongest between 5 and 30 minutes ($-14.74$ bps at $h = 5$, $-22.03$ at $h = 30$), and attenuates thereafter ($-14.27$ at $h = 120$, with much wider standard errors); the profile is hump-shaped, peaking near 30 minutes. The stance$\times$novelty interaction on abnormal VIX returns (Figure~\ref{fig:R10_abn_interaction}) is significant from $h = 5$ through $h = 120$, with the $t$-statistic reaching $-5.06$.
78 +
79 +The ES coefficient on stance grows monotonically from near zero at $h = 5$ to 12.0 bps at $h = 120$ (standard errors also grow). A frictionless model would predict an immediate level shift; the gradual increase we observe is more consistent with sequential adjustment by heterogeneous investors.
80 +
81 +\subsubsection{Cross-Asset Patterns (H4)}
82 +
83 +Risk assets (ES, CL) both rise after dovish tone, with similar timing (H4a). Safe havens (GC, ZN, ZF) fall, consistent with risk-on/risk-off. But the volatility responses differ across assets (H4b): VIX responds most to the stance$\times$novelty interaction, Treasury volatility responds mainly to stance alone, and commodity volatility responds to both. The return homogeneity and volatility heterogeneity suggest that the two channels operate with different relative strength across asset classes.
84 +
85 +\subsubsection{Pre-Announcement Placebo Tests}
86 +
87 +Figures~\ref{fig:placebo_stance}--\ref{fig:placebo_novelty} report pre-announcement placebo tests: we estimate the same IRF specification for cumulative returns from $-h$ to 0. If statement content is not anticipated, all pre-announcement coefficients should be zero. None is significant at 10\% across any ticker, horizon, or semantic measure. This rules out information leakage and pre-existing trends.
88 +
89 +\subsubsection{Multi-Method Robustness}
90 +
91 +Tables~\ref{tab:robustness_A}--\ref{tab:robustness_B} report five alternative inference methods for every event-level coefficient. We call a result ``robust'' if at least 3 of 5 methods give $p < 0.10$. The most robust results are: stance on $\Delta\log\text{RV}$ for ES, CL, GC, ZF, and ZN (each $\#\text{Sig} = 3$); stance on $\Delta\text{RV}$ for ES, CL, and DX (each $\#\text{Sig} = 3$); and novelty on $\Delta\text{RV}$ for ES and CL ($\#\text{Sig} = 4$). Figure~\ref{fig:rob_heatmap} displays these counts. Sub-period analysis (Figure~\ref{fig:rob_subperiod}) shows consistent coefficient signs across six Fed regimes, though magnitudes vary.
92 +
93 +With 7 assets and multiple dependent variables, we test many coefficients; results with $\#\text{Sig} = 1$ should be treated with caution. We rely on the Benjamini--Hochberg correction and the robustness counts to limit false discovery.
94 +
95 +\subsection{Economic Interpretation: Temporal Dynamics and Half-Lives}
96 +\label{sec:halflife}
97 +
98 +We summarize the speed of adjustment using half-lives: the time for an initial effect to fall to half its peak. For exponential decay, $t_{1/2} = \ln(2)/\lambda$.
99 +
100 +The VIX response ($+11.2$ bps on impact) falls to half its peak within 15--20 minutes; reading the decay as approximately exponential gives a half-life of about 17.5 minutes. Once the statement text is known, uncertainty about the Fed's message dissipates quickly.
101 +
102 +Equity returns show the opposite pattern. The ES response to dovish tone \textit{grows} steadily over the post-announcement window, roughly tripling between $h = 30$ (3.8 bps) and $h = 120$ (12.0 bps). Fundamental repricing---revising expected cash flows and discount rates---takes longer than uncertainty resolution, consistent with gradual diffusion across heterogeneous investors.
103 +
104 +The two patterns together---fast volatility decay and slow return amplification---are hard to reconcile with a single channel. The informational dimension (is the statement new?) resolves quickly; the directional dimension (is it hawkish or dovish?) takes longer to price in.
105 +
added PHD_chapitre3_theses_20260731/sections/05_conclusion.tex +61 −0
@@ -0,0 +1,61 @@
1 +\section{Conclusion}
2 +\label{sec:conclusion}
3 +
4 +\subsection{Summary of Findings}
5 +
6 +We summarize our findings under two headings: what we learn about the economics of monetary policy transmission, and what we contribute methodologically.
7 +
8 +\textbf{Tone predicts directional returns.} Dovish tone is associated with higher equity returns that build to about 12 bps over two hours, and with lower safe-haven returns (gold: $-3.1$ bps; Treasuries: $-1.1$ bps at 15 minutes), supporting H1a--b. This is consistent with the standard transmission mechanism in \citet{bernanke2005} and \citet{gurkaynak2005}---accommodative expectations lower discount rates and shift portfolios toward risk assets---but we show that the channel operates through the \textit{qualitative} content of the statement, not only through the rate decision.
9 +
10 +Tone effects grow stronger over time: the ES coefficient rises from 0.2 bps at $h = 5$ to 12.0 bps at $h = 120$ (H1c). \citet{rosa2013} documents continued price adjustment over hours after FOMC releases; our minute-level data show that this amplification is consistent with heterogeneous processing speeds across market participants \citep{veldkamp2011}.
11 +
12 +\textbf{Novelty predicts lower volatility, not higher.} Novelty is associated with \textit{reduced} implied volatility (VIX: $-5.57$ bps, $p < 0.05$; interaction $t = -5.06$), which reverses the sign we predicted under H2a. Prior work on disagreement-driven volatility \citep{patton2013} and news-based uncertainty \citep{manela2017} would predict the opposite. The difference, we believe, is that in the specific setting of FOMC statements, a departure from prior language acts as a \textit{signal of clarity}: the Fed is actively updating its message, which narrows the set of plausible policy paths. A repetitive statement, conversely, leaves open whether the Fed's views have changed but the language has not been updated. Novelty effects dissipate in under 10 minutes (H2b) and do not predict returns (H2c), confirming that this channel operates on volatility rather than on prices.
13 +
14 +\textbf{The composite surprise replicates known patterns.} Hawkish surprises are associated with 1--7 bps lower equity returns (H3a) and 11.2 bps higher VIX (H3b), with volatility effects concentrated in the first minutes (H3c). That our textual MPS produces results consistent with the rate-surprise literature \citep{kuttner2001,bernanke2005,gurkaynak2005,savor2014} reassures us that the NLP pipeline captures economically relevant variation.
15 +
16 +\textbf{Cross-asset patterns differ between returns and volatility.} Risk assets (ES, CL) respond to tone with the same sign and similar timing (H4a). Safe havens (GC, ZN, ZF) move in the opposite direction. But volatility responses vary: VIX loads on the interaction, Treasury volatility on stance alone, commodity volatility on both. This pattern is consistent with \citet{fleming1999} and \citet{balduzzi2001}, who document heterogeneous announcement effects across asset classes.
17 +
18 +\textbf{Volatility resolves fast; returns adjust slowly.} The VIX half-life is about 17.5 minutes; the ES tone effect continues to build for two hours after the announcement. A single-channel model cannot produce both patterns simultaneously. The fast channel (novelty $\to$ volatility) reflects the resolution of uncertainty about what the Fed will say. The slow channel (tone $\to$ returns) reflects the time needed to revise discount rates, growth expectations, and portfolio allocations across a heterogeneous investor base.
19 +
20 +\subsection{Methodological Contributions}
21 +
22 +On the measurement side, we make three contributions. First, using two architecturally different models (MiniLM, 33M parameters; BERT, 110M parameters) and comparing their outputs reduces the risk that any finding is an artifact of one particular architecture. The inter-model confidence (mean 0.837 for novelty) flags statements where the two models disagree. Second, two-stage domain adaptation (TSDAE then MNRL) yields axis separations of 0.248--0.302, well above the 0.04 minimum. Third, PCA-based reference selection replaces subjective date choices with an algorithm, removing a source of researcher discretion.
23 +
24 +Working at the minute level (148 events $\times$ $\pm$120 minutes) matters for two reasons: it allows causal identification from the pre-determined release time (validated by the placebo tests), and it reveals the fast-volatility / slow-return asymmetry that daily data would average away.
25 +
26 +\subsection{Implications for Policy and Communication Strategy}
27 +
28 +Three findings have implications for how the Fed drafts its statements.
29 +
30 +First, novelty reduces volatility. The conventional view is that central banks should change their language gradually to avoid ``surprising'' markets. Our estimates say the opposite: substantive changes in wording are associated with \textit{lower} implied volatility, presumably because a clear departure from the prior statement narrows the range of plausible interpretations. When the Fed needs to signal a regime change---a new framework, a pivot from tightening to easing---a decisive rewrite of the statement may be less destabilizing than incremental edits.
31 +
32 +Second, tone has persistent valuation effects. Because the equity response to tone grows over 2 hours rather than being absorbed immediately, the hawkish-dovish framing of the statement has real wealth consequences: a one-standard-deviation dovish shift is associated with up to 12 bps of equity returns, which, on a market capitalization of roughly \$40 trillion, is economically meaningful.
33 +
34 +Third, the interaction matters. The stance$\times$novelty interaction on VIX ($t = -5.06$, persistent from 5 to 120 minutes) implies that a directionally clear statement packaged in new language reduces uncertainty more than the same message delivered in boilerplate.
35 +
36 +\subsection{Limitations}
37 +
38 +We note several limitations.
39 +
40 +\textit{Sample period.} Our 2008--2025 sample is dominated by extraordinary monetary policy regimes: the zero lower bound (2008--2015), quantitative easing, and the post-pandemic tightening cycle. While our sub-period stability analysis (Figure~\ref{fig:rob_subperiod}) shows that coefficient signs are generally consistent across six Fed policy regimes, we cannot rule out that the magnitude of communication effects differs in more ``normal'' policy environments. In particular, the ZLB period may overstate the importance of qualitative communication (since rate decisions were constrained, statements became the primary policy instrument), while the recent tightening cycle may understate novelty effects (since rate hikes were widely anticipated, reducing the scope for genuine communication surprises).
41 +
42 +\textit{Measurement error.} Our NLP-based tone and novelty measures are subject to measurement error whose properties are difficult to characterize fully. Classical measurement error would attenuate our coefficient estimates toward zero (creating an ``errors-in-variables'' bias), making our significant findings conservative. However, if measurement error is correlated with meeting characteristics (e.g., if our models systematically misclassify statements from certain Fed Chairs or policy regimes), the bias could go in either direction. The inter-model agreement diagnostics (mean confidence 0.837 for novelty, 0.562--0.806 for tone) help identify statements where measurement may be less reliable, but they do not fully resolve this concern.
43 +
44 +\textit{Multiple testing.} With 7 assets, 5 dependent variables, and multiple inference methods, we test a large number of coefficients. We address this through Benjamini--Hochberg FDR correction within each family of tests and by reporting multi-method robustness counts. Nevertheless, some individually significant results---particularly those with $\#\text{Sig} = 1$ in the robustness tables---should be interpreted cautiously as potentially reflecting false discovery.
45 +
46 +\textit{Identification threats.} Our identification strategy exploits the pre-determined timing and content of FOMC statements, validated by pre-announcement placebo tests. However, we cannot fully rule out confounding from simultaneous information releases. FOMC announcements sometimes coincide with the Summary of Economic Projections (``dot plot''), which provides quantitative rate path forecasts that may interact with our textual measures. Similarly, market expectations about the subsequent press conference (beginning 30 minutes after the statement) may influence post-announcement price dynamics within our event window. These concurrent information flows are not fully separable from the statement text itself.
47 +
48 +\textit{Generalizability.} We study U.S. futures markets around FOMC statements only. Whether the same tone-return and novelty-volatility patterns hold for other central banks (ECB, BOJ, BOE), other Fed communications (minutes, speeches), or other asset classes (corporate bonds, emerging-market equities) is an open question.
49 +
50 +\subsection{Future Directions}
51 +
52 +Several extensions seem natural.
53 +
54 +\textit{More semantic dimensions.} We measure tone and novelty; FOMC statements also vary in uncertainty language, temporal focus (forward- vs.\ backward-looking), and internal consensus (voting dissents). The PCA axis construction can accommodate additional dimensions.
55 +
56 +\textit{International markets.} FOMC announcements move global asset prices \citep{wongswan2009,ehrmann2011}. Applying the tone-novelty decomposition to international data would show whether both channels transmit across borders or whether one dominates.
57 +
58 +\textit{Structural models.} Our evidence is reduced-form. A model with heterogeneous agents and differential processing of directional versus informational content could rationalize the fast-volatility / slow-return asymmetry and generate further predictions.
59 +
60 +\textit{Joint analysis with rate surprises.} Our textual MPS and the rate-based surprise of \citet{kuttner2001} measure different things. Estimating both jointly would reveal whether they contain complementary information and whether the information-versus-policy decomposition of \citet{jarocinski2020,nakamura2018} maps onto our tone-versus-novelty decomposition.
61 +
added PHD_chapitre3_theses_20260731/tables/tableD1_desc_1min_returns.tex +28 −0
@@ -0,0 +1,28 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Descriptive Statistics: 1-Minute Log Returns on FOMC Days}
6 +\label{tab:desc_1min_ret}
7 +\begin{threeparttable}
8 +\begin{tabular}{lrrrrrrrrrr}
9 +\toprule
10 +Ticker & $N$ & Mean & SD & Skew & Kurt & JB & Min & Median & Max & AC(1) \\
11 +\midrule
12 +E-mini S\&P 500 & 42,180 & -0.030 & 3.70 & 0.96 & 208.0 & 73858777$^{***}$ & -118.7 & 0.000 & 148.0 & -0.022 \\
13 +VIX Futures & 37,050 & 0.072 & 14.12 & 0.02 & 77.9 & 8665376$^{***}$ & -466.1 & 0.000 & 419.8 & -0.077 \\
14 +10Y T-Note & 42,180 & 0.009 & 1.07 & 1.21 & 44.4 & 3027292$^{***}$ & -13.8 & 0.000 & 30.8 & -0.103 \\
15 +5Y T-Note & 42,180 & 0.004 & 0.68 & -4.70 & 444.1 & 342093523$^{***}$ & -43.8 & 0.000 & 17.6 & -0.059 \\
16 +Dollar Index & 42,180 & -0.000 & 1.69 & -1.74 & 1508.6 & 3984214777$^{***}$ & -119.2 & 0.000 & 115.4 & 0.194 \\
17 +Crude Oil WTI & 42,180 & -0.006 & 5.14 & -0.64 & 166.7 & 47098526$^{***}$ & -191.8 & 0.000 & 143.9 & -0.062 \\
18 +Gold & 41,895 & 0.001 & 2.91 & 0.24 & 121.3 & 24424589$^{***}$ & -91.5 & 0.000 & 108.8 & -0.017 \\
19 +\bottomrule
20 +\end{tabular}
21 +\begin{tablenotes}[flushleft]
22 +\small
23 +\item \textit{Notes:} All statistics in basis points except $N$, AC(1), and JB. Sample includes all 1-minute observations on FOMC event days $\pm 1$ day with grid regularization and forward-filling of short gaps. Skewness and excess kurtosis are sample moments. JB is the Jarque--Bera test statistic. AC(1) denotes the first-order autocorrelation coefficient. Negative AC(1) values reflect bid--ask bounce effects typical of high-frequency data \citep{andersen1997}.
24 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
25 +\end{tablenotes}
26 +\end{threeparttable}
27 +\end{table}
28 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableD3_desc_rolling.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Descriptive Statistics: Rolling Realized Measures (5-Minute Window)}
6 +\label{tab:desc_rolling}
7 +\begin{threeparttable}
8 +\begin{tabular}{lrrrrrrrrrrr}
9 +\toprule
10 + & $N$ & \multicolumn{5}{c}{Realized Volatility (bps)} & \multicolumn{2}{c}{$\log(\text{RV})$} & \multicolumn{3}{c}{Realized Beta vs.\ ES} \\
11 +\cmidrule(lr){3-7} \cmidrule(lr){8-9} \cmidrule(lr){10-12}
12 +Contract & & Mean & SD & P5 & P50 & P95 & Mean & SD & Mean & SD & P50 \\
13 +\midrule
14 +E-mini S\&P 500 & 35,668 & 3.43 & 5.68 & 0.00 & 2.41 & 10.38 & -11.78 & 6.55 & 1.000 & 0.000 & 1.000 \\
15 +10Y T-Note & 35,668 & 1.41 & 1.63 & 0.00 & 1.31 & 3.81 & -14.11 & 7.09 & -0.018 & 0.395 & 0.000 \\
16 +5Y T-Note & 35,668 & 0.79 & 1.15 & 0.00 & 0.66 & 2.36 & -14.77 & 6.86 & -0.007 & 0.225 & 0.000 \\
17 +Dollar Index & 35,668 & 0.78 & 3.45 & 0.00 & 0.00 & 3.53 & -19.32 & 6.34 & -0.010 & 0.482 & 0.000 \\
18 +Crude Oil WTI & 35,668 & 5.86 & 9.47 & 0.00 & 3.97 & 17.90 & -11.24 & 6.67 & 0.184 & 1.472 & 0.006 \\
19 +Gold & 35,427 & 3.64 & 4.64 & 0.00 & 2.73 & 10.51 & -11.29 & 6.30 & 0.044 & 0.958 & 0.000 \\
20 +VIX Futures & 31,330 & 11.16 & 23.33 & 0.00 & 0.00 & 52.40 & -18.31 & 7.74 & -0.714 & 4.451 & 0.000 \\
21 +\bottomrule
22 +\end{tabular}
23 +\begin{tablenotes}[flushleft]
24 +\small
25 +\item \textit{Notes:} Statistics computed across all minute-level observations within $\pm 30$ minutes of FOMC announcements ($K = 5$ min rolling window, NA-tolerant with up to 20\% missing data). RV is the NA-tolerant rolling realized variance defined in Section~\ref{sec:methodology}, displayed in square-root (volatility) units in basis points; $\log(\text{RV})$ is computed on the realized variance in raw decimal-return units, so its level is not directly comparable to the bps columns. $\beta_t^{\text{real}} = \text{RCov}(r_i, r_{\text{ES}}) / \text{RVar}(r_{\text{ES}})$. ES beta is 1.000 by construction. $\log(\text{RV})$ is near-Gaussian, validating its use as a regression dependent variable.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableP_panel_beta_novelty.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: Beta on Post $\times$ Novelty (5-min)}
6 +\label{tab:panel_beta_novelty}
7 +\begin{threeparttable}
8 +\begin{tabular}{lcccccc}
9 +\toprule
10 + & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -0.6861*** & -0.0144 & -0.0114 & -0.0017 & 0.1629*** & 0.0538** \\
13 + & (0.1540) & (0.0120) & (0.0070) & (0.0011) & (0.0582) & (0.0268) \\
14 +Post $\times$ Novelty & -0.1361 & -0.0116 & -0.0064 & -0.0037 & -0.0295 & -0.0445 \\
15 + & (0.1765) & (0.0080) & (0.0048) & (0.0030) & (0.0706) & (0.0204) \\
16 +\midrule
17 +$N$ & 4,798 & 5,724 & 5,724 & 5,724 & 5,724 & 5,663 \\
18 +$R^2$ & 0.001 & 0.001 & 0.001 & 0.005 & 0.000 & 0.002 \\
19 +Adj.\ $R^2$ & 0.000 & 0.000 & 0.000 & 0.005 & 0.000 & 0.001 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: Realized Beta (5-min). Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableP_panel_beta_stance.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: Beta on Post $\times$ Stance (5-min)}
6 +\label{tab:panel_beta_stance}
7 +\begin{threeparttable}
8 +\begin{tabular}{lcccccc}
9 +\toprule
10 + & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -0.5789*** & -0.0206 & -0.0175** & -0.0023 & 0.0992 & 0.0405 \\
13 + & (0.1988) & (0.0153) & (0.0087) & (0.0016) & (0.0644) & (0.0342) \\
14 +Post $\times$ Stance & 0.1780 & -0.0179 & -0.0165 & -0.0023 & -0.1643 & -0.0415 \\
15 + & (0.2433) & (0.0178) & (0.0098) & (0.0017) & (0.0782) & (0.0404) \\
16 +\midrule
17 +$N$ & 4,798 & 5,724 & 5,724 & 5,724 & 5,724 & 5,663 \\
18 +$R^2$ & 0.000 & 0.001 & 0.002 & 0.001 & 0.003 & 0.001 \\
19 +Adj.\ $R^2$ & 0.000 & 0.001 & 0.002 & 0.001 & 0.003 & 0.000 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: Realized Beta (5-min). Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableP_panel_logrv_novelty.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: $\log$(RV) on Post $\times$ Novelty (5-min, $\pm$30 min)}
6 +\label{tab:panel_logrv_novelty}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -13.4894*** & -18.9261*** & -16.4259*** & -16.9510*** & -22.6505*** & -13.4467*** & -13.4834*** \\
13 + & (0.2892) & (0.3813) & (0.3396) & (0.3423) & (0.1015) & (0.2887) & (0.2946) \\
14 +Post $\times$ Novelty & -0.5169 & 0.3020 & -1.8142*** & -1.2188* & 0.8439 & -0.3328 & -0.4821 \\
15 + & (0.5996) & (0.8881) & (0.5683) & (0.5493) & (0.4881) & (0.6298) & (0.6316) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.003 & 0.001 & 0.031 & 0.015 & 0.069 & 0.001 & 0.002 \\
19 +Adj.\ $R^2$ & 0.002 & 0.001 & 0.031 & 0.015 & 0.068 & 0.001 & 0.002 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: $\log(\text{RV}_t(5))$. Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableP_panel_logrv_stance.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: $\log$(RV) on Post $\times$ Stance (5-min, $\pm$30 min)}
6 +\label{tab:panel_logrv_stance}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -14.8474*** & -19.9268*** & -17.7088*** & -17.9767*** & -22.6514*** & -14.9591*** & -15.0663*** \\
13 + & (0.4794) & (0.4462) & (0.4275) & (0.4466) & (0.1175) & (0.4704) & (0.4845) \\
14 +Post $\times$ Stance & -5.3084*** & -3.2805*** & -5.2530*** & -4.1583*** & 0.1481 & -5.8682*** & -6.0348*** \\
15 + & (0.8271) & (0.7803) & (0.6490) & (0.7052) & (0.2836) & (0.8526) & (0.8609) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.134 & 0.047 & 0.133 & 0.091 & 0.001 & 0.149 & 0.166 \\
19 +Adj.\ $R^2$ & 0.134 & 0.047 & 0.132 & 0.091 & 0.001 & 0.149 & 0.166 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: $\log(\text{RV}_t(5))$. Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableP_panel_return_novelty.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: Returns on Post $\times$ Novelty (5-min)}
6 +\label{tab:panel_return_novelty}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -0.02 & 0.00 & 0.01 & 0.00 & 0.01 & -0.07 & 0.04 \\
13 + & (0.03) & (0.19) & (0.01) & (0.01) & (0.01) & (0.10) & (0.03) \\
14 +Post $\times$ Novelty & -0.04 & 0.62 & 0.06 & 0.01 & 0.01 & -0.18 & 0.03 \\
15 + & (0.10) & (0.79) & (0.05) & (0.01) & (0.05) & (0.42) & (0.06) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.000 & 0.001 & 0.002 & 0.000 & 0.000 & 0.000 & 0.000 \\
19 +Adj.\ $R^2$ & -0.000 & 0.001 & 0.002 & -0.000 & -0.000 & 0.000 & -0.000 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: 1-Min Return (bps, 5-min). Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Coefficients $\times 10,000$. Semantic measures from MiniLM--BERT ensemble.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableP_panel_return_stance.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: Returns on Post $\times$ Stance (5-min)}
6 +\label{tab:panel_return_stance}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 0.01 & -0.18 & 0.01 & 0.01 & 0.03** & -0.02 & 0.04 \\
13 + & (0.04) & (0.16) & (0.01) & (0.01) & (0.01) & (0.05) & (0.03) \\
14 +Post $\times$ Stance & 0.11 & -0.41 & -0.02 & 0.02 & 0.05 & 0.15 & 0.02 \\
15 + & (0.09) & (0.52) & (0.04) & (0.01) & (0.02) & (0.30) & (0.05) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.000 & 0.000 & 0.000 & 0.000 & 0.001 & 0.000 & 0.000 \\
19 +Adj.\ $R^2$ & 0.000 & 0.000 & -0.000 & 0.000 & 0.000 & 0.000 & -0.000 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: 1-Min Return (bps, 5-min). Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Coefficients $\times 10,000$. Semantic measures from MiniLM--BERT ensemble.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableP_panel_rv_novelty.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: RV on Post $\times$ Novelty (5-min)}
6 +\label{tab:panel_rv_novelty}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 3.26*** & 10.79*** & 1.09*** & 0.64*** & 0.25*** & 5.85*** & 3.49*** \\
13 + & (0.33) & (1.54) & (0.09) & (0.07) & (0.08) & (0.62) & (0.27) \\
14 +Post $\times$ Novelty & 3.09 & 8.32 & 0.14 & 0.25 & 0.63 & 4.94 & 1.93 \\
15 + & (1.37) & (5.78) & (0.35) & (0.30) & (0.36) & (2.40) & (1.01) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.097 & 0.040 & 0.003 & 0.013 & 0.043 & 0.071 & 0.050 \\
19 +Adj.\ $R^2$ & 0.097 & 0.040 & 0.003 & 0.012 & 0.042 & 0.071 & 0.050 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: RV (bps, 5-min). Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableP_panel_rv_stance.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: RV on Post $\times$ Stance (5-min)}
6 +\label{tab:panel_rv_stance}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 2.79*** & 7.86*** & 0.88*** & 0.54*** & 0.23** & 4.32*** & 2.92*** \\
13 + & (0.40) & (1.56) & (0.10) & (0.08) & (0.09) & (0.56) & (0.30) \\
14 +Post $\times$ Stance & -1.27 & -7.49* & -0.76** & -0.35 & 0.05 & -4.97** & -1.79* \\
15 + & (0.91) & (3.41) & (0.28) & (0.24) & (0.22) & (1.92) & (0.85) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.008 & 0.021 & 0.042 & 0.012 & 0.000 & 0.036 & 0.021 \\
19 +Adj.\ $R^2$ & 0.008 & 0.021 & 0.042 & 0.012 & 0.000 & 0.036 & 0.021 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: RV (bps, 5-min). Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_delta_beta_30min.tex +33 −0
@@ -0,0 +1,33 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Rolling Realized Beta Change Regressions (30-min Window)}
6 +\label{tab:rolling_delta_beta_30min}
7 +\begin{threeparttable}
8 +\begin{tabular}{lcccccc}
9 +\toprule
10 + & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -0.2364 & 0.0098 & -0.0025 & 0.0058 & -0.0933 & 0.1275* \\
13 + & (0.1651) & (0.0192) & (0.0134) & (0.0074) & (0.0650) & (0.0758) \\
14 +Stance & -0.0190 & 0.0118 & 0.0195 & 0.0137 & -0.0008 & 0.1244 \\
15 + & (0.2653) & (0.0198) & (0.0158) & (0.0093) & (0.0891) & (0.0869) \\
16 +Novelty & 0.1096 & -0.0096 & -0.0064 & -0.0086 & -0.1290 & -0.0539 \\
17 + & (0.0917) & (0.0102) & (0.0073) & (0.0075) & (0.1020) & (0.0344) \\
18 +Stance $\times$ Novelty & -0.0818 & 0.0064 & 0.0074 & 0.0011 & -0.1772 & -0.0852 \\
19 + & (0.1018) & (0.0118) & (0.0074) & (0.0080) & (0.2106) & (0.0755) \\
20 +\midrule
21 +$N$ & 71 & 88 & 88 & 88 & 88 & 87 \\
22 +$R^2$ & 0.009 & 0.003 & 0.008 & 0.045 & 0.064 & 0.047 \\
23 +Adj.\ $R^2$ & -0.036 & -0.033 & -0.027 & 0.010 & 0.031 & 0.013 \\
24 +\bottomrule
25 +\end{tabular}
26 +\begin{tablenotes}[flushleft]
27 +\small
28 +\item \textit{Notes:} Dependent variable: $\Delta\hat{\beta}$ (change in rolling realized beta relative to ES, 30-min window). Coefficients are in beta units; a coefficient of $0.12$ means that a one-standard-deviation increase in the regressor is associated with a $0.12$ increase in co-movement with the S\&P~500. All regressors z-scored. Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Semantic measures from MiniLM--BERT ensemble.
29 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
30 +\end{tablenotes}
31 +\end{threeparttable}
32 +\end{table}
33 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_delta_log_rv_30min.tex +33 −0
@@ -0,0 +1,33 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Rolling Delta log(RV) Regressions (30-min Window, Interaction Model)}
6 +\label{tab:rolling_delta_log_rv_30min}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 6.1744*** & 6.8238*** & 7.1287*** & 6.9380*** & -0.3664 & 8.4210*** & 8.4805*** \\
13 + & (1.0374) & (1.6303) & (1.0670) & (1.0305) & (0.3671) & (0.8868) & (0.8682) \\
14 +Stance & -2.9143 & -0.8191 & -2.5521 & -2.3930 & -0.9082 & -2.1849 & -1.7794 \\
15 + & (1.0198) & (1.8314) & (1.1059) & (1.0661) & (0.5066) & (1.0687) & (1.0675) \\
16 +Novelty & 0.3441 & -0.2304 & -0.7730 & -0.6468 & 0.5076 & 0.8902 & 0.7810 \\
17 + & (0.6037) & (0.9547) & (0.5429) & (0.5328) & (0.5352) & (0.4145) & (0.3970) \\
18 +Stance $\times$ Novelty & -0.6767 & -1.5107 & -1.0145 & -1.0850 & -0.3540 & -0.7406 & -0.6972 \\
19 + & (0.6652) & (0.7161) & (0.5278) & (0.5025) & (0.7771) & (0.6279) & (0.6369) \\
20 +\midrule
21 +$N$ & 148 & 130 & 148 & 148 & 148 & 148 & 147 \\
22 +$R^2$ & 0.064 & 0.018 & 0.075 & 0.071 & 0.055 & 0.052 & 0.041 \\
23 +Adj.\ $R^2$ & 0.044 & -0.005 & 0.056 & 0.051 & 0.036 & 0.033 & 0.021 \\
24 +\bottomrule
25 +\end{tabular}
26 +\begin{tablenotes}[flushleft]
27 +\small
28 +\item \textit{Notes:} Dependent variable: $\Delta\log(\text{RV})$ (Rolling 30-min). Coefficients are in log-units; a coefficient of $-2.91$ on Stance for ES means that a one-standard-deviation hawkish shift is associated with a $2.91$ log-unit decrease in realized volatility. All regressors z-scored. Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Semantic measures from MiniLM--BERT ensemble.
29 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
30 +\end{tablenotes}
31 +\end{threeparttable}
32 +\end{table}
33 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_delta_rv_30min.tex +33 −0
@@ -0,0 +1,33 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Rolling Delta RV Regressions (30-min Window, Interaction Model)}
6 +\label{tab:rolling_delta_rv_30min}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 5.54*** & 26.00*** & 2.23*** & 1.65*** & -0.40 & 12.32*** & 9.96*** \\
13 + & (1.67) & (6.51) & (0.64) & (0.49) & (0.62) & (2.02) & (1.51) \\
14 +Stance & -4.94** & -2.80 & -1.50* & -1.30* & -1.65 & -7.78** & -1.05 \\
15 + & (1.53) & (6.03) & (0.71) & (0.64) & (0.90) & (2.65) & (1.92) \\
16 +Novelty & 6.82* & 3.42 & 0.70 & 0.94 & 1.15 & 8.20* & 3.34* \\
17 + & (3.42) & (5.98) & (0.81) & (0.83) & (1.09) & (3.63) & (1.35) \\
18 +Stance $\times$ Novelty & -5.32 & -7.96** & -1.66* & -2.48* & -1.20 & -8.61* & -4.68 \\
19 + & (3.15) & (2.55) & (0.83) & (1.12) & (1.39) & (3.60) & (3.13) \\
20 +\midrule
21 +$N$ & 148 & 130 & 148 & 148 & 148 & 148 & 147 \\
22 +$R^2$ & 0.286 & 0.054 & 0.131 & 0.245 & 0.137 & 0.268 & 0.128 \\
23 +Adj.\ $R^2$ & 0.271 & 0.032 & 0.113 & 0.229 & 0.119 & 0.253 & 0.109 \\
24 +\bottomrule
25 +\end{tabular}
26 +\begin{tablenotes}[flushleft]
27 +\small
28 +\item \textit{Notes:} Dependent variable: Delta RV (bps) (Rolling 30-min). Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Coefficients $\times 10,000$. Semantic measures from MiniLM--BERT ensemble.
29 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
30 +\end{tablenotes}
31 +\end{threeparttable}
32 +\end{table}
33 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_irf_novelty_abn_h.tex +36 −0
@@ -0,0 +1,36 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Novelty on Abnormal Cumulative Return}
6 +\label{tab:irf_novelty_abn_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & -- & -- & -- & -- & -- \\
13 +VIX Futures & 10.21 & 1.84 & -4.80 & -10.09 & 7.18 \\
14 + & (9.77) & (10.47) & (8.32) & (6.08) & (10.24) \\
15 +10Y T-Note & 1.81 & 2.59 & 2.17 & 2.48 & 7.61 \\
16 + & (0.97) & (2.14) & (1.87) & (2.07) & (3.41) \\
17 +5Y T-Note & 0.38 & 1.26 & 0.81 & 0.61 & 5.00 \\
18 + & (0.58) & (1.71) & (1.37) & (1.91) & (2.57) \\
19 +Dollar Index & 1.40 & 1.72 & 1.26 & 1.56 & 4.51 \\
20 + & (1.48) & (2.77) & (2.28) & (2.92) & (3.68) \\
21 +Crude Oil WTI & 0.47 & 0.60 & -3.13 & -1.42 & 2.62 \\
22 + & (3.34) & (3.64) & (7.50) & (6.75) & (5.73) \\
23 +Gold & -0.30 & 2.34 & 1.43 & -0.21 & 6.27 \\
24 + & (1.46) & (3.08) & (2.70) & (2.63) & (2.70) \\
25 +\midrule
26 +$N$ & 148 & 148 & 148 & 148 & 148 \\
27 +\bottomrule
28 +\end{tabular}
29 +\begin{tablenotes}[flushleft]
30 +\small
31 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Novelty reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. ES entries are omitted because abnormal returns are defined relative to the ES benchmark and are zero by construction. Pre-announcement placebo test passed: no significant pre-event coefficients.
32 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
33 +\end{tablenotes}
34 +\end{threeparttable}
35 +\end{table}
36 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_irf_novelty_ret_h.tex +37 −0
@@ -0,0 +1,37 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Novelty on Cumulative Return}
6 +\label{tab:irf_novelty_ret_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & -0.99 & -1.42 & -0.65 & -0.61 & -3.91 \\
13 + & (0.56) & (1.70) & (1.34) & (1.82) & (2.25) \\
14 +VIX Futures & 10.24 & 4.84 & -2.55 & -6.02 & 6.22 \\
15 + & (9.19) & (9.31) & (7.62) & (5.36) & (8.33) \\
16 +10Y T-Note & 0.82 & 1.17 & 1.52 & 1.87 & 3.70 \\
17 + & (0.60) & (0.97) & (0.91) & (1.09) & (2.15) \\
18 +5Y T-Note & -0.61 & -0.16 & 0.16 & -0.00 & 1.08 \\
19 + & (0.41) & (0.10) & (0.17) & (0.27) & (0.47) \\
20 +Dollar Index & 0.41 & 0.30 & 0.61 & 0.94 & 0.60 \\
21 + & (1.34) & (1.74) & (1.51) & (1.62) & (2.22) \\
22 +Crude Oil WTI & -0.52 & -0.82 & -3.78 & -2.03 & -1.29 \\
23 + & (3.54) & (4.29) & (7.83) & (6.83) & (6.20) \\
24 +Gold & -1.30 & 0.84 & 0.55 & -1.13 & 2.00 \\
25 + & (1.42) & (2.62) & (2.33) & (1.89) & (2.36) \\
26 +\midrule
27 +$N$ & 148 & 148 & 148 & 148 & 148 \\
28 +\bottomrule
29 +\end{tabular}
30 +\begin{tablenotes}[flushleft]
31 +\small
32 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Novelty reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. Pre-announcement placebo test passed: no significant pre-event coefficients.
33 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
34 +\end{tablenotes}
35 +\end{threeparttable}
36 +\end{table}
37 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_irf_stance_abn_h.tex +36 −0
@@ -0,0 +1,36 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Policy Stance on Abnormal Cumulative Return}
6 +\label{tab:irf_stance_abn_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & -- & -- & -- & -- & -- \\
13 +VIX Futures & -15.04 & -22.74 & -28.86 & -28.65 & -30.54 \\
14 + & (7.07) & (9.76) & (10.84) & (16.52) & (37.21) \\
15 +10Y T-Note & -0.85 & -3.12 & -4.67 & -8.29 & -15.86 \\
16 + & (1.20) & (2.81) & (3.59) & (4.03) & (8.44) \\
17 +5Y T-Note & 0.66 & -1.35 & -2.86 & -6.03 & -12.68 \\
18 + & (0.85) & (2.20) & (3.10) & (3.63) & (7.84) \\
19 +Dollar Index & 0.78 & -0.55 & -2.72 & -7.27 & -15.99 \\
20 + & (0.92) & (1.86) & (2.63) & (3.73) & (8.39) \\
21 +Crude Oil WTI & -2.33 & -2.79 & 0.99 & -3.76 & -2.65 \\
22 + & (3.71) & (5.05) & (7.17) & (8.73) & (8.44) \\
23 +Gold & -1.77 & -5.75 & -5.34 & -8.84** & -13.76 \\
24 + & (1.25) & (2.57) & (2.54) & (2.66) & (5.73) \\
25 +\midrule
26 +$N$ & 148 & 148 & 148 & 148 & 148 \\
27 +\bottomrule
28 +\end{tabular}
29 +\begin{tablenotes}[flushleft]
30 +\small
31 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Policy Stance reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. ES entries are omitted because abnormal returns are defined relative to the ES benchmark and are zero by construction. Pre-announcement placebo test passed: no significant pre-event coefficients.
32 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
33 +\end{tablenotes}
34 +\end{threeparttable}
35 +\end{table}
36 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_irf_stance_ret_h.tex +37 −0
@@ -0,0 +1,37 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Policy Stance on Cumulative Return}
6 +\label{tab:irf_stance_ret_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & 0.23 & 1.98 & 3.76 & 6.82 & 11.98 \\
13 + & (0.92) & (2.11) & (2.76) & (3.31) & (6.76) \\
14 +VIX Futures & -14.74 & -19.11 & -22.03 & -17.73 & -14.27 \\
15 + & (6.26) & (8.33) & (9.47) & (14.62) & (31.57) \\
16 +10Y T-Note & -0.61 & -1.14 & -0.90 & -1.47 & -3.88 \\
17 + & (0.43) & (0.85) & (0.93) & (0.92) & (2.20) \\
18 +5Y T-Note & 0.89 & 0.63 & 0.90 & 0.79 & -0.71 \\
19 + & (0.34) & (0.35) & (0.55) & (0.51) & (1.34) \\
20 +Dollar Index & 1.01 & 1.43 & 1.05 & -0.45 & -4.01 \\
21 + & (0.43) & (0.61) & (0.45) & (0.89) & (2.14) \\
22 +Crude Oil WTI & -2.10 & -0.81 & 4.75 & 3.06 & 9.32 \\
23 + & (4.28) & (6.28) & (8.46) & (10.23) & (9.69) \\
24 +Gold & -1.47 & -3.08 & 0.51 & 0.62 & 1.31 \\
25 + & (1.30) & (1.42) & (2.04) & (2.46) & (3.96) \\
26 +\midrule
27 +$N$ & 148 & 148 & 148 & 148 & 148 \\
28 +\bottomrule
29 +\end{tabular}
30 +\begin{tablenotes}[flushleft]
31 +\small
32 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Policy Stance reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. Pre-announcement placebo test passed: no significant pre-event coefficients.
33 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
34 +\end{tablenotes}
35 +\end{threeparttable}
36 +\end{table}
37 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_irf_stance_x_novelty_abn_h.tex +36 −0
@@ -0,0 +1,36 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Stance $\times$ Novelty on Abnormal Cumulative Return}
6 +\label{tab:irf_stance_x_novelty_abn_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & -- & -- & -- & -- & -- \\
13 +VIX Futures & -31.74** & -58.80*** & -72.89*** & -65.76*** & -53.38* \\
14 + & (9.50) & (14.98) & (18.63) & (13.00) & (17.64) \\
15 +10Y T-Note & -4.54 & -10.44 & -11.95 & -14.24 & -12.05 \\
16 + & (2.09) & (5.70) & (6.03) & (6.30) & (6.89) \\
17 +5Y T-Note & -1.63 & -6.94 & -8.03 & -9.66 & -5.72 \\
18 + & (0.99) & (4.18) & (4.36) & (4.58) & (4.66) \\
19 +Dollar Index & 0.18 & -2.85 & -5.11 & -6.15 & -3.52 \\
20 + & (1.65) & (3.98) & (4.37) & (5.05) & (4.70) \\
21 +Crude Oil WTI & 4.84 & 0.86 & 10.90 & 6.59 & 12.00 \\
22 + & (6.44) & (8.45) & (14.43) & (16.04) & (15.12) \\
23 +Gold & -1.30 & -12.83 & -10.31 & -7.35 & -6.73 \\
24 + & (1.32) & (5.31) & (4.61) & (3.60) & (4.95) \\
25 +\midrule
26 +$N$ & 148 & 148 & 148 & 148 & 148 \\
27 +\bottomrule
28 +\end{tabular}
29 +\begin{tablenotes}[flushleft]
30 +\small
31 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Stance $\times$ Novelty reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. ES entries are omitted because abnormal returns are defined relative to the ES benchmark and are zero by construction. Pre-announcement placebo test passed: no significant pre-event coefficients.
32 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
33 +\end{tablenotes}
34 +\end{threeparttable}
35 +\end{table}
36 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_irf_stance_x_novelty_ret_h.tex +37 −0
@@ -0,0 +1,37 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Stance $\times$ Novelty on Cumulative Return}
6 +\label{tab:irf_stance_x_novelty_ret_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & 2.78 & 6.91 & 8.27 & 9.47 & 6.35 \\
13 + & (1.36) & (4.14) & (4.39) & (4.61) & (4.49) \\
14 +VIX Futures & -28.02** & -47.97** & -61.74*** & -51.80*** & -44.06* \\
15 + & (8.62) & (12.75) & (15.54) & (10.66) & (14.53) \\
16 +10Y T-Note & -1.76 & -3.53 & -3.68 & -4.77 & -5.70 \\
17 + & (0.85) & (1.68) & (1.74) & (1.88) & (2.84) \\
18 +5Y T-Note & 1.15 & -0.03 & 0.24 & -0.19 & 0.63 \\
19 + & (0.57) & (0.10) & (0.16) & (0.28) & (0.46) \\
20 +Dollar Index & 2.96 & 4.05 & 3.16 & 3.32 & 2.83 \\
21 + & (1.15) & (1.53) & (1.27) & (1.41) & (2.21) \\
22 +Crude Oil WTI & 7.62 & 7.77 & 19.17 & 16.07 & 18.35 \\
23 + & (7.45) & (11.46) & (18.20) & (19.84) & (18.66) \\
24 +Gold & 1.48 & -5.97* & -2.19 & 1.94 & -0.60 \\
25 + & (1.59) & (2.08) & (1.57) & (2.34) & (2.13) \\
26 +\midrule
27 +$N$ & 148 & 148 & 148 & 148 & 148 \\
28 +\bottomrule
29 +\end{tabular}
30 +\begin{tablenotes}[flushleft]
31 +\small
32 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Stance $\times$ Novelty reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. Pre-announcement placebo test passed: no significant pre-event coefficients.
33 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
34 +\end{tablenotes}
35 +\end{threeparttable}
36 +\end{table}
37 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_log_rv_ratio_30min.tex +33 −0
@@ -0,0 +1,33 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Rolling log(RV Ratio) Regressions (30-min Window, Interaction Model)}
6 +\label{tab:rolling_log_rv_ratio_30min}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 0.2577*** & -0.6826 & -0.1044 & -0.0595 & -2.2662*** & 0.2341 & 0.4248 \\
13 + & (0.0810) & (0.6770) & (0.1403) & (0.0703) & (0.7194) & (0.3526) & (0.3908) \\
14 +Stance & -0.2575 & 0.4071 & -0.1965 & -0.1674 & -4.7532 & -0.7852 & -0.8420 \\
15 + & (0.1249) & (0.5125) & (0.1874) & (0.1886) & (1.5983) & (0.5421) & (0.5977) \\
16 +Novelty & -0.0173 & -0.1680 & 0.0871 & 0.1068 & -0.0110 & 0.3561 & 0.4141 \\
17 + & (0.0437) & (0.6219) & (0.1386) & (0.1650) & (0.3329) & (0.1975) & (0.1945) \\
18 +Stance $\times$ Novelty & 0.0890 & 0.1548 & 0.1046 & -0.0441 & 1.2119 & 0.1608 & -0.0041 \\
19 + & (0.0779) & (0.4344) & (0.0892) & (0.0755) & (0.7615) & (0.3288) & (0.3215) \\
20 +\midrule
21 +$N$ & 88 & 41 & 40 & 40 & 17 & 89 & 88 \\
22 +$R^2$ & 0.069 & 0.004 & 0.183 & 0.078 & 0.366 & 0.086 & 0.083 \\
23 +Adj.\ $R^2$ & 0.036 & -0.077 & 0.114 & 0.002 & 0.220 & 0.054 & 0.050 \\
24 +\bottomrule
25 +\end{tabular}
26 +\begin{tablenotes}[flushleft]
27 +\small
28 +\item \textit{Notes:} Dependent variable: log(RV Ratio) (Rolling 30-min). Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Semantic measures from MiniLM--BERT ensemble.
29 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
30 +\end{tablenotes}
31 +\end{threeparttable}
32 +\end{table}
33 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_robustness_30min_A.tex +63 −0
@@ -0,0 +1,63 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\scriptsize
5 +\caption{Robustness: Event-Level Regressions --- $\Delta$RV and $\Delta\log$RV (30-Minute Window, Interaction Model)}
6 +\label{tab:robustness_A}
7 +\begin{threeparttable}
8 +\begin{tabular}{lllrrllllll}
9 +\toprule
10 +Contract & Dep.\ Var. & Regressor & $\hat{\beta}$ & $t$ & $p_{\text{NW}}$ & $p_{\text{Boot}}$ & $p_{\text{Clust}}$ & $p_{\text{QReg}}$ & $p_{\text{Perm}}$ & \#Sig \\
11 +\midrule
12 +Crude Oil WTI & $\Delta$RV & Stance & -7.78 & -2.94 & 0.004 & 0.691 & 0.001 & -- & 0.007 & 3 \\
13 +Crude Oil WTI & $\Delta$RV & Novelty & 8.20 & 2.26 & 0.025 & 0.068 & 0.009 & -- & 0.000 & 4 \\
14 +Crude Oil WTI & $\Delta$RV & St. $\times$ Nov. & -8.61 & -2.39 & 0.018 & 0.212 & 0.014 & -- & 0.002 & 3 \\
15 +Dollar Index & $\Delta$RV & Stance & -1.65 & -1.83 & 0.069 & 0.260 & 0.095 & -- & 0.014 & 3 \\
16 +Dollar Index & $\Delta$RV & Novelty & 1.15 & 1.06 & 0.293 & 0.069 & 0.366 & -- & 0.014 & 2 \\
17 +Dollar Index & $\Delta$RV & St. $\times$ Nov. & -1.20 & -0.86 & 0.392 & 0.124 & 0.469 & -- & 0.047 & 1 \\
18 +E-mini S\&P 500 & $\Delta$RV & Stance & -4.94 & -3.22 & 0.002 & 0.584 & 0.011 & -- & 0.019 & 3 \\
19 +E-mini S\&P 500 & $\Delta$RV & Novelty & 6.82 & 2.00 & 0.048 & 0.013 & 0.028 & -- & 0.002 & 4 \\
20 +E-mini S\&P 500 & $\Delta$RV & St. $\times$ Nov. & -5.32 & -1.69 & 0.094 & 0.150 & 0.167 & -- & 0.024 & 2 \\
21 +Gold & $\Delta$RV & Stance & -1.05 & -0.55 & 0.583 & 0.692 & 0.602 & -- & 0.561 & 0 \\
22 +Gold & $\Delta$RV & Novelty & 3.34 & 2.48 & 0.014 & 0.318 & 0.056 & -- & 0.021 & 3 \\
23 +Gold & $\Delta$RV & St. $\times$ Nov. & -4.68 & -1.49 & 0.138 & 0.266 & 0.117 & -- & 0.025 & 1 \\
24 +VIX Futures & $\Delta$RV & Stance & -2.80 & -0.46 & 0.644 & 0.659 & 0.580 & -- & 0.619 & 0 \\
25 +VIX Futures & $\Delta$RV & Novelty & 3.42 & 0.57 & 0.568 & 0.492 & 0.486 & -- & 0.382 & 0 \\
26 +VIX Futures & $\Delta$RV & St. $\times$ Nov. & -7.96 & -3.12 & 0.002 & 0.601 & 0.058 & -- & 0.067 & 3 \\
27 +5Y T-Note & $\Delta$RV & Stance & -1.30 & -2.04 & 0.043 & 0.236 & 0.235 & -- & 0.050 & 2 \\
28 +5Y T-Note & $\Delta$RV & Novelty & 0.94 & 1.13 & 0.260 & 0.203 & 0.235 & -- & 0.041 & 1 \\
29 +5Y T-Note & $\Delta$RV & St. $\times$ Nov. & -2.48 & -2.22 & 0.028 & 0.070 & 0.149 & -- & 0.004 & 3 \\
30 +10Y T-Note & $\Delta$RV & Stance & -1.50 & -2.10 & 0.038 & 0.210 & 0.210 & -- & 0.025 & 2 \\
31 +10Y T-Note & $\Delta$RV & Novelty & 0.70 & 0.87 & 0.388 & 0.314 & 0.313 & -- & 0.082 & 1 \\
32 +10Y T-Note & $\Delta$RV & St. $\times$ Nov. & -1.66 & -1.99 & 0.048 & 0.135 & 0.250 & -- & 0.013 & 2 \\
33 +\midrule
34 +Crude Oil WTI & $\Delta\log$RV & Stance & -2.1849 & -2.04 & 0.043 & 0.442 & 0.035 & -- & 0.022 & 3 \\
35 +Crude Oil WTI & $\Delta\log$RV & Novelty & 0.8902 & 2.15 & 0.033 & 0.614 & 0.086 & -- & 0.135 & 2 \\
36 +Crude Oil WTI & $\Delta\log$RV & St. $\times$ Nov. & -0.7406 & -1.18 & 0.240 & 0.650 & 0.203 & -- & 0.326 & 0 \\
37 +Dollar Index & $\Delta\log$RV & Stance & -0.9082 & -1.79 & 0.075 & 0.337 & 0.121 & -- & 0.041 & 2 \\
38 +Dollar Index & $\Delta\log$RV & Novelty & 0.5076 & 0.95 & 0.345 & 0.209 & 0.478 & -- & 0.064 & 1 \\
39 +Dollar Index & $\Delta\log$RV & St. $\times$ Nov. & -0.3540 & -0.46 & 0.649 & 0.464 & 0.673 & -- & 0.247 & 0 \\
40 +E-mini S\&P 500 & $\Delta\log$RV & Stance & -2.9143 & -2.86 & 0.005 & 0.716 & 0.001 & -- & 0.004 & 3 \\
41 +E-mini S\&P 500 & $\Delta\log$RV & Novelty & 0.3441 & 0.57 & 0.570 & 0.639 & 0.609 & -- & 0.577 & 0 \\
42 +E-mini S\&P 500 & $\Delta\log$RV & St. $\times$ Nov. & -0.6767 & -1.02 & 0.311 & 0.644 & 0.249 & -- & 0.389 & 0 \\
43 +Gold & $\Delta\log$RV & Stance & -1.7794 & -1.67 & 0.098 & 0.433 & 0.093 & -- & 0.066 & 3 \\
44 +Gold & $\Delta\log$RV & Novelty & 0.7810 & 1.97 & 0.051 & 0.583 & 0.120 & -- & 0.168 & 1 \\
45 +Gold & $\Delta\log$RV & St. $\times$ Nov. & -0.6972 & -1.09 & 0.275 & 0.618 & 0.236 & -- & 0.334 & 0 \\
46 +VIX Futures & $\Delta\log$RV & Stance & -0.8191 & -0.45 & 0.655 & 0.610 & 0.563 & -- & 0.558 & 0 \\
47 +VIX Futures & $\Delta\log$RV & Novelty & -0.2304 & -0.24 & 0.810 & 0.816 & 0.794 & -- & 0.817 & 0 \\
48 +VIX Futures & $\Delta\log$RV & St. $\times$ Nov. & -1.5107 & -2.11 & 0.037 & 0.781 & 0.037 & -- & 0.192 & 2 \\
49 +5Y T-Note & $\Delta\log$RV & Stance & -2.3930 & -2.24 & 0.026 & 0.545 & 0.012 & -- & 0.014 & 3 \\
50 +5Y T-Note & $\Delta\log$RV & Novelty & -0.6468 & -1.21 & 0.227 & 0.624 & 0.164 & -- & 0.293 & 0 \\
51 +5Y T-Note & $\Delta\log$RV & St. $\times$ Nov. & -1.0850 & -2.16 & 0.032 & 0.827 & 0.024 & -- & 0.163 & 2 \\
52 +10Y T-Note & $\Delta\log$RV & Stance & -2.5521 & -2.31 & 0.022 & 0.504 & 0.010 & -- & 0.012 & 3 \\
53 +10Y T-Note & $\Delta\log$RV & Novelty & -0.7730 & -1.42 & 0.157 & 0.697 & 0.090 & -- & 0.224 & 1 \\
54 +10Y T-Note & $\Delta\log$RV & St. $\times$ Nov. & -1.0145 & -1.92 & 0.057 & 0.791 & 0.044 & -- & 0.205 & 2 \\
55 +\bottomrule
56 +\end{tabular}
57 +\begin{tablenotes}[flushleft]
58 +\scriptsize
59 +\item \textit{Notes:} This table reports multi-method robustness results for the event-level interaction model $\Delta Y_i = \alpha + \beta_1 \text{Stance}_i + \beta_2 \text{Novelty}_i + \beta_3 (\text{Stance}_i \times \text{Novelty}_i) + \varepsilon_i$ using 30-minute rolling windows. The upper panel reports $\Delta$RV coefficients in basis points; the lower panel reports $\Delta\log$RV coefficients in native units. All regressors are z-scored. Five inference methods are compared: $p_{\text{NW}}$ = Newey--West HAC; $p_{\text{Boot}}$ = wild bootstrap (1,999 replications, Rademacher weights); $p_{\text{Clust}}$ = clustered by event date (HC1); $p_{\text{QReg}}$ = quantile regression at median ($\tau = 0.5$); $p_{\text{Perm}}$ = permutation test (4,999 replications). \#Sig = number of methods yielding $p < 0.10$. A coefficient is considered robust when $\#\text{Sig} \geq 3$. Stance effects on $\Delta\log$RV are significant for 5 of 7 contracts with $\#\text{Sig} \geq 3$, confirming the volatility channel.
60 +\end{tablenotes}
61 +\end{threeparttable}
62 +\end{table}
63 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_robustness_30min_B.tex +60 −0
@@ -0,0 +1,60 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\scriptsize
5 +\caption{Robustness: Event-Level Regressions --- RV Ratio and $\Delta\beta$ (30-Minute Window, Interaction Model)}
6 +\label{tab:robustness_B}
7 +\begin{threeparttable}
8 +\begin{tabular}{lllrrllllll}
9 +\toprule
10 +Contract & Dep.\ Var. & Regressor & $\hat{\beta}$ & $t$ & $p_{\text{NW}}$ & $p_{\text{Boot}}$ & $p_{\text{Clust}}$ & $p_{\text{QReg}}$ & $p_{\text{Perm}}$ & \#Sig \\
11 +\midrule
12 +Crude Oil WTI & RV Ratio & Stance & 0.0611 & 0.13 & 0.895 & 0.894 & 0.894 & -- & 0.886 & 0 \\
13 +Crude Oil WTI & RV Ratio & Novelty & 0.4481 & 1.16 & 0.249 & 0.479 & 0.232 & -- & 0.123 & 0 \\
14 +Crude Oil WTI & RV Ratio & St. $\times$ Nov. & -0.1615 & -0.28 & 0.784 & 0.777 & 0.782 & -- & 0.694 & 0 \\
15 +Dollar Index & RV Ratio & Stance & -1.1473 & -1.50 & 0.158 & -- & -- & -- & -- & 0 \\
16 +Dollar Index & RV Ratio & Novelty & -0.4007 & -2.15 & 0.051 & -- & -- & -- & -- & 1 \\
17 +Dollar Index & RV Ratio & St. $\times$ Nov. & 0.3796 & 1.66 & 0.121 & -- & -- & -- & -- & 0 \\
18 +E-mini S\&P 500 & RV Ratio & Stance & -0.4896 & -1.62 & 0.110 & 0.846 & 0.021 & -- & 0.167 & 1 \\
19 +E-mini S\&P 500 & RV Ratio & Novelty & -0.1825 & -1.58 & 0.118 & 0.851 & 0.079 & -- & 0.403 & 1 \\
20 +E-mini S\&P 500 & RV Ratio & St. $\times$ Nov. & 0.2511 & 1.20 & 0.232 & 0.856 & 0.088 & -- & 0.382 & 1 \\
21 +Gold & RV Ratio & Stance & 0.0845 & 0.16 & 0.872 & 0.896 & 0.881 & -- & 0.876 & 0 \\
22 +Gold & RV Ratio & Novelty & 0.6432 & 1.41 & 0.162 & 0.432 & 0.159 & -- & 0.093 & 1 \\
23 +Gold & RV Ratio & St. $\times$ Nov. & -0.5527 & -1.08 & 0.284 & 0.650 & 0.254 & -- & 0.296 & 0 \\
24 +VIX Futures & RV Ratio & Stance & 0.4669 & 1.79 & 0.081 & 0.655 & 0.142 & -- & 0.214 & 1 \\
25 +VIX Futures & RV Ratio & Novelty & -0.3809 & -1.33 & 0.191 & 0.641 & 0.289 & -- & 0.290 & 0 \\
26 +VIX Futures & RV Ratio & St. $\times$ Nov. & -0.0118 & -0.08 & 0.940 & 0.984 & 0.954 & -- & 0.963 & 0 \\
27 +5Y T-Note & RV Ratio & Stance & -0.0861 & -0.65 & 0.523 & 0.559 & 0.486 & -- & 0.360 & 0 \\
28 +5Y T-Note & RV Ratio & Novelty & 0.0596 & 0.42 & 0.675 & 0.717 & 0.666 & -- & 0.667 & 0 \\
29 +5Y T-Note & RV Ratio & St. $\times$ Nov. & -0.0415 & -0.57 & 0.574 & 0.716 & 0.624 & -- & 0.666 & 0 \\
30 +10Y T-Note & RV Ratio & Stance & -0.0692 & -0.63 & 0.536 & 0.642 & 0.605 & -- & 0.476 & 0 \\
31 +10Y T-Note & RV Ratio & Novelty & 0.0436 & 0.36 & 0.717 & 0.781 & 0.775 & -- & 0.744 & 0 \\
32 +10Y T-Note & RV Ratio & St. $\times$ Nov. & 0.1231 & 1.54 & 0.133 & 0.562 & 0.182 & -- & 0.234 & 0 \\
33 +\midrule
34 +Crude Oil WTI & $\Delta\beta$ & Stance & -0.0008 & -0.01 & 0.993 & 0.998 & 0.994 & -- & 0.996 & 0 \\
35 +Crude Oil WTI & $\Delta\beta$ & Novelty & -0.1290 & -1.26 & 0.210 & 0.586 & 0.213 & -- & 0.149 & 0 \\
36 +Crude Oil WTI & $\Delta\beta$ & St. $\times$ Nov. & -0.1772 & -0.84 & 0.403 & 0.442 & 0.393 & -- & 0.157 & 0 \\
37 +Dollar Index & $\Delta\beta$ & Stance & 0.0137 & 1.48 & 0.144 & 0.456 & 0.162 & -- & 0.100 & 1 \\
38 +Dollar Index & $\Delta\beta$ & Novelty & -0.0086 & -1.15 & 0.254 & 0.391 & 0.310 & -- & 0.118 & 0 \\
39 +Dollar Index & $\Delta\beta$ & St. $\times$ Nov. & 0.0011 & 0.13 & 0.894 & 0.935 & 0.889 & -- & 0.813 & 0 \\
40 +Gold & $\Delta\beta$ & Stance & 0.1244 & 1.43 & 0.156 & 0.579 & 0.116 & -- & 0.157 & 0 \\
41 +Gold & $\Delta\beta$ & Novelty & -0.0539 & -1.56 & 0.121 & 0.721 & 0.127 & -- & 0.335 & 0 \\
42 +Gold & $\Delta\beta$ & St. $\times$ Nov. & -0.0852 & -1.13 & 0.262 & 0.646 & 0.177 & -- & 0.270 & 0 \\
43 +VIX Futures & $\Delta\beta$ & Stance & -0.0190 & -0.07 & 0.943 & 0.944 & 0.946 & -- & 0.936 & 0 \\
44 +VIX Futures & $\Delta\beta$ & Novelty & 0.1096 & 1.20 & 0.236 & 0.704 & 0.272 & -- & 0.497 & 0 \\
45 +VIX Futures & $\Delta\beta$ & St. $\times$ Nov. & -0.0818 & -0.80 & 0.425 & 0.757 & 0.495 & -- & 0.642 & 0 \\
46 +5Y T-Note & $\Delta\beta$ & Stance & 0.0195 & 1.23 & 0.221 & 0.583 & 0.393 & -- & 0.409 & 0 \\
47 +5Y T-Note & $\Delta\beta$ & Novelty & -0.0064 & -0.88 & 0.381 & 0.786 & 0.481 & -- & 0.639 & 0 \\
48 +5Y T-Note & $\Delta\beta$ & St. $\times$ Nov. & 0.0074 & 0.99 & 0.325 & 0.820 & 0.381 & -- & 0.693 & 0 \\
49 +10Y T-Note & $\Delta\beta$ & Stance & 0.0118 & 0.59 & 0.554 & 0.772 & 0.693 & -- & 0.714 & 0 \\
50 +10Y T-Note & $\Delta\beta$ & Novelty & -0.0096 & -0.94 & 0.349 & 0.736 & 0.403 & -- & 0.665 & 0 \\
51 +10Y T-Note & $\Delta\beta$ & St. $\times$ Nov. & 0.0064 & 0.55 & 0.585 & 0.849 & 0.599 & -- & 0.830 & 0 \\
52 +\bottomrule
53 +\end{tabular}
54 +\begin{tablenotes}[flushleft]
55 +\scriptsize
56 +\item \textit{Notes:} Continuation of Table~\ref{tab:robustness_A}. The upper panel reports RV Ratio coefficients (post/pre RV); the lower panel reports $\Delta\beta$ coefficients (change in realized beta relative to ES). All regressors are z-scored. The same five inference methods are used: Newey--West HAC, wild bootstrap, clustered SE, quantile regression, and permutation test. RV Ratio and $\Delta\beta$ results are generally weaker than the $\Delta$RV and $\Delta\log$RV results in Table~\ref{tab:robustness_A}, suggesting that the volatility channel operates primarily through level changes rather than ratio adjustments or co-movement shifts.
57 +\end{tablenotes}
58 +\end{threeparttable}
59 +\end{table}
60 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tableR_rv_ratio_30min.tex +33 −0
@@ -0,0 +1,33 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Rolling RV Ratio Regressions (30-min Window, Interaction Model)}
6 +\label{tab:rolling_rv_ratio_30min}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 1.6276*** & 1.6260*** & 0.9991*** & 1.0135*** & 1.8017** & 2.9633*** & 3.7617*** \\
13 + & (0.1997) & (0.2639) & (0.0987) & (0.0537) & (0.6352) & (0.3545) & (0.4085) \\
14 +Stance & -0.4896 & 0.4669 & -0.0692 & -0.0861 & -1.1473 & 0.0611 & 0.0845 \\
15 + & (0.3029) & (0.2605) & (0.1108) & (0.1334) & (0.7665) & (0.4605) & (0.5241) \\
16 +Novelty & -0.1825 & -0.3809 & 0.0436 & 0.0596 & -0.4007 & 0.4481 & 0.6432 \\
17 + & (0.1157) & (0.2858) & (0.1196) & (0.1413) & (0.1864) & (0.3861) & (0.4561) \\
18 +Stance $\times$ Novelty & 0.2511 & -0.0118 & 0.1231 & -0.0415 & 0.3796 & -0.1615 & -0.5527 \\
19 + & (0.2085) & (0.1563) & (0.0801) & (0.0731) & (0.2289) & (0.5868) & (0.5128) \\
20 +\midrule
21 +$N$ & 88 & 41 & 40 & 40 & 17 & 89 & 88 \\
22 +$R^2$ & 0.050 & 0.050 & 0.129 & 0.025 & 0.145 & 0.035 & 0.044 \\
23 +Adj.\ $R^2$ & 0.016 & -0.027 & 0.057 & -0.056 & -0.053 & 0.001 & 0.010 \\
24 +\bottomrule
25 +\end{tabular}
26 +\begin{tablenotes}[flushleft]
27 +\small
28 +\item \textit{Notes:} Dependent variable: RV Ratio (Rolling 30-min). Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Semantic measures from MiniLM--BERT ensemble.
29 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
30 +\end{tablenotes}
31 +\end{threeparttable}
32 +\end{table}
33 +\end{landscape}
added PHD_chapitre3_theses_20260731/tables/tables.tex +33 −0
@@ -0,0 +1,33 @@
1 +\clearpage
2 +\section*{Tables}
3 +\addcontentsline{toc}{section}{Tables}
4 +
5 +% Table 1: Descriptive Statistics - 1-Minute Returns
6 +\input{tables/tableD1_desc_1min_returns.tex}
7 +
8 +% Table 2: Descriptive Statistics - Rolling Realized Measures
9 +\input{tables/tableD3_desc_rolling.tex}
10 +
11 +% Table 3-4: Panel Regressions (Main Results: log(RV) on Stance and Novelty)
12 +\input{tables/tableP_panel_logrv_stance.tex}
13 +\input{tables/tableP_panel_logrv_novelty.tex}
14 +
15 +% Table 5-9: Event-Level Regressions (Main Results)
16 +\input{tables/tableR_delta_rv_30min.tex}
17 +\input{tables/tableR_delta_log_rv_30min.tex}
18 +\input{tables/tableR_delta_beta_30min.tex}
19 +
20 +% Table 10-11: IRF Tables (Main Results: Stance and Novelty on Returns)
21 +\input{tables/tableR_irf_stance_ret_h.tex}
22 +\input{tables/tableR_irf_novelty_ret_h.tex}
23 +\input{tables/tableR_irf_stance_x_novelty_ret_h.tex}
24 +
25 +% Table 12-13: IRF Tables (Main Results: Abnormal Returns)
26 +\input{tables/tableR_irf_stance_abn_h.tex}
27 +\input{tables/tableR_irf_novelty_abn_h.tex}
28 +\input{tables/tableR_irf_stance_x_novelty_abn_h.tex}
29 +
30 +% Table 14-15: Multi-Method Robustness (Split into two manageable tables)
31 +\input{tables/tableR_robustness_30min_A.tex}
32 +\input{tables/tableR_robustness_30min_B.tex}
33 +
added README.md +144 −0
@@ -0,0 +1,144 @@
1 +# 🎓 Three Essays on High-Frequency Return and Volatility Dynamics in Commodities and Financial Futures Markets
2 +
3 +<p align="center">
4 + <img alt="PhD" src="https://img.shields.io/badge/PhD-Finance-8B0000?style=for-the-badge&logo=googlescholar&logoColor=white">
5 + <img alt="University" src="https://img.shields.io/badge/Universit%C3%A9-Laval-E30513?style=for-the-badge&logo=academia&logoColor=white">
6 + <img alt="LaTeX" src="https://img.shields.io/badge/Made%20with-LaTeX-008080?style=for-the-badge&logo=latex&logoColor=white">
7 + <img alt="Status" src="https://img.shields.io/badge/D%C3%A9p%C3%B4t-2026-blueviolet?style=for-the-badge">
8 +</p>
9 +
10 +<p align="center">
11 + <img alt="Chapters" src="https://img.shields.io/badge/chapters-3-success?style=flat-square">
12 + <img alt="Pages" src="https://img.shields.io/badge/thesis-188%20pages-informational?style=flat-square">
13 + <img alt="Data" src="https://img.shields.io/badge/data-high--frequency%20(1--5%20min)-orange?style=flat-square">
14 + <img alt="Markets" src="https://img.shields.io/badge/markets-energy%20%7C%20metals%20%7C%20equity%20%7C%20rates-yellow?style=flat-square">
15 + <img alt="Compiled" src="https://img.shields.io/badge/latexmk-0%20errors%20%7C%200%20warnings-brightgreen?style=flat-square">
16 +</p>
17 +
18 +> **Doctoral thesis by articles (thèse par articles)** — Simon-Pierre Boucher,
19 +> Faculté des sciences de l'administration (FSA), Université Laval.
20 +> Co-authors: Marie-Hélène Gagnon & Gabriel J. Power.
21 +
22 +---
23 +
24 +## 🗂️ Repository Contents
25 +
26 +| 📁 Folder | 📄 Content | 🏷️ Status |
27 +|---|---|---|
28 +| [`phd_chap1_20260731/`](phd_chap1_20260731/) | 🛢️ **Chapter 1** — Speculative trading in energy markets | 🟢 Revised for *The Energy Journal* |
29 +| [`phd_chap2_20260731/`](phd_chap2_20260731/) | 📈 **Chapter 2** — iNAV & commodity volatility transmission | 🟡 Submission version, *Journal of Futures Markets* |
30 +| [`PHD_chapitre3_theses_20260731/`](PHD_chapitre3_theses_20260731/) | 🏦 **Chapter 3** — FOMC tone & novelty (NLP) | 🟡 Manuscript (2026-03-06) |
31 +| [`these-ulaval/`](these-ulaval/) | 📕 **Global thesis** — full ULaval document (`ulthese`/memoir) | 🟢 Compiles clean, 188 pp. |
32 +
33 +---
34 +
35 +## 🛢️ Chapter 1 — Speculative Trading in Energy Markets: Evidence from Macroeconomic Surprises
36 +
37 +![Data](https://img.shields.io/badge/data-5--min%20futures%2C%202007--2024-orange?style=flat-square)
38 +![Markets](https://img.shields.io/badge/contracts-CL%20%C2%B7%20NG%20%C2%B7%20GC%20%C2%B7%20SI%20%C2%B7%20HG%20%C2%B7%20PA-blue?style=flat-square)
39 +![Method](https://img.shields.io/badge/methods-WLS--EWMA%20%C2%B7%20GARCH%20%C2%B7%20CFTC%20COT-9cf?style=flat-square)
40 +
41 +**Question:** Does speculative trading amplify or dampen the impact of macroeconomic news on commodity futures?
42 +
43 +Using high-frequency data and 26 macroeconomic announcement releases, this essay measures how the intensity of speculation (an NLS proxy built from the CFTC disaggregated Commitments of Traders — money managers vs. swap dealers) conditions the reaction of energy (crude oil, natural gas) and metals (gold, silver, copper, palladium) futures to standardized macro surprises.
44 +
45 +**Key findings** 🎯
46 +- 🧯 Increased speculative trading **dampens** the impact of macro surprises on price drift, volatility, and bid-ask spreads.
47 +- 💧 Speculators **improve liquidity and price discovery** while reducing volatility.
48 +- ⚡ The damping effect is stronger for **procyclical commodities** (oil, gas) than for safe havens (gold).
49 +- 👤 The beneficial effects are driven by **money managers**, not swap dealers.
50 +
51 +📦 *In the folder:* `main.tex` (monolithic article), `tables.tex` (14 tables), `figures.tex` (6 figures), COVID & ZLB appendices, `master.bib`.
52 +
53 +---
54 +
55 +## 📈 Chapter 2 — Seeing Through the ETF: Indicative NAV and Commodity Volatility Transmission
56 +
57 +![Data](https://img.shields.io/badge/data-%E2%89%8845M%20tick%20obs%2C%202010--2023-orange?style=flat-square)
58 +![Markets](https://img.shields.io/badge/ETFs-GLD%20%C2%B7%20SLV%20%C2%B7%20USO%20%C2%B7%20UNG-blue?style=flat-square)
59 +![Method](https://img.shields.io/badge/methods-HAR--X%20%C2%B7%20jump%20decomposition%20%C2%B7%20Minnesota%20BVAR-9cf?style=flat-square)
60 +
61 +**Question:** How does volatility flow between commodity ETFs and their underlying assets — and what does the intraday **indicative NAV (iNAV)** reveal that daily data cannot?
62 +
63 +This essay builds a **novel minute-level iNAV dataset** for four single-commodity ETFs (gold, silver, oil, natural gas), decomposes realized variance into continuous and jump components (Barndorff-Nielsen–Shephard), and estimates HAR-X / HAR-CJ-X models at 1, 5 and 30 minutes plus a Bayesian VAR.
64 +
65 +**Key findings** 🎯
66 +- 🪞 The iNAV gives a **sharper image** of the ETF–underlying volatility relationship.
67 +- 💥 Transmission runs primarily through **jumps**, not diffusion.
68 +- ⏱️ **Sampling frequency matters**: 1-minute estimates are up to **2× larger** than 30-minute ones.
69 +- 🥇 Precious metals: **unidirectional** (iNAV → ETF, passive arbitrage); ⛽ energy: **bidirectional & asymmetric**.
70 +
71 +📦 *In the folder:* `main.tex` + `sections/` (intro, data, methods, results, conclusion), 13 tables, 8 figures (RV & IRF plots), `master.bib`, compiled `main.pdf`.
72 +
73 +---
74 +
75 +## 🏦 Chapter 3 — Returns and Volatility Around FOMC Announcements: A High-Frequency Analysis of Policy Tone and Novelty
76 +
77 +![Data](https://img.shields.io/badge/data-1--min%2C%20148%20FOMC%20events%2C%202008--2025-orange?style=flat-square)
78 +![Markets](https://img.shields.io/badge/contracts-ES%20%C2%B7%20VX%20%C2%B7%20ZN%20%C2%B7%20ZF%20%C2%B7%20DX%20%C2%B7%20CL%20%C2%B7%20GC-blue?style=flat-square)
79 +![Method](https://img.shields.io/badge/methods-NLP%20ensemble%20(MiniLM%2BBERT)%20%C2%B7%20Jord%C3%A0%20LPs%20%C2%B7%20panels-9cf?style=flat-square)
80 +![JEL](https://img.shields.io/badge/JEL-E52%20%C2%B7%20E58%20%C2%B7%20G12%20%C2%B7%20G14-lightgrey?style=flat-square)
81 +
82 +**Question:** When the Fed speaks, what moves markets — **what** it says (tone) or **how new** it is (novelty)?
83 +
84 +FOMC statements (217 releases, 2000–2025) are decomposed into **policy tone** (hawkish/dovish) and **informational novelty** using a dual-model NLP ensemble (MiniLM + BERT, TSDAE+MNRL fine-tuning, PCA-based reference selection), then linked to 1-minute futures data via event regressions, minute-level panels and Jordà local projections, with placebo tests and five inference methods.
85 +
86 +**Key findings** 🎯
87 +- 🗣️ **Tone predicts directional returns**: a 1σ dovish shift → equity gains building to ≈ **+12 bps within two hours**.
88 +- 🆕 **Novelty predicts volatility**: the stance × novelty interaction on VIX persists 5–120 min (*t* = −5.06).
89 +- 📊 Stance moves realized volatility in **6 of 7 contracts** (*p* < 0.01).
90 +- 🧪 Pre-announcement placebos are null — effects are announcement-driven.
91 +
92 +📦 *In the folder:* `chapitre3.tex` + `sections/`, 24 tables, 19 figures, mathematical proofs appendix, `master.bib`, compiled `chapitre3.pdf`.
93 +
94 +---
95 +
96 +## 📕 Global Thesis — `these-ulaval/`
97 +
98 +![Class](https://img.shields.io/badge/class-ulthese%20(memoir)-008080?style=flat-square)
99 +![Build](https://img.shields.io/badge/build-latexmk%20%C2%B7%20pdfLaTeX-brightgreen?style=flat-square)
100 +![Pages](https://img.shields.io/badge/output-188%20pages-informational?style=flat-square)
101 +
102 +The full **thesis-by-articles document** assembled for the FESP (Faculté des études supérieures et postdoctorales) requirements:
103 +
104 +- 🇫🇷 French front matter (résumé, remerciements, avant-propos) + 🇬🇧 English abstracts per chapter
105 +- 📖 General introduction · 3 chapters · general conclusion · appendices A–B (Chapter 3 proofs & extras)
106 +- 📚 Consolidated bibliography (`bib/these.bib`, 271 unique keys merged from the three articles)
107 +- 🧾 [`INVENTAIRE.md`](these-ulaval/INVENTAIRE.md) — exhaustive audit of the sources, conflicts and every mechanical fix (labels prefixed `chN:`, package conflicts, BibTeX dedup…)
108 +
109 +```bash
110 +cd these-ulaval
111 +latexmk # → main.pdf (0 errors, 0 undefined refs, 0 missing citations)
112 +```
113 +
114 +---
115 +
116 +## 🧭 Thread of the Thesis
117 +
118 +The three essays share one lens: **high-frequency data around information events**.
119 +
120 +```
121 + 🛢️ Ch. 1 📈 Ch. 2 🏦 Ch. 3
122 + macro announcements → ETF ↔ underlying → FOMC statements
123 + × speculation volatility tone × novelty
124 + (who trades matters) transmission (what & how new)
125 + └──────────── returns · volatility · liquidity ────────────┘
126 +```
127 +
128 +---
129 +
130 +## ✍️ Authors
131 +
132 +| | |
133 +|---|---|
134 +| **Simon-Pierre Boucher** | PhD candidate in Finance, Université Laval — [simon-pierre.boucher.1@ulaval.ca](mailto:simon-pierre.boucher.1@ulaval.ca) |
135 +| **Marie-Hélène Gagnon** | Professor of Finance, CRREP, Université Laval |
136 +| **Gabriel J. Power** | IG Wealth Management Chairholder, Professor of Finance, CRREP & CRIB, Université Laval |
137 +
138 +> 🙏 Funding: Social Sciences and Humanities Research Council (SSHRC) & Chaire Industrielle-Alliance Groupe financier.
139 +
140 +---
141 +
142 +<p align="center">
143 + <sub>📌 Folder suffix <code>20260731</code> = frozen snapshot of each article (July 31, 2026 versions). Sources are never edited in place — all thesis adaptations live in <code>these-ulaval/</code>.</sub>
144 +</p>
added phd_chap1_20260731/FIG_MSCT.pdf +0 −0

Binary file not shown.

added phd_chap1_20260731/FIG_NLS.pdf +0 −0

Binary file not shown.

added phd_chap1_20260731/FIG_WT.pdf +0 −0

Binary file not shown.

added phd_chap1_20260731/appendix_covid.tex +519 −0
@@ -0,0 +1,519 @@
1 +\section{Appendix B - Results using the COVID Sample Period}
2 +\renewcommand{\thetable}{B-\arabic{table}}
3 +\setcounter{table}{0} % Reset table counter
4 +
5 +\begin{sidewaystable}
6 +\caption{Effects of Macro Announcements and Financialization on Futures Returns - COVID Sample Period}
7 +\label{tab:macro_fin_fut_returns_covid}
8 +\centering
9 +\resizebox{\linewidth}{!}{%
10 +\begin{tabular}{@{}lllllllllllll@{}}
11 +\toprule
12 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
13 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
14 +
15 +\textbf{Initial jobless claims} & -0.02432** & -0.01818 & 0.12506*** & -2.80729*** & -0.17005*** & 2.79627*** & -0.01710 & -1.80177*** & -0.04697* & 0.50617 & 0.02130 & 0.00395 \\
16 +\textbf{ADP Employment} & 0.10493** & -0.01992 & -0.07145*** & 0.03843*** & 0.20312*** & -0.07302*** & 0.05359** & -0.00743 & 0.00293 & -0.01197 & 0.06487 & 0.05085* \\
17 +\textbf{CB Consumer} & -0.00014 & 0.01730 & -0.01156 & -0.03514*** & 0.01126 & -0.12638*** & -0.01154 & 0.01428 & -0.00428 & -0.00635 & 0.05444 & 0.01849 \\
18 +\textbf{Advance retail sales} & 0.01675 & 0.00934 & 0.00999 & -0.02615*** & 0.05658*** & -0.00972 & 0.00271 & 0.01182 & -0.00801 & 0.00407 & -0.01226 & -0.00540 \\
19 +\textbf{Building permit} & 0.04472*** & -0.00473 & -0.00506 & 0.02308*** & -0.02296 & 0.06065*** & 0.01689* & -0.00460 & -0.00525 & -0.01228 & 0.08097** & 0.02061 \\
20 +\textbf{Construction spending} & 0.07403 & -0.08007*** & -0.02819* & 0.00141 & -0.01867 & -0.00880 & -0.00176 & -0.01373* & 0.04320 & 0.04773** & -0.02909 & -0.01744 \\
21 +\textbf{Consumer credit} & -0.01203 & 0.00357 & -0.00482 & -0.00376 & -0.00308 & 0.00661 & 0.00403 & -0.01316 & 0.01510 & 0.03158 & -0.00888 & 0.00622 \\
22 +\textbf{Consumer price index} & 0.02370 & 0.02270* & 0.02254 & -0.00306 & -0.08259*** & -0.00465 & -0.03278** & 0.00641 & -0.04731*** & 0.01463 & -0.06911 & 0.03523 \\
23 +\textbf{Durable goods orders} & -0.01326 & -0.00519 & -0.03861** & 0.09163*** & -0.00732 & -0.11911*** & -0.01810 & 0.02209 & -0.03421 & 0.00352 & 0.01278 & -0.02896 \\
24 +\textbf{Existing home sales} & -0.02893 & -0.02409 & 0.03032* & -0.03031*** & 0.00957 & -0.05879** & -0.01252 & 0.00328 & 0.00121 & 0.03252 & 0.10149* & -0.04166 \\
25 +\textbf{Factory orders} & 0.06701 & -0.02383 & -0.02140 & -0.01821 & -0.01981 & -0.08170** & 0.05127* & 0.01387 & -0.01113 & -0.01429 & 0.20189*** & -0.00073 \\
26 +\textbf{Gross domestic product} & -0.02344 & 0.01999** & -0.03597*** & 0.01002 & -0.03173 & 0.00734 & -0.02807*** & 0.00406 & -0.05014** & -0.00290 & -0.04468 & 0.01072 \\
27 +\textbf{Housing starts} & 0.01845 & 0.02643* & -0.00186 & -0.03386*** & 0.04042** & 0.02865 & -0.00735 & 0.02018** & -0.02342 & 0.01822 & 0.03539 & -0.00961 \\
28 +\textbf{Industrial production} & 0.03643** & 0.00465 & -0.03529*** & -0.00053 & 0.05792*** & 0.00069 & -0.01500 & 0.01881 & 0.00841 & -0.00974 & -0.03503 & -0.00202 \\
29 +\textbf{New home sales} & -0.01988 & 0.02155 & -0.01107 & -0.00021 & 0.05154** & -0.01592 & -0.00762 & 0.01283 & 0.02024 & -0.00408 & -0.03319 & -0.01629 \\
30 +\textbf{Non-farm employment} & 0.02498 & 0.00862 & -0.27690*** & 0.02188** & 0.17649*** & -0.04451** & 0.06183 & 0.00160 & 0.02000** & -0.00112 & 0.08142 & 0.04914 \\
31 +\textbf{Pending home sales} & 0.00688 & 0.01007 & -0.01427 & -0.16695*** & 0.00245 & -0.09205*** & 0.00528 & 0.01944 & 0.00860 & -0.07946** & -0.02308 & -0.00338 \\
32 +\textbf{Personal consumption} & -0.00795 & -0.01160 & -0.03914*** & -0.00496 & -0.03188 & -0.01156 & -0.00356 & -0.00346 & 0.03274 & -0.00329 & -0.06953* & -0.02143 \\
33 +\textbf{Personal income} & -0.01321 & -0.00959 & -0.02591 & -0.01454*** & -0.03079 & -0.02411* & -0.04359 & -0.00410 & 0.02955 & 0.00582 & 0.07648 & 0.03243 \\
34 +\textbf{Producer price index} & 0.01504 & 0.00858 & -0.03052** & -0.01257 & -0.03537 & -0.01580 & 0.00541 & -0.01962 & 0.04519* & 0.04732 & -0.06727* & 0.00949 \\
35 +\textbf{Trade balance} & 0.29865*** & 0.03278 & -0.12200*** & -0.01377* & 0.41798*** & -0.05178*** & -0.11417** & 0.00071 & -0.56894*** & -0.01681 & -0.36628** & -0.36300** \\
36 +\textbf{Crude Oil Weekly inventory} & -0.21132*** & -0.01899*** & & & & & & & & & & \\
37 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.01789 & 0.01571 \\ \midrule
38 +
39 +\textbf{Observations} & \multicolumn{2}{c}{167663} & \multicolumn{2}{c}{167808} & \multicolumn{2}{c}{167513} & \multicolumn{2}{c}{167800} & \multicolumn{2}{c}{99525} & \multicolumn{2}{c}{167472} \\
40 +\textbf{$R^2$} & \multicolumn{2}{c}{0.004106} & \multicolumn{2}{c}{0.004110} & \multicolumn{2}{c}{0.002538} & \multicolumn{2}{c}{0.000898} & \multicolumn{2}{c}{0.001077} & \multicolumn{2}{c}{0.001355} \\ \bottomrule
41 +\end{tabular}
42 +}
43 +\begin{tablenotes}
44 +\singlespacing
45 +\footnotesize
46 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and the financialization variable $PCA_t$. The period covered is from 2020-01-31 to 2022-06-10. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred and especially if that announcement was unanticipated. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of financialization.
47 +\end{tablenotes}
48 +\end{sidewaystable}
49 +
50 +
51 +
52 +
53 +\begin{sidewaystable}
54 +\caption{Effects of Macro Announcements and Financialization on Futures Conditional Variance - COVID Sample Period}
55 +\label{tab:macro_fin_fut_var_covid}
56 +\centering
57 +\resizebox{\linewidth}{!}{%
58 +\begin{tabular}{@{}lllllllllllll@{}}
59 +\toprule
60 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
61 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
62 +
63 +\textbf{Initial jobless claims} & 0.00877 & 0.06854** & 0.04889*** & 0.00632 & 0.10143*** & -0.02561* & 0.01633*** & -0.00186 & 0.08097*** & -0.00257 & -0.00080 & 0.03698 \\
64 +\textbf{ADP Employment} & 0.03197 & 0.05389 & 0.00108 & 0.00335 & 0.03112 & 0.01241 & 0.06538*** & -0.00894 & -0.09486** & -0.02727 & 0.01188 & -0.00604 \\
65 +\textbf{CB Consumer} & 0.05354 & -0.00523 & 0.07158*** & -0.01573** & 0.06761*** & 0.00890 & -0.00622 & -0.00047 & 0.02592 & 0.00496 & -0.01753 & -0.01483 \\
66 +\textbf{Advance retail sales} & -0.00851 & -0.04267 & 0.04067*** & -0.01819*** & 0.10993*** & -0.00152 & 0.01390 & -0.02434** & -0.01311 & -0.05964** & 0.05261 & 0.00172 \\
67 +\textbf{Building permit} & -0.00704 & 0.04270 & 0.00467 & -0.00389 & -0.00659 & 0.01853 & -0.00975 & 0.00774 & 0.03353 & -0.00250 & 0.05338* & -0.01865 \\
68 +\textbf{Construction spending} & 0.02353 & -0.01352 & 0.03937*** & 0.00092 & 0.03030 & 0.00426 & 0.00374 & 0.00524 & -0.00435 & 0.02602 & 0.04897 & 0.01006 \\
69 +\textbf{Consumer credit} & -0.00584 & -0.02756 & 0.00412 & 0.02848*** & 0.04301** & -0.05662*** & -0.00637 & -0.03798*** & 0.03674 & 0.00690 & 0.01900 & -0.06251 \\
70 +\textbf{Consumer price index} & -0.00862 & 0.00919 & 0.09025*** & 0.00549 & 0.08811*** & -0.04065*** & 0.09950*** & 0.01802* & -0.02377 & 0.03608 & 0.05494 & 0.04110* \\
71 +\textbf{Durable goods orders} & -0.01320 & 0.00783 & 0.01782* & -0.02205*** & 0.04292** & -0.00940 & -0.01289 & -0.00039 & 0.05048 & -0.18229*** & -0.05984* & -0.01398 \\
72 +\textbf{Existing home sales} & 0.09568* & 0.02457 & 0.04677*** & -0.00119 & 0.05161** & -0.03212** & 0.04829*** & -0.00778 & 0.16009*** & -0.00301 & 0.08823*** & 0.05824** \\
73 +\textbf{Factory orders} & 0.04917 & -0.00674 & 0.03061*** & -0.00275 & 0.02062 & -0.05376*** & 0.02066 & 0.02086* & 0.01577 & -0.00984 & 0.04968 & 0.03284 \\
74 +\textbf{Gross domestic product} & -0.02008 & -0.04953 & 0.05825*** & -0.01752*** & 0.10358*** & -0.00124 & 0.00580 & -0.03097*** & 0.00318 & -0.06473** & -0.06185* & 0.01394 \\
75 +\textbf{Housing starts} & -0.00072 & 0.03780 & 0.00500 & -0.00374 & -0.00688 & -0.02454 & -0.00968 & -0.00918 & 0.03057 & -0.00948 & 0.06142** & -0.00844 \\
76 +\textbf{Industrial production} & -0.03464 & 0.00114 & -0.01280 & -0.01032** & 0.03812* & -0.00325 & -0.01843 & -0.00983 & 0.04435 & 0.04172** & -0.03347 & -0.00747 \\
77 +\textbf{New home sales} & 0.00487 & 0.00237 & 0.07711*** & -0.05791*** & 0.10903*** & -0.11668*** & -0.00918 & -0.00837 & 0.02354 & 0.00263 & -0.00242 & -0.05764** \\
78 +\textbf{Non-farm employment} & 0.08122 & 0.01522 & 0.18711*** & -0.01610*** & 0.31928*** & -0.02718* & 0.12234*** & 0.00857 & 0.05419 & -0.00820 & 0.01478 & 0.03549 \\
79 +\textbf{Pending home sales} & 0.01072 & -0.01184 & 0.03844*** & -0.01432** & 0.03845* & -0.03944** & 0.01314 & -0.00949 & 0.07867* & 0.03292 & 0.06944** & 0.09557*** \\
80 +\textbf{Personal consumption} & -0.01919 & -0.02455 & 0.02154** & -0.01287** & 0.04426* & 0.01001 & -0.00224 & -0.00698 & 0.07037 & 0.02141 & 0.08558** & 0.06103*** \\
81 +\textbf{Personal income} & 0.00244 & -0.01211 & 0.02834*** & -0.02153*** & 0.05301** & -0.03660** & -0.00017 & 0.00224 & 0.07217* & -0.03518 & 0.05645* & -0.01085 \\
82 +\textbf{Producer price index} & 0.02955 & 0.03470 & 0.04263*** & -0.01819*** & 0.09391*** & 0.02112 & -0.00732 & -0.00549 & 0.03624 & 0.02379 & -0.04616 & 0.01102 \\
83 +\textbf{Trade balance} & 0.06625 & 0.02541 & 0.02594*** & 0.00010 & 0.03557* & 0.00010 & 0.02995** & 0.00010 & 0.10040** & 0.00010 & 0.01924 & -0.00631 \\
84 +\textbf{Crude Oil Weekly inventory} & 0.15709*** & 0.00010 & & & & & & & & & & \\
85 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.55047*** & 0.00010 \\ \midrule
86 +
87 +\textbf{Observations} & \multicolumn{2}{c}{167663} & \multicolumn{2}{c}{167808} & \multicolumn{2}{c}{167513} & \multicolumn{2}{c}{167800} & \multicolumn{2}{c}{99525} & \multicolumn{2}{c}{167472} \\
88 +\textbf{$R^2$} & \multicolumn{2}{c}{0.024286} & \multicolumn{2}{c}{0.119026} & \multicolumn{2}{c}{0.078046} & \multicolumn{2}{c}{0.084932} & \multicolumn{2}{c}{0.051116} & \multicolumn{2}{c}{0.138467} \\ \bottomrule
89 +\end{tabular}
90 +}
91 +\begin{tablenotes}
92 +\singlespacing
93 +\footnotesize
94 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the financialization variable $PCA_t$. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2020-01-31 to 2022-06-10. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of financialization.
95 +\end{tablenotes}
96 +\end{sidewaystable}
97 +
98 +
99 +
100 +\begin{sidewaystable}
101 +\caption{Effects of Macro Announcements and Financialization (Money Managers only) on Futures Returns - COVID Sample Period}
102 +\label{tab:macro_fin_mm_fut_returns_covid}
103 +\centering
104 +\resizebox{\linewidth}{!}{%
105 +\begin{tabular}{@{}lllllllllllll@{}}
106 +\toprule
107 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
108 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
109 +
110 +\textbf{Initial jobless claims} & 0.01284 & -0.42931 & 0.18523*** & -0.60124*** & -0.18006*** & 1.43876*** & -0.01157 & -0.07340 & -0.01816 & 0.06554 & -0.01707 & -0.54064 \\
111 +\textbf{ADP Employment} & 0.29525 & -1.27353 & -0.20069*** & 0.83071** & -0.00499 & -0.26184 & 0.02301*** & 0.13008 & 0.00089 & 0.00840 & -0.02549* & -0.41182 \\
112 +\textbf{CB Consumer} & 0.07655 & -0.51606 & -0.09771** & 0.47099* & -0.06772 & 0.46189 & -0.04213 & 0.17658 & -0.03088 & 0.10234 & -0.01077 & 1.92141* \\
113 +\textbf{Advance retail sales} & -0.02923 & 0.34904 & -0.07305*** & 0.43390*** & -0.01274 & 0.22680 & 0.01271 & -0.06940 & -0.01833 & 0.09338 & -0.00873 & 0.00984 \\
114 +\textbf{Building permit} & -0.09509 & 1.00771 & 0.02568 & -0.19516 & -0.09011* & 0.39078 & -0.04005** & 0.13851* & 0.00371 & 0.04582 & 0.08024** & 0.86233** \\
115 +\textbf{Construction spending} & 0.41652** & -2.77195** & -0.04190 & 0.08539 & 0.02789 & -0.34852 & 0.01859 & -0.13908 & 0.01249 & 0.09305 & -0.01210 & -1.19446** \\
116 +\textbf{Consumer credit} & -0.00923 & -0.01745 & -0.02585 & 0.12077 & 0.05161** & -0.29881** & -0.00532 & 0.06380 & -0.03173 & 0.19807 & -0.01666 & 0.08127 \\
117 +\textbf{Consumer price index} & -0.24917 & 1.86806* & -0.43977*** & 2.23921*** & 0.19333*** & -0.58014** & -0.07806*** & 0.27258* & -0.10369** & 0.20287 & -0.03498 & 0.23866 \\
118 +\textbf{Durable goods orders} & 0.09373 & -0.74428 & -0.13019*** & 0.56867** & -0.04308 & 0.18076 & -0.03350 & 0.07762 & 0.02374 & -0.23976 & 0.04727 & -0.54832 \\
119 +\textbf{Existing home sales} & 0.74825*** & -5.48168*** & -0.10481 & 0.80092* & -0.13335 & 0.75410 & -0.01008 & 0.04062 & 0.03378 & -0.12891 & 0.10720 & -0.39774 \\
120 +\textbf{Factory orders} & 0.00864 & 0.26749 & -0.05923 & 0.18085 & -0.00838 & -0.08036 & 0.04497 & 0.03837 & 0.08834 & -0.51740** & 0.21529*** & 1.18248 \\
121 +\textbf{Gross domestic product} & -0.01706 & -0.08802 & -0.33648*** & 1.71065*** & -0.00324 & -0.10068 & -0.06030*** & 0.19003*** & -0.04116 & -0.06590 & 0.01248 & -0.86484 \\
122 +\textbf{Housing starts} & 0.21757** & -1.76668*** & -0.01221 & 0.05853 & 0.00456 & -0.24308 & -0.03423* & 0.15038* & 0.03046 & -0.22306 & 0.03863 & 0.92630*** \\
123 +\textbf{Industrial production} & 0.34103** & -2.13714* & -0.00293 & -0.20915* & -0.07816 & 0.16609 & -0.02928* & 0.09407 & 0.00574 & 0.03016 & -0.02946 & -0.32844 \\
124 +\textbf{New home sales} & -0.23837 & 1.50728 & -0.12033*** & 0.78060*** & 0.04347 & -0.47414** & -0.00786 & 0.00541 & 0.03361 & -0.07157 & -0.05379 & 0.74563 \\
125 +\textbf{Non-farm employment} & -0.38931 & 2.35649* & -0.68355*** & 2.98988*** & 0.41050*** & -2.25335*** & -0.01543* & 0.49231* & -0.07231** & 0.68623* & 0.01033 & 1.53154 \\
126 +\textbf{Pending home sales} & -0.00301 & 0.15189 & -0.04072 & 0.13567 & -0.05453 & 0.29881 & 0.00433 & 0.02053 & -0.01675 & 0.15366 & -0.01676 & -0.80029 \\
127 +\textbf{Personal consumption} & 0.04548 & -0.40212 & -0.12306*** & 0.46241*** & 0.01365 & -0.07518 & 0.00234 & -0.00958 & 0.02166 & 0.04800 & -0.03182 & 0.06593 \\
128 +\textbf{Personal income} & -0.05495 & 0.14342 & 0.00497 & -0.03505 & 0.14836 & -0.94084 & -0.01037 & -0.25330* & -0.05332 & 0.26052 & 0.01097 & -0.98839 \\
129 +\textbf{Producer price index} & -0.10278 & 0.85437 & -0.06206 & 0.22680 & 0.03505 & -0.32221 & 0.01070 & -0.04785 & 0.02766 & 0.12770 & -0.06587* & 0.44089 \\
130 +\textbf{Trade balance} & -1.09678* & 5.61848 & 0.00987 & -0.77170 & -0.28472* & 0.23575 & -0.02639 & -0.50428 & -0.67818*** & 0.80536** & -0.66975*** & 7.55158*** \\
131 +\textbf{Crude Oil Weekly inventory} & 0.03025 & -1.67095*** & & & & & & & & & & \\
132 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.18997*** & 0.63710 \\ \midrule
133 +
134 +\textbf{Observations} & \multicolumn{2}{c}{167663} & \multicolumn{2}{c}{167808} & \multicolumn{2}{c}{167513} & \multicolumn{2}{c}{167800} & \multicolumn{2}{c}{99525} & \multicolumn{2}{c}{167472} \\
135 +\textbf{$R^2$} & \multicolumn{2}{c}{0.004087} & \multicolumn{2}{c}{0.003488} & \multicolumn{2}{c}{0.001728} & \multicolumn{2}{c}{0.000983} & \multicolumn{2}{c}{0.001166} & \multicolumn{2}{c}{0.001653} \\ \bottomrule
136 +\end{tabular}
137 +}
138 +\begin{tablenotes}
139 +\singlespacing
140 +\footnotesize
141 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and the financialization variable $MM_NLS_t$, calculated with the money manager positions. The period covered is from 2020-01-31 to 2022-06-10. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred and especially if that announcement was unanticipated. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of financialization.
142 +\end{tablenotes}
143 +\end{sidewaystable}
144 +
145 +
146 +
147 +\begin{sidewaystable}
148 +\caption{Effects of Macro Announcements and Financialization (Swap Dealers only) on Futures Returns - COVID Sample Period}
149 +\label{tab:macro_fin_sd_fut_returns_covid}
150 +\centering
151 +\resizebox{\linewidth}{!}{%
152 +\begin{tabular}{@{}lllllllllllll@{}}
153 +\toprule
154 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
155 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
156 +
157 +\textbf{Initial jobless claims} & -0.35559 & -1.52603 & 0.01394 & 0.02196 & -0.00342 & -2.17155*** & 0.08446* & 0.41894* & -0.19160*** & -1.05912*** & -0.08847** & 1.77686** \\
158 +\textbf{ADP Employment} & 0.08035 & 0.06661 & -0.31471** & -0.78958* & -0.11333** & -1.67280 & -0.22634*** & -1.05459*** & 0.00289 & -0.00433 & -0.05779** & 1.40793 \\
159 +\textbf{CB Consumer} & 0.54866 & 2.62593 & -0.24794* & -0.70080* & 0.06550 & 1.59409* & 0.08546 & -0.61999 & 0.02718 & -0.20183 & 0.04790 & 0.01052 \\
160 +\textbf{Advance retail sales} & 0.83956*** & 4.05137*** & -0.30872*** & -0.94688*** & 0.03121* & 0.27971 & -0.03042 & 0.22433 & 0.02659 & -0.15492 & -0.01285 & 0.09216 \\
161 +\textbf{Building permit} & 0.44978** & 2.31848** & -0.13443 & -0.39083 & 0.04930** & 0.60257* & -0.06394** & -0.53905** & 0.01289 & -0.00590 & 0.16126*** & -1.02460** \\
162 +\textbf{Construction spending} & 0.27632 & 1.39045 & 0.23373 & 0.72075 & -0.00773 & 0.60794 & -0.14991** & -0.88770** & 0.05101 & -0.12208 & -0.11196 & 1.69267* \\
163 +\textbf{Consumer credit} & 0.00461 & 0.08012 & -0.04758 & -0.12187 & 0.01797 & 0.32756* & 0.03715 & -0.19518 & 0.02984 & -0.17111 & -0.02980 & 0.23179 \\
164 +\textbf{Consumer price index} & -0.05112 & -0.32176 & -0.57345*** & -1.58847** & 0.12469*** & 0.69010* & -0.28822*** & -1.98064*** & 0.00680 & -0.41770* & -0.00893 & -0.14206 \\
165 +\textbf{Durable goods orders} & 0.13207 & 0.71558 & -0.38513*** & -1.10042*** & -0.03065 & -0.83509 & 0.06604 & -0.59511 & 0.01982 & -0.40849 & 0.01972 & -0.35887 \\
166 +\textbf{Existing home sales} & -0.04459 & -0.02126 & 0.11253 & 0.26883 & -0.04172 & -1.15384 & -0.09829* & -0.67851* & -0.01171 & 0.10709 & 0.11297 & -0.25142 \\
167 +\textbf{Factory orders} & 0.18355 & 0.64725 & 0.25872 & 0.84187 & -0.02900 & -0.14117 & 0.08340 & -0.20274 & -0.14787* & -0.80132* & 0.21908** & -0.23554 \\
168 +\textbf{Gross domestic product} & 0.46231*** & 2.30788*** & 0.19991 & 0.68373 & 0.05351 & 1.71279 & -0.05994** & -0.56015*** & -0.05200 & 0.00596 & -0.05958* & 0.28580 \\
169 +\textbf{Housing starts} & -0.32851 & -1.71555 & -0.07149 & -0.21937 & -0.03948 & 0.03702 & -0.01791 & 0.10276 & -0.10283* & 0.58656* & 0.16249*** & -1.56023***\\
170 +\textbf{Industrial production} & 0.17043 & 1.00224 & 0.08398 & 0.37057* & 0.06373*** & -0.34937 & -0.08247* & 0.46599 & 0.04412 & -0.20798 & -0.00941 & -0.37177 \\
171 +\textbf{New home sales} & -0.03078 & -0.06343 & 0.18592 & 0.58193 & -0.01074 & 0.53075 & 0.01912 & -0.18141 & 0.01301 & 0.02973 & 0.02364 & -0.82941* \\
172 +\textbf{Non-farm employment} & 2.07613* & 10.49944* & -3.88548*** & -10.45671***& 0.14035** & -2.11552 & -0.39987** & -1.96803** & -0.38964** & -1.75421** & -0.04663 & 1.68979 \\
173 +\textbf{Pending home sales} & 0.13424 & 0.54767 & 0.02888 & 0.11045 & 0.02128 & 0.43282 & 0.00606 & 0.00543 & 0.06637 & -0.30888 & -0.01983 & -0.02780 \\
174 +\textbf{Personal consumption} & 0.23334 & 1.13462 & -0.34704*** & -0.96377*** & 0.01301 & 0.48689 & -0.02546 & 0.13933 & 0.14246** & -0.56744 & 0.00804 & -1.01067 \\
175 +\textbf{Personal income} & 0.48466 & 2.52886 & -0.07014 & -0.18887 & -0.08014 & -1.76992 & -0.20808* & -1.03538* & 0.00040 & -0.07478 & -0.09644 & 3.03041* \\
176 +\textbf{Producer price index} & -0.37035 & -1.91965 & 0.22676 & 0.72619 & 0.00020 & 0.30589 & -0.02940 & 0.22461 & 0.04012 & 0.04467 & -0.04055 & -0.30514 \\
177 +\textbf{Trade balance} & -2.43297 & -9.82726 & -0.22310 & -0.34760 & 0.22495** & 0.04987 & -0.31868* & 1.15728 & -0.17144 & -1.82171*** & -0.13828 & -3.43949* \\
178 +\textbf{Crude Oil Weekly inventory} & -0.71319*** & -2.29368*** & & & & & & & & & & \\
179 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.13552** & -0.31667 \\ \midrule
180 +
181 +\textbf{Observations} & \multicolumn{2}{c}{167663} & \multicolumn{2}{c}{167808} & \multicolumn{2}{c}{167513} & \multicolumn{2}{c}{167800} & \multicolumn{2}{c}{99525} & \multicolumn{2}{c}{167472} \\
182 +\textbf{$R^2$} & \multicolumn{2}{c}{0.004380} & \multicolumn{2}{c}{0.003420} & \multicolumn{2}{c}{0.001651} & \multicolumn{2}{c}{0.001399} & \multicolumn{2}{c}{0.001579} & \multicolumn{2}{c}{0.001734} \\ \bottomrule
183 +\end{tabular}
184 +}
185 +\begin{tablenotes}
186 +\singlespacing
187 +\footnotesize
188 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and the financialization variable $SWAP_NLS_t$, calculated with the swap dealer positions. The period covered is from 2020-01-31 to 2022-06-10. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred and especially if that announcement was unanticipated. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of financialization.
189 +\end{tablenotes}
190 +\end{sidewaystable}
191 +
192 +
193 +\begin{sidewaystable}
194 +\caption{Effects of Macro Announcements and Financialization (Money Managers only) on Futures Conditional Variance - COVID Sample Period}
195 +\label{tab:macro_fin_mm_var_covid}
196 +\centering
197 +\resizebox{\linewidth}{!}{%
198 +\begin{tabular}{@{}lllllllllllll@{}}
199 +\toprule
200 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
201 +\textbf{Announcements} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} \\ \midrule
202 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
203 +
204 +\textbf{Initial jobless claims} & -0.04461 & 2.85159* & -0.00467 & -0.05895 & 0.15242*** & -0.16088 & 0.01131 & 0.00230 & 0.08837*** & -0.09377 & -0.00605 & 0.21895 \\
205 +\textbf{ADP Employment} & -0.34572 & 4.05873** & 0.01157 & -0.11669 & 0.05935 & 0.18762 & 0.06509***& -0.05024 & -0.08643* & -0.20231 & 0.00600 & 0.14602 \\
206 +\textbf{CB Consumer} & -0.49745* & 1.32170 & 0.09255***& -0.17445 & 0.03300 & -1.30168***& 0.00077 & -0.05129 & 0.05362 & -0.20067 & -0.02037 & 0.32092 \\
207 +\textbf{Advance retail sales} & -0.18724 & -0.76763 & 0.07227** & -0.53535*** & 0.34259*** & -0.62714**& 0.02209 & -0.22841***& 0.01804 & -0.26755 & 0.04779 & 0.50548 \\
208 +\textbf{Building permit} & 0.10048 & 1.66375 & -0.09165***& 0.14266 & 0.10536** & 0.33833 & 0.02134 & 0.06065 & 0.07337 & -0.07325 & 0.04532 & -0.42110 \\
209 +\textbf{Construction spending} & -0.19798 & 0.62185 & 0.01384 & 0.02905 & -0.03106 & 0.10504 & -0.00470 & 0.06039 & 0.00351 & 0.21481 & 0.05824* & 0.03839 \\
210 +\textbf{Consumer credit} & -0.09134 & 1.21640 & -0.00171 & 0.00941 & 0.02411 & -0.86118***& -0.01479 & -0.26289***& 0.00540 & 0.13574 & 0.01831 & 0.13310 \\
211 +\textbf{Consumer price index} & -0.16864 & 2.37100 & 0.09632***& -0.40220** & 0.25226*** & 0.04154 & 0.13703***& -0.14927* & -0.04150 & -0.13397 & 0.03385 & -0.36319 \\
212 +\textbf{Durable goods orders} & -0.33507 & -3.41535* & 0.08977***& -0.29773* & 0.03393 & -0.50163* & -0.03340* & -0.06481 & 0.06564 & -0.19562 & -0.05252 & 0.06643 \\
213 +\textbf{Existing home sales} & 0.56294** & 1.05599 & -0.00734 & 0.18055 & 0.13902*** & -0.09197 & 0.05677***& -0.10169 & 0.21109***& -0.12753 & 0.08619***& -0.60224 \\
214 +\textbf{Factory orders} & -0.08984 & 1.60278 & -0.00245 & 0.27924 & 0.03794 & 0.72946**& 0.03449** & -0.22021**& 0.03417 & 0.13810 & 0.05311* & -0.39455 \\
215 +\textbf{Gross domestic product} & -0.23953 & -0.41313 & 0.01217 & 0.52640*** & -0.03592 & -0.69722***& -0.02595 & -0.27360***& -0.01096 & -0.27690 & -0.04993 & 0.03640 \\
216 +\textbf{Housing starts} & 0.05990 & 1.01861 & -0.08970***& 0.19485 & 0.11759** & 0.16098 & 0.02760 & -0.06027 & 0.07192 & -0.38680* & 0.06048* & 0.68109 \\
217 +\textbf{Industrial production} & -0.17177 & -3.74853*** & -0.04685 & -0.35670***& -0.06541 & -0.53071***& -0.01066 & -0.06344 & 0.10981** & 0.03150 & -0.03748 & -0.14918 \\
218 +\textbf{New home sales} & -0.01003 & 0.17466 & 0.08614***& -0.57164***& 0.19530*** & 0.03185 & -0.02271 & -0.00031 & 0.03605 & 0.38720** & -0.00739 & 0.72281* \\
219 +\textbf{Non-farm employment} & 0.05604 & 3.04568 & 0.28587***& -0.41916** & 0.30708*** & 0.00215 & 0.12292***& 0.10831 & 0.00414 & 0.23142 & 0.00350 & 0.36589 \\
220 +\textbf{Pending home sales} & -0.40328 & -0.36436 & 0.11405***& -0.34236* & 0.03729 & -0.29147 & -0.00213 & -0.07714 & 0.04750 & -0.06213 & 0.06356* & -2.05880***\\
221 +\textbf{Personal consumption} & 0.03999 & -2.13154 & 0.08202** & -0.28124* & 0.09221 & -0.17986 & 0.00857 & -0.04301 & 0.07635 & -0.09191 & 0.13395***& -0.64605 \\
222 +\textbf{Personal income} & 0.29453 & -0.50060 & 0.07845***& -0.74149***& 0.08435 & -0.51294* & 0.00589 & 0.00579 & 0.08636* & -0.41902* & 0.06267* & 0.26214 \\
223 +\textbf{Producer price index} & 0.09115 & -0.16130 & 0.09431***& -0.52717***& 0.18854*** & -0.68457**& -0.00799 & 0.01655 & 0.12402** & -0.19304 & -0.05209 & 0.16588 \\
224 +\textbf{Trade balance} & 0.08689 & -1.05789 & 0.12133***& 0.00010 & 0.15697*** & 0.00010 & 0.02785 & 0.00010 & 0.12659***& 0.00010 & 0.01636 & 0.97422*** \\
225 +\textbf{Crude Oil Weekly inventory} & 0.31257 & 0.00010 & & & & & & & & & & \\
226 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.53582*** & 0.00010 \\ \midrule
227 +
228 +\textbf{Observations} & \multicolumn{2}{c}{167663} & \multicolumn{2}{c}{167808} & \multicolumn{2}{c}{167513} & \multicolumn{2}{c}{167800} & \multicolumn{2}{c}{99525} & \multicolumn{2}{c}{167472} \\
229 +\textbf{$R^2$} & \multicolumn{2}{c}{0.036079} & \multicolumn{2}{c}{0.110085} & \multicolumn{2}{c}{0.072735} & \multicolumn{2}{c}{0.087602} & \multicolumn{2}{c}{0.047790} & \multicolumn{2}{c}{0.135707} \\ \bottomrule
230 +\end{tabular}
231 +}
232 +\begin{tablenotes}
233 +\singlespacing
234 +\footnotesize
235 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the financialization variable $MM_NLS_t$, calculated with the money manager positions. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2020-01-31 to 2022-06-10. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of financialization.
236 +\end{tablenotes}
237 +\end{sidewaystable}
238 +
239 +
240 +\begin{sidewaystable}
241 +\caption{Effects of Macro Announcements and Financialization (Swap Dealers only) on Futures Conditional Variance - COVID Sample Period}
242 +\label{tab:macro_fin_sd_var_covid}
243 +\centering
244 +\resizebox{\linewidth}{!}{%
245 +\begin{tabular}{@{}lllllllllllll@{}}
246 +\toprule
247 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
248 +\textbf{Announcements} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} & \multicolumn{1}{c}{$\Phi_m$} & \multicolumn{1}{c}{$\phi_m$} \\ \midrule
249 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
250 +
251 +\textbf{Initial jobless claims} & -0.02922 & 4.03858 & -0.00383 & -0.44581* & 0.13311*** & 0.68878 & 0.01537 & 0.61188** & 0.08235** & 0.30350 & -0.00311 & -0.38407 \\
252 +\textbf{ADP Employment} & 0.89022 & -1.01674 & -0.14995* & -0.24589 & 0.06432** & -0.03518 & -0.03627 & 0.12394 & -0.15721* & 0.71733* & 0.04367 & -0.88879** \\
253 +\textbf{CB Consumer} & -0.16354 & 2.14791 & -0.01169 & 0.24884 & 0.06540** & 1.38311*** & -0.02742 & 0.18772 & -0.11676 & 0.71702 & 0.05968 & 0.12444 \\
254 +\textbf{Advance retail sales} & 0.44938 & -0.51159 & 0.12493 & -0.71539* & 0.18362*** & 0.93335** & -0.01610 & 0.80474*** & -0.14029 & 0.38779 & 0.04598 & 0.72141* \\
255 +\textbf{Building permit} & -0.11462 & 1.52827 & -0.23660* & 0.13656 & 0.04223 & -0.36732 & 0.13936*** & -0.15939 & -0.03245 & 0.18738 & -0.00720 & 0.68219 \\
256 +\textbf{Construction spending} & 0.35162 & -0.30770 & 0.08577 & 0.04062 & 0.01138 & 0.26210 & 0.02996 & -0.18551 & -0.04229 & -0.26997 & -0.00935 & 0.07279 \\
257 +\textbf{Consumer credit} & -0.06546 & -0.55449 & 0.01866 & -1.87127*** & 0.05736* & 0.01452 & 0.02409 & 2.08176*** & 0.08719 & 0.01319 & 0.01232 & 0.49723 \\
258 +\textbf{Consumer price index} & -0.11343 & 1.27139 & -0.52340*** & -0.22151 & 0.11112*** & -0.57925 & 0.25742*** & -0.01637 & -0.02735 & 0.16253 & -0.01576 & 0.43095 \\
259 +\textbf{Durable goods orders} & 0.25844 & -3.94367 & -0.05658 & 0.27322 & 0.01283 & 0.81324** & -0.00975 & -0.03792 & 0.01667 & 0.88570** & -0.09599* & -0.18731 \\
260 +\textbf{Existing home sales} & -0.77611 & 4.12467 & 0.13657 & -0.31927 & 0.09330*** & -0.22595 & 0.05463 & 0.31807 & 0.01878 & 0.02580 & 0.10396** & 0.92352** \\
261 +\textbf{Factory orders} & 0.91990 & 1.70409 & -0.07701 & -0.03624 & 0.01073 & -0.67484 & -0.03092 & -0.44596* & 0.01082 & -0.22664 & -0.03744 & 1.07310** \\
262 +\textbf{Gross domestic product} & 0.34501 & -1.88465 & 0.04424 & -0.64959* & 0.06082* & 1.02371*** & 0.08284* & 1.10052*** & 0.05347 & 0.40694 & -0.15514*** & 0.94600** \\
263 +\textbf{Housing starts} & -0.40274 & 0.07790 & -0.21414* & -0.03151 & 0.04666 & -0.53977 & -0.18703*** & 0.05743 & -0.03861 & 0.80121* & -0.01860 & -0.43125 \\
264 +\textbf{Industrial production} & -0.01230 & -2.49043 & -0.02319 & 0.20445 & -0.06506** & 0.25795 & -0.02684 & 0.15867 & -0.08525 & 0.04713 & 0.00632 & 0.29616 \\
265 +\textbf{New home sales} & -0.07277 & 1.40281 & 0.41370*** & 1.62105*** & 0.16036*** & 0.36746 & 0.02375 & 0.47451* & 0.03870 & -0.38979 & -0.01679 & -0.90365** \\
266 +\textbf{Non-farm employment} & 0.38153 & -0.79927 & 0.73582*** & 0.87849*** & 0.33138*** & 1.02249** & 0.04335 & -0.45437* & 0.13090 & -0.34504 & 0.09655* & 0.66310 \\
267 +\textbf{Pending home sales} & -0.16135 & 2.17731 & 0.33625*** & -0.09161 & 0.08945*** & -0.37429 & 0.08860** & 0.17491 & 0.14610* & 0.30697 & 0.01209 & 2.49867*** \\
268 +\textbf{Personal consumption} & 0.44670 & -0.26628 & -0.00970 & -0.25217 & 0.01905 & -0.42262 & -0.03202 & 0.21254 & 0.00779 & 0.24570 & -0.10497** & 2.07186*** \\
269 +\textbf{Personal income} & -0.05879 & -3.64050 & -0.05739 & 1.06157*** & 0.03155 & 1.21226*** & -0.03398 & -0.03706 & 0.02667 & 0.69591 & -0.11986** & -0.22351 \\
270 +\textbf{Producer price index} & -0.73060 & 3.86034 & 0.40256*** & -0.23286 & 0.16360*** & 1.00912** & -0.00112 & 0.23935 & -0.07333 & 0.37691 & -0.02853 & -0.98772** \\
271 +\textbf{Trade balance} & 0.88337 & 0.15135 & -0.05401 & 0.00010 & 0.08634*** & 0.00010 & -0.00930 & 0.00010 & 0.02746 & 0.00010 & 0.10671** & -0.02417 \\
272 +\textbf{Crude Oil Weekly inventory} & 0.18957 & 0.00010 & & & & & & & & & & \\
273 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.55306*** & 0.00010 \\ \midrule
274 +
275 +\textbf{Observations} & \multicolumn{2}{c}{167663} & \multicolumn{2}{c}{167808} & \multicolumn{2}{c}{167513} & \multicolumn{2}{c}{167800} & \multicolumn{2}{c}{99525} & \multicolumn{2}{c}{167472} \\
276 +\textbf{$R^2$} & \multicolumn{2}{c}{0.031995} & \multicolumn{2}{c}{0.09131} & \multicolumn{2}{c}{0.073455} & \multicolumn{2}{c}{0.085618} & \multicolumn{2}{c}{0.043910} & \multicolumn{2}{c}{0.135931} \\ \bottomrule
277 +\end{tabular}
278 +}
279 +\begin{tablenotes}
280 +\singlespacing
281 +\footnotesize
282 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the financialization variable $SWAP_NLS_t$, calculated with the swap dealer positions. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2020-01-31 to 2022-06-10. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of financialization.
283 +\end{tablenotes}
284 +\end{sidewaystable}
285 +
286 +
287 +\begin{sidewaystable}
288 +\caption{Effects of Macro Announcements and Financialization on Futures Bid-Ask Spreads - COVID Sample Period}
289 +\label{tab:macro_fin_fut_spread_covid}
290 +\centering
291 +\resizebox{\linewidth}{!}{%
292 +\begin{tabular}{@{}lllllllllllll@{}}
293 +\toprule
294 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
295 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule
296 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
297 +
298 +\textbf{Initial jobless claims} & -8.45457* & 32.63860 & -32.15185 & 68.58676 & 2.69425 & -10.34599 & -3.69442 & -0.68497 & 17.97636 & -38.23165 & -2.18300 & 16.07627 \\
299 +\textbf{ADP Employment} & -7.01380 & 27.57371 & 2.75782 & -15.96412 & -6.32736 & 23.38047 & -1.04991 & 5.47512 & 10.51878 & -83.44112 & -6.04361 & -15.95091 \\
300 +\textbf{CB Consumer} & 1.72193 & -8.29055 & 0.20613 & 0.67657 & -1.33193 & 8.92511 & -3.85447 & 14.70581 & -12.61134 & -0.93962 & -2.02269 & 32.67489 \\
301 +\textbf{Advance retail sales} & 0.47142 & -2.97107 & -12.26401 & 22.01365 & 6.34823 & -43.34305 & -5.43414 & 3.75827 & 31.56759 & -174.90854* & -8.11642 & 16.35424 \\
302 +\textbf{Building permit} & 6.89017 & -27.68956 & 3.39690 & -7.73748 & -1.74443 & 13.41171 & -4.95466 & 3.63697 & -14.84855 & 0.76456 & -3.87804 & 1.60257 \\
303 +\textbf{Construction spending} & -2.61608 & 7.03623 & -13.45280 & 4.24283 & 1.96312 & -29.53599 & 0.94003 & 46.09774 & -11.47392 & 32.35733 & -12.96641*** & 1.97118 \\
304 +\textbf{Consumer credit} & 2.41314 & -9.94837 & 7.78231 & -15.71571 & -2.71320 & 7.74770 & 0.59937 & 3.02055 & -25.74707 & -20.31381 & -2.07031 & 23.61450 \\
305 +\textbf{Consumer price index} & 7.20494 & -25.29491 & -3.42521 & -1.65148 & -39.28089*** & -145.43158*** & -11.52600 & 38.83663 & -32.67489 & -46.75416 & 2.77011 & 24.58328 \\
306 +\textbf{Durable goods orders} & -9.89192 & 37.28075 & -29.75619 & 69.17601 & 0.87449 & -13.73658 & -0.82563 & -54.12952* & 20.87349 & 128.40773* & -11.41486** & -52.47868 \\
307 +\textbf{Existing home sales} & -0.46522 & 0.34583 & 3.92832 & -6.46341 & 16.71948 & -67.10920 & 0.31434 & 9.70195 & 16.71922 & -8.48132 & -17.03243*** & -178.54859* \\
308 +\textbf{Factory orders} & -15.48047 & 74.63383 & 44.16864 & -105.93270 & 15.54016 & -74.27283 & 4.97664 & 45.23353 & -2.59182 & 11.01312 & -3.79898 & 64.35242 \\
309 +\textbf{Gross domestic product} & -1.58999 & 5.70127 & -43.82863 & 88.03500 & -1.47836 & -1.12038 & -0.70136 & -5.97657 & -7.17208 & 6.66107 & -11.27067* & -130.33774 \\
310 +\textbf{Housing starts} & 6.12910 & -24.96061 & -2.81247 & 10.02873 & -3.53717 & 21.79738 & -5.20147 & 5.35973 & -16.49801 & -7.01299 & -4.03108 & -5.86341 \\
311 +\textbf{Industrial production} & 2.91449 & -12.96496 & 48.87900 & -105.74691 & 0.00122 & -2.30832 & 4.95540 & 35.35167 & 2.15082 & -11.17841 & 1.06505 & 73.51929 \\
312 +\textbf{New home sales} & -24.01000* & -99.88216* & -11.07115 & 30.45099 & -9.11509 & 62.63078 & -1.48689 & 0.20277 & 10.26320 & -73.53129 & 7.81899 & 185.08587* \\
313 +\textbf{Non-farm employment} & -2.96793 & 11.23798 & 43.03459 & -112.42904 & -63.21266*** & -338.05257*** & -3.33554 & 12.29767 & -16.32851 & 41.11426 & 0.03300 & 79.67313 \\
314 +\textbf{Pending home sales} & -0.84391 & 2.97208 & -164.00384*** & -385.97966*** & 14.75059 & -72.72820 & -5.07796 & 9.09510 & -27.61151* & 15.32589 & -2.42626 & 5.97021 \\
315 +\textbf{Personal consumption} & 1.86327 & -0.94651 & -140.08397*** & -394.38524*** & -39.02233** & -135.70025* & 1.60913 & -11.74366 & -6.64960 & -88.63991 & -3.03578 & 67.46156 \\
316 +\textbf{Personal income} & -10.61629 & 46.60744 & -83.54386* & -227.64691** & -59.19209*** & -215.96703*** & -4.67215 & -5.26524 & -11.94332 & -78.55460 & -2.97292 & 41.70955 \\
317 +\textbf{Producer price index} & 1.70179 & -16.86221 & 7.01259 & -17.88897 & -10.70080 & 15.75020 & 0.00117 & -10.70497 & -9.80993 & 56.90673 & 0.98973 & -59.24571 \\
318 +\textbf{Trade balance} & -3.97475 & 14.39262 & -92.13706* & -218.06267* & 2.84071 & -15.06097 & -4.84260 & 9.71728 & 35.51784 & 133.79336 & -0.60358 & 16.84096 \\
319 +\textbf{Crude Oil Weekly inventory} & -10.02190** & -36.44271** & & & & & & & & & & \\
320 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -4.57990* & -7.32267 \\ \midrule
321 +\textbf{Observations} & \multicolumn{2}{c}{166685} & \multicolumn{2}{c}{166719} & \multicolumn{2}{c}{166731} & \multicolumn{2}{c}{166815} & \multicolumn{2}{c}{99034} & \multicolumn{2}{c}{166546} \\
322 +\textbf{$R^2$} & \multicolumn{2}{c}{0.001149} & \multicolumn{2}{c}{0.001423} & \multicolumn{2}{c}{0.000525} & \multicolumn{2}{c}{0.000218} & \multicolumn{2}{c}{0.000397} & \multicolumn{2}{c}{0.003331} \\ \bottomrule
323 +\end{tabular}
324 +}
325 +\begin{tablenotes}
326 + \singlespacing
327 + \footnotesize
328 + This table presents estimates of the equation $R_{\text{SPREAD}}_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{\text{SPREAD}}_{t-\tau}^{t}+\epsilon_{t}$, analyzing the effects of financialization and macroeconomic announcements on the bid-ask spread using the financialization variable $PCA_t$. The period covered is from 2020-01-31 to 2022-06-10. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the financialization variable $PCA_t$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and financialization. The $\beta$ coefficient represents the effect of the lagged bid-ask spread.
329 +\end{tablenotes}
330 +\end{sidewaystable}
331 +
332 +
333 +\begin{sidewaystable}
334 +\caption{Effects of Macro Announcements and Financialization (Money Managers only) on Futures Bid-Ask Spreads - COVID Sample Period}
335 +\label{tab:macro_fin_mm_fut_spread_covid}
336 +\centering
337 +\resizebox{\linewidth}{!}{%
338 +\begin{tabular}{@{}lllllllllllll@{}}
339 +\toprule
340 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
341 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule
342 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
343 +\textbf{Initial jobless claims} & -6.53749 & 36.99070 & -5.09100 & 8.51443 & 3.41174 & -17.48035 & -3.40548 & -2.41912 & 19.37294 & -11.70774 & -2.81559 & -11.19065 \\
344 +\textbf{ADP Employment} & -2.54564 & 12.16488 & -0.27605 & -23.53318 & 0.20027 & -4.49908 & -1.23394 & 6.01069 & 25.27774 & -127.94447* & -5.27134 & -23.83603 \\
345 +\textbf{CB Consumer} & -2.18130 & 12.93870 & -45.84963** & -249.22187** & -0.02745 & 4.28165 & -4.10521 & 13.23897 & -12.91633 & 1.48671 & -3.68272 & -21.06597 \\
346 +\textbf{Advance retail sales} & -0.35744 & 0.60072 & 12.70640 & -101.23214 & -0.61295 & -16.17703 & -6.01125 & 5.65043 & 18.50305 & -273.57349*** & -5.83752 & -135.71818 \\
347 +\textbf{Building permit} & 1.99415 & -9.94633 & -2.29627 & 13.91138 & 4.82895 & -18.77987 & -5.04469 & 3.49561 & -16.69684 & 13.34416 & -4.15163 & -17.87371 \\
348 +\textbf{Construction spending} & -2.75408 & 10.65215 & -11.46947 & -2.50924 & -2.52379 & -11.13408 & 0.78212 & 36.65732 & -16.15920 & 44.82008 & -13.10610*** & -59.57002 \\
349 +\textbf{Consumer credit} & 0.70337 & -4.51235 & 11.79169 & -57.03598 & -1.09255 & 1.01969 & 0.49895 & 3.20891 & -22.23673 & -35.21557 & -3.29733 & -8.11439 \\
350 +\textbf{Consumer price index} & -4.69999 & 51.54822 & -5.81072 & 7.74688 & -26.55744*** & -122.68944*** & -13.34444 & 40.37926 & -27.29220 & -60.56031 & 1.93294 & 1.46551 \\
351 +\textbf{Durable goods orders} & -19.22690** & -118.46268** & 17.25772 & -109.15007 & -1.05906 & -6.50919 & -0.64667 & -42.91178 & 5.24254 & -165.86851** & -8.48567* & -53.43413 \\
352 +\textbf{Existing home sales} & -1.22338 & 4.97090 & 3.87388 & -14.80845 & 15.17943* & -79.97844* & 0.10610 & 9.01045 & 21.46227 & -33.72955 & -8.30693 & -79.67399 \\
353 +\textbf{Factory orders} & -4.76485 & 41.39189 & 19.49683 & -102.68878 & 5.78866 & -44.82403 & 5.03609 & 33.39901 & -3.90920 & 7.89477 & -6.88273 & -23.95548 \\
354 +\textbf{Gross domestic product} & -1.04247 & 5.06662 & -8.41631 & 9.68084 & 0.71472 & -12.75077 & -0.58138 & -5.05292 & -9.61105 & 50.03523 & -2.92402 & -125.74563 \\
355 +\textbf{Housing starts} & 2.05682 & -11.80710 & -2.95333 & 26.21003 & 1.34557 & 0.91365 & -5.35574 & 5.12633 & -17.75742 & 5.21089 & -3.97261 & -18.35345 \\
356 +\textbf{Industrial production} & 0.35193 & -1.63075 & 23.54541 & -104.71462 & -1.09340 & 3.33982 & 4.44746 & 33.67293 & 3.49871 & -10.35028 & -1.81734 & -4.53699 \\
357 +\textbf{New home sales} & 7.92804 & -45.17302 & -1.18304 & 16.09251 & -4.08212 & 54.53273 & -1.44261 & -0.18358 & 18.08228 & -82.06347 & -0.78691 & 10.34050 \\
358 +\textbf{Non-farm employment} & -0.82310 & 1.81903 & 1.58473 & -30.55903 & -23.72445*** & -227.51036*** & -3.86881 & 13.19033 & -25.23743 & 79.33754 & -3.17593 & -0.45983 \\
359 +\textbf{Pending home sales} & -0.02295 & -1.11790 & -46.49598*** & -246.06299** & 3.57038 & -29.84477 & -5.31654 & 8.30615 & -30.47064 & 23.53088 & -2.51547 & 13.68996 \\
360 +\textbf{Personal consumption} & 0.01625 & 13.82409 & -29.32162** & -315.62711*** & -47.47106*** & -204.48894*** & 2.26731 & -13.85110 & 3.13005 & -100.63543 & -10.62189* & -143.20008** \\
361 +\textbf{Personal income} & -4.67095 & 35.99125 & -12.11663 & -128.20470* & -63.31274*** & -289.00279*** & -4.44401 & -6.07123 & -0.37210 & -110.28908 & -6.55370 & 55.53102 \\
362 +\textbf{Producer price index} & 0.59074 & -21.44445 & 1.19636 & -8.48787 & -8.07756 & 6.48243 & 0.41099 & -11.28126 & -14.63389 & 62.18375 & 3.17994 & -21.68273 \\
363 +\textbf{Trade balance} & -1.32886 & 3.12729 & 6.92187 & -37.58044 & 1.17240 & -11.13297 & -4.80578 & 7.07448 & 20.95441 & 150.55476 & -1.27918 & -1.62928 \\
364 +\textbf{Crude Oil Weekly inventory} & -8.79628 & 45.75521 & & & & & & & & & & \\
365 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -4.19533* & -26.06610 \\
366 +\midrule
367 +\textbf{Observations} & \multicolumn{2}{c}{166685} & \multicolumn{2}{c}{166719} & \multicolumn{2}{c}{166731} & \multicolumn{2}{c}{166815} & \multicolumn{2}{c}{99034} & \multicolumn{2}{c}{166546} \\
368 +\textbf{$R^2$} & \multicolumn{2}{c}{0.000479} & \multicolumn{2}{c}{0.000691} & \multicolumn{2}{c}{0.000759} & \multicolumn{2}{c}{0.000220} & \multicolumn{2}{c}{0.000466} & \multicolumn{2}{c}{0.001290} \\
369 +\bottomrule
370 +\end{tabular}
371 +}
372 +\begin{tablenotes}
373 + \singlespacing
374 + \footnotesize
375 + This table presents estimates of the equation $R_{\text{SPREAD}}_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{\text{SPREAD}}_{t-\tau}^{t}+\epsilon_{t}$, analyzing the effects of financialization and macroeconomic announcements on the bid-ask spread using the financialization variable $MM_NLS_t$, calculated with the money manager positions. The period covered is from 2020-01-31 to 2022-06-10. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the financialization variable $MM_NLS_t$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and financialization. The $\beta$ coefficient represents the effect of the lagged bid-ask spread.
376 +\end{tablenotes}
377 +\end{sidewaystable}
378 +
379 +
380 +\begin{sidewaystable}
381 +\caption{Effects of Macro Announcements and Financialization (Swap Dealers only) on Futures Bid-Ask Spreads - COVID Sample Period}
382 +\label{tab:macro_fin_sd_fut_spread_covid}
383 +\centering
384 +\resizebox{\linewidth}{!}{%
385 +\begin{tabular}{@{}lllllllllllll@{}}
386 +\toprule
387 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
388 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule
389 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
390 +\textbf{Initial jobless claims} & 10.24187 & 54.86243 & -19.79900 & -47.94679 & 1.04395 & 11.95376 & 0.62635 & -29.73601 & -0.94417 & 97.30547 & -1.47409 & -24.50216 \\
391 +\textbf{ADP Employment} & 10.29355 & 54.30388 & -20.42843 & -51.03246 & -3.41656 & -41.32937 & 9.60909 & -68.44492 & -43.34301 & 300.61380* & -2.79075 & -30.87692 \\
392 +\textbf{CB Consumer} & -93.69279*** & 458.93758*** & -112.66663* & 330.74749* & 0.70623 & 0.26801 & 11.04775 & -89.06956 & -5.85562 & -24.73806 & -0.82104 & -41.16415 \\
393 +\textbf{Advance retail sales} & 34.00074 & 169.06349 & 48.22101 & 157.71656 & -3.04485 & 6.14432 & 2.72265 & -50.99529 & -174.25599*** & 632.68492*** & -5.57590 & -33.72617 \\
394 +\textbf{Building permit} & 2.40051 & 10.26005 & -21.64306 & -65.43971 & -2.03091 & -69.35593 & 11.64191 & -113.75751 & -20.81458 & 32.81489 & -1.59690 & -35.80282 \\
395 +\textbf{Construction spending} & 11.35242 & 60.90826 & -57.88134 & -138.98892 & -4.73548 & -5.03573 & 15.79675 & -79.49194 & -8.78460 & -8.46941 & -10.92327*** & 3.32543 \\
396 +\textbf{Consumer credit} & -11.51170 & -58.97665 & -4.14107 & -16.16466 & -1.95637 & -19.32312 & 3.04256 & -16.03746 & -46.57863 & 102.99198 & -1.22397 & -31.05955 \\
397 +\textbf{Consumer price index} & 45.05825 & 218.46376 & 14.38145 & 57.41763 & -9.65485* & 88.71944 & 8.75704 & -97.35713 & -57.20775 & 120.35114 & 5.58330 & -66.19648 \\
398 +\textbf{Durable goods orders} & 10.51353 & 58.94425 & -36.54239 & -103.95797 & -1.95416 & 4.36544 & -35.37134** & 194.14464* & 77.56131 & -270.22198 & -7.09413 & -13.91533 \\
399 +\textbf{Existing home sales} & 5.62694 & 29.94909 & -20.82648 & -68.15208 & 3.39694 & 48.06118 & 4.48249 & -22.91333 & 6.25310 & 53.85716 & -12.91688** & 66.49211 \\
400 +\textbf{Factory orders} & -9.76512 & -51.85161 & -74.81358 & -229.86249 & 0.09464 & 46.53674 & -44.22186*** & 231.73259** & 16.83545 & -90.41738 & 0.67131 & -111.96233 \\
401 +\textbf{Gross domestic product} & 1.38827 & 7.88813 & -11.18460 & -11.74147 & 5.21688 & 126.46891 & 4.77117 & -45.00653 & 10.67219 & -77.56816 & -4.94355 & -26.94017 \\
402 +\textbf{Housing starts} & -0.79973 & -4.70638 & -49.82587 & -154.17195 & -1.92575 & -65.07468 & 11.45273 & -113.04466 & -25.09012 & 45.89718 & -1.66228 & -32.69808 \\
403 +\textbf{Industrial production} & -9.04338 & -46.23321 & -192.28574** & 569.08575** & -0.38470 & 2.10315 & 0.66317 & 45.23953 & -9.28254 & 66.17973 & -0.00280 & -32.91286 \\
404 +\textbf{New home sales} & 1.81114 & 2.99737 & -32.14284 & -102.23884 & 2.49281 & -37.07252 & -0.25835 & -7.75060 & -19.21790 & 143.30837 & 8.11365 & -136.05671 \\
405 +\textbf{Non-farm employment} & -0.09881 & 1.38195 & -41.41392 & -110.67017 & -0.41960 & 319.36321*** & 8.29837 & -71.76987 & -23.71044 & 37.26705 & 0.03030 & -60.23125 \\
406 +\textbf{Pending home sales} & 3.39829 & 16.95587 & -230.69323*** & 682.50102*** & -5.85800 & 155.38021** & 5.92134 & -65.65184 & -4.37531 & -109.72612 & -4.79465 & 43.67730 \\
407 +\textbf{Personal consumption} & 21.90717 & 95.65292 & -46.37971 & -197.44690 & 11.39043 & 93.80840 & 17.45292 & -111.85795 & -46.22900 & 185.56535 & -1.86717 & -64.54461 \\
408 +\textbf{Personal income} & 18.53030 & 87.08505 & -37.00806 & -142.20539 & -15.22461** & 134.75459** & 9.42308 & -96.61453 & -53.05446 & 204.31813 & -1.77619 & -49.21373 \\
409 +\textbf{Producer price index} & 23.95203 & 125.03180 & 12.29022 & 38.56243 & -7.92362 & -14.81001 & -16.94245 & 109.19877 & 21.59151 & -144.39402 & 1.19073 & 28.92005 \\
410 +\textbf{Trade balance} & 6.96130 & 37.26328 & 30.02162 & 89.82014 & -1.29340 & -11.27194 & 13.88305 & -119.16198 & 69.05956 & -131.54404 & -0.98567 & -3.43898 \\
411 +\textbf{Crude Oil Weekly inventory} & -22.17392** & 117.15902** & & & & & & & & & & \\
412 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -1.46102 & -44.32457 \\
413 +\midrule
414 +\textbf{Observations} & \multicolumn{2}{c}{166685} & \multicolumn{2}{c}{166719} & \multicolumn{2}{c}{166731} & \multicolumn{2}{c}{166815} & \multicolumn{2}{c}{99034} & \multicolumn{2}{c}{166546} \\
415 +\textbf{$R^2$} & \multicolumn{2}{c}{0.00114} & \multicolumn{2}{c}{0.001013} & \multicolumn{2}{c}{0.000521} & \multicolumn{2}{c}{0.002449} & \multicolumn{2}{c}{0.000819} & \multicolumn{2}{c}{0.008131} \\
416 +\bottomrule
417 +\end{tabular}
418 +}
419 +\begin{tablenotes}
420 + \singlespacing
421 + \footnotesize
422 + This table presents estimates of the equation $R_{\text{SPREAD}}_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{\text{SPREAD}}_{t-\tau}^{t}+\epsilon_{t}$, analyzing the effects of financialization and macroeconomic announcements on the bid-ask spread using the financialization variable $SWAP_NLS_t$, calculated with the swap dealer positions. The period covered is from 2020-01-31 to 2022-06-10. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the financialization variable $SWAP_NLS_t$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and financialization. The $\beta$ coefficient represents the effect of the lagged bid-ask spread.
423 +\end{tablenotes}
424 +\end{sidewaystable}
425 +
426 +
427 +\begin{sidewaystable}
428 +\caption{Effects of Macro Announcements and Financialization (NLS) on Futures Returns - COVID Sample Period}
429 +\label{tab:macro_fin_nls_fut_returns_covid}
430 +\centering
431 +\resizebox{\linewidth}{!}{%
432 +\begin{tabular}{@{}lllllllllllll@{}}
433 +\toprule
434 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
435 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule
436 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
437 +\textbf{Initial jobless claims} & -0.05304 & 0.15319 & 0.67456*** & -1.29179*** & 0.07436 & -0.34916 & -0.01207 & -0.07707 & -0.01584 & 0.07139 & -0.11231** & -1.56397*** \\
438 +\textbf{ADP Employment} & 0.22435 & -0.60967 & -0.44274*** & 0.90844*** & 0.70483*** & -3.01802*** & 0.02568 & 0.16286 & 0.00351 & -0.01967 & -0.12050** & -2.93345** \\
439 +\textbf{CB Consumer} & 0.01593 & -0.05564 & -0.31780*** & 0.71482** & -0.00624 & 0.06684 & -0.02688 & 0.13241 & -0.02079 & 0.10325 & 0.06404 & 0.39286 \\
440 +\textbf{Advance retail sales} & -0.05187 & 0.33324 & -0.41640*** & 0.96100*** & 0.20226** & -0.82729* & 0.01210 & -0.07264 & -0.01557 & 0.09594 & -0.02402 & -0.34742 \\
441 +\textbf{Building permit} & 0.03941 & 0.03705 & -0.05159 & 0.11245 & -0.11363* & 0.40468 & -0.03852** & 0.16867** & 0.01120 & 0.00524 & 0.14075*** & 0.99814* \\
442 +\textbf{Construction spending} & 0.65898** & -2.88345** & -0.03630 & 0.03066 & 0.07823 & -0.48804 & 0.01588 & -0.16719 & 0.01435 & 0.11400 & -0.13218* & -2.26882** \\
443 +\textbf{Consumer credit} & -0.03940 & 0.12984 & -0.04807 & 0.10213 & 0.09214** & -0.39242** & -0.00315 & 0.06116 & -0.01061 & 0.13497 & -0.01933 & -0.12264 \\
444 +\textbf{Consumer price index} & 0.31416* & -1.56755** & -1.09536*** & 2.65481*** & -0.19543** & 0.46196 & -0.08642*** & 0.46350** & -0.08641*** & 0.22573* & -0.01456 & 0.15191 \\
445 +\textbf{Durable goods orders} & 0.35070 & -1.58708 & -0.48512*** & 1.08308*** & 0.22685* & -1.10379* & -0.03552 & 0.11180 & 0.02170 & -0.34540 & 0.00889 & -0.33582 \\
446 +\textbf{Existing home sales} & 0.37743 & -1.89630* & -0.10240 & 0.30048 & 0.24303** & -1.09853** & -0.02365 & 0.10840 & 0.03241 & -0.18899 & 0.08941 & -0.06124 \\
447 +\textbf{Factory orders} & 0.20435 & -0.68398 & -0.02856 & 0.01114 & -0.06706 & 0.21191 & 0.04305* & 0.09729 & 0.02654 & -0.31984 & 0.26264** & 0.86038 \\
448 +\textbf{Gross domestic product} & -0.20342 & 0.86461 & -0.67894*** & 1.57151*** & -0.18048 & 0.75512 & -0.05325*** & 0.21315** & -0.05130** & -0.03363 & -0.05500** & -0.40989 \\
449 +\textbf{Housing starts} & 0.36735** & -1.80380** & -0.03907 & 0.08773 & -0.04173 & 0.00353 & -0.02956 & 0.16777 & 0.02437 & -0.28613 & 0.16296*** & 2.02325*** \\
450 +\textbf{Industrial production} & -0.17304 & 0.64564 & -0.17486** & 0.50872*** & -0.00921 & -0.19491 & -0.02577 & 0.08019 & 0.00533 & 0.04843 & -0.04705 & -0.36049 \\
451 +\textbf{New home sales} & -0.23104 & 0.98392 & -0.25247** & 0.63549** & 0.06011 & -0.42542 & -0.00855 & 0.01137 & 0.02505 & -0.06303 & 0.00508 & 1.08808* \\
452 +\textbf{Non-farm employment} & -0.27310 & 1.19546 & -2.14386*** & 4.54338*** & 0.78667*** & -3.34173** & -0.02520*** & 0.52836* & 0.02156 & -0.18188 & -0.19984** & -5.81269** \\
453 +\textbf{Pending home sales} & -0.06442 & 0.34858 & -0.03672 & 0.05687 & -0.00501 & 0.03113 & 0.00715 & -0.00186 & -0.01095 & 0.15398 & -0.05828 & -0.67781 \\
454 +\textbf{Personal consumption} & 0.11481 & -0.56463 & -0.24542*** & 0.49336*** & -0.00707 & 0.05437 & 0.00234 & -0.00638 & 0.01645 & 0.12192 & 0.05777 & 2.02525** \\
455 +\textbf{Personal income} & -0.04283 & 0.05470 & 0.39444 & -0.86742 & -0.01019 & 0.01571 & -0.01784* & 0.27411* & -0.00134 & -0.13281 & -0.13135 & -3.18790 \\
456 +\textbf{Producer price index} & -0.16712 & 0.84336 & -0.15421 & 0.30151 & -0.03584 & 0.11642 & 0.00967 & -0.04937 & 0.02245 & 0.20957 & -0.03629 & 0.43642 \\
457 +\textbf{Trade balance} & -0.73246 & 2.12028 & -0.33167 & 0.53286 & -0.21421 & -0.08428 & -0.04745 & -0.53524 & -0.58781*** & 0.60070** & -0.07978 & 6.76062** \\
458 +\textbf{Crude Oil Weekly inventory} & -0.06204 & -0.71763** & & & & & & & & & & \\
459 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.14237*** & 0.47267 \\
460 +\midrule
461 +\textbf{Observations} & \multicolumn{2}{c}{167663} & \multicolumn{2}{c}{167808} & \multicolumn{2}{c}{167513} & \multicolumn{2}{c}{167800} & \multicolumn{2}{c}{99525} & \multicolumn{2}{c}{167472} \\
462 +\textbf{$R^2$} & \multicolumn{2}{c}{0.004142} & \multicolumn{2}{c}{0.003897} & \multicolumn{2}{c}{0.001678} & \multicolumn{2}{c}{0.000988} & \multicolumn{2}{c}{0.001109} & \multicolumn{2}{c}{0.001805} \\
463 +\bottomrule
464 +\end{tabular}
465 +}
466 +\begin{tablenotes}
467 +\singlespacing
468 +\footnotesize
469 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and financialization variable $X_{t,2}=NLS_t$. The period covered is from 2020-01-31 to 2022-06-10. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred and especially if that announcement was unanticipated. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of financialization.
470 +\end{tablenotes}
471 +\end{sidewaystable}
472 +
473 +
474 +
475 +\begin{sidewaystable}
476 +\caption{Effects of Macro Announcements and Financialization (NLS) on Futures Conditional Variance - COVID Sample Period}
477 +\label{tab:macro_fin_nls_var_covid}
478 +\centering
479 +\resizebox{\linewidth}{!}{%
480 +\begin{tabular}{@{}lllllllllllll@{}}
481 +\toprule
482 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
483 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule
484 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
485 +\textbf{Initial jobless claims} & -0.09636 & 3.35298* & 0.02294 & -0.15107 & 0.19983*** & -1.07585*** & 0.01309 & -0.04209 & 0.08340*** & -0.14983 & 0.00017 & 0.15059 \\
486 +\textbf{ADP Employment} & 0.65757* & -4.72734*** & 0.06518 & -0.34747* & 0.28162*** & -0.14248 & 0.06963*** & -0.06428 & -0.09531** & -0.20166 & 0.01880 & 0.66031 \\
487 +\textbf{CB Consumer} & -0.94417*** & -0.62403 & 0.21887*** & -0.52652** & 0.10017 & -1.30271*** & 0.00055 & -0.06044 & 0.03359 & -0.23366 & 0.02124 & 0.04480 \\
488 +\textbf{Advance retail sales} & 0.12243 & -2.33292 & 0.26516** & -0.68949*** & 0.41279*** & -0.52480 & 0.02151 & -0.28203*** & 0.00377 & -0.22670 & 0.05886 & -0.29694 \\
489 +\textbf{Building permit} & 0.48651 & 1.59467 & -0.28847*** & -0.00757 & 0.11495 & 0.26274 & 0.02049 & 0.04902 & 0.04828 & -0.03058 & 0.03644 & -0.69603 \\
490 +\textbf{Construction spending} & -0.30513 & -1.01085 & 0.04250 & -0.05809 & -0.03163 & -0.22089 & -0.00162 & 0.07112 & -0.00531 & 0.16762 & 0.00908 & -0.03468 \\
491 +\textbf{Consumer credit} & 0.20470 & -1.53043 & 0.02835 & -0.10382 & 0.09498 & -0.71454* & -0.01401 & -0.32437*** & 0.02753 & 0.08909 & 0.01673 & -0.33224 \\
492 +\textbf{Consumer price index} & 0.30876 & 0.61810 & 0.14219 & -0.02822 & 0.26772*** & 0.15355 & 0.13431*** & -0.17237* & -0.02796 & -0.04975 & 0.01102 & -0.71745 \\
493 +\textbf{Durable goods orders} & -0.14714 & 1.43927 & 0.03029 & 0.21013 & 0.00619 & -0.58963* & -0.03130* & -0.07651 & 0.04970 & -0.06524 & -0.10203** & -0.07702 \\
494 +\textbf{Existing home sales} & -0.21432 & 0.77755 & -0.04377 & 0.29942 & 0.18678** & 0.04353 & 0.05590*** & -0.09597 & 0.18962*** & -0.06904 & 0.08217* & -1.33778** \\
495 +\textbf{Factory orders} & -0.10802 & -0.49357 & -0.09774 & 0.19930 & 0.01168 & 0.57755 & 0.03064* & -0.21455* & 0.01980 & 0.13061 & -0.03477 & -0.97256 \\
496 +\textbf{Gross domestic product} & 0.08325 & -2.59059 & -0.02549 & 0.62377*** & -0.03792 & -0.55955* & -0.01721 & -0.34547*** & 0.00297 & -0.23786 & -0.11512** & -0.71962 \\
497 +\textbf{Housing starts} & 0.54768 & 1.97644 & -0.26025*** & 0.05228 & 0.12271 & 0.48892 & 0.02738* & -0.06923 & 0.04611 & -0.37882* & 0.01934 & 0.67079 \\
498 +\textbf{Industrial production} & -0.45243 & -0.19230 & -0.03451 & 0.27323* & -0.15184 & -0.36886 & -0.01130 & -0.07301 & 0.08228* & 0.01327 & 0.00786 & -0.48328 \\
499 +\textbf{New home sales} & 0.08668 & -0.41910 & 0.07095 & -1.44824*** & 0.23089*** & -0.99621*** & -0.01913 & -0.04275 & 0.02849 & 0.32940** & -0.01434 & 1.36646*** \\
500 +\textbf{Non-farm employment} & 0.17025 & 1.42351 & 0.80207*** & -0.65071*** & 0.54701*** & -0.73662** & 0.12699*** & 0.12890 & 0.04275 & 0.27753 & 0.09921** & -0.59763 \\
501 +\textbf{Pending home sales} & -0.29046 & -0.79520 & 0.31423*** & -0.29805 & 0.20897*** & -0.03802 & -0.00090 & -0.06165 & 0.06603 & -0.06209 & 0.03494 & -3.15790*** \\
502 +\textbf{Personal consumption} & 0.14833 & -1.21877 & 0.14734 & -0.22859 & 0.04742 & 0.24194 & 0.00385 & -0.02860 & 0.06989 & -0.03114 & -0.07238 & -2.00898*** \\
503 +\textbf{Personal income} & 0.25973 & -0.28073 & 0.12412 & 0.46855** & -0.00291 & -0.54811 & 0.00321 & -0.00955 & 0.07410* & -0.51366** & -0.06398 & 0.07551 \\
504 +\textbf{Producer price index} & 0.08816 & 0.43065 & -0.15708 & -0.35724* & 0.22355*** & -0.59363* & -0.00614 & -0.03986 & 0.09817** & -0.15742 & -0.04277 & 0.93243* \\
505 +\textbf{Trade balance} & -0.01952 & 0.74449 & 0.17818** & 0.00010 & 0.17335** & 0.00010 & 0.03429** & 0.00010 & 0.10701*** & 0.00010 & 0.07544* & -0.17694 \\
506 +\textbf{Crude Oil Weekly inventory} & -0.00049 & 0.00010 & & & & & & & & & & \\
507 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.53992*** & 0.00010 \\
508 +\midrule
509 +\textbf{Observations} & \multicolumn{2}{c}{167663} & \multicolumn{2}{c}{167808} & \multicolumn{2}{c}{167513} & \multicolumn{2}{c}{167800} & \multicolumn{2}{c}{99525} & \multicolumn{2}{c}{167472} \\
510 +\textbf{$R^2$} & \multicolumn{2}{c}{0.026858} & \multicolumn{2}{c}{0.104799} & \multicolumn{2}{c}{0.072185} & \multicolumn{2}{c}{0.086944} & \multicolumn{2}{c}{0.047558} & \multicolumn{2}{c}{0.137333} \\
511 +\bottomrule
512 +\end{tabular}
513 +}
514 +\begin{tablenotes}
515 +\singlespacing
516 +\footnotesize
517 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the financialization variable $NLS_t$. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2020-01-31 to 2022-06-10. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of financialization.
518 +\end{tablenotes}
519 +\end{sidewaystable}
added phd_chap1_20260731/appendix_full.tex +93 −0
@@ -0,0 +1,93 @@
1 +\section{Appendix A - Full Sample Results using the NLS Proxy}
2 +\renewcommand{\thetable}{A-\arabic{table}}
3 +\setcounter{table}{0} % Reset table counter
4 +\begin{sidewaystable}
5 +\caption{Effects of Macro Announcements and Financialization (NLS) on Futures Returns - Full Sample}
6 +\label{tab:macro_fin_nls_fut_returns_full}
7 +\centering
8 +\resizebox{\linewidth}{!}{%
9 +\begin{tabular}{@{}lllllllllllll@{}}
10 +\toprule
11 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
12 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
13 +
14 +\textbf{Initial jobless claims} & -0.19763*** & 0.87888*** & 0.70022*** & -1.33955*** & -0.44930*** & 1.98801*** & -0.00740 & -0.01540 & 0.02913** & -0.22057*** & -0.00266 & -0.10938 \\
15 +\textbf{ADP Employment} & 0.34291*** & -1.06756** & -1.18741*** & 2.52498*** & 0.58107*** & -2.36989*** & -0.02591*** & 0.16900 & 0.00801 & -0.14170 & -0.11464*** & -2.79764*** \\
16 +\textbf{CB Consumer} & 0.11672*** & -0.46895*** & -0.05938*** & 0.07030* & 0.07957*** & -0.15159** & 0.00049 & 0.00567 & -0.06841*** & 0.11631** & 0.02933 & -0.01440 \\
17 +\textbf{Advance retail sales} & 0.19991*** & -0.74542*** & -0.22861*** & 0.44747*** & 0.10409*** & -0.42536** & 0.01333 & -0.08638 & -0.01767 & 0.00664 & -0.02725 & -0.47149 \\
18 +\textbf{Building permit} & -0.00037 & 0.03995 & -0.02651** & 0.03697 & 0.06270*** & -0.15653** & 0.00698 & -0.00358 & -0.00233 & -0.03748 & 0.04272* & 0.36164** \\
19 +\textbf{Construction spending} & 0.02190 & -0.14609 & -0.04079** & 0.09211** & 0.04044* & -0.14376* & -0.00070 & -0.18868*** & 0.00951 & -0.00257 & -0.02886 & -0.17899 \\
20 +\textbf{Consumer credit} & -0.01646 & 0.08344 & -0.01359* & 0.02955 & -0.01554* & 0.06217 & 0.00217 & 0.02437 & 0.01087 & -0.01043 & -0.01175 & -0.11173 \\
21 +\textbf{Consumer price index} & 0.08391*** & -0.29308** & -0.17224*** & 0.31022*** & 0.26865*** & -0.75945*** & -0.03002*** & -0.02265 & -0.14233*** & 0.23064*** & 0.01302 & 0.15012 \\
22 +\textbf{Durable goods orders} & 0.15300*** & -0.73816*** & -0.07265*** & 0.14703*** & 0.06435*** & -0.14930* & -0.01052* & 0.00024 & -0.05354*** & 0.14525*** & -0.00328 & -0.09972 \\
23 +\textbf{Existing home sales} & 0.08722*** & -0.59238*** & -0.02000 & 0.05026 & -0.02244 & 0.10822 & -0.02238*** & -0.06905 & 0.01177 & -0.04488 & 0.04083 & 0.21897 \\
24 +\textbf{Factory orders} & -0.00112 & 0.01036 & -0.02949 & 0.00285 & -0.05976** & 0.14591 & 0.00357 & 0.02488 & -0.01887 & 0.00026 & 0.10729*** & 0.68503*** \\
25 +\textbf{Gross domestic product} & 0.04571* & -0.18390 & -0.16085*** & 0.24165*** & 0.16953*** & -0.33056*** & 0.00920 & -0.00478 & -0.05405*** & 0.00789 & -0.03601 & -0.33730* \\
26 +\textbf{Housing starts} & 0.03224 & -0.12690 & -0.06263*** & 0.11383*** & 0.07118*** & -0.14876** & 0.00503 & 0.00563 & 0.00159 & -0.03648* & 0.02340 & 0.38343** \\
27 +\textbf{Industrial production} & 0.02075 & -0.15376 & -0.00085 & -0.07097 & -0.01332 & -0.11571 & -0.00947 & -0.00797 & -0.02227 & 0.02109 & -0.02279 & -0.02294 \\
28 +\textbf{New home sales} & 0.09853*** & -0.43638** & -0.06940*** & 0.13595*** & -0.03309* & -0.04095 & -0.02368*** & 0.08652* & -0.00796 & 0.02235 & -0.00278 & -0.08427 \\
29 +\textbf{Non-farm employment} & 1.40422*** & -5.11125*** & -3.19414*** & 6.81185*** & 1.38897*** & -6.04367*** & -0.02702*** & 0.62417** & 0.00013 & 0.15340 & -0.21810*** & -6.29434*** \\
30 +\textbf{Pending home sales} & 0.07333*** & -0.34943** & -0.01840 & 0.00310 & -0.01292 & -0.02301 & -0.01712** & -0.11126 & -0.00877 & 0.03099 & -0.02417 & -0.39913 \\
31 +\textbf{Personal consumption} & -0.01478 & 0.06799 & -0.03594** & 0.06306* & 0.00926 & -0.05828 & 0.00274 & 0.08690 & 0.01300 & -0.05027 & 0.00370 & 0.36897 \\
32 +\textbf{Personal income} & 0.00873 & -0.13628 & -0.07190 & 0.15000 & 0.25364*** & -1.05220** & -0.01709* & 0.27874** & -0.01271 & -0.02240 & -0.02828 & -0.82024 \\
33 +\textbf{Producer price index} & 0.03513* & -0.23561** & -0.07698*** & 0.15026*** & -0.01476 & -0.04153 & -0.00480 & -0.01293 & -0.02035 & 0.04451 & -0.04862* & -0.22346 \\
34 +\textbf{Trade balance} & 0.00320 & -0.05635 & -0.05207** & 0.11911** & -0.01111 & 0.04144 & -0.00354 & -0.13232 & -0.03871 & 0.07731 & -0.04667 & -0.45064 \\
35 +\textbf{Crude Oil Weekly inventory} & -0.07659*** & -0.39633*** & & & & & & & & & & \\
36 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.36472*** & -0.07672 \\ \midrule
37 +
38 +\textbf{Observations} & \multicolumn{2}{c}{1193455} & \multicolumn{2}{c}{1190001} & \multicolumn{2}{c}{1180816} & \multicolumn{2}{c}{1138696} & \multicolumn{2}{c}{749168} & \multicolumn{2}{c}{1101836} \\
39 +\textbf{$R^2$} & \multicolumn{2}{c}{0.001975} & \multicolumn{2}{c}{0.002043} & \multicolumn{2}{c}{0.001014} & \multicolumn{2}{c}{0.000290} & \multicolumn{2}{c}{0.000246} & \multicolumn{2}{c}{0.001176} \\ \bottomrule
40 +\end{tabular}
41 +}
42 +\begin{tablenotes}
43 +\singlespacing
44 +\footnotesize
45 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and financialization variable $X_{t,2}=NLS_t$. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred and especially if that announcement was unanticipated. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of financialization.
46 +\end{tablenotes}
47 +\end{sidewaystable}
48 +
49 +
50 +\begin{sidewaystable}
51 +\caption{Effects of Macro Announcements and Financialization (NLS) on Futures Conditional Variance - Full Sample}
52 +\label{tab:macro_fin_nls_var_full}
53 +\centering
54 +\resizebox{\linewidth}{!}{%
55 +\begin{tabular}{@{}lllllllllllll@{}}
56 +\toprule
57 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
58 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
59 +
60 +\textbf{Initial jobless claims} & 0.10116*** & -0.43618*** & 0.04743*** & 0.01989 & 0.11651*** & -0.18846*** & 0.02474*** & -0.05555*** & 0.11786*** & -0.11355*** & -0.01167 & -0.02107 \\
61 +\textbf{ADP Employment} & 0.02792 & -0.20043 & 0.06377*** & -0.08427*** & 0.05327*** & -0.03006 & 0.02831*** & 0.03382 & -0.02858* & 0.01538 & 0.01600 & 0.05689 \\
62 +\textbf{CB Consumer} & 0.10356*** & -0.41440*** & 0.03139*** & 0.01209 & 0.04540*** & 0.02788 & 0.02838*** & -0.10290** & 0.06091*** & -0.09895*** & -0.00770 & -0.01276 \\
63 +\textbf{Advance retail sales} & 0.16868*** & -0.86165*** & 0.09604*** & -0.05612* & 0.12379*** & -0.05358 & 0.02819*** & -0.15572*** & 0.11351*** & -0.16926*** & 0.05769*** & 0.25105* \\
64 +\textbf{Building permit} & 0.10470*** & -0.54035*** & 0.04592*** & -0.06034* & 0.06133*** & -0.19301*** & 0.00484 & -0.07408* & 0.07672*** & -0.06975*** & 0.02190 & -0.02627 \\
65 +\textbf{Construction spending} & 0.15338*** & -0.62724*** & 0.09518*** & -0.11889*** & 0.08606*** & -0.09793 & 0.04108*** & -0.18959*** & 0.04626*** & 0.01330 & -0.01904 & -0.32089** \\
66 +\textbf{Consumer credit} & 0.04798* & -0.26103* & 0.00778 & -0.00318 & 0.01661 & 0.00378 & 0.00280 & 0.03858 & -0.00633 & 0.03412 & 0.00152 & 0.01392 \\
67 +\textbf{Consumer price index} & 0.07237** & -0.06680 & 0.13915*** & -0.07513* & 0.26627*** & -0.52898*** & 0.09389*** & -0.37452*** & 0.18224*** & -0.27404*** & 0.02706 & 0.03922 \\
68 +\textbf{Durable goods orders} & 0.09826*** & -0.41079*** & 0.02310** & 0.01561 & 0.04156*** & 0.02205 & 0.00773 & -0.02411 & 0.06781*** & -0.03730 & -0.02225 & -0.25576* \\
69 +\textbf{Existing home sales} & 0.04757* & -0.05948 & 0.01879* & 0.04102 & 0.03838*** & -0.02961 & 0.02231*** & -0.13249*** & 0.04815*** & -0.06749** & 0.02201 & -0.11251 \\
70 +\textbf{Factory orders} & 0.08762*** & -0.31063** & -0.00678 & 0.16162*** & 0.01842 & 0.15052** & 0.02641*** & -0.13159*** & 0.03966*** & -0.04030 & -0.03041 & -0.69494*** \\
71 +\textbf{Gross domestic product} & 0.11632*** & -0.61837*** & 0.04659*** & 0.05303 & 0.09959*** & -0.06323 & 0.02464*** & 0.01653 & 0.11025*** & -0.12093*** & -0.05116** & -0.30947** \\
72 +\textbf{Housing starts} & 0.11313*** & -0.60745*** & 0.03660*** & -0.03112 & 0.06213*** & -0.19012*** & 0.00848 & -0.09479** & 0.05963*** & -0.04924 & 0.01829 & 0.04047 \\
73 +\textbf{Industrial production} & 0.09578*** & -0.63833*** & 0.01928* & 0.00308 & 0.02059 & -0.00706 & 0.00182 & -0.04794 & 0.01964 & -0.07145** & -0.03281 & -0.18705 \\
74 +\textbf{New home sales} & 0.10149*** & -0.52688*** & 0.04068*** & 0.01515 & 0.05778*** & 0.02800 & 0.02485*** & -0.00096 & 0.06768*** & -0.11522*** & -0.02516 & -0.22929* \\
75 +\textbf{Non-farm employment} & 0.36475*** & -1.20583*** & 0.23592*** & 0.04682 & 0.38126*** & -0.14107** & 0.12881*** & -0.15602*** & 0.17336*** & -0.05048 & 0.02807 & -0.35383** \\
76 +\textbf{Pending home sales} & 0.10920*** & -0.49393*** & 0.01999* & -0.01624 & 0.02548* & 0.02096 & 0.02833*** & -0.00831 & 0.05870*** & -0.08511*** & -0.01374 & -0.30813** \\
77 +\textbf{Personal consumption} & 0.00638 & 0.02122 & 0.06203*** & -0.08145** & 0.10177*** & -0.26781*** & 0.00662 & -0.00781 & 0.11081*** & -0.14798*** & 0.02410 & -0.37604** \\
78 +\textbf{Personal income} & 0.00650 & 0.04657 & 0.09149*** & -0.15016*** & 0.12450*** & -0.34101*** & 0.01322** & -0.02338 & 0.09417*** & -0.10866*** & 0.00283 & -0.43167*** \\
79 +\textbf{Producer price index} & 0.09735*** & -0.34959** & 0.01720 & 0.08462** & 0.09045*** & -0.09926 & 0.01106* & -0.05096 & 0.11987*** & -0.06696** & 0.00810 & 0.00477 \\
80 +\textbf{Trade balance} & 0.07844*** & -0.12829 & -0.00650 & 0.20643*** & 0.00273 & 0.39295*** & 0.00919 & 0.02977 & 0.10323*** & -0.08686*** & 0.01821 & 0.11158 \\
81 +\textbf{Crude Oil Weekly inventory} & 0.00456 & 0.62373*** & & & & & & & & & & \\
82 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.53064*** & -0.68665*** \\ \midrule
83 +
84 +\textbf{Observations} & \multicolumn{2}{c}{1193455} & \multicolumn{2}{c}{1190001} & \multicolumn{2}{c}{1180816} & \multicolumn{2}{c}{1138696} & \multicolumn{2}{c}{749168} & \multicolumn{2}{c}{1101836} \\
85 +\textbf{$R^2$} & \multicolumn{2}{c}{0.070652} & \multicolumn{2}{c}{0.080638} & \multicolumn{2}{c}{0.073961} & \multicolumn{2}{c}{0.070571} & \multicolumn{2}{c}{0.076293} & \multicolumn{2}{c}{0.135964} \\ \bottomrule
86 +\end{tabular}
87 +}
88 +\begin{tablenotes}
89 +\singlespacing
90 +\footnotesize
91 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the financialization variable $NLS_t$. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2007-04-01 to 2024-02-11. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of financialization.
92 +\end{tablenotes}
93 +\end{sidewaystable}
added phd_chap1_20260731/appendix_zlb.tex +508 −0
@@ -0,0 +1,508 @@
1 +\section{Appendix C - Results using the ZLB Sample Period}
2 +
3 +\renewcommand{\thetable}{C-\arabic{table}}
4 +\setcounter{table}{0} % Reset table counter
5 +\begin{sidewaystable}
6 +\caption{Effects of Macro Announcements and Financialization on Futures Returns - ZLB Sample Period}
7 +\label{tab:macro_fin_fut_returns_zlb}
8 +\centering
9 +\resizebox{\linewidth}{!}{%
10 +\begin{tabular}{@{}lllllllllllll@{}}
11 +\toprule
12 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
13 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule
14 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
15 +\textbf{Initial jobless claims} & -0.13544*** & 0.03145 & 0.38937*** & -2.20180*** & -0.52935*** & 1.69862*** & 0.11665*** & -0.93702*** & 0.10728** & 0.09333 & -0.07764 & 0.02027 \\
16 +\textbf{ADP Employment} & 0.40401*** & -0.34661** & -1.04667*** & 0.17884*** & 1.09465*** & -0.30735*** & 0.18981 & 0.02337 & -0.27832** & -0.04634 & 0.35566* & -0.30378* \\
17 +\textbf{CB Consumer} & 0.04738*** & 0.01175 & -0.02012** & 0.62504*** & -0.00886 & -0.59504*** & -0.04047*** & 0.21153** & -0.01632 & 0.00898 & 0.04587** & -0.00829 \\
18 +\textbf{Advance retail sales} & 0.19924*** & -0.13052*** & -0.21536*** & -0.00301 & 0.21829*** & -0.02132** & -0.06358** & 0.01148 & -0.04149 & -0.01470 & 0.07127 & -0.08261** \\
19 +\textbf{Building permit} & 0.00876 & 0.01262 & -0.02796*** & 0.09826*** & 0.03448* & -0.14706*** & 0.01076 & -0.06447*** & -0.03238*** & -0.01135 & -0.01339 & -0.01139 \\
20 +\textbf{Construction spending} & -0.01690 & 0.01949 & 0.00174 & 0.00232 & 0.01964 & -0.00415 & -0.00216 & -0.01157 & 0.02190 & -0.00176 & -0.03652 & 0.01486 \\
21 +\textbf{Consumer credit} & -0.00477 & 0.01247 & -0.00297 & 0.00767 & 0.01331 & 0.00498 & 0.00779 & 0.02053** & 0.00688 & 0.01732 & 0.01528 & 0.00689 \\
22 +\textbf{Consumer price index} & -0.02604* & -0.01915 & -0.01121 & -0.00595* & 0.02882* & -0.01229*** & 0.00962 & -0.01451** & -0.03467** & 0.00074 & 0.01985 & 0.00602 \\
23 +\textbf{Durable goods orders} & 0.01955* & 0.00840 & -0.00926 & -0.00451 & 0.01189 & -0.00811 & 0.04012*** & 0.00826 & 0.04348*** & 0.01607 & 0.02851* & -0.01551 \\
24 +\textbf{Existing home sales} & 0.05956*** & -0.00741 & -0.03056*** & 0.02765*** & -0.02193 & -0.03688*** & -0.05200*** & 0.04313*** & -0.00767 & 0.00402 & -0.00295 & -0.01494 \\
25 +\textbf{Factory orders} & 0.00032 & 0.00381 & -0.02526*** & 0.03390*** & 0.03445** & -0.03750*** & 0.00079 & 0.00265 & -0.01269 & -0.01291 & 0.00259 & 0.01178 \\
26 +\textbf{Gross domestic product} & 0.03011** & -0.03081** & -0.09874*** & -0.00417 & 0.14231*** & -0.00916 & 0.04358*** & 0.00699 & -0.07660*** & -0.00398 & 0.03823* & -0.02532** \\
27 +\textbf{Housing starts} & 0.01897 & 0.00464 & -0.06460*** & 0.02261*** & 0.08712*** & -0.03600*** & 0.04787*** & -0.01312 & -0.01460 & -0.03896** & 0.01486 & -0.03128** \\
28 +\textbf{Industrial production} & 0.03864* & 0.02468 & -0.01743* & 0.01087* & 0.03365* & -0.01217 & 0.00807 & -0.03854*** & -0.02556 & 0.01029 & 0.02772 & -0.00121 \\
29 +\textbf{New home sales} & 0.04508** & 0.00345 & -0.05411*** & 0.00318 & 0.04529** & 0.00117 & 0.08631*** & 0.00136 & -0.01694 & 0.00085 & 0.02382 & 0.00223 \\
30 +\textbf{Non-farm employment} & 2.88465*** & -0.58559*** & -3.26273*** & 0.03767*** & 2.54446*** & -0.02702** & -1.63634*** & 0.03802*** & 0.00194 & -0.02640 & 1.13985*** & -0.86544*** \\
31 +\textbf{Pending home sales} & 0.03575** & 0.00827 & -0.00068 & -1.18804*** & 0.00061 & -1.77992*** & -0.03715*** & 0.72042*** & -0.01184 & -0.46442** & 0.02185 & -0.02241 \\
32 +\textbf{Personal consumption} & 0.05077*** & -0.06441** & -0.01743 & 0.00769 & 0.00016 & -0.00554 & 0.01817 & 0.01877** & -0.01690 & -0.03080* & -0.01447 & -0.00431 \\
33 +\textbf{Personal income} & -0.05408 & -0.07328 & -0.01315 & -0.02674*** & -0.07365 & 0.00118 & 0.02456 & -0.00143 & 0.00748 & 0.00283 & -0.07876 & -0.18097** \\
34 +\textbf{Producer price index} & 0.02545** & 0.01051 & -0.01703** & -0.00020 & -0.01533 & -0.05688 & 0.00989 & 0.07844** & 0.02613** & -0.03985 & -0.02221 & -0.00095 \\
35 +\textbf{Trade balance} & -0.02379 & -0.03072*** & -0.02416** & -0.00750 & -0.00933 & -0.00533 & -0.01005 & -0.00455 & 0.00269 & -0.00794 & 0.02307 & 0.00533 \\
36 +\textbf{Crude Oil Weekly inventory} & -0.18216*** & -0.04537*** & & & & & & & & & & \\
37 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.01427 & 0.00311 \\
38 +\midrule
39 +\textbf{Observations} & \multicolumn{2}{c}{495652} & \multicolumn{2}{c}{500857} & \multicolumn{2}{c}{492438} & \multicolumn{2}{c}{485244} & \multicolumn{2}{c}{315201} & \multicolumn{2}{c}{448530} \\
40 +\textbf{$R^2$} & \multicolumn{2}{c}{0.00258} & \multicolumn{2}{c}{0.003119} & \multicolumn{2}{c}{0.001676} & \multicolumn{2}{c}{0.000838} & \multicolumn{2}{c}{0.000442} & \multicolumn{2}{c}{0.000339} \\
41 +\bottomrule
42 +\end{tabular}
43 +}
44 +\begin{tablenotes}
45 +\singlespacing
46 +\footnotesize
47 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and the financialization variable $PCA_t$. The period covered is from 2008-12-22 to 2015-12-21. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred and especially if that announcement was unanticipated. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of financialization.
48 +\end{tablenotes}
49 +\end{sidewaystable}
50 +
51 +
52 +
53 +\begin{sidewaystable}
54 +\caption{Effects of Macro Announcements and Financialization on Futures Conditional Variance - ZLB Sample Period}
55 +\label{tab:macro_fin_fut_var_zlb}
56 +\centering
57 +\resizebox{\linewidth}{!}{%
58 +\begin{tabular}{@{}lllllllllllll@{}}
59 +\toprule
60 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
61 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
62 +\textbf{Initial jobless claims} & 0.05230*** & -0.02390*** & 0.05120*** & -0.00208 & 0.06948*** & 0.00435 & 0.03286*** & -0.00654** & 0.07873*** & -0.01437*** & -0.01883*** & -0.01405*** \\
63 +\textbf{ADP Employment} & -0.00064 & -0.02039** & 0.03401*** & -0.01693*** & 0.01975* & -0.01844*** & 0.03716*** & -0.01102* & 0.01213 & 0.00286 & 0.01214 & -0.01709 \\
64 +\textbf{CB Consumer} & 0.05423*** & 0.00382 & 0.03491*** & -0.00805** & 0.06318*** & -0.00753 & 0.04855*** & -0.00729 & 0.03441*** & 0.00766 & -0.00136 & -0.00468 \\
65 +\textbf{Advance retail sales} & 0.08163*** & 0.01716 & 0.06425*** & -0.02067*** & 0.08051*** & -0.01868** & 0.05908*** & -0.00111 & 0.06138*** & 0.00211 & 0.01204 & -0.01596 \\
66 +\textbf{Building permit} & 0.02824*** & -0.02616** & 0.04710*** & -0.01613*** & 0.05076*** & -0.02120*** & 0.03100*** & -0.00311 & 0.05838*** & -0.00125 & 0.00152 & -0.02018* \\
67 +\textbf{Construction spending} & 0.07983*** & -0.03904*** & 0.05463*** & -0.01756*** & 0.05670*** & 0.00613 & 0.06941*** & -0.02122*** & 0.04758*** & -0.00686 & 0.02882* & -0.01822 \\
68 +\textbf{Consumer credit} & 0.01569 & 0.01426 & 0.00817 & 0.00565 & 0.02038* & 0.00663 & 0.00402 & -0.00019 & -0.00170 & -0.00942 & 0.00357 & 0.00090 \\
69 +\textbf{Consumer price index} & 0.00563 & 0.01145 & 0.05637*** & -0.01680*** & 0.07700*** & -0.04432*** & 0.00890 & -0.00318 & 0.03889** & 0.00686 & -0.00879 & -0.01715 \\
70 +\textbf{Durable goods orders} & 0.05961*** & -0.05183*** & 0.04597*** & -0.00691* & 0.06557*** & -0.00012 & 0.03034*** & -0.00654 & 0.07662*** & 0.01014 & 0.01540 & -0.00200 \\
71 +\textbf{Existing home sales} & 0.01855* & -0.02477** & 0.04511*** & -0.01162*** & 0.05889*** & 0.00880 & 0.03025*** & -0.01350** & 0.00448 & -0.00450 & 0.02979** & -0.02428** \\
72 +\textbf{Factory orders} & 0.02384** & -0.00138 & 0.02625*** & 0.00064 & 0.02110* & -0.00514 & 0.01934** & -0.00737 & 0.02307* & 0.00569 & 0.04423*** & -0.00473 \\
73 +\textbf{Gross domestic product} & 0.04255*** & -0.05356*** & 0.06410*** & -0.01349*** & 0.10056*** & -0.01680** & 0.04971*** & -0.00819 & 0.07056*** & -0.03732*** & -0.03347** & -0.00852 \\
74 +\textbf{Housing starts} & 0.02871*** & -0.02024** & 0.03890*** & -0.01092*** & 0.04433*** & -0.01277* & 0.03447*** & 0.00137 & 0.05830*** & -0.00097 & -0.00431 & -0.01496 \\
75 +\textbf{Industrial production} & -0.00494 & -0.00948 & 0.01734*** & 0.00079 & 0.01097 & 0.00834 & 0.00637 & -0.00028 & -0.01547 & -0.01559* & -0.01885 & -0.03405*** \\
76 +\textbf{New home sales} & 0.04713*** & -0.03780*** & 0.05702*** & 0.00197 & 0.06054*** & 0.00135 & 0.04559*** & -0.01446** & 0.05664*** & -0.05001*** & 0.00841 & -0.02391** \\
77 +\textbf{Non-farm employment} & 0.25386*** & -0.00842** & 0.27310*** & -0.03339*** & 0.36578*** & -0.03915*** & 0.17894*** & -0.01511*** & 0.17564*** & 0.00139 & 0.09638*** & -0.01587* \\
78 +\textbf{Pending home sales} & 0.02915*** & -0.03815*** & 0.01095* & 0.00048 & 0.02361** & -0.01508** & 0.04939*** & -0.01620*** & 0.00786 & 0.00298 & 0.02930** & -0.07122*** \\
79 +\textbf{Personal consumption} & -0.01296 & -0.03662*** & 0.02088*** & 0.00532 & 0.03304*** & 0.00595 & -0.01157 & -0.01046* & 0.06353*** & 0.00710 & 0.04514*** & -0.01013 \\
80 +\textbf{Personal income} & 0.01238 & -0.03356*** & 0.05491*** & -0.01713*** & 0.07143*** & -0.03372*** & 0.01392 & 0.00363 & 0.06359*** & 0.01845 & 0.02882* & -0.01777 \\
81 +\textbf{Producer price index} & 0.06409*** & 0.01709 & 0.04481*** & 0.00272 & 0.05388*** & -0.02738*** & 0.02235** & -0.00973 & 0.14576*** & -0.02135** & 0.02488 & -0.01853 \\
82 +\textbf{Trade balance} & 0.05802*** & -0.01613*** & 0.04319*** & 0.00406 & 0.04435*** & -0.00285 & 0.01022 & -0.00094 & 0.04279*** & 0.00306 & -0.01591 & -0.01452* \\
83 +\textbf{Crude Oil Weekly inventory} & 0.11581*** & 0.07104*** & & & & & & & & & & \\
84 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.77055*** & -0.02633*** \\
85 +\midrule
86 +\textbf{Observations} & \multicolumn{2}{c}{495652} & \multicolumn{2}{c}{500857} & \multicolumn{2}{c}{492438} & \multicolumn{2}{c}{485244} & \multicolumn{2}{c}{315201} & \multicolumn{2}{c}{448530} \\
87 +\textbf{$R^2$} & \multicolumn{2}{c}{0.143674} & \multicolumn{2}{c}{0.084117} & \multicolumn{2}{c}{0.074892} & \multicolumn{2}{c}{0.068345} & \multicolumn{2}{c}{0.057901} & \multicolumn{2}{c}{0.163648} \\
88 +\bottomrule
89 +\end{tabular}
90 +}
91 +\begin{tablenotes}
92 +\singlespacing
93 +\footnotesize
94 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the financialization variable $PCA_t$. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2008-12-22 to 2015-12-21. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of financialization.
95 +\end{tablenotes}
96 +\end{sidewaystable}
97 +
98 +\begin{sidewaystable}
99 +\caption{Effects of Macro Announcements and Financialization (Money Managers only) on Futures Returns - ZLB Sample Period}
100 +\label{tab:macro_fin_mm_fut_returns_zlb}
101 +\centering
102 +\resizebox{\linewidth}{!}{%
103 +\begin{tabular}{@{}lllllllllllll@{}}
104 +\toprule
105 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
106 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
107 +\textbf{Initial jobless claims} & -0.61873*** & 3.79363*** & 0.82122*** & -1.83823*** & -0.94610*** & 3.51646*** & 0.13481*** & 0.26240 & -0.30845 & 0.89588* & -0.06368 & 0.28495 \\
108 +\textbf{ADP Employment} & 1.44785*** & -7.85026*** & -2.63505*** & 6.54340*** & 1.65861*** & -5.12131*** & -0.29183** & -0.98879 & 0.09768 & -0.81895 & 0.66288** & 0.35649 \\
109 +\textbf{CB Consumer} & 0.23286*** & -1.72586*** & -0.02786* & 0.03013 & 0.03949* & 0.20351 & -0.03721*** & -0.12280 & -0.10851* & 0.19894* & 0.06576** & 0.22021 \\
110 +\textbf{Advance retail sales} & 0.72061*** & -5.21570*** & -0.48559*** & 1.07889*** & 0.40723*** & -1.59806*** & -0.04838* & 0.51979** & -0.14490 & 0.22726 & 0.10462* & 0.03384 \\
111 +\textbf{Building permit} & 0.00943 & -0.03507 & -0.01738 & -0.04424 & 0.03626* & -0.02107 & -0.00067 & -0.01652 & 0.00852 & -0.09906 & -0.03031 & -0.14419 \\
112 +\textbf{Construction spending} & -0.01686 & -0.07735 & 0.00554 & -0.01680 & 0.03991 & -0.13778 & 0.01419 & -0.25300*** & 0.09235 & -0.15374 & -0.04047 & -0.07342 \\
113 +\textbf{Consumer credit} & -0.00935 & 0.00035 & -0.02739** & 0.10286** & 0.01537 & -0.18194*** & 0.00156 & 0.05545 & 0.05113 & -0.09891 & 0.02655 & 0.14118 \\
114 +\textbf{Consumer price index} & 0.00068 & -0.14800 & -0.04091** & 0.12628 & -0.04450* & 0.12259 & 0.00586 & -0.31484*** & 0.00768 & -0.09833 & 0.03047 & 0.18925 \\
115 +\textbf{Durable goods orders} & 0.20737*** & -1.59397*** & -0.08826*** & 0.32878*** & 0.06089*** & -0.45767*** & -0.02528** & -0.04901 & 0.07290 & -0.06165 & 0.00919 & -0.30145 \\
116 +\textbf{Existing home sales} & 0.09446** & -0.43011 & -0.07796*** & 0.23737*** & -0.04781 & 0.30091 & -0.05891*** & -0.13818 & -0.05208 & 0.10284 & 0.00204 & -0.03008 \\
117 +\textbf{Factory orders} & -0.05469 & 0.45466 & -0.05182* & 0.10148 & 0.07530** & 0.27928 & 0.00053 & -0.02256 & 0.02337 & -0.07138 & 0.02626 & 0.38077 \\
118 +\textbf{Gross domestic product} & 0.09203** & -0.51635** & -0.16170*** & 0.24657** & 0.25379*** & -0.84989*** & -0.02338* & 0.24025** & -0.23846*** & 0.38337** & 0.01852 & -0.11700 \\
119 +\textbf{Housing starts} & 0.10174** & -0.67073** & -0.08595*** & 0.07723 & 0.12082*** & 0.23702 & -0.02456** & -0.00819 & -0.06445 & 0.10574 & -0.01387 & -0.26555 \\
120 +\textbf{Industrial production} & 0.13054** & -0.83082* & -0.03781 & 0.08352 & -0.05807* & 0.16764 & 0.01605 & -0.17091 & -0.15723** & 0.27747* & 0.03047 & 0.02433 \\
121 +\textbf{New home sales} & 0.11875** & -0.61516 & -0.16013*** & 0.41718*** & 0.08742*** & 0.26801 & 0.07452*** & -0.01334 & -0.13515 & 0.25862 & 0.02251 & -0.06486 \\
122 +\textbf{Non-farm employment} & 8.42886*** & -45.72262*** & -7.18381*** & 15.65980*** & 5.95620*** & -24.71473*** & -1.55073*** & 2.37769** & -0.02971 & 0.41305 & 0.48504 & -11.46876*** \\
123 +\textbf{Pending home sales} & 0.23739*** & -1.67951*** & 0.01256 & -0.06539 & -0.00922 & 0.07800 & -0.03930*** & -0.06399 & -0.15640* & 0.32581* & -0.00562 & -0.42268 \\
124 +\textbf{Personal consumption} & -0.01615 & 0.35868 & -0.06144* & 0.19471* & 0.02187 & -0.16030 & 0.00970 & 0.19646 & -0.08276 & 0.13089 & -0.02085 & -0.11183 \\
125 +\textbf{Personal income} & -0.12451 & 1.06590 & 0.07096 & -0.31886 & -0.21668 & 1.21137 & 0.02078 & -0.32997 & -0.38298 & 0.75093 & -0.29178** & -3.32571*** \\
126 +\textbf{Producer price index} & 0.12110*** & -0.87303*** & -0.05265*** & 0.13217** & -0.04707* & 0.21052 & 0.01299 & -0.11666 & 0.09948* & -0.16284 & -0.03937 & -0.19326 \\
127 +\textbf{Trade balance} & -0.02921 & 0.14179 & -0.02691 & 0.02610 & -0.02325 & 0.10813 & -0.01952 & 0.11285 & 0.08551 & -0.16591 & 0.02784 & 0.06101 \\
128 +\textbf{Crude Oil Weekly inventory} & -0.14249*** & -0.10192 & & & & & & & & & & \\
129 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.69681*** & -0.87379*** \\ \midrule
130 +\textbf{Observations} & \multicolumn{2}{c}{495652} & \multicolumn{2}{c}{500857} & \multicolumn{2}{c}{492438} & \multicolumn{2}{c}{485244} & \multicolumn{2}{c}{315201} & \multicolumn{2}{c}{448530} \\
131 +\textbf{$R^2$} & \multicolumn{2}{c}{0.002881} & \multicolumn{2}{c}{0.003122} & \multicolumn{2}{c}{0.001578} & \multicolumn{2}{c}{0.000642} & \multicolumn{2}{c}{0.000452} & \multicolumn{2}{c}{0.001913} \\
132 +\bottomrule
133 +\end{tabular}
134 +}
135 +\begin{tablenotes}
136 +\singlespacing
137 +\footnotesize
138 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and the financialization variable $MM_NLS_t$, calculated with the money manager positions. The period covered is from 2008-12-22 to 2015-12-21. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred and especially if that announcement was unanticipated. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of financialization.
139 +\end{tablenotes}
140 +\end{sidewaystable}
141 +
142 +
143 +
144 +\begin{sidewaystable}
145 +\caption{Effects of Macro Announcements and Financialization (Swap Dealers only) on Futures Returns - ZLB Sample Period}
146 +\label{tab:macro_fin_sd_fut_returns_zlb}
147 +\centering
148 +\resizebox{\linewidth}{!}{%
149 +\begin{tabular}{@{}lllllllllllll@{}}
150 +\toprule
151 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
152 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule
153 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
154 +\textbf{Initial jobless claims} & -0.32056*** & -1.72794*** & 0.38264*** & 0.42661 & -0.40779*** & -1.27055 & 0.13737 & -0.87888 & 0.14847** & 0.45145 & -0.07010 & -0.10390 \\
155 +\textbf{ADP Employment} & 0.69846*** & 3.71649*** & -0.85488*** & -1.58718 & 0.85300*** & 0.48351 & -0.99839 & 4.17881* & -0.16141 & 1.79936* & 0.90271* & -1.93898 \\
156 +\textbf{CB Consumer} & 0.10301*** & 0.87977*** & -0.01011 & 0.10979 & -0.01216 & -0.10324 & -0.14402* & 0.57572** & -0.02866* & -0.14917 & 0.11438 & -0.43482 \\
157 +\textbf{Advance retail sales} & 0.24843*** & 1.51216*** & -0.28012*** & -0.97235*** & 0.13895*** & -0.56864 & -0.63271*** & -2.12932*** & -0.07521** & -0.39294 & 0.46335*** & -2.78721*** \\
158 +\textbf{Building permit} & 0.02936 & 0.18263 & -0.02238** & 0.03441 & 0.03317** & 0.25235 & -0.22253*** & -0.75451*** & -0.03283*** & 0.00983 & 0.01802 & -0.28407 \\
159 +\textbf{Construction spending} & 0.00556 & 0.30199** & 0.01642 & 0.13233 & 0.01630 & 0.12278 & -0.29005*** & -0.96048*** & 0.03267 & 0.13762 & -0.01452 & -0.12698 \\
160 +\textbf{Consumer credit} & -0.00969 & -0.01432 & -0.00832 & -0.07044 & 0.01001 & -0.11920 & 0.02716 & -0.07626 & 0.00857 & 0.07192 & 0.04016 & -0.17524 \\
161 +\textbf{Consumer price index} & -0.04760*** & -0.23974** & 0.02015 & 0.31684 & -0.01748 & -0.44052* & -0.11054 & 0.39000 & -0.03451** & 0.05952 & 0.05187 & -0.23282 \\
162 +\textbf{Durable goods orders} & 0.11917*** & 0.59693*** & -0.05002*** & -0.26004*** & 0.00327 & 0.25498* & -0.08716** & -0.20937* & 0.09130*** & 0.25771*** & 0.00296 & 0.12176 \\
163 +\textbf{Existing home sales} & 0.05460*** & 0.01145 & -0.02656 & -0.13171 & -0.00135 & 0.15546 & -0.07654 & 0.40998* & -0.01008 & -0.08721 & 0.15042* & -0.85454* \\
164 +\textbf{Factory orders} & -0.00835 & -0.14771 & -0.04126* & -0.14039 & 0.02815* & -0.11486 & -0.06044 & 0.19988 & -0.01459 & -0.03536 & 0.07999 & -0.53160 \\
165 +\textbf{Gross domestic product} & 0.06167*** & 0.31740*** & -0.04749** & 0.37778*** & 0.14238*** & 0.34776* & -0.03946 & 0.23063 & -0.07962*** & -0.16292 & -0.03689 & 0.44542 \\
166 +\textbf{Housing starts} & 0.09313*** & 0.48485*** & -0.09264*** & -0.16993 & 0.09609*** & -0.25287 & -0.02460 & 0.15814 & -0.03424** & -0.19375** & -0.01233 & 0.00095 \\
167 +\textbf{Industrial production} & 0.07388*** & 0.42044** & 0.00084 & 0.14587 & 0.03551* & 0.23188 & -0.00288 & 0.03430 & -0.03194 & -0.07516 & 0.07229 & -0.28270 \\
168 +\textbf{New home sales} & 0.09002*** & 0.44011*** & -0.03834 & 0.11355 & 0.04672** & 0.06064 & 0.01671 & 0.17876 & -0.00305 & 0.01911 & 0.09846 & -0.50693 \\
169 +\textbf{Non-farm employment} & 3.88185*** & 13.80768*** & -4.82381*** & -15.84916*** & 2.10300*** & 8.70416** & -0.61228 & 6.83749*** & 0.35630 & 2.69980 & 0.22581 & 5.96351 \\
170 +\textbf{Pending home sales} & 0.05149*** & 0.45606*** & 0.00041 & 0.05162 & 0.01186 & 0.51632** & -0.00387 & 0.12369 & -0.00455 & -0.07283 & 0.08726 & -0.27124 \\
171 +\textbf{Personal consumption} & 0.03372 & -0.02065 & 0.00836 & 0.12445 & 0.00736 & 0.54898* & 0.10110 & -0.27521 & -0.01750 & -0.00691 & -0.05347 & 0.27416 \\
172 +\textbf{Personal income} & -0.05994 & -0.50016 & -0.02996 & -0.16180 & 0.01430 & 2.79718** & -0.10297 & 0.41100 & -0.05688 & -0.65726 & -0.46663 & 3.32969 \\
173 +\textbf{Producer price index} & 0.02268* & 0.11458 & -0.00540 & 0.08145 & -0.01398 & 0.34543** & -0.00170 & 0.03633 & 0.01875 & -0.08175 & -0.06659 & 0.30030 \\
174 +\textbf{Trade balance} & -0.02017 & -0.09883 & -0.00403 & 0.14655 & -0.00768 & -0.21584 & -0.06485 & 0.18213 & 0.01558 & 0.10698 & 0.01481 & 0.04146 \\
175 +\textbf{Crude Oil Weekly inventory} & -0.25063*** & -0.60613*** & & & & & & & & & & \\
176 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -1.20271*** & 3.39611*** \\
177 +\midrule
178 +\textbf{Observations} & \multicolumn{2}{c}{495652} & \multicolumn{2}{c}{500857} & \multicolumn{2}{c}{492438} & \multicolumn{2}{c}{485244} & \multicolumn{2}{c}{315201} & \multicolumn{2}{c}{448530} \\
179 +\textbf{$R^2$} & \multicolumn{2}{c}{0.002955} & \multicolumn{2}{c}{0.002525} & \multicolumn{2}{c}{0.001046} & \multicolumn{2}{c}{0.000681} & \multicolumn{2}{c}{0.000478} & \multicolumn{2}{c}{0.002085} \\
180 +\bottomrule
181 +\end{tabular}
182 +}
183 +\begin{tablenotes}
184 +\singlespacing
185 +\footnotesize
186 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and the financialization variable $SWAP_NLS_t$, calculated with the swap dealer positions. The period covered is from 2008-12-22 to 2015-12-21. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred and especially if that announcement was unanticipated. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of financialization.
187 +\end{tablenotes}
188 +\end{sidewaystable}
189 +
190 +
191 +
192 +\begin{sidewaystable}
193 +\caption{Effects of Macro Announcements and Financialization (Money Managers only) on Futures Conditional Variance - ZLB Sample Period}
194 +\label{tab:macro_fin_mm_var_zlb}
195 +\centering
196 +\resizebox{\linewidth}{!}{%
197 +\begin{tabular}{@{}lllllllllllll@{}}
198 +\toprule
199 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
200 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule
201 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
202 +\textbf{Initial jobless claims} & 0.11042*** & -0.51631*** & 0.05263*** & -0.00595 & 0.08731*** & -0.12781** & 0.03485*** & -0.06463* & 0.17234*** & -0.20324*** & -0.02869*** & -0.13137 \\
203 +\textbf{ADP Employment} & 0.06053** & -0.52779** & 0.07769*** & -0.17422*** & 0.06459*** & -0.31473*** & 0.03757*** & -0.03711 & 0.00838 & 0.00886 & 0.01206 & -0.00492 \\
204 +\textbf{CB Consumer} & 0.20124*** & -1.34580*** & 0.05019*** & -0.06305 & 0.05986*** & 0.02708 & 0.05096*** & -0.13200* & 0.09902** & -0.14213 & -0.00238 & -0.01103 \\
205 +\textbf{Advance retail sales} & 0.20892*** & -1.16165*** & 0.12186*** & -0.22994*** & 0.12102*** & -0.28523** & 0.06519*** & -0.25509*** & 0.06152 & -0.00004 & -0.00614 & -0.23023 \\
206 +\textbf{Building permit} & 0.10335*** & -0.67730** & 0.08754*** & -0.16183*** & 0.10387*** & -0.38658*** & 0.03230*** & -0.08438 & 0.05805 & 0.00050 & -0.01453 & -0.22983 \\
207 +\textbf{Construction spending} & 0.30662*** & -2.00920*** & 0.10043*** & -0.17977*** & 0.09162*** & -0.24762** & 0.07129*** & -0.15578** & 0.03585 & 0.02620 & 0.01936 & -0.11956 \\
208 +\textbf{Consumer credit} & 0.06781** & -0.45377* & 0.01954 & -0.04524 & 0.02045 & -0.00112 & 0.00041 & 0.13979** & -0.07108 & 0.14854 & 0.00592 & 0.03352 \\
209 +\textbf{Consumer price index} & 0.03264 & -0.30328 & 0.10382*** & -0.19722*** & 0.09979*** & -0.21962 & 0.01185 & -0.02195 & 0.02337 & 0.03283 & -0.02505 & -0.27462 \\
210 +\textbf{Durable goods orders} & 0.16495*** & -0.98185*** & 0.05274*** & -0.02911 & 0.03975** & 0.19593 & 0.03304*** & -0.11622 & 0.11101** & -0.07750 & 0.02164 & 0.07533 \\
211 +\textbf{Existing home sales} & 0.06024** & -0.37169 & 0.06853*** & -0.09569** & 0.07145*** & -0.09328 & 0.03615*** & -0.19790*** & 0.02413 & -0.04280 & 0.03945* & 0.08293 \\
212 +\textbf{Factory orders} & 0.08018*** & -0.53355** & 0.00713 & 0.07431 & -0.00389 & 0.18423 & 0.02073** & 0.01816 & 0.10843** & -0.18796* & 0.00491 & -0.51457*** \\
213 +\textbf{Gross domestic product} & 0.19148*** & -1.31834*** & 0.08641*** & -0.09203* & 0.13275*** & -0.23199* & 0.05266*** & -0.04615 & 0.24948*** & -0.39107*** & -0.04221* & -0.10882 \\
214 +\textbf{Housing starts} & 0.07005** & -0.38038 & 0.05814*** & -0.07651 & 0.07579*** & -0.22709* & 0.03668*** & -0.11181 & 0.04081 & 0.03840 & -0.01655 & -0.17388 \\
215 +\textbf{Industrial production} & 0.07962** & -0.78674*** & 0.01453 & 0.01179 & 0.04212** & -0.22839* & 0.00789 & -0.05580 & 0.07086 & -0.18734* & -0.04399** & -0.34835** \\
216 +\textbf{New home sales} & 0.13000*** & -0.72125*** & 0.05974*** & -0.01109 & 0.06094*** & -0.00298 & 0.04619*** & -0.06623 & 0.27489*** & -0.48294*** & -0.00206 & -0.15513 \\
217 +\textbf{Non-farm employment} & 0.41187*** & -1.49559*** & 0.38777*** & -0.47071*** & 0.46246*** & -0.70966*** & 0.17870*** & 0.07823 & 0.13212*** & 0.09361 & 0.07222*** & -0.36383** \\
218 +\textbf{Pending home sales} & 0.16519*** & -1.17845*** & 0.02729** & -0.06490 & -0.00125 & 0.17222 & 0.04910*** & -0.05658 & 0.04843 & -0.08794 & -0.02888 & -0.72767*** \\
219 +\textbf{Personal consumption} & 0.00594 & -0.09037 & 0.06558*** & -0.17108*** & 0.07976*** & -0.31621** & -0.01151 & 0.07871 & 0.16883*** & -0.22576* & 0.01929 & -0.26873 \\
220 +\textbf{Personal income} & 0.03350 & -0.15354 & 0.15047*** & -0.37937*** & 0.17321*** & -0.73187*** & 0.01627* & -0.11945 & 0.17525*** & -0.24260** & -0.00037 & -0.36836** \\
221 +\textbf{Producer price index} & 0.17157*** & -1.00764*** & 0.07182*** & -0.10851** & 0.12842*** & -0.53013*** & 0.02160** & 0.04167 & 0.28218*** & -0.29799*** & 0.03465 & 0.10017 \\
222 +\textbf{Trade balance} & 0.11056*** & -0.48706** & 0.06293*** & -0.08213* & 0.03800** & 0.04965 & 0.01111 & -0.02654 & 0.02059 & 0.04861 & -0.02653 & -0.17166 \\
223 +\textbf{Crude Oil Weekly inventory} & 0.10301*** & 0.10001 & & & & & & & & & & \\
224 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.80936*** & 0.99041*** \\
225 +\midrule
226 +\textbf{Observations} & \multicolumn{2}{c}{495652} & \multicolumn{2}{c}{500857} & \multicolumn{2}{c}{492438} & \multicolumn{2}{c}{485244} & \multicolumn{2}{c}{315201} & \multicolumn{2}{c}{448530} \\
227 +\textbf{$R^2$} & \multicolumn{2}{c}{0.16859} & \multicolumn{2}{c}{0.085873} & \multicolumn{2}{c}{0.071917} & \multicolumn{2}{c}{0.072993} & \multicolumn{2}{c}{0.05873} & \multicolumn{2}{c}{0.162315} \\
228 +\bottomrule
229 +\end{tabular}
230 +}
231 +\begin{tablenotes}
232 +\singlespacing
233 +\footnotesize
234 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the financialization variable $MM_NLS_t$, calculated with the money manager positions. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2008-12-22 to 2015-12-21. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of financialization.
235 +\end{tablenotes}
236 +\end{sidewaystable}
237 +
238 +
239 +\begin{sidewaystable}
240 +\caption{Effects of Macro Announcements and Financialization (Swap Dealers only) on Futures Conditional Variance - ZLB Sample Period}
241 +\label{tab:macro_fin_sd_var_zlb}
242 +\centering
243 +\resizebox{\linewidth}{!}{%
244 +\begin{tabular}{@{}lllllllllllll@{}}
245 +\toprule
246 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
247 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule
248 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
249 +\textbf{Initial jobless claims} & 0.06549*** & 0.23925*** & 0.03655*** & -0.12678*** & 0.06619*** & 0.16700** & -0.08182*** & 0.37478*** & 0.09328*** & 0.22900*** & 0.00441 & -0.14759 \\
250 +\textbf{ADP Employment} & 0.01409 & 0.23268*** & 0.05449*** & 0.17291** & 0.01648 & 0.17252 & 0.09995** & 0.43807*** & 0.00677 & -0.08376 & 0.01720 & -0.03058 \\
251 +\textbf{CB Consumer} & 0.07658*** & 0.44580*** & 0.05110*** & 0.14388* & 0.06286*** & 0.03792 & 0.11840** & 0.53857*** & 0.03987*** & 0.10670 & 0.01192 & -0.08440 \\
252 +\textbf{Advance retail sales} & 0.10253*** & 0.41360*** & 0.06959*** & 0.03131 & 0.07934*** & 0.12412 & 0.16067*** & 0.71939*** & 0.06466*** & 0.05069 & 0.00636 & 0.03670 \\
253 +\textbf{Building permit} & 0.04852*** & 0.34467*** & 0.05713*** & 0.07698 & 0.04958*** & 0.14126 & 0.16538*** & 0.65673*** & 0.05851*** & 0.00501 & -0.01757 & 0.12566 \\
254 +\textbf{Construction spending} & 0.11014*** & 0.54991*** & 0.09045*** & 0.30367*** & 0.04787*** & 0.38803*** & 0.17841*** & 0.78982*** & 0.05540*** & 0.11190 & 0.16948*** & -0.89328*** \\
255 +\textbf{Consumer credit} & 0.02295** & 0.10146 & 0.02228** & 0.11800 & 0.02205** & -0.07699 & 0.09392* & 0.29470* & -0.01962 & -0.24147* & 0.01390 & -0.06912 \\
256 +\textbf{Consumer price index} & 0.00803 & 0.17941 & 0.06524*** & 0.10223 & 0.08070*** & 0.35766* & -0.05375 & 0.21154 & 0.03108* & -0.11694 & 0.01028 & -0.08624 \\
257 +\textbf{Durable goods orders} & 0.07999*** & 0.42086*** & 0.03571*** & -0.08651 & 0.06175*** & 0.14099 & -0.00805 & 0.12492 & 0.08454*** & 0.14655 & 0.01360 & 0.01322 \\
258 +\textbf{Existing home sales} & 0.01941* & 0.00449 & 0.06046*** & 0.13692* & 0.05950*** & -0.05527 & -0.09607** & 0.42236*** & 0.00571 & 0.02091 & 0.09926** & -0.42834 \\
259 +\textbf{Factory orders} & 0.03087*** & 0.21435** & 0.03428*** & 0.06602 & 0.02560** & -0.14294 & 0.11869*** & 0.45413*** & 0.04081*** & 0.31705*** & -0.04324 & 0.59690** \\
260 +\textbf{Gross domestic product} & 0.06672*** & 0.47858*** & 0.04323*** & -0.16305* & 0.09343*** & 0.38106*** & -0.00081 & 0.16193 & 0.09609*** & 0.43490*** & -0.02224 & -0.06610 \\
261 +\textbf{Housing starts} & 0.04340*** & 0.28794*** & 0.03510*** & -0.03896 & 0.04387*** & 0.05641 & -0.12478*** & 0.53005*** & 0.05811*** & -0.00276 & -0.03612 & 0.21804 \\
262 +\textbf{Industrial production} & 0.01151 & 0.33085*** & 0.02356** & 0.05644 & 0.00956 & 0.08248 & 0.02927 & -0.07343 & -0.00602 & 0.11675 & 0.00339 & -0.13793 \\
263 +\textbf{New home sales} & 0.06516*** & 0.30707*** & 0.09172*** & 0.30519*** & 0.05405*** & 0.28987** & 0.00163 & 0.14223 & 0.08315*** & 0.44694*** & 0.00018 & 0.06456 \\
264 +\textbf{Non-farm employment} & 0.28112*** & 0.64937*** & 0.27301*** & 0.02822 & 0.32165*** & 2.05556*** & 0.10034** & 0.25963** & 0.20289*** & 0.43864*** & 0.13651*** & -0.23850 \\
265 +\textbf{Pending home sales} & 0.05843*** & 0.46806*** & 0.03808*** & 0.22974*** & 0.02312** & 0.01365 & 0.15574*** & 0.65543*** & 0.01164 & 0.05860 & 0.03240 & -0.01779 \\
266 +\textbf{Personal consumption} & -0.00771 & -0.05499 & 0.06986*** & 0.40150*** & 0.02991** & 0.30442* & 0.02375 & -0.11179 & 0.08672*** & 0.33367*** & -0.03480 & 0.49927* \\
267 +\textbf{Personal income} & 0.01155 & -0.08770 & 0.15783*** & 0.89515*** & 0.05059*** & 0.66424*** & 0.09887* & 0.27659* & 0.07954*** & 0.29818** & -0.09135* & 0.79020*** \\
268 +\textbf{Producer price index} & 0.07622*** & 0.29168*** & 0.05587*** & 0.10034 & 0.04884*** & 0.21410 & -0.02427 & 0.15638 & 0.16635*** & 0.31874*** & 0.12565*** & -0.65777** \\
269 +\textbf{Trade balance} & 0.06486*** & 0.08942 & 0.00256 & -0.36379*** & 0.05518*** & 0.56239*** & -0.01684 & 0.09220 & 0.04085*** & -0.03389 & -0.02124 & 0.04951 \\
270 +\textbf{Crude Oil Weekly inventory} & 0.27209*** & 1.11802*** & & & & & & & & & & \\
271 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.91556*** & -1.06191*** \\
272 +\midrule
273 +\textbf{Observations} & \multicolumn{2}{c}{495652} & \multicolumn{2}{c}{500857} & \multicolumn{2}{c}{492438} & \multicolumn{2}{c}{485244} & \multicolumn{2}{c}{315201} & \multicolumn{2}{c}{448530} \\
274 +\textbf{$R^2$} & \multicolumn{2}{c}{0.164857} & \multicolumn{2}{c}{0.084522} & \multicolumn{2}{c}{0.080360} & \multicolumn{2}{c}{0.081774} & \multicolumn{2}{c}{0.060560} & \multicolumn{2}{c}{0.159830} \\
275 +\bottomrule
276 +\end{tabular}
277 +}
278 +\begin{tablenotes}
279 +\singlespacing
280 +\footnotesize
281 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the financialization variable $SWAP_NLS_t$, calculated with the swap dealer positions. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2008-12-22 to 2015-12-21. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of financialization.
282 +\end{tablenotes}
283 +\end{sidewaystable}
284 +
285 +
286 +
287 +
288 +
289 +\begin{sidewaystable}
290 +\caption{Effects of Macro Announcements and Financialization on Futures Bid-Ask Spreads - ZLB Sample Period}
291 +\label{tab:macro_fin_fut_spread_zlb}
292 +\centering
293 +\resizebox{\linewidth}{!}{%
294 +\begin{tabular}{@{}lllllllllllll@{}}
295 +\toprule
296 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
297 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule
298 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
299 +\textbf{Initial jobless claims} & -16.00333*** & -51.10653* & -1.73127 & -12.56477 & 1.85348 & -15.59903 & -5.64733** & -33.84972** & -32.4102* & 48.40736 & 5.70478 & 42.48549 \\
300 +\textbf{ADP Employment} & -16.35851* & 85.82022 & -16.78598 & 50.52782 & -5.37368 & 30.51418 & 0.03847 & 3.65756 & -24.67035 & 61.81186 & -5.12503 & 5.96857 \\
301 +\textbf{CB Consumer} & -1.23711 & -69.2872 & -14.06577 & 0.64039 & -0.87123 & -52.32945 & -4.03997 & 18.31622 & -51.39933 & 115.90883 & -0.40159 & 94.69497 \\
302 +\textbf{Advance retail sales} & 1.40018 & -45.6879 & -13.77611 & 35.45859 & 3.27477 & -71.30641* & 4.89159 & 56.34027 & -30.45098 & 39.79882 & -0.23822 & -8.98071 \\
303 +\textbf{Building permit} & 1.0965 & -42.88952 & -0.0073 & -29.39759 & -2.79939 & -5.92466 & 2.84472 & 16.29514 & 16.96645 & -77.52439 & 4.78917 & 36.56556 \\
304 +\textbf{Construction spending} & -8.1099 & -31.87598 & -1.89998 & -36.67751 & -3.10591 & -33.29713 & -1.16889 & -6.621 & 35.00149 & -38.30712 & -12.96835 & -3.34615 \\
305 +\textbf{Consumer credit} & -2.58426 & -14.29559 & -10.61788 & 10.6241 & -3.40728 & 0.15245 & 1.11525 & 2.04455 & 31.20187 & -24.57443 & -8.41912 & -26.36806 \\
306 +\textbf{Consumer price index} & -5.30815 & -2.13596 & 1.81553 & -36.01225 & 3.80651 & -73.91784 & -3.18302 & -23.15434 & -53.65038 & 69.63912 & 14.61196 & 122.18279 \\
307 +\textbf{Durable goods orders} & -5.0545 & -16.78362 & 7.73155 & -27.70292 & 0.12963 & -33.61572 & -12.15565*** & -73.07976** & -14.32433 & 28.10486 & 24.58134 & 167.9539* \\
308 +\textbf{Existing home sales} & -41.33384*** & -177.251*** & 1.47027 & -38.49211 & 1.6915 & -50.53013 & -7.88794* & -25.39868 & -20.46618 & 13.35423 & -2.40443 & 63.14923 \\
309 +\textbf{Factory orders} & -12.09431 & 47.58704 & -24.32356** & 47.49917 & -4.94705 & -1.58757 & -9.77857** & -66.24584* & -21.516 & 74.87304 & 11.00565 & 93.9416 \\
310 +\textbf{Gross domestic product} & -0.59704 & -48.89464 & 1.80948 & -34.2578 & -3.86269 & -10.78592 & -8.14099* & -50.41495 & 54.1126 & -83.66291 & 4.47733 & 49.12674 \\
311 +\textbf{Housing starts} & 2.37208 & -58.50889 & -2.18171 & -26.1717 & 1.79126 & -27.81315 & 4.17817 & 23.10012 & 25.26078 & -95.61864 & 5.16356 & 51.09392 \\
312 +\textbf{Industrial production} & 7.91328 & -85.82509 & 6.1417 & 13.16061 & 1.64513 & -5.01436 & 0.31643 & -25.28908 & 25.85394 & -48.9534 & -5.10113 & 3.66837 \\
313 +\textbf{New home sales} & -12.50029 & 5.15278 & -15.74503 & 26.09201 & 0.73836 & -60.33626* & -8.0408* & -53.98368 & -18.52912 & 19.2442 & -22.81888** & -55.61635 \\
314 +\textbf{Non-farm employment} & -23.73484** & 101.78191 & 10.92711 & -67.74338* & -1.21567 & -68.55908* & -2.87725 & 0.98461 & 39.44131 & -120.72932* & -1.08631 & 37.70346 \\
315 +\textbf{Pending home sales} & -12.36887 & 46.59198 & -8.89564 & -5.03952 & -6.91436 & 8.49183 & -0.90003 & -1.19892 & 58.60186 & -83.49949 & -17.93609 & -40.79346 \\
316 +\textbf{Personal consumption} & -3.16096 & -45.86163 & -42.77794*** & -94.24372** & 2.38111 & -44.88402 & -4.99405 & -28.83845 & 36.92665 & -93.51525 & 15.47596 & 99.25116 \\
317 +\textbf{Personal income} & -17.44851 & 9.28027 & -5.8513 & -6.0265 & -0.74131 & -32.70676 & -6.09333 & -8.80177 & 72.22104 & -168.1134* & 1.77599 & 43.97083 \\
318 +\textbf{Producer price index} & -4.18185 & -3.40316 & 4.39087 & -59.49314 & 2.38309 & -49.41511 & 3.10567 & 13.27673 & 43.1821 & -96.27467 & -1.31215 & -8.00578 \\
319 +\textbf{Trade balance} & -2.89459 & 0.12539 & 16.28516 & -68.83361* & -12.28214** & -73.47717** & -5.13375 & -68.19869* & -34.95634 & 52.70414 & 19.49949 & 108.20832 \\
320 +\textbf{Crude Oil Weekly inventory} & -3.78018 & 2.94781 & & & & & & & & & & \\
321 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -15.92054*** & 8.01229 \\
322 +\midrule
323 +\textbf{Observations} & \multicolumn{2}{c}{442484} & \multicolumn{2}{c}{423623} & \multicolumn{2}{c}{425419} & \multicolumn{2}{c}{423604} & \multicolumn{2}{c}{263717} & \multicolumn{2}{c}{402138} \\
324 +\textbf{$R^2$} & \multicolumn{2}{c}{0.00217} & \multicolumn{2}{c}{0.004576} & \multicolumn{2}{c}{0.003518} & \multicolumn{2}{c}{0.006533} & \multicolumn{2}{c}{0.000792} & \multicolumn{2}{c}{0.001498} \\
325 +\bottomrule
326 +\end{tabular}
327 +}
328 +\begin{tablenotes}
329 + \singlespacing
330 + \footnotesize
331 + This table presents estimates of the equation $R_{\text{SPREAD}}_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{\text{SPREAD}}_{t-\tau}^{t}+\epsilon_{t}$, analyzing the effects of financialization and macroeconomic announcements on the bid-ask spread using the financialization variable $PCA_t$. The period covered is from 2008-12-22 to 2015-12-21. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the financialization variable $PCA_t$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and financialization. The $\beta$ coefficient represents the effect of the lagged bid-ask spread.
332 +\end{tablenotes}
333 +\end{sidewaystable}
334 +
335 +
336 +
337 +\begin{sidewaystable}
338 +\caption{Effects of Macro Announcements and Financialization (Money Managers only) on Futures Bid-Ask Spreads - ZLB Sample Period}
339 +\label{tab:macro_fin_mm_fut_spread_zlb}
340 +\centering
341 +\resizebox{\linewidth}{!}{%
342 +\begin{tabular}{@{}lllllllllllll@{}}
343 +\toprule
344 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
345 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule \textbf{Initial jobless claims} & -20.68427*** & -105.00327** & -2.37886 & -13.88928 & 0.85187 & -15.81395 & -3.84721** & -25.38919** & -29.32423* & 47.61619 & 0.23161 & 14.08545 \\
346 +\textbf{ADP Employment} & -10.70784 & 61.35073 & -7.9992 & 27.61237 & -1.95043 & 14.46688 & -0.4924 & -5.75308 & -37.17904 & 96.00867 & -6.46641 & -6.80614 \\
347 +\textbf{CB Consumer} & 8.24743 & -172.08814** & -13.0998 & -3.37821 & -3.5525 & -53.8779** & -6.70922* & 7.23005 & -54.23384 & -136.78321* & -14.0705** & 17.40477 \\
348 +\textbf{Advance retail sales} & 4.64624 & -89.95338 & -11.26009 & 36.68171 & -2.43298 & -53.85158 & 1.42316 & 30.63149 & -29.36105 & 42.45732 & 1.12961 & 1.01811 \\
349 +\textbf{Building permit} & -2.89125 & -17.897 & -3.43736 & -22.37716 & -1.63887 & -21.8388 & 2.48143 & 18.48554 & 14.75744 & -82.69896 & -1.05121 & -3.58565 \\
350 +\textbf{Construction spending} & 3.29696 & -134.29107 & -3.84404 & -39.90423 & -3.09324 & -49.3878* & -0.87022 & 1.11585 & 35.44588 & -43.58201 & -13.8704** & -21.05458 \\
351 +\textbf{Consumer credit} & -11.22279 & 49.53703 & -8.56694 & 5.44352 & -1.76439 & -15.71136 & 1.39558 & -19.33393 & 28.63734 & -22.10787 & -9.13929** & -83.23667** \\
352 +\textbf{Consumer price index} & -13.93216 & 76.50589 & -0.21954 & -39.20841 & -0.10717 & -75.22832 & -1.62078 & -11.4688 & -42.60754 & 56.08179 & 1.209 & 71.71343 \\
353 +\textbf{Durable goods orders} & -7.54133 & 3.55354 & -2.83085 & 12.56572 & -1.73257 & -31.56256 & -7.59164** & -41.68739 & -25.03093 & 53.77892 & 7.50562 & -156.29075** \\
354 +\textbf{Existing home sales} & -3.34707 & -81.4449 & -3.20009 & -29.63337 & 0.14226 & -61.46683 & -5.86357* & -18.99988 & -27.87553 & 30.56141 & -10.1933* & 29.23645 \\
355 +\textbf{Factory orders} & -6.16365 & 17.32965 & -13.98287 & 16.73252 & -4.13087 & -9.12439 & -5.97947 & -44.06619 & -39.12619 & 124.28877 & 1.88641 & 69.96703 \\
356 +\textbf{Gross domestic product} & -13.76375 & 46.47701 & -0.30476 & -35.97655 & -3.04203 & -21.34015 & -6.41353* & -52.79619** & 64.36866 & -115.3362 & -2.09212 & 20.63409 \\
357 +\textbf{Housing starts} & -0.80429 & -42.1111 & -5.76125 & -16.85222 & 2.23226 & -54.12491 & 3.20416 & 23.16712 & 19.64747 & -95.49455 & -2.45369 & 1.93243 \\
358 +\textbf{Industrial production} & 3.31097 & -68.28671 & 1.25975 & 42.47721 & 0.19297 & 10.95381 & 2.18291 & -9.1981 & 29.09147 & -61.27499 & -6.44157 & -10.15044 \\
359 +\textbf{New home sales} & 1.73875 & -111.66443 & -12.67987* & 19.533 & -2.59729 & -59.86757** & -5.65855* & -62.00041* & -25.48035 & 37.20469 & -16.79411*** & -41.26684 \\
360 +\textbf{Non-farm employment} & -6.05015 & -8.55708 & -0.50504 & -34.62659 & -4.55827 & -71.49574** & -3.2698 & -37.44957 & 20.57392 & -94.2021 & -6.84786 & -6.63895 \\
361 +\textbf{Pending home sales} & 0.18598 & -44.49616 & -7.31374 & -14.32549 & -4.21469 & -9.86129 & -1.38583 & 3.41728 & -66.17146* & -109.8488 & -13.62246** & -27.54183 \\
362 +\textbf{Personal consumption} & -6.45243 & -19.81889 & -35.93527*** & -91.85734*** & -1.08025 & -36.35222 & -3.68456 & -24.59092 & 24.11914 & -77.60203 & 6.26835 & 90.18945 \\
363 +\textbf{Personal income} & -15.8152 & 4.34974 & -7.30566 & -1.51559 & -2.23903 & -34.31172 & -5.35807 & -2.78512 & 50.25133 & -143.72241 & -3.64513 & 20.74326 \\
364 +\textbf{Producer price index} & -1.3336 & -23.78791 & -4.20022 & -36.81842 & -1.21427 & -42.38487 & 2.1662 & -1.09006 & 27.0739 & -73.16995 & 1.83996 & 27.28103 \\
365 +\textbf{Trade balance} & 12.0524 & -124.28376 & 3.85427 & -35.0069 & 3.29726 & -32.51734 & -1.18248 & -50.97158* & -34.48335 & 57.20098 & 8.274 & 79.76658 \\
366 +\textbf{Crude Oil Weekly inventory}& 0.03916 & -27.99856 & & & & & & & & & & \\ \textbf{Natural Gas Weekly inventory}& & & & & & & & & & & -16.29304*** & 14.84561 \\ \midrule \textbf{Observations} & \multicolumn{2}{c}{442484} & \multicolumn{2}{c}{423623} & \multicolumn{2}{c}{425419} & \multicolumn{2}{c}{423604} & \multicolumn{2}{c}{263717} & \multicolumn{2}{c}{402138} \\ \textbf{$R^2$} & \multicolumn{2}{c}{0.002641} & \multicolumn{2}{c}{0.005741} & \multicolumn{2}{c}{0.006596} & \multicolumn{2}{c}{0.005428} & \multicolumn{2}{c}{0.000793} & \multicolumn{2}{c}{0.001472} \\ \bottomrule
367 +\end{tabular}
368 +}
369 +\begin{tablenotes}
370 + \singlespacing
371 + \footnotesize
372 + This table presents estimates of the equation $R_{\text{SPREAD}}_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{\text{SPREAD}}_{t-\tau}^{t}+\epsilon_{t}$, analyzing the effects of financialization and macroeconomic announcements on the bid-ask spread using the financialization variable $MM_NLS_t$, calculated with the money manager positions. The period covered is from 2008-12-22 to 2015-12-21. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the financialization variable $MM_NLS_t$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and financialization. The $\beta$ coefficient represents the effect of the lagged bid-ask spread.
373 +\end{tablenotes}
374 +\end{sidewaystable}
375 +
376 +
377 +\begin{sidewaystable}
378 +\caption{Effects of Macro Announcements and Financialization (Swap Dealers only) on Futures Bid-Ask Spreads - ZLB Sample Period}
379 +\label{tab:macro_fin_sd_fut_spread_zlb}
380 +\centering
381 +\resizebox{\linewidth}{!}{%
382 +\begin{tabular}{@{}lllllllllllll@{}}
383 +\toprule
384 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
385 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule \textbf{Initial jobless claims} & -9.70982*** & -15.54702 & -7.41405* & -16.99933 & -0.87426 & -12.05957 & -28.39027** & 79.51247* & -12.8026** & 78.8092** & 6.89901 & -50.86415 \\
386 +\textbf{ADP Employment} & -5.04102 & -19.02927 & 7.12669 & 62.36418 & -0.99439 & -48.34049 & -37.24995 & 117.49739 & -3.46974 & 166.90156** & -1.98372 & -28.54869 \\
387 +\textbf{CB Consumer} & -7.71462** & 38.80791 & -18.99452** & -44.64697 & -10.14746*** & 74.74238* & 1.32911 & -24.44071 & -0.8234 & 150.76645* & -5.04671 & -65.45442 \\
388 +\textbf{Advance retail sales} & -2.71514 & 29.49461 & -8.28581 & -31.79646 & -8.07342** & 49.69185 & 31.88235 & -108.56536 & -17.28082 & -105.20463 & -3.91055 & 33.44669 \\
389 +\textbf{Building permit} & -2.4287 & 30.95294 & -10.20073 & -19.85955 & -5.85395 & 93.58691* & 12.59912 & -35.79649 & -19.13448 & 59.08388 & 0.70383 & -10.01426 \\
390 +\textbf{Construction spending} & -10.28971*** & 28.67396 & -5.69355 & 58.78217 & -11.26693*** & 129.75726*** & 19.93967 & -74.54055 & 18.80882 & 66.45357 & -9.79129 & -16.70764 \\
391 +\textbf{Consumer credit} & -4.64729 & 2.7322 & -12.81252* & -39.6375 & -5.31435* & -50.33509 & -2.88824 & 16.66229 & 18.75699 & 13.70968 & -14.94415 & 69.68635 \\
392 +\textbf{Consumer price index} & -3.58355 & 20.93 & -17.80927 & -91.09082 & -7.40821 & -1.58168 & 19.04109 & -73.64789 & -19.25252 & -30.63022 & 8.67378 & -81.47668 \\
393 +\textbf{Durable goods orders} & -6.29166 & 13.71442 & -9.92567 & -84.27418 & -5.34766 & -26.81333 & -76.51265*** & 227.76389** & -0.84517 & -23.54654 & 19.76025 & -141.30414 \\
394 +\textbf{Existing home sales} & -21.96591*** & 93.33814*** & -11.92521 & -21.92123 & -6.6351* & -26.137 & -29.97805 & 80.20644 & -17.34199 & -63.01144 & -0.81818 & -74.29811 \\
395 +\textbf{Factory orders} & -5.37916 & -9.29571 & -15.11789 & -42.07155 & -6.45161 & 80.14459* & 66.45919*** & 252.39659*** & 12.77376 & -50.47565 & 13.08624 & -108.75549 \\
396 +\textbf{Gross domestic product} & -2.95996 & 55.29713* & -20.52585** & -94.82327 & -7.02991* & -52.11982 & -30.72292 & 80.88127 & 2.39599 & -40.31041 & 5.75672 & -57.21379 \\
397 +\textbf{Housing starts} & -2.83051 & 39.84226 & -12.8538 & -31.36574 & -4.50277 & -79.94489 & 3.45189 & 0.18174 & -18.02096 & 79.79165 & 3.6474 & -42.64445 \\
398 +\textbf{Industrial production} & -0.3606 & 50.91209 & 15.8752* & 53.38602 & 2.89946 & 103.28414* & -9.79112 & 43.35057 & 0.8095 & 2.70895 & -3.74984 & -12.27813 \\
399 +\textbf{New home sales} & -12.33853*** & -6.63989 & -9.29849 & -3.74621 & -9.385*** & -34.76467 & -20.88664 & 54.30035 & -11.73976 & -56.28 & -26.8981** & 81.549 \\
400 +\textbf{Non-farm employment} & -10.22689** & -26.44439 & -2.96963 & 40.02359 & -12.62304*** & 56.40337 & -48.90015*** & 144.63828** & -17.01717 & 77.99757 & -2.36315 & -28.38375 \\
401 +\textbf{Pending home sales} & -8.15393* & -37.19652 & -11.84533 & -15.34974 & -6.43336* & -50.29698 & 31.06753 & -112.98075 & 19.11827* & 70.14828 & -26.46345** & 94.41996 \\
402 +\textbf{Personal consumption} & -5.91986 & 57.1833* & -35.50488*** & 176.31043*** & -4.3936 & 12.90128 & -24.06962 & 60.49793 & -12.75231 & -12.28368 & 25.56888 & -163.91677 \\
403 +\textbf{Personal income} & -14.16262*** & 22.91791 & -15.21694 & -63.02097 & -6.14842 & -11.83513 & -12.65687 & 18.54016 & -13.20959 & 71.40471 & 7.32959 & -79.47057 \\
404 +\textbf{Producer price index} & -3.10533 & 18.07292 & -7.05299 & 35.47537 & -5.95269 & -17.13207 & 53.30866 & -174.17692 & -6.29783 & -12.09179 & -0.82314 & 5.00915 \\
405 +\textbf{Trade balance} & -3.33891 & -5.69184 & -2.22867 & 7.51202 & 1.15567 & 51.57873 & 51.27467** & 177.58139** & -17.16892 & -110.72695 & 20.11087 & -112.49114 \\
406 +\textbf{Crude Oil Weekly inventory}& -4.02655 & -5.05036 & & & & & & & & & & \\
407 +\textbf{Natural Gas Weekly inventory}& & & & & & & & & & & -13.95719** & 21.27953 \\ \midrule \textbf{Observations} &\multicolumn{2}{c}{442484} & \multicolumn{2}{c}{423623} & \multicolumn{2}{c}{425419} & \multicolumn{2}{c}{423604} & \multicolumn{2}{c}{263717} & \multicolumn{2}{c}{402138} \\ \textbf{$R^2$} &\multicolumn{2}{c}{0.001695} & \multicolumn{2}{c}{0.003782} & \multicolumn{2}{c}{0.004444} & \multicolumn{2}{c}{0.001106} & \multicolumn{2}{c}{0.000875} & \multicolumn{2}{c}{0.001482} \\ \bottomrule
408 +\end{tabular}
409 +}
410 +\begin{tablenotes}
411 + \singlespacing
412 + \footnotesize
413 + This table presents estimates of the equation $R_{\text{SPREAD}}_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{\text{SPREAD}}_{t-\tau}^{t}+\epsilon_{t}$, analyzing the effects of financialization and macroeconomic announcements on the bid-ask spread using the financialization variable $SWAP_NLS_t$, calculated with the swap dealer positions. The period covered is from 2008-12-22 to 2015-12-21. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the financialization variable $SWAP_NLS_t$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and financialization. The $\beta$ coefficient represents the effect of the lagged bid-ask spread.
414 +\end{tablenotes}
415 +\end{sidewaystable}
416 +
417 +
418 +
419 +\begin{sidewaystable}
420 +\caption{Effects of Macro Announcements and Financialization (NLS) on Futures Returns - ZLB Sample Period}
421 +\label{tab:macro_fin_nls_fut_returns_zlb}
422 +\centering
423 +\resizebox{\linewidth}{!}{%
424 +\begin{tabular}{@{}lllllllllllll@{}}
425 +\toprule
426 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
427 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule \textbf{Initial jobless claims} & -0.70429*** & 3.78529*** & 1.09087*** & -2.25841*** & -1.39832*** & 4.71356*** & 0.12351*** & 0.17344 & 0.12053 & -0.00437 & -0.0518 & 0.20917 \\
428 +\textbf{ADP Employment} & 1.39435*** & -7.12987*** & -3.84009*** & 8.90483*** & 1.98621*** & -5.40179*** & 0.24898** & -1.03637 & 0.04779 & -0.64322 & 1.06705* & 2.42274 \\
429 +\textbf{CB Consumer} & 0.25256*** & -1.62456*** & -0.04746** & 0.08402 & 0.06657** & -0.29594* & -0.03378*** & -0.06595 & -0.11484** & 0.19093* & 0.0321 & -0.07378 \\
430 +\textbf{Advance retail sales} & 0.50785*** & -2.82851*** & -0.65198*** & 1.35297*** & 0.51345*** & -1.65641*** & 0.00957 & 0.64197*** & -0.15308 & 0.2109 & 0.22209* & 0.79246 \\
431 +\textbf{Building permit} & 0.03023 & -0.13412 & -0.01199 & -0.05338 & -0.04015 & 0.00913 & -0.00318 & -0.03226 & -0.01608 & -0.03873 & -0.00869 & 0.09105 \\
432 +\textbf{Construction spending} & 0.03806 & -0.44559* & 0.04202 & -0.121 & 0.06492 & -0.22008 & -0.00302 & 0.37014*** & 0.09459 & -0.13789 & -0.11753* & -0.51691 \\
433 +\textbf{Consumer credit} & -0.01035 & 0.01027 & -0.03464** & 0.10232** & 0.02475* & -0.18162*** & 0.00614 & 0.10737 & 0.04239 & -0.0683 & 0.04942 & 0.2199 \\
434 +\textbf{Consumer price index} & -0.05367* & 0.22486 & -0.05491** & 0.14857* & -0.0711** & 0.2472 & -0.00286 & -0.27025** & 0.02297 & -0.1113 & 0.05088 & 0.20721 \\
435 +\textbf{Durable goods orders} & 0.22761*** & -1.12323*** & -0.1184*** & 0.34599*** & 0.10638*** & -0.58869*** & -0.02373*** & 0.15346 & 0.1488*** & -0.18161** & -0.03788 & -0.34522 \\
436 +\textbf{Existing home sales} & 0.06734*** & -0.1611 & -0.09425*** & 0.24307*** & -0.07379 & 0.34111* & -0.05257*** & -0.15687* & -0.05244 & 0.0895 & 0.094 & 0.47253 \\
437 +\textbf{Factory orders} & -0.02188 & 0.16758 & -0.06297* & 0.11496 & -0.07732* & 0.21643 & -0.00072 & -0.03245 & 0.00867 & -0.03692 & 0.08812 & 0.56325 \\
438 +\textbf{Gross domestic product} & 0.09399*** & -0.41028** & -0.19392*** & 0.29743*** & 0.33782*** & -1.01923*** & -0.04529*** & 0.16728 & -0.26445*** & 0.37966** & -0.01533 & -0.24002 \\
439 +\textbf{Housing starts} & 0.15596*** & -0.7602*** & -0.11085*** & 0.14525** & 0.1609*** & -0.39369** & -0.02807** & 0.1126 & -0.01944 & 0.01178 & -0.02819 & -0.12451 \\
440 +\textbf{Industrial production} & 0.12147** & -0.60138** & -0.04582 & 0.08981 & -0.08024* & 0.23464 & 0.00612 & -0.08287 & -0.14273* & 0.21769 & 0.029 & -0.01512 \\
441 +\textbf{New home sales} & 0.15372*** & -0.69244*** & -0.20306*** & 0.48779*** & 0.11643*** & -0.33441* & -0.0736*** & -0.00415 & -0.17056 & 0.28999 & 0.04644 & 0.13686 \\
442 +\textbf{Non-farm employment} & 6.45167*** & -25.2801*** & -8.95331*** & 18.07887*** & 7.96384*** & -27.72306*** & -1.57712*** & 0.35965 & -0.74642 & 1.70491 & -1.55898** & -17.77143*** \\
443 +\textbf{Pending home sales} & 0.14767*** & -0.90112*** & 0.00245 & -0.02759 & -0.01856 & 0.09937 & -0.03845*** & -0.20744* & -0.11909 & 0.21728 & -0.01862 & -0.31759 \\
444 +\textbf{Personal consumption} & 0.00221 & 0.19129 & -0.07321* & 0.19355* & 0.02273 & -0.12291 & 0.02338 & 0.18267 & -0.03361 & 0.02948 & -0.04455 & -0.20038 \\
445 +\textbf{Personal income} & -0.16673 & 1.15899 & 0.0385 & -0.14958 & -0.25816 & 1.09273 & 0.01267 & -0.53643 & -0.31864 & 0.56008 & -0.62609*** & -4.00139*** \\
446 +\textbf{Producer price index} & 0.07284*** & -0.42142*** & -0.06599*** & 0.14536** & 0.06633** & -0.24047* & 0.00701 & -0.08332 & 0.02887 & -0.00862 & -0.0945** & -0.41908 \\
447 +\textbf{Trade balance} & -0.00985 & -0.00227 & -0.04424 & 0.07583 & -0.02616 & 0.09349 & -0.0116 & 0.03755 & 0.07156 & -0.1219 & -0.02821 & -0.32178 \\
448 +\textbf{Crude Oil Weekly inventory} & -0.14108*** & -0.0825 & & & & & & & & & & \\
449 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.98452*** & -2.09043*** \\ \midrule
450 +\textbf{Observations} & \multicolumn{2}{c}{ 495652 } & \multicolumn{2}{c}{ 500857 } & \multicolumn{2}{c}{ 492438 } & \multicolumn{2}{c}{ 485244 } & \multicolumn{2}{c}{ 315201 } & \multicolumn{2}{c}{ 448530 } \\
451 +\textbf{$R^2$} & \multicolumn{2}{c}{ 0.003123 } & \multicolumn{2}{c}{ 0.003413 } & \multicolumn{2}{c}{ 0.001762 } & \multicolumn{2}{c}{ 0.000693 } & \multicolumn{2}{c}{ 0.000439 } & \multicolumn{2}{c}{ 0.002058 } \\
452 +\bottomrule
453 +\end{tabular}
454 +}
455 +\begin{tablenotes}
456 +\singlespacing
457 +\footnotesize
458 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and financialization variable $X_{t,2}=NLS_t$. The period covered is from 2008-12-22 to 2015-12-21. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred and especially if that announcement was unanticipated. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of financialization.
459 +\end{tablenotes}
460 +\end{sidewaystable}
461 +
462 +
463 +
464 +
465 +
466 +\begin{sidewaystable}
467 +\caption{Effects of Macro Announcements and Financialization (NLS) on Futures Conditional Variance - ZLB Sample Period}
468 +\label{tab:macro_fin_nls_var_zlb}
469 +\centering
470 +\resizebox{\linewidth}{!}{%
471 +\begin{tabular}{@{}lllllllllllll@{}}
472 +\toprule
473 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
474 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule \textbf{Initial jobless claims} & 0.10694*** & -0.42408*** & 0.05269*** & -0.00488 & 0.10043*** & -0.16325*** & 0.03223*** & -0.03834 & 0.16713*** & -0.16984*** & -0.02784 & -0.05767 \\
475 +\textbf{ADP Employment} & 0.06941*** & -0.53219*** & 0.10252*** & -0.2171*** & 0.07215*** & -0.272** & 0.03539*** & -0.05173 & 0.00482 & 0.01455 & -0.00141 & -0.08407 \\
476 +\textbf{CB Consumer} & 0.16537*** & -0.87541*** & 0.05819*** & -0.07511 & 0.04967** & 0.07337 & 0.0439*** & -0.14431** & 0.09397* & -0.1162 & -0.00335 & -0.01169 \\
477 +\textbf{Advance retail sales} & 0.18938*** & -0.86228*** & 0.13967*** & -0.23979*** & 0.13996*** & -0.30885*** & 0.05161*** & -0.25845*** & 0.06489 & -0.00631 & -0.04253 & -0.33072 \\
478 +\textbf{Building permit} & 0.09698*** & -0.52935*** & 0.12162*** & -0.24222*** & 0.14168*** & -0.4931*** & 0.02762*** & -0.11219 & 0.03091 & 0.05324 & -0.04738 & -0.30239 \\
479 +\textbf{Construction spending} & 0.2325*** & -1.18295*** & 0.15062*** & -0.30249*** & 0.12038*** & -0.33192*** & 0.066*** & -0.07152 & 0.02884 & 0.03642 & 0.03301 & 0.02156 \\
480 +\textbf{Consumer credit} & 0.0453** & -0.22174 & 0.02772* & -0.06164 & 0.03052 & -0.05207 & 0.00668 & 0.10916 & -0.07063 & 0.13114 & 0.01441 & 0.06766 \\
481 +\textbf{Consumer price index} & 0.04682 & -0.40441* & 0.14873*** & -0.30026*** & 0.11292*** & -0.23301 & 0.01046 & -0.05358 & 0.05319 & -0.0272 & -0.00815 & -0.03531 \\
482 +\textbf{Durable goods orders} & 0.14852*** & -0.70578*** & 0.04542*** & 0.00033 & 0.01576 & 0.27348** & 0.02838*** & -0.06921 & 0.1074** & -0.06132 & 0.01688 & 0.0068 \\
483 +\textbf{Existing home sales} & 0.02751 & -0.05921 & 0.06524*** & -0.06524 & 0.05876** & -0.00113 & 0.02689*** & -0.17408** & 0.00283 & 0.00325 & 0.04728 & 0.08737 \\
484 +\textbf{Factory orders} & 0.06571*** & -0.3652** & -0.00722 & 0.10438** & -0.02243 & 0.23381** & 0.02224** & 0.03656 & 0.10924** & -0.16832* & -0.06361 & -0.68095*** \\
485 +\textbf{Gross domestic product} & 0.16121*** & -0.91541*** & 0.11461*** & -0.16196*** & 0.15722*** & -0.29682** & 0.04996*** & -0.04091 & 0.26447*** & -0.37431*** & -0.06994 & -0.2124 \\
486 +\textbf{Housing starts} & 0.08035*** & -0.40519** & 0.08746*** & -0.15567*** & 0.10646*** & -0.32943*** & 0.0307*** & -0.14564** & 0.02419 & 0.06669 & -0.05308 & -0.30261 \\
487 +\textbf{Industrial production} & 0.0652*** & -0.56456*** & 0.0062 & 0.03609 & 0.03259 & -0.11575 & 0.00655 & -0.00002 & 0.0695 & -0.16238* & -0.04995 & -0.19915 \\
488 +\textbf{New home sales} & 0.11635*** & -0.51095*** & 0.07002*** & -0.04171 & 0.06764*** & -0.03879 & 0.04642*** & 0.04233 & 0.32761*** & -0.53084*** & -0.01708 & -0.16199 \\
489 +\textbf{Non-farm employment} & 0.41625*** & -1.34407*** & 0.42629*** & -0.49897*** & 0.5661*** & -1.06553*** & 0.18526*** & -0.15465** & 0.18194*** & -0.01315 & 0.02026 & -0.49134** \\
490 +\textbf{Pending home sales} & 0.14487*** & -0.8662*** & 0.03822** & -0.08605* & -0.00309 & 0.1383 & 0.04684*** & -0.03212 & 0.05059 & -0.0823 & -0.08788** & -0.70701*** \\
491 +\textbf{Personal consumption} & -0.02318 & 0.15519 & 0.05627*** & -0.109* & 0.09287*** & -0.29668** & -0.00731 & 0.10063 & 0.14937** & -0.16345 & -0.06411 & -0.60997** \\
492 +\textbf{Personal income} & -0.00477 & 0.16598 & 0.16396*** & -0.34493*** & 0.2105*** & -0.73824*** & 0.01295 & -0.02388 & 0.16505*** & -0.19673* & -0.09811** & -0.76453*** \\
493 +\textbf{Producer price index} & 0.12583*** & -0.51275*** & 0.04942*** & -0.01475 & 0.14451*** & -0.47612*** & 0.02331** & 0.02659 & 0.2776*** & -0.25324*** & 0.07495* & 0.29216 \\
494 +\textbf{Trade balance} & 0.08875*** & -0.24 & 0.04125*** & 0.00637 & 0.02838 & 0.08788 & 0.01139 & 0.03537 & 0.01691 & 0.05017 & -0.03151 & -0.11329 \\
495 +\textbf{Crude Oil Weekly inventory} & 0.0962*** & 0.15058 & & & & & & & & & & \\
496 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.9177*** & 1.06866*** \\ \midrule
497 +\textbf{Observations} & \multicolumn{2}{c}{ 495652 } & \multicolumn{2}{c}{ 500857 } & \multicolumn{2}{c}{ 492438 } & \multicolumn{2}{c}{ 485244 } & \multicolumn{2}{c}{ 315201 } & \multicolumn{2}{c}{ 448530 } \\
498 +\textbf{$R^2$} & \multicolumn{2}{c}{ 0.15738 } & \multicolumn{2}{c}{ 0.084362 } & \multicolumn{2}{c}{ 0.071708 } & \multicolumn{2}{c}{ 0.064643 } & \multicolumn{2}{c}{ 0.058782 } & \multicolumn{2}{c}{ 0.161297 } \\
499 +\bottomrule
500 +\end{tabular}
501 +}
502 +\begin{tablenotes}
503 +\singlespacing
504 +\footnotesize
505 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the financialization variable $NLS_t$. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2008-12-22 to 2015-12-21. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of financialization.
506 +\end{tablenotes}
507 +\end{sidewaystable}
508 +
added phd_chap1_20260731/figures.tex +49 −0
@@ -0,0 +1,49 @@
1 +
2 + \begin{figure}[h]
3 + \centering
4 + \includegraphics[scale=0.4]{pi_cl.png}}
5 + \caption{Cumulative Abnormal Return of Crude Oil
6 +60 Minutes Before and After Macroeconomic Announcements}
7 + \label{fig:cl}
8 + \end{figure}
9 +
10 + \begin{figure}[h]
11 + \centering
12 + \includegraphics[scale=0.4]{pi_gc.png}}
13 + \caption{Cumulative Abnormal Return of Gold
14 +60 Minutes Before and After Macroeconomic Announcements}
15 + \label{fig:gc}
16 + \end{figure}
17 +
18 + \begin{figure}[h]
19 + \centering
20 + \includegraphics[scale=0.4]{pi_hg.png}}
21 + \caption{Cumulative Abnormal Return of Copper
22 +60 Minutes Before and After Macroeconomic Announcements}
23 + \label{fig:hg}
24 + \end{figure}
25 +
26 + \begin{figure}[h]
27 + \centering
28 + \includegraphics[scale=0.4]{pi_si.png}}
29 + \caption{Cumulative Abnormal Return of Silver
30 +60 Minutes Before and After Macroeconomic Announcements}
31 + \label{fig:pi}
32 + \end{figure}
33 +
34 + \begin{figure}[h]
35 + \centering
36 + \includegraphics[scale=0.4]{pi_pa.png}}
37 + \caption{Cumulative Abnormal Return of Palladium
38 +60 Minutes Before and After Macroeconomic Announcements}
39 + \label{fig:pa}
40 + \end{figure}
41 +
42 + \begin{figure}[h]
43 + \centering
44 + \includegraphics[scale=0.4]{pi_ng.png}}
45 + \caption{Cumulative Abnormal Return of Natural Gas
46 +60 Minutes Before and After Macroeconomic Announcements}
47 + \label{fig:ng}
48 + \end{figure}
49 +
\ No newline at end of file
added phd_chap1_20260731/main.tex +558 −0
@@ -0,0 +1,558 @@
1 +
2 +\documentclass[11pt]{article}
3 +
4 +% Packages
5 +\usepackage{amssymb, amsmath, amsfonts, eurosym, geometry, ulem, graphicx, caption, color, setspace, sectsty, comment, footmisc, pdflscape, subfigure, array, hyperref}
6 +\usepackage{rotating, multirow, tabularx, apalike, environ, atbegshi, booktabs, lscape, xcolor, tablefootnote, ragged2e, nicematrix, enumitem, wasysym, framed, pgfgantt, subfloat, blindtext, colortbl, palatino, mathpazo, threeparttable, sourcesanspro, natbib, tocloft}
7 +
8 +\usepackage{subcaption}
9 +\usepackage{siunitx}
10 +\sisetup{
11 + round-mode = places,
12 + round-precision = 3,
13 + detect-weight = true,
14 + detect-inline-family = math
15 +}
16 +
17 +% Hyperref setup
18 +\hypersetup{
19 + colorlinks,
20 + citecolor=blue,
21 + linkcolor=magenta
22 +}
23 +
24 +% Custom colors
25 +\definecolor{mypink1}{rgb}{0.858, 0.188, 0.478}
26 +
27 +% Theorem and proof environments
28 +\newtheorem{theorem}{Theorem}
29 +\newtheorem{corollary}[theorem]{Corollary}
30 +\newtheorem{proposition}{Proposition}
31 +\newenvironment{proof}[1][Proof]{\noindent\textbf{#1.} }{\ \rule{0.5em}{0.5em}}
32 +
33 +\newtheorem{hyp}{Hypothesis}
34 +\newtheorem{subhyp}{Hypothesis}[hyp]
35 +\renewcommand{\thesubhyp}{\thehyp\alph{subhyp}}
36 +
37 +% Citation style
38 +\setcitestyle{authoryear, open={(}, close={)}}
39 +\bibliographystyle{chicago}
40 +
41 +% Geometry settings
42 +\geometry{left=1.0in, right=1.0in, top=1.0in, bottom=1.0in}
43 +
44 +% Section font
45 +\allsectionsfont{\sffamily}
46 +
47 +% Normal text settings
48 +\normalem
49 +
50 +% Double spacing
51 +\doublespacing
52 +
53 +% Title and author information
54 +\title{\textbf{ Speculative trading in energy markets:\\ Evidence from macroeconomic surprises\\\small{Revised version for \emph{The Energy Journal}}}\footnote{The authors thank participants at seminars at Humboldt University Berlin, South Dakota State University (Ness School), University of Illinois, Urbana-Champaign (ACE), and at meetings of the Commodity \& Energy Markets Association (2021),World Finance \& Banking Association (2021), Société canadienne de sciences économiques (2022), 4th Ethical Finance and Sustainability (EFS) conference (2022), and Multinational Finance Society (2022). The authors also thank the editor George Filis, two anonymous referees, as well as Jocelyn Grira, Joseph Marks, Alessandro Melone (discussants) and Scott Irwin, Michel Robe, and Zhiguang Wang. For financial support, the authors thank the Social Sciences and Humanities Research Council and the Chaire Industrielle-Alliance Groupe financier. Any remaining errors are ours alone.}}
55 +%\author{Author names removed}
56 +\author{Simon-Pierre Boucher\footnote{PhD student in finance, Université Laval, Quebec City QC Canada G1V 0A6, email: \texttt{simon-pierre.boucher.1@ulaval.ca}}\and Marie-H{\'e}l{\`e}ne Gagnon\footnote{Professor of Finance and Research Fellow, CRREP, Université Laval, email: \texttt{marie-helene.gagnon@fsa.ulaval.ca}}\and Gabriel J. Power\footnote{IG Wealth Management Chairholder, Professor of Finance and Research Fellow, CRREP and CRIB, Université Laval, email: \texttt{gabriel.power@fsa.ulaval.ca}}}
57 +\date{\today}
58 +
59 +\begin{document}
60 +\pagenumbering{roman}
61 +% Title page
62 +\begin{titlepage}
63 +\maketitle
64 +\singlespacing
65 +\begin{abstract}
66 +% MH Perso, l'abstract je le trouve ok pour les speculative trading....
67 +\noindent Speculative trading in energy and commodity markets has been blamed for increased volatility, price distortions and market inefficiency, with negative effects on the real economy. We take a new approach to investigate the impact of speculative trading using macroeconomic announcements and high-frequency data. We study the impact of 26 macroeconomic announcement releases on energy commodities (crude oil, natural gas) as our baseline case, which we contrast with metals (gold, silver, copper, and palladium). We find that increased speculative trading lessens the impact of macroeconomic surprises on futures markets, as measured by price drift, volatility, and bid-ask spreads. Our full-sample results show that increased trading by speculators improves liquidity and price discovery, while reducing volatility. We document a damping effect on volatility that is stronger for procyclical commodities such as crude oil and natural gas than for precious metals such as gold, which is a safe haven. In sub-sample analysis where we separate the effects of money managers and swap dealers, we find that the positive effects that we document are driven by money managers. Since traditional market participants prefer stability, our results suggest a beneficial impact of increased trading and speculation.
68 +%Financialization in energy and commodity markets has been blamed for increased volatility, price distortions, and market inefficiency. Taking a new approach using high-frequency data, we investigate the impact of financialization by testing whether macroeconomic surprises have a greater impact on energy and commodity futures when financial investors are more active in these markets. We study the impact of 26 macroeconomic announcement releases on energy commodities (crude oil, natural gas) as our baseline case, which we contrast with metals (gold, silver, copper, and palladium). We find that greater participation of financial investors dampens the impact of macroeconomic release surprises on futures markets, as measured by price drift, volatility, and bid-ask spreads. Critics argue that increased participation of financial investors raises prices and volatility, which has negative effects on the real economy, especially in energy markets. However, our results show that financial investors improve liquidity and price discovery while reducing volatility. We document a damping effect on volatility that is stronger for procyclical commodities such as crude oil and natural gas than for precious metals such as gold, which is a safe haven. Since traditional market participants prefer stability, our results suggest a beneficial impact of financialization. When we disaggregate the results, we find that the positive effects of financial investor participation are linked to money managers rather than to swap dealers.
69 +
70 +\vspace{0.2in}
71 +
72 +\noindent\textbf{Keywords:} energy markets, crude oil, futures, high-frequency, speculation, trading, commercial, volatility, macroeconomic, announcements.\
73 +\bigskip
74 +\end{abstract}
75 +
76 +\setcounter{page}{0}
77 +\thispagestyle{empty}
78 +\end{titlepage}
79 +
80 +\newpage
81 +\doublespacing
82 +
83 +%% Pas besoin de table of contents
84 +%\tableofcontents
85 +%\newpage
86 +%\listoftables
87 +%\newpage
88 +\pagenumbering{arabic}
89 +
90 +\section{Introduction}
91 +\label{sec:introduction}
92 +
93 +%SETUP: EASIER TO TRADE COMMO FUTURES, IT’S BIG BUSINESS PROFITABLE, IT MATTERS
94 +The appeal of energy and other commodities as an asset class has grown since the Commodity Futures Modernization Act of 2000 (CFMA). By partially deregulating derivatives, the CFMA has made it easier to trade commodity futures contracts for investment purposes. Rather than invest in physicals or in shares of commodity-linked firms, investors can use futures to gain exposure to "commodity beta" \citep{boons2014price}. This evolution is particularly relevant for energy markets, where futures trading volume has grown tremendously. For instance, the total trading volume of commodity derivatives was 137.3 bn contracts in 2023, which is 64\% more than in 2022 (Futures Industry Association, 2024).
95 +An important reason for this trend is that commodities have periodically benefited from bull cycles (most notably in 2004-2008), attracting a growing number of speculators and institutional investors. As a result, the commodities asset class has become an important but volatile revenue source for trading firms and investment banks. %\footnote{Gross margins from commodities trading reached about \$57 bn during the 2007-09 period. While profits were more modest in the following years, they increased once more from \$36 bn in 2018 to \$100 bn in 2021 as a result of increased worldwide demand and supply chain bottlenecks, and up to \$150 bn in 2022. A substantial portion of these profits came from energy trading, particularly during periods of high oil price volatility.}
96 +Three firms, Goldman Sachs, Citi, and Macquarie earned together \$20 bn from commodities trading in 2022, much of it from energy-related contracts.\footnote{Sources: The Financial Times, Bloomberg, S\&P Global and Euronews.}
97 +
98 +
99 +%ARE FINANCIAL INVESTORS BAD? PRESSURE ON POLICYMAKERS, DESPITE WEAK EVIDENCE
100 +Is the presence of more financial investors harmful to traditional market participants, such as hedgers? While many think so, the evidence is unclear. The idea that poorly informed investors can disrupt markets has a long history \citep{shleifer1990noise} and has been revived in a recent theoretical literature on financialization \citep{basak2016model,goldstein2022commodity}. These papers are motivated by the commodity price run-up of 2004-2008, which occurred shortly after the CFMA was passed \citep{domanski2007financial}. Critics argue that the activities of financial investors, who are not directly involved in producing or processing commodities, can distort prices and increase volatility.\footnote{A high-profile example is \citet{masters2009testimony}, who testified before the U.S. Congress in 2008 and before the CFTC in 2009 about ``Ending excessive speculation in commodity markets.'' While his argument is not supported by empirical evidence, as shown by \citet{irwin2012testing}, the Masters hypothesis reflects beliefs held at the time by many market participants.} These critics claim that energy and commodity markets have become more sensitive to financial market fluctuations, and less to supply and demand fundamentals.
101 +Energy futures markets have attracted attention due to the popular perception that large price swings affect the real economy (e.g., through higher gasoline and heating costs) \citep{cheng2014financialization}. Research, however, generally does not support this claim \citep{baumeister2014oil}. Whether or not these fears are justified, policymakers have taken notice and the CFTC has progressively implemented rule changes such as new position limits.%\footnote{In 2009 and 2010, the CFTC held public hearings to review the impact of speculative trading on market integrity. In 2010, the CFTC proposed new regulations to impose position limits on commodity derivatives (e.g., number of contracts held by any single trader). The ``Final rule on position limits'' was only adopted on March 15, 2021.}
102 +
103 +
104 +%COMMODITY BULL CYCLE WHO IS TO BLAME, NOT CLEAR
105 +
106 +A large empirical literature debates the causes of periodic price and volatility run-ups in commodity and energy markets. Researchers emphasize the importance of differentiating between speculative traders and passive investors (e.g., index traders). The latter category of traders is more recent and trades energy and commodity contracts for diversification purposes rather than for speculative profit. While \citet{singleton2014investor} suggests that financial investors may be to blame for higher energy prices, \citet{kilian2014role} use a structural model to show that speculation can be ruled out as a cause of the oil price surge during 2003-2008 -- even though speculative demand played a role in previous oil price spikes. Further evidence against the hypothesis that index traders are responsible for the sharp increase in commodity prices is provided by \citet{irwin2011index} and \citet{irwin2012testing}. In a different strand of the literature, \citet{buyukcsahin2011speculators} use Granger causality tests and daily data to investigate whether speculators increase crude oil futures prices. They find little evidence to support that claim. Also using daily position-level data, \citet{brunetti2016speculators} show that speculators reduce price volatility in commodity and energy markets. Reviewing this early literature, \citet{fattouh2013role} conclude that speculation is unlikely to explain the commodity and energy bull cycle of 2004-2008.
107 +
108 +
109 +%RECENT FINANCIALIZATION PAPERS THEORY AND EMPIRICAL
110 +Recent research provides new theoretical grounds to establish how trading activity could affect energy and commodity prices \citep{basak2016model,goldstein2022commodity}. The subsequent empirical literature, however, does not reach a consensus. \citet{henderson2015new} use data on commodity-linked notes to show that uninformed trading flows affect commodity prices, but \citet{ready2022order} argue that the economic magnitude of this effect is too small to matter. Other recent papers find instances of futures price overshooting, reversals, and greater noise in markets \citep{da2024financialization}. They also find that commodities seem to display higher correlations with equities and with each other \citep{kang2023financialization}.
111 +
112 +
113 +%POSITIONING OUR PAPER AS ABOUT SPECULATIVE TRADING, NOT FINANCIALIZATION (REFEREE 2) %MH J'ai collé un petit paragraphe du haut avec celui-ci. Je pense que ça va et que c'est ensemble.
114 +
115 +
116 +
117 +%Although financialization has been linked to the trading activities of any non-traditional, financial investors in commodity markets (e.g., hedge funds), a more commonly accepted definition focuses on the activities of passive (index) traders \citep{tang2012index}. In contrast with speculation, which is likely to reflect informed trades, index traders are considered uninformed. Since this paper is motivated by the economic role played by all non-commercial traders, we will refer to speculative trading as a general description of this broad class of market participants. It is important to distinguish, however, speculators from index or institutional investors. It is the latter who are the more recent financial actors in energy and commodity markets \citep{irwin2011index}. Our paper therefore investigates the impact of speculative trading as well as sub-categories of non-commercial traders.%, rather than the narrower class of index traders.
118 +
119 +
120 +%WHAT WE DO AND WHY IT IS INNOVATIV
121 +Thus, our main contribution is to provide sharply identified evidence on the impact of speculative trading on energy (crude oil and natural gas) and metal markets (gold, silver, copper, and palladium), with additional evidence on sub-categories of traders. We focus on speculation rather than financialization, which has been linked to the trading activities of passive (index) traders \citep{tang2012index}. Specifically, our paper investigates the impact of speculative trading of sub-categories of non-commercial traders. Our rationale is that speculative trading is likely to reflect informed trades, which is our focus, in contrast to index traders who are considered uninformed. Energy commodities serve as our baseline case due to their economic importance and high trading volumes, while metals offer a useful comparison, particularly as gold is perceived as a safe-haven asset. We use high-frequency (5-minute) data to measure the instantaneous reaction of commodity futures returns, volatility, and bid-ask spreads to the surprise component in macroeconomic announcement releases \citep{andersen2007real, kurov2019price}. The data runs from April 4th, 2007, to February 11th, 2024. Our framework also accounts for the time-varying intensity of speculative trading activity. This study builds on \citet{kilian2011energy}, who find no evidence, at a daily frequency, that energy prices react to macroeconomic announcements. By using intraday data, we can better identify the impact of specific macro surprises. We also avoid a common criticism of event study methods, namely that using daily frequency data may reduce the power of statistical tests and could lead the researcher to misattribute the effect of a specific announcement, as other market events occur the same day \citep*{kothari2007econometrics}.
122 +
123 +
124 +%OUR FINDINGS, ONE BY ONE, BRIEFLY
125 +We find evidence of beneficial effects (price stability and market efficiency) from increased trading activity in energy and commodity markets. Our first finding is a damping effect on price reactions: while macro surprises generate a positive abnormal return for good news (and negative for bad news), the magnitude of this reaction is significantly weaker when speculative trading is higher. Second, we find a similar damping effect on volatility reactions. While all surprises (good or bad) generate a volatility increase, this reaction is lessened when the futures market shows more speculative trading. Third, we document lower bid-ask spreads when speculative trading is higher, controlling for the surprise environment. Fourth and last, these beneficial effects are linked to the trading activities of money managers. In contrast, increased trading by swap dealers appears to have an amplifying effect on reactions to macro surprises. These new insights are made possible by investigating this issue using a new angle, namely their sensitivity to macroeconomic surprises, and with high-frequency data. Our findings have important implications for energy market investment and regulation, and to the broader debate about speculation in energy and commodity markets.
126 +
127 +
128 +By investigating a broad range of traders in energy and commodity markets, our findings also extend the work of \citet{brunetti2016speculators}. They find that financial investors, especially money managers and hedge funds, help commodity markets by supplying liquidity, reducing volatility, and generally improving market efficiency. Moreover, our results relate to \citet{cheng2015convective} who show that financial investors, being better informed about markets, contribute to price discovery and liquidity. This is particularly relevant for energy markets, where accurate price discovery is crucial for physical market participants and investors. Thus, speculative traders help markets by distributing and assimilating new information into prices. These insights are valuable given the ongoing energy transition and the importance of efficient price discovery in energy markets. Two papers are probably closest to ours. First, \citet{brunetti2016speculators} who find that hedge funds add liquidity to commodity markets, resulting in more efficient prices and lower volatility. They argue that it is merchant positions (i.e., hedgers) that are linked to greater volatility, and that the presence of hedge funds allows for faster and more efficient price discovery. Second, using daily data, \citet{kilian2011energy} study how energy prices react to macroeconomic announcements. Our paper extends this work to high frequency data.
129 +
130 +
131 +
132 +
133 +
134 +
135 +
136 +
137 +
138 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
139 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
140 +\section{Background}
141 +% personnellement, je pense qu'on peut enlever le paragraphe en entier. le reste de la section s'applique bien je pense. et on en a parlé dans l'intro de ça.
142 +%\alert{\subsection{Speculative trading activity in energy and commodity futures markets}
143 +%In the literature, financialization usually refers to the process by which an asset class, such as commodities, attracts significantly more attention from financial investors. As a result, prices and volatility may no longer be determined only by commodity supply and demand \citep{cheng2014financialization}. \citet{brunetti2016speculators} refers to financialization as ``the changing mix of participant positions''. These financial investors include hedge funds, commodity trading firms, swap dealers and index traders, but specifically exclude hedgers. Traders may take commodity futures positions for purposes of speculation, diversification, or factor exposure to commodity beta. REVISE TO AVOID FINANCIALIZATION.}
144 +
145 +
146 +Price discovery in commodity markets occurs mainly in futures markets and is affected by informational frictions around supply and demand. Thus, risk sharing and information discovery represent a potential channel for financial investors to generate distortions in energy and commodity markets \citep{cheng2014financialization}. In their model, \citet{basak2016model} predict that the increased presence of financial actors can increase commodity futures volatility, as well as correlations between commodity and equity returns. \citet{goldstein2022commodity} also argue that under some conditions, a greater presence of financial investors can be harmful to commodity markets.
147 + Theory shows how a change in the participant mix in commodity markets could generate undesirable distortions, but the empirical literature is far from settled. \citet{singleton2014investor} argues that trading activity by financial investors creates informational frictions, leading commodity prices to become more volatile and to diverge from their fundamental values. Using a no-arbitrage argument, however, \citet*{hamilton2014risk} show that the positions of commodity traders included in index funds cannot be used to achieve excess returns in futures markets. \citet{ready2022order} show that while index traders do have a positive price impact, it is much too small to explain the apparent price distortions or bull cycles observed since 2004. In addition, financial investors do not have a uniform impact on market liquidity. Investors affect liquidity risk by either providing liquidity to meet the hedging needs of other traders or consuming liquidity when they trade for their own needs \citep{kang2020tale}. Indeed, \citet{brunetti2014commodity} show using data on commodity trader positions that index traders provide insurance against price risk.
148 +
149 +\subsection{Macroeconomic announcements}
150 +Surprises in macroeconomic announcements affect financial markets, whether in stocks \citep*{scholtus2014speed} or in bonds \citep{fleming1997moves}. In a key study, \citet*{balduzzi2001economic} find that 17 public news releases affect bond prices, trading volume, and bid-ask spreads. \citet{karali2014macro} show that energy futures markets exhibit asymmetric responses to macroeconomic news, with significant volatility spillovers between natural gas and crude oil markets. \citet{cao2024us} document a time-varying relationship between U.S. monetary policy and crude oil prices, finding that unexpected oil price increases can push monetary policy from expansionary to restrictive stance. \citet{kang2020economic} further show that after 2004, short-term oil price volatility is driven by industrial production, term spreads, and credit spreads, along with traditional market factors.
151 +
152 +
153 +The literature on commodity-specific announcements is smaller and less conclusive. \citet*{hollstein2020volatility} look at how different economic variables affect the term structure of commodity futures volatility. They show that speculation and jobs-related macro variables have the largest impact on volatility. \citet{zhu2022oil} further show that stock market anomalies can be explained to some extent by oil price shocks, separately from the effect of other macroeconomic variables and investor sentiment.
154 +While the literature finds a clear impact of macroeconomic announcements on stock and bond prices, there is no clear answer as to whether they affect commodity futures prices, or whether increased trading by financial participants accentuates these reactions. This issue is especially relevant for energy markets, given their macroeconomic importance. Our research provides new insights by using high-frequency data, expanding the set of announcements, and considering a time-varying measure of speculative trading intensity to capture trading activities for each of the commodities in the sample.
155 +
156 +
157 +
158 +
159 +
160 +
161 +
162 +
163 +
164 +
165 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
166 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
167 +\section{Data}
168 +
169 +We now present a detailed description of our data. Since this paper relies on several types of data, we describe: i) how to obtain the macroeconomic announcement surprises, ii) the commodity futures data, and iii) how to capture speculative trading activity.%, and iv) a PCA analysis for financial investor activity.
170 +
171 +\subsection{Data on macroeconomic announcements}
172 +
173 +The macroeconomic announcement release data are obtained from Bloomberg and Refinitiv Eikon. We collect information on 22 announcements that are standard to the literature \citep[see e.g.,][]{andersen2003micro}. Our sample for macroeconomic announcements is matched to our high-frequency data and therefore runs from April 2nd, 2007 to February 11th, 2024. The announcements belong to ten categories: Income, Employment, Industrial Activity, Investment, Consumption, Housing Sector, Government, Net Exports, Inflation, and Forward-looking. Most of the announcements are released on a monthly basis. Table \ref{tab:stat1} summarizes the announcements and provides more detail such as the number of observations, release frequency, source, unit of measure, and time of release. Bloomberg provides analyst forecasts for all announcements, as well as the actual value of the announcement release. For all announcements except the Consumer Price Index and Initial Jobless Claims releases, a positive surprise will be interpreted by investors as signaling a strong economy \citep*{fleming1997moves}.
174 +%Based on a survey of U.S. announcements and their impacts on financial markets, the only ones that signal a weaker economy are \citep*{fleming1997moves}. For the other announcements, a positive surprise will be interpreted by investors as signaling a strong economy.
175 +In addition, we include energy sector-specific announcements published by the U.S. Energy Information Administration. The first is the weekly crude oil storage report, which provides an update on the quantity of crude oil held in storage in the U.S. The second is the weekly natural gas storage report. We do not include OPEC announcements, as they cannot be reliably used in a high-frequency econometric design \citep{10.1257/aer.20190964}.\footnote{There are a few issues with the OPEC announcements: First, they are not released at a specific time. Second, it is impossible to know precisely when a given OPEC announcement was made available to investors. Third, OPEC's influence has weakened since the 1980s.}
176 +
177 +It is common practice in this literature to use the standardized surprise of an announcement rather than its realized value to quantify the unexpected component of the release. To calculate surprises, we follow \citet*{balduzzi2001economic}. Let $A_{kt}$ be the realized value (i.e., release) of macroeconomic announcement $k$ at time $t$, and let $E_{kt}$ be the median value of all Bloomberg analyst forecasts for announcement $k$ at time $t$. To standardize the surprise, we divide the raw surprise $(A_{kt} - E_{kt})$ by $\sigma_k$, the sample standard deviation of the surprise for announcement $k$. Thus, equation (\ref{eqn:SURPRISE}) describes the standardized surprise for announcement $k$ at time $t$:
178 +
179 +\begin{equation}\label{eqn:SURPRISE}
180 +S_{kt} = \frac{A_{kt} - E_{kt}}{\sigma_k}
181 +\end{equation}
182 +
183 +The sample period is used to compute $\sigma_k$, as in \citet{balduzzi2001economic} and \citet{kurov2019price}.\footnote{The literature argues that measuring $\sigma_k$ in this way is reasonable because the standardized surprise is not used for forecasting purposes. Using raw surprises is not recommended due to scaling issues, nor is using analyst dispersion for $\sigma_k$ because announcement coverage sometimes involves only a few analysts. For robustness, we also estimate our models using surprises where $\sigma_k$ is computed using only past observations. The main findings are unchanged. In this case, we exclude the first $M$ observations (e.g., $M=10$) to get a reasonable sample size for $\sigma_k$.} Table \ref{tab:stat2} presents the minimum, 1st quartile, median, mean, 3rd quartile, and maximum of the surprise for each announcement.
184 +
185 +
186 +
187 +
188 +
189 +
190 +
191 +
192 +
193 +
194 +\subsection{Commodity futures price data}
195 +
196 +For intraday data on commodity futures prices, we use Barchart's API.\footnote{See the \url{https://www.barchart.com/futures} website.} Our dataset for prices contains some of the most economically significant commodity futures contracts traded in the U.S. We use a high-frequency price series that runs from April 2nd, 2007 to February 11, 2024. Among these contracts, crude oil and natural gas are pro-cyclical, while gold and silver behave as safe havens. High-grade copper and palladium are industrial metals used in the manufacturing of consumer products.
197 +
198 +For each of the commodities in our sample, price returns $R_t$ are calculated as the log return over a 5-minute period $(\tau=5)$ beginning at time $t$. The database provides the futures contract close price ($p_{t}^{close}$) of each 5-minute period. Thus, $R_t$ is obtained as in equation (\ref{eqn:RETURN}):
199 +
200 +\begin{equation}\label{eqn:RETURN}
201 +R_t^{t+\tau} = \ln \left( \frac{p_{t+\tau}^{close}}{p_{t}^{close}} \right) = \ln (p_{t+\tau}^{close}) - \ln(p_{t}^{close})
202 +\end{equation}
203 +
204 +Descriptive statistics for the 5-minute log returns are presented in Table \ref{tab:stat4}. The most extreme outlier observations belong to crude oil, while gold has the fewest outliers.\footnote{The main findings are robust to using different window lengths by estimating equation \ref{eqn:RETURN} using 30-minute returns.}
205 +
206 +\subsection{Measures of speculative trading activity and trader categories}
207 +%To measure the impact of financialization, we need a measure that captures the intensity of speculation in commodity markets.
208 +
209 +
210 +%We investigate the impact of speculative trading activity increasing relative to productive activity. % We consider proxies designed to capture the intensity of speculation in commodity markets and we use them as indicators of financial investor activity.
211 +%For instance, swap dealers are not speculators, but they belong to our broad definition of financialization.
212 +The index of speculative trading is constructed using data in the \emph{Commitment of Traders (CoT) Report} published weekly by the Commodity Futures Trading Commission (CFTC). The data provided by the CFTC includes the number of positions held by different types of participants in commodity markets. The CFTC separates trader types as follows: \textit{Commercials} refer to trader-reported futures positions which the trader claims are used for hedging purposes, while \textit{Non-Commercials} is obtained by subtracting the total long and short commercial positions from the total open interest.\footnote{The CFTC defines commercial traders as participants in commodity markets who primarily use futures contracts to hedge their business activities (e.g., buying or selling commodities). All traders who are not classified as Commercial are automatically classified as Non-Commercial traders. To obtain the number of long positions held by Non-Commercial traders, we subtract the total long Commercial positions from the total open interest. For the number of short positions held by Non-Commercial traders, we subtract the total short Commercial Positions from the total open interest.}
213 +%Given that $CoT$ reports are not available in real time, they are not considered to be sources of information that traders could act upon, but rather a way to capture the state of the market.
214 +We use the following information presented in the $CoT$ report: for a futures contract $i$, the number of long and short positions held by Non-Commercial traders are $SL_i$ and $SS_i$, respectively, while for Commercial traders they are $HL_i$ and $HS_i$.\footnote{In an earlier draft, we also reported results based on two alternative proxies as well as a proxy constructed using principal component analysis. The alternative proxies are Working's $T$ \citep{working1960speculation} and the market share of non-commercial traders \citep*{buyukcsahin2014speculators}. These results, which are available upon request, are consistent with our main findings and do not change the paper's implications.}
215 +
216 +%The first proxy we consider to assess levels of speculative and hedging activity is Working’s $T$ \citep{working1960speculation}. This index compares the activity levels of Non-Commercial commodity futures traders (e.g., speculators) to those of Commercial traders (e.g., hedgers). Working’s $T$ measures the extent to which speculation exceeds the level required to offset any unbalanced hedging at the market clearing price. This index, denoted $WT_i$, is computed as follows:
217 +%Historically this is true but today more nuanced Typically, Commercial traders take short positions in futures contracts while Non-Commercial traders take long positions.
218 +
219 +%\begin{equation} \label{eqn:Working}
220 +%WT_i = \left\{
221 +%\begin{matrix}
222 +%1 + \frac{SS_i}{HL_i + HS_i} & \mbox{if} & HS_i \ge HL_i \\
223 +%1 + \frac{SL_i}{HL_i + HS_i} & \mbox{if} & HS_i < HL_i
224 +%\end{matrix}
225 +%\right.
226 +%\end{equation}
227 +
228 + %\citet*{buyukcsahin2014speculators} suggest a measure that emphasizes the \emph{market share of Non-Commercial traders (MSCT)}. This ratio is expressed as the sum of the short and long positions of Non-Commercial traders over twice the total open interest in a market:
229 +%%Instead of Working’s $T$,
230 +%
231 +%
232 +%\begin{equation} \label{eqn:MSCT}
233 +%MSCT_i = \frac{SL_i + SS_i}{2 \times OI_i}
234 +%\end{equation}
235 +
236 +The specific measure we use follows \citet{hedegaard2011margins}, who suggests an index of speculative activity computed as the ratio of net long speculative positions over total open interest ($NLS_i$):
237 +
238 +\begin{equation} \label{eqn:NLS}
239 +NLS_i = \frac{SL_i - SS_i}{OI_i}
240 +\end{equation}
241 +
242 +In addition to computing $NLS$ using the full sample data, we use disaggregated data from the CFTC to compute the NLS index separately for money manager (MM) and swap dealer (SD) positions, which allows for additional empirical analysis. The data on money manager and swap dealer positions come from Quandl's API.\footnote{See the \url{https://data.nasdaq.com/data/CFTC-commodity-futures-trading-commission-reports} website.}
243 +\emph{Money managers} typically refer to Non-Commercial market participants who are involved in managing funds and investing in commodity futures and options markets \citep{fishe2012identifying}.\footnote{This category is also called ``Managed money.'' The CFTC writes that they are ``registered commodity trading advisor (CTA); a registered commodity pool operator (CPO); or an unregistered fund identified by CFTC.'' There is some overlap between Money managers and hedge funds, but they are distinct.} Their activities are influenced by financial and economic factors related to commodities. Money managers are often considered to be more informed investors because they actively manage portfolios and adjust their positions based on market information and analysis.
244 +% factors such as speculative activity, imperfect information about real economic activity as well as supply, demand, and inventory accumulation in commodity markets \citep{singleton2014}.
245 +\emph{Swap dealers} are considered as Non-Commercial traders by the CFTC. They typically use futures contracts to hedge risk generated by their swap positions. Swap dealers have been studied in relation to index investors, as their positions are distinct from those of other market participants. Their activities are influenced by the need to manage large exposures and to facilitate trading for clients. While money managers regularly take long or short futures positions, swap dealers mainly take long positions \citep{fishe2012identifying}. Hedgers, who we exclude from the analysis, tend to take short positions.
246 +%often leading them to take significant speculative positions in the market.
247 +%and contribute to price discovery and liquidity \citep{brunetti2016speculators}
248 +%\citep{sandrs2016bubbles}
249 +
250 +
251 +%\subsection{A financialization proxy using principal component analysis}
252 +%
253 +%Descriptive statistics for the three financialization variables are shown in Table \ref{tab:stat5} and computed separately for each of the six commodities in our sample, based on the number of open positions for a given futures contract. The variables themselves are scale-free, unlike the number of open positions. The MSCT variable fluctuates between 0 and 0.5, while NLS varies between -0.4 and 0.8, and Working's $T$ between 1 and 2. Since the literature does not show that one is superior to the others, and for brevity’s sake, we use Principal Component Analysis to construct a new proxy using information from the three variables. This step allows us to present results based on a single variable.\footnote{In an earlier draft, we showed results using the three different variables and found that our main findings were similar.}
254 +%Our approach is as follows: First, since the variables have different scales, we standardize them to have a mean of zero and a variance of one. Second, we compute the covariance matrix. Third, we calculate the eigenvalues and eigenvectors. Fourth, we obtain the principal components as the eigenvectors of the covariance matrix, and we use the first principal component as our new proxy for financialization. Table \ref{tab:PCA} shows the outcome of the PCA analysis and in particular, that the first principal component explains a large part of the individual variances.
255 +
256 +
257 +
258 +
259 +
260 +
261 +
262 +
263 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
264 +\section{Econometric framework and methods}
265 +
266 +\subsection{Modeling the impact of surprises on returns}\label{return}
267 +
268 +Our high-frequency regression model is based on \citet{kurov2019price}.\footnote{In unreported results, we run the regressions using the approach shown in \citet{andersen2003micro}. The results are similar.} We run the following regression using the specification in equation (\ref{eq:Model 1}):
269 +
270 +\begin{equation}\label{eq:Model 1}
271 +R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{j} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_j)+\beta R_{t-\tau}^{t}+\epsilon_{t}
272 +\end{equation}
273 +%%% MH :donc on enlève les notes en rouge???
274 +where $R_{t}^{t+\tau}$ is the continuously compounded futures return from time $t$ to $t+\tau$, $S_{mt}$ is the surprise for macroeconomic announcement $m$ published at time $t$, and $X_{j}$ is the NLS speculative trading intensity variable, which is updated at a weekly frequency, with $j$ the index for the week.
275 +%$X_{t}$ is the value of the speculative intensity proxy using the $NLS$ variable.
276 +The impact of macro announcements on commodity futures returns can be assessed by looking at the $\gamma_m$ coefficient in the mean equation, while the $\delta$ coefficient controls for the level of speculative trading intensity as it relates to futures returns. The key coefficient to help answer our main research question is $\theta_m$, which relates the effect of time-varying speculative trading intensity on the impact of the news release.% This last coefficient is the most important one to help answer our main research question.
277 +The regression is estimated using a two-step weighted least squares (WLS) procedure. %For robustness, appendix A presents results using the NLS variable.
278 +To account for heteroskedasticity, we construct a volatility estimate by means of an exponential moving average, using the regression residuals obtained in the first step. This auxiliary regression is presented in equation~(\ref{eqn:auxiliary 2}), with a smoothing parameter $\alpha=0.9$ and a starting parameter value set to $\sigma_1=\epsilon_t$:
279 +
280 +\begin{equation}\label{eqn:auxiliary 2}
281 +\sigma_t=\alpha \sigma_{t-1}+(1-\alpha) \mid \epsilon_t \mid
282 +\end{equation}
283 +
284 +After obtaining $\sigma_t$ for each observation, we apply the transformation $w_t = \hat{\sigma_t}^{-2}$ to obtain the WLS regression weight. Then, we multiply each variable by $w_t$ and run an OLS regression to estimate the model.
285 +
286 +
287 +
288 +
289 +
290 +\subsection{Modeling the impact of surprises on volatility}\label{variance}
291 +
292 +To estimate the volatility equation, we use a GARCH specification, as it is well known that the variance of commodity futures returns displays time variation and clustering \citep*[see e.g.,][]{brunetti2014commodity}. We specify a GARCH (1,1) model and extend the equation by including our NLS speculative intensity proxy as well as the macroeconomic news surprise variables.
293 +First, we estimate the mean equation (\ref{eqn:MeanEqn}):
294 +
295 +\begin{equation}\label{eqn:MeanEqn}
296 +%R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+\beta R_{t-\tau}^{t}+\epsilon_{t} ICI À VOUS AVEC sp PARCE QUE JE NE VOIS PAS CE QUI A CHANGÉ
297 +R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+\beta R_{t-\tau}^{t}+\epsilon_{t}
298 +\end{equation}
299 +
300 +Then, we estimate the following equation for conditional variance:
301 +
302 +\begin{equation}\label{eqn:VarianceEqn}
303 +%\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h
304 +\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{j}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h
305 +\end{equation}
306 +%j,ai change le j en bas
307 +
308 +
309 +
310 +where $I_{k,t}=D_{m,t} \cdot X_{j}$ and $D_{m,t}$ is a dummy variable for macro announcement $m$. The latter equals 1 if an announcement takes place at time $t$ (5-minute frequency) and equals 0 otherwise. $X_{j}$ is the NLS speculative intensity variable as defined earlier. The $\rho_h$ coefficient captures intraday periodicity, while the $D_h$ dummy equals 1 at hour $h$ and 0 otherwise. The impact of macro announcement $m$ on conditional variance is captured by the $\Phi_m$ coefficient in equation~(\ref{eqn:VarianceEqn}), while the $\beta$ coefficient shows the impact of the speculative intensity variable $X_{j}$. Finally, the $\phi_k$ coefficient shows the interaction effect from speculative trading and the macro surprise $m$.
311 +The standard errors are computed using the Newey-West heteroskedasticity and autocorrelation consistent (HAC) estimator with automatic lag selection, following the procedure outlined in \citet{newey1994automatic}.%\footnote{This approach accounts for both heteroskedasticity and serial correlation in the residuals, which is standard practice in high-frequency financial data analysis.}
312 + This methodology mirrors the approach used in \citet{andersen2003micro, andersen2007real} and \citet{kurov2019price} to study announcement effects in other asset classes. In unpublished results, we consider the mixed-data sampling (MIDAS) approach proposed by \citet{ghysels2004midas}. We find that the results are similar.%\footnote{We thank a reviewer for suggesting a mixed-frequency approach. Results are available upon request.}
313 +
314 +% Among the announcements in our sample, all but one are ``good news.'' Only a positive surprise in Initial Jobless Claims indicates a deterioration in economic conditions. Therefore, the surprise coefficient is expected to be positive for all pro-cyclical commodities (i.e., all but gold and silver) for all announcements except Initial Jobless Claims, for which it should be negative (since a positive surprise is ``bad news’’). In the case of gold and silver, which are safe-haven commodities, the reverse is expected for coefficient signs.
315 +
316 +
317 +
318 +
319 +
320 +
321 +
322 +
323 +
324 +\subsection{Modeling the impact on bid-ask spreads}
325 + Speculative trading could make markets more efficient by improving information. We test this hypothesis by measuring the effect of macro surprises on the futures price bid-ask spread in high-frequency regressions.
326 +%We aim to investigate whether increased financialization affects the responsiveness of commodity futures returns to macroeconomic surprises by analyzing the impact on the bid-ask spread.
327 +The bid-ask spread is widely recognized as a measure of market efficiency. A narrower spread suggests less uncertainty about the asset's true value and reflects lower transaction costs, improved liquidity, and lower information asymmetry \citep{Roll1984}.
328 +%A narrower bid-ask spread typically indicates higher liquidity and better informational efficiency, suggesting less uncertainty about the asset's value.
329 +Furthermore, \citet{chordia2008liquidity} show that the bid-ask spread is an indicator of market quality and market efficiency. Since a smaller spread is associated with a more efficient price discovery process, an increase in the quality of market information should decrease the spread. Therefore, we estimate the following equation, where the relative bid-ask spread is defined as $(Ask_t-Bid_t)/Mid_t$:
330 +
331 +\begin{equation}\label{eq:Model 2}
332 +%\text{Spread}_{t}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta \text{Spread}_{t-\tau}+\epsilon_{t}
333 +\text{Spread}_{t}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{j} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_j)+\beta \: \text{Spread}_{t-\tau}+\epsilon_{t}.
334 +\end{equation}
335 +
336 +In equation ~(\ref{eq:Model 2}) , $\text{Spread}_{t}$ is the relative bid-ask spread measured at 5-minute frequency $t$ using the high-frequency data and $\text{Spread}_{t-\tau}$ is the lagged spread. In addition, $\sum_{m=1}^{22} \gamma_m D_{m,t}$ accounts for the macroeconomic announcements, where each dummy variable $D_{m,t}$ is multiplied by its respective coefficient $\gamma_m$. We also include the speculative trading intensity variable $X_{j}$ with its coefficient $\delta$. The interaction terms $\sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_j)$ capture the effect of speculative trading intensity on the bid-ask spread at the time of a release, with each term multiplied by its respective coefficient $\theta_m$. To test whether increased speculative trading activity improves informational efficiency at the time of a macroeconomic announcement, we check whether the sign on $\theta_m$ is negative and significant, thus reducing the spread. %Indeed, the interaction terms $\theta_m (D_{m,t} \cdot X_t)$ capture the effect of financialization on the spread at the time of a release.% macroeconomic announcements.
337 +%(measured at daily frequency, where $j$ represents the trading day)
338 +
339 +
340 +
341 +
342 +
343 +
344 +
345 +
346 +
347 +
348 +%\subsection{Impact according to the type of non-commercial trader}
349 +%\alert{TOO SHORT AS A STAND ALONE SECTION. FIGURE OUT HOW TO INTEGRATE IT IN THE TEXT.}
350 +
351 +
352 +
353 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
354 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
355 +\section{Results} \label{sec:result}
356 +
357 +\subsection{The impact of surprises on cumulative abnormal returns}
358 +
359 +We begin by documenting the impact of macroeconomic announcement surprises on commodity futures returns. The impact of surprises is shown in the following graphs of high-frequency cumulative abnormal returns (CARs). Figures \ref{fig:cl} to \ref{fig:ng} show CARs for each commodity as measured over a window of 60 minutes before to 60 minutes after a macroeconomic announcement release. The figures are constructed similarly to those shown in \citet{kurov2019price}. The magnitude of the CARs after announcement releases is comparable to those shown in their paper for stock index and Treasury futures. Unlike them, however, we do not see evidence of a pre-announcement drift.
360 +
361 +
362 +A red line denotes the average CAR for announcement releases that are seen as negative surprises (i.e., worse than anticipated news), while a green line denotes the average CAR for positive surprises.
363 +For brevity, we discuss only the CARs for crude oil and gold, as they are representative of pro-cyclical energy markets and safe haven assets. Figure \ref{fig:cl} shows the average CAR for crude oil futures. The CAR increases following a positive surprise and decreases following a negative one, confirming that crude oil is a pro-cyclical commodity. In contrast, figure \ref{fig:gc} shows that gold futures react in the opposite manner. The red line indicates that CAR is positive after bad news, while the green line shows that CAR is negative after good news. %These results support the hypothesis that gold is a safe haven asset.%finanthe one presented for stocks in their paper.\footnote{A difference is that the commodity markets do not seem to display pre-announcement drift whereas \citet{kurov2019price} documented one for the S\&P 500 futures.}
364 +%The increase (decrease) in CAR in reaction to a positive (negative) surprise confirms that crude oil is a pro-cyclical commodity.
365 +
366 +
367 +
368 +
369 +\subsection{Macroeconomic surprises, speculative trading activity and futures returns}
370 +
371 +Table~\ref{tab:macro_fin_nls_fut_returns_full} presents the results of high-frequency regressions that explain commodity futures returns immediately after a macroeconomic announcement release. Our discussion focuses on coefficients that are statistically significant at the 5\% level.\footnote{In an earlier draft, we also reported results for different sub-periods, such as the Zero Lower Bound period, the 2008-2010 financial crisis and Great Recession, and the COVID-19 period. These results do not materially affect our findings or conclusions, and they are available upon request. } %For brevity, the table only reports results for regressions in which the PCA proxy is used.\footnote{That being said, there is no material difference in the results if we use one of the other measures.}
372 +We begin with energy commodities, which serve as our baseline case. The $\gamma_m$ coefficient shows the immediate impact of a macro surprise on commodity futures returns. For crude oil futures, we find that several macroeconomic announcements exhibit significant effects. Consider for instance Initial Jobless Claims, for which a greater than expected value indicates bad economic news. The table shows that for this announcement, $\gamma_m$ is negative, indicating that crude oil prices tend to drop in response to unexpected increases in jobless claims. This result is consistent with the expectation that higher jobless claims signal weaker economic conditions, which reduce the demand for crude oil. The corresponding $\theta_m$ coefficient for Initial Jobless Claims is positive, however, suggesting that increased speculative trading mitigates the negative impact of bad macroeconomic news on crude oil prices. This damping effect suggests that markets are better informed as a result of the increased participation of speculators. Indeed, the announcement release creates a smaller surprise and a smaller shock.
373 +
374 +%TOUT CA EST A CHANGER AVEC NLS car les coefficients changent de signes parfois et ne sont plus significatifs au mêmes endroits. For natural gas, the ADP Employment announcements show a pattern similar to crude oil, with a positive $\gamma_m$ coefficient implying that better than expected employment figures boost energy prices, reflecting increased economic activity and demand. The negative $\theta_m$ coefficient indicates that speculative trading dampens this positive reaction. The CB Consumer Confidence and Advance Retail Sales announcements also show positive $\gamma_m$ coefficients and negative $\theta_m$ coefficients for both energy commodities, supporting the claim that increased speculative trading intensity smooths out the market's response to macroeconomic surprises.
375 +
376 +In the case of natural gas futures, we find that the results across announcements are less frequently significant. Increased speculative trading tends to increase the magnitude of the surprise's effects, when the results are significant. An exception is the natural gas market-specific announcement release (inventories), for which the coefficient is negative but not significant. Overall for natural gas futures, we do not find as much support that speculative trading dampens the effect of macroeconomic announcements on returns. The reason why the results for natural gas futures are less conclusive is most likely that this futures contract displays greater volatility, that it has a lower trading volume \citep{irwin2012testing}, and it has less speculative trading activity (thus, fewer informed traders) \citep{buyukcsahin2014speculators}.
377 +%results confirm the specific nature of natural gas, in terms of lower trading volume , and less speculative trading activity such as international traders \citep{buyukcsahin2014speculators}.
378 +
379 +Copper is a pro-cyclical, industrial commodity, so it is expected that the results for copper futures should resemble those for crude oil. The results are highly significant for many announcements. If we look at announcements such as ADP Employment and CB Consumer Confidence, which are ``good news'', the $\gamma_m$ coefficients are positive, confirming copper's pro-cyclical nature and its ties to industrial production. The corresponding $\theta_m$ coefficients are negative, consistent with the damping effect that we document for energy commodities. Thus, the main finding for copper futures is that, as with crude oil, increased speculative trading intensity has the effect of weakening the impact of a macro surprise.
380 +%Comparing these results with other commodities, the behavior of copper futures closely resembles that of crude oil -- indeed, both are pro-cyclical. For
381 +
382 +
383 +Gold is considered to be a safe haven asset \citep{baur2010gold}. Therefore, it is expected that the reaction of gold futures returns to macro surprises will be the opposite to what we have found for crude oil, natural gas, and copper. This is indeed what we find: gold futures returns are lower after "good news" and higher after "bad news". These results support the idea that energy commodities are pro-cyclical, while gold is a safe haven asset. In particular, the positive $\gamma_m$ coefficient for Initial Jobless Claims indicates that gold prices rise in response to unexpected increases in jobless claims, as investors seek safety in gold positions during economic uncertainty. The $\theta_m$ coefficient is negative, indicating that speculative trading activity tempers this flight to safety, leading to less pronounced price increases.%Turning to gold futures, the reaction to macroeconomic surprises is opposite to what we find for our energy baseline case:
384 +In the case of the ADP Employment release, the $\gamma_m$ coefficient is negative for gold futures, suggesting that strong employment figures reduce gold prices as investors move away from safe-haven assets towards pro-cyclical assets, such as energy commodities. The positive $\theta_m$ coefficient suggests that speculative trading reduces the extent of this price drop. The CB Consumer Confidence and Advance Retail Sales announcements also generate negative $\gamma_m$ coefficients for gold futures and positive $\theta_m$ coefficients at a 10 percent level, providing additional evidence of a moderating influence of speculative trading on the reaction of gold to economic news. The main finding is therefore that whether a commodity is pro-cyclical or a safe haven asset, increased speculative trading has a damping effect on the reactions to macro surprises.
385 +
386 +The last two commodities in our sample, silver and palladium futures, behave more like gold futures. We find that for Initial Jobless Claims for palladium and ADP Employment in silver, significant $\gamma_m$ coefficients are found in directions consistent with their status as safe haven assets. The $\theta_m$ coefficients for silver and palladium also indicate a damping effect on price reactions. Taken together, our results show that the damping effect of speculative trading is stronger in energy markets, suggesting that the beneficial effects of financial participants are particularly important for energy commodities.
387 +
388 +
389 +
390 +
391 +
392 +
393 +\subsection{Macroeconomic surprises, speculative trading and volatility}
394 +
395 +Table~\ref{tab:macro_fin_nls_var_full} presents regression results to explain the conditional variance of high-frequency commodity futures returns after macroeconomic announcements. We first examine our baseline assets, energy commodities, as volatility in energy markets is of particular concern given their economic importance and direct impact on consumer prices.
396 +For crude oil, macroeconomic surprises generally lead to an increase in conditional variance, as the $\Phi_m$ coefficients are consistently positive. For instance, a surprise in Initial Jobless Claims significantly increases crude oil volatility. Natural gas exhibits similar patterns, with some variations in magnitude and a greater impact on inventory-related announcements. This result suggests that unexpected economic news generates greater uncertainty and price fluctuations in energy markets.
397 +In contrast, the interaction coefficients $\phi_m$, which inform us about the impact of speculative trading, tend to be negative for the two energy commodities. The implication is that increased trading activity by speculative traders lowers volatility following a macro surprise. Taking Initial Jobless Claims as an example, we find that higher levels of speculation reduce the impact of news on volatility in crude oil futures markets. Thus, speculative trading can act as a stabilizing force by damping the heightened volatility that occurs after a surprise in macroeconomic news. %\textcolor{red}{This stabilizing effect is especially valuable in energy markets, where excessive volatility can have significant economic consequences}.
398 +
399 +Comparing these results with those for other commodities, we find that gold, copper, silver, and palladium also show positive $\Phi_m$ coefficients across various announcements, indicating increased volatility following macro surprises. However, the magnitude of these effects is generally smaller than what we see in energy markets, especially in the case of precious metals. The corresponding $\phi_m$ coefficients, measuring the damping effect of financial investor activity, are negative across commodities and announcements, and the strongest effects are found in crude oil futures markets. This pattern holds for other macroeconomic announcements as well. The coefficients for ADP Employment, CB Consumer Confidence, and Advance Retail Sales, for instance, generally indicate increased volatility after surprises, as shown by the positive $\Phi_m$ coefficients, while the corresponding $\phi_m$ coefficients are negative, supporting the finding of a stabilizing effect of increased speculative trading. %\textcolor{red}{The consistency of these results across different types of announcements is particularly noteworthy in energy markets, where price stability is crucial for both consumers and industrial users}.
400 +
401 +Therefore, our findings highlight the valuable role of speculative traders in reducing volatility, particularly in crude oil futures markets, where price stability has important implications for the broader economy. While macroeconomic news tends to increase volatility across commodity markets, the damping effect of speculative trading appears strongest in energy markets. We show that this relationship is consistent across different types of announcements and remains robust when controlling for various market conditions.
402 +
403 +
404 +
405 +
406 +\subsection{Macroeconomic surprises, speculation and bid-ask spreads}
407 +
408 +Table~\ref{tab:return-fin-full} shows our results for the impact of macroeconomic surprises and speculative trading intensity on futures prices bid-ask spreads. This empirical analysis provides a test of informational efficiency. We first focus on energy markets. In the bid-ask spread regressions, the $\gamma_m$ coefficient denotes the effect of surprises on the spread, while the $\theta_m$ coefficient shows the interaction effect between surprises and speculative trading intensity. The main hypothesis is whether $\theta_m < 0$, which would indicate that greater speculative activity improves informational efficiency through narrower spreads. Such a finding would be consistent with what we have documented above for futures returns and volatility. In the case of crude oil futures, we find that $\gamma_m$ is negative for the initial jobless claims announcement, which means that the market becomes more efficient immediately after a news release. This is consistent with the resolution of uncertainty. The $\theta_m$ coefficient is always negative when it is significant, indicating that the bid-ask spread narrows even more (implying greater informational efficiency) after a macro announcement if crude oil futures markets benefit from greater speculative activity relative to hedging activity, as measured by the NLS proxy. The results for natural gas futures are similar to those for crude oil. The relationship between speculative trading and market efficiency is clearest during periods of higher trading volume, such as the release of storage reports and weather-related announcements.
409 +%\textcolor{red}{CHECK TO MAKE SURE THAT THIS IS STILL ACCURATE.}
410 +
411 +Comparing these results to those obtained for the other commodities in our sample, we find that the results for $\theta_m$ (speculative intensity) are less often significant, but that they are negative when they are significant. The $\theta_m$ coefficient being negative suggests that greater speculative trading intensity improves market efficiency, as bid-ask spreads tend to be lower when the coefficient is significant. This improvement in market efficiency appears to be most pronounced in energy markets, where accurate price discovery is particularly important given their relevance for the real economy.
412 +%The stronger effects we observe in energy markets may reflect their greater integration with the broader financial system and higher trading volumes, which allow for more efficient price discovery. This is particularly relevant given the increasing importance of energy price stability for economic policy and planning}.
413 +Overall, the results for the bid-ask spread provide additional support for our claim that greater trading activity has beneficial effects on commodity derivatives markets, with these benefits being especially notable in energy markets where efficient price discovery has important implications for both market participants and the broader economy.
414 +
415 +
416 +
417 +
418 +\subsection{Differences in results according to trader type}
419 +
420 +To investigate whether differences in trader type are relevant in explaining our findings, we provide disaggregated results in this section. To this end, we estimate equations (\ref{eqn:MeanEqn}), (\ref{eqn:VarianceEqn}) and (\ref{eq:Model 2}) for two categories of Non-Commercial traders, namely, swap dealers (SD) and money managers (MM). For each of the two, the CFTC reports the number of long and short positions in their disaggregated Commitment of Traders (COT) reports. We compute the NLS index for each trader category over time, allowing us to separately quantify the intensity of trading activity by money managers and swap dealers.%\footnote{We use the NLS variable for this disaggregated analysis because it is the only one we can construct using the available data on MM and SD positions.}
421 +
422 +
423 +%This section shows disaggregated results for the two types of financial participants that are reported separately, namely swap dealers and money managers. %We continue to use the NLS variable, but it is computed separately for MM and SD futures positions. %\footnote{The main reason is that the MSCT and Working's $T$ variables cannot be computed only for MM or SD because the calculations involve the positions of commercial traders.}
424 + First, we examine the returns equation for money manager positions, as shown in table \ref{tab:macro_fin_mm_fut_returns_full}. Increased trading activity by money managers has the same effect as in our baseline results. If we consider crude oil futures, for example, the $\gamma_m$ macro surprise coefficient is positive while the $\theta_m$ coefficient for speculative trading is negative. Since $\theta$ has the opposite sign to $\gamma$, the implication is that increased futures trading activity by money managers lowers the impact of surprises on futures returns. This is similar to our baseline, aggregate findings.
425 + Next, table \ref{tab:macro_fin_sd_fut_returns_full} presents results using only swap dealer positions. Here we find a notable difference relative to money managers. This table shows that the speculative intensity coefficient $\theta$ for swap dealers has the same sign as the macro surprise coefficient $\gamma$. This result means that increased swap dealer trading activity seems to amplify the reaction of futures returns to macro surprises. The exception to these results is in the case of natural gas futures, where increased trading by swap dealers for some announcements appears to have the opposite effect on returns. To contrast this finding with prior research, \citet{brunetti2016speculators} find, using daily data, that the positions of swap dealers are not correlated with contemporaneous returns and volatility in commodity futures markets. They further show that hedge funds decrease, and hedgers increase, volatility. Our empirical analysis extends their findings using high-frequency data and the setting of macro surprises as a source of new information affecting energy and commodity futures markets.
426 +
427 +Tables \ref{tab:macro_fin_mm_var_full} and \ref{tab:macro_fin_sd_var_full} show our disaggregated results for the variance equation using sample data for money managers and swap dealers, respectively. The money manager results are similar to what we find in the aggregate sample, namely that both good and bad surprises increase volatility (as shown by $\Phi_{mm}>0$), while greater money manager trading activity lowers the impact of news on volatility (as shown by $\phi_{mm}<0$). This result strengthens prior evidence in the literature about beneficial effects of speculators, which were based on daily data \citep{brunetti2016speculators}. In contrast, our results suggest that increased activity by swap dealers appears to increase volatility after macro news (since $\phi_{sd}>0$). Together, the results line up with our findings for the returns equation and with our main message, which is that informed traders stabilize markets by contributing new information. Swap dealers, as intermediaries, typically do not trade based on information but rather following the needs of their clients. This economic motivation can explain why our results suggest that their trading activities amplify price and volatility reactions to news.
428 +%. that the signs for $\gamma_m$ and $\theta_m$ are equal. Therefore, unlike money managers, it seems that more trading activity by swap dealers causes even more strong market reactions to macroeconomic news announcements. Swap dealers' trading operations are motivated by supporting their clients' needs rather than information.
429 +%Findings for the variance equation using only money managers and swap dealers can be found in tables respectively. We find that whereas for swap dealers the financialization interaction coefficient $\phi_k$ is positive, for money managers it is negative. These findings support the economic interpretation of previous findings for returns. Although rising trade by money managers lessens the impact of macro surprises on volatility, a higher presence of swap dealers seems to enhance the influence of surprises on volatility. Therefore, the trading actions of money managers help to minimize market volatility after market news; yet, the activities of swap dealers exacerbate this effect.
430 +Although our discussion centers on crude oil as a benchmark commodity, the results for the other pro-cyclical commodities support our interpretation of the findings. In the case of gold futures, the disaggregated results continue to support a safe haven interpretation \citep{erb2013golden}. Indeed, gold has features of a commodity and a currency, but earlier studies have also found that its value increases with investor risk aversion, since it is perceived as a safe haven during times of economic uncertainty and market volatility.
431 +
432 +
433 +%\subsubsection{Bid-ask spread analysis for money managers and swap dealers}
434 +
435 +%The examination of bid-ask spreads following macroeconomic announcements is expanded to include a NLS financialization proxy computed using just money manager (MM) or swap dealer (SD) positions, to account for potential differences in the impact of their trading activities.
436 +Lastly, we examine whether the effects on bid-ask spreads differ between trader types.
437 +Table \ref{tab:macro_fin_mm_fut_spread_full} presents the results for money managers. As the $\theta_m$ coefficients tend to be negative, it seems that money managers enhance market efficiency by lowering bid-ask spreads. The results for swap dealers are shown in Table \ref{tab:macro_fin_sd_fut_spread_full}. Unlike for money managers, we find that $\theta_m$ coefficients for swap dealers tend to be positive, indicating that greater swap dealer activity increases the bid-ask spread after a macroeconomic release. The results for swap dealers therefore suggest a decline in market efficiency due to their increased trading activities. %meaning that a higher degree of activity by swap dealers reduces market efficiency by increasing bid-ask spreads following a macroeconomic announcement, indicating a decline in market efficiency.
438 +These findings are in line with our main results and further support the claim that traders in energy and commodity markets do not all have the same effect on market efficiency. The results point to the importance of information acquisition and investor attention as an economic channel.
439 +%These results also relate to \citet{rakowski2021twitter} who show that Twitter activity, by serving as a distinct form of investor attention, has significant effects on trading volume and asset prices.
440 +
441 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
442 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
443 +
444 +%Important: keep this in LaTex format; use % to comment out; keep references as is
445 +\section{Discussion and implications}
446 +
447 +
448 +
449 +Our findings contribute new insights to an unsettled literature on the impact of different types of financial participants in energy and commodity markets \citep{ready2022order}. In addition to being an important alternative asset class, energy commodities are central to economic activity, while energy futures trading is important for price stability. Being closely related to macroeconomic risk \citep{cheng2015convective}, energy commodities therefore play a key role in macro-finance. By taking a novel angle of high-frequency market reactions to macroeconomic surprises, our results contribute a more nuanced picture of the impact of speculative trading and help to reconcile previous findings in the literature.
450 +%Our analysis builds on the models proposed by \citet{goldstein2022commodity} and \citet{basak2016model}, which explain how institutional investors in particular can influence commodity prices, volatility and market efficiency.
451 +
452 +%Our paper is motivated by recent theoretical advances.
453 +\citet{goldstein2022commodity} develop a theoretical model to examine how financial traders improve information transmission between futures and spot markets. They show how market efficiency should benefit from the presence of informed financial traders, whose actions lead to increases in pricing transparency and to lower information asymmetry. Our empirical results provide support for this model. We show that increased speculative trading activity lowers bid-ask spreads after a macro news release, in addition to reducing the magnitude of price and volatility reactions to macro surprises. This improvement in informational efficiency is especially valuable in energy markets, where price discovery has direct implications for industrial users and consumers. We find that this effect is driven by money managers. In addition, our results point to smaller, negative effects linked to swap dealers. The difference between the two sets of disaggregated empirical results can be explained by the fact that they trade for different purposes. Money managers aim to make a profit based on information, while swap dealers are intermediaries who, while perhaps informed, primarily provide services to customers such as institutional investors, in addition to hedging their swap positions. In line with this interpretation, our results also build on \citet{fishe2012identifying}, who show that some financial traders (e.g., money managers) are better informed than others in commodity markets.
454 +%\textcolor{red}{This information advantage appears particularly valuable in energy markets, where complex supply-demand dynamics and geopolitical factors make price discovery especially challenging}.
455 +
456 +%DO NOT INCLUDE THIS PART SINCE WE WANT TO DOWNPLAY INDEX TRADERS
457 +%Our findings also provide evidence to support the predictions in the model proposed by \citet{basak2016model}. They argue that financialization should increase commodity futures prices and volatility, especially for indexed futures. They argue that the activities of institutional investors enhance the link between commodities and equities. This relationship is particularly relevant in energy markets, where crude oil and natural gas prices have become increasingly correlated with financial markets. We find some evidence to this effect, as increased trading by swap dealers, who tend to be long and deal with index funds, increases price reactions and volatility. However, our results suggest that overall, these effects are offset by the stabilizing presence of informed traders such as money managers.
458 +
459 +Our findings also extend and build on the results of an earlier empirical literature that uses daily-level data. This literature includes \citet{brunetti2016speculators} who find that speculative traders, especially money managers and hedge funds, are helpful to energy and commodity markets, as well as \citet{buyukcsahin2011speculators} and \citet{alquist2013role} who show that the futures positions of financial firms such as hedge funds do not predict next-day changes in crude oil prices.
460 +Moreover, our results relate to \citet{cheng2015convective} who show that financial investors, being better informed about markets, contribute to price discovery and liquidity. The key message is therefore that speculative trading is helpful to markets by distributing and assimilating new information into prices.
461 +%This is particularly relevant for energy markets, where accurate price discovery is crucial not only for market participants but also for energy policy and planning}.
462 +The findings shown in this paper have important implications for energy markets regulation and policy. First, they suggest that attempts to limit speculative trading in energy markets could, in fact, increase price volatility and reduce informational efficiency and price discovery. Second, they indicate that different types of financial participants have distinct effects on market quality, suggesting that regulatory frameworks should pay careful attention to market composition. Third, they highlight the importance of maintaining a robust price discovery mechanism in energy markets, given their crucial role in the economy.
463 +%REDONDANT
464 +%To summarize, the evidence we present suggests that financial investors play an important role in improving the functioning of energy markets, which is especially important given current challenges including the energy transition, geopolitical tensions, and the need for stable energy prices to support economic growth.
465 +
466 +
467 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
468 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
469 +
470 +%\section{Robustness in sub-periods}
471 +%
472 +%To rule out the possibility that our main results are driven by a specific period, we present in the following robustness checks our results for different sub-sample periods, namely the Zero Lower Bound period from December 22nd, 2008 to December 21st, 2015, and the COVID-19 pandemic period from January 31st, 2020 to June 10th, 2022. The start and end dates are selected following the literature and sources such as the NBER and the CDC. Our main finding is that the sub-sample results are consistent with our full sample results, despite some minor discrepancies. Therefore, our main findings are not explained by any particular period, but rather hold for the entire sample period from 2007 to 2024.
473 +%
474 +%% The full tables are presented in appendices B and C, respectively. In order to investigate whether our results are period-specific, we now present the results for different sub-periods. The results presented in the appendix are for specific sub-periods, namely the Zero Lower Bound (ZLB) (2008-12-22 to 2015-12-21) and COVID-19 sub-periods (2020-01-31 to 2022-06-10). The main finding is that the sub-period results are consistent with the full-sample results presented earlier.
475 +
476 +%\subsection{Price reactions during the Zero Lower Bound and COVID-19 sub-periods}
477 +%
478 +%Tables~\ref{tab:macro_fin_fut_returns_covid} and \ref{tab:macro_fin_fut_returns_zlb} present the results of high-frequency regressions to explain returns after macroeconomic announcements, estimated separately for the Zero Lower Bound (ZLB) and COVID-19 sub-periods. We discuss a few major announcements here. Looking at Initial Jobless Claims, the surprise coefficient $\gamma_m$ is negative for crude oil and positive for gold in both sub-periods, consistent with the full sample results.
479 +%%indicating that more jobless claims than expected decrease crude oil prices and increase gold prices.
480 +%
481 +%The financialization $\theta_m$ coefficient is negative for crude oil during the COVID-19 sub-period and positive during the ZLB sub-period, suggesting that the interaction effect of financialization varies to some degree over the years. In the case of ADP Employment, $\gamma_m$ is positive for crude oil and copper while negative for gold in both sub-periods. The $\theta_m$ coefficient has the opposite sign to $\gamma_m$, similar to our baseline results. In the case of CB Consumer Confidence, $\gamma_m$ is negative for gold during the ZLB sub-period but is not significant during COVID-19, while the $\theta_m$ coefficients have the opposite sign, consistent with the damping effect we discuss. Lastly, for Advance Retail Sales, $\gamma_m$ is significant for crude oil, gold, copper, and silver in both sub-periods while $\theta_m$ generally has the opposite sign to $\gamma_m$. Thus, our sub-period results are for the most part similar to the full sample results, and we do not see any important differences in the economic interpretation of the coefficient signs.
482 +%
483 +%\subsection{Volatility reactions during the Zero Lower Bound and COVID-19 sub-periods}
484 +%
485 +%Table \ref{tab:macro_fin_fut_var_covid} and Table \ref{tab:macro_fin_fut_var_zlb} present the results of regressions to estimate the reaction of conditional variance to macro surprises in different sub-periods. This analysis confirms our full sample results, namely that surprises increase volatility while financialization acts to lower this impact. This can be seen from the $\Phi_m$ surprise coefficients, which remain positive, while the financialization coefficients $\phi_m$ are negative. Thus, our main findings are not driven by a particular sub-period.
486 +%
487 +%\subsection{Bid-ask spreads during the Zero Lower Bound and COVID-19 sub-periods}
488 +%
489 +%%This section examines the impact of macroeconomic surprises and financialization on the bid-ask spread during the Zero Lower Bound (ZLB) and COVID-19 sub-periods, aiming to determine whether the relationship between financialization and market efficiency, as measured by the bid-ask spread, remains consistent across different economic environments.
490 +%
491 +%Tables \ref{tab:macro_fin_fut_spread_zlb} and \ref{tab:macro_fin_fut_spread_covid} present our regression results for the bid-ask spread during the ZLB and COVID-19 sub-periods. We find that in both sub-periods the $\gamma_m$ coefficients tend to be negative, indicating that macroeconomic surprises generally reduce the bid-ask spread across our sample of commodities. This is the case, for instance, for Initial Jobless Claims for crude oil and copper futures in the COVID-19 sub-period, which implies increased efficiency. Furthermore, the $\theta_m$ coefficients tend to be negative in the two sub-periods, suggesting that financial investor activity improves efficiency by narrowing spreads after the release of macro news.
492 +%
493 +%%during the COVID-19 period are generally negative, indicating that increased financialization amplifies the efficiency improvements induced by macroeconomic surprises. Specifically, for crude oil, the negative $\theta_m$ coefficient suggests that the participation of financial investors, such as money managers, enhances the market's ability to process new information, further reducing the bid-ask spread.
494 +%%% POURQUOI ON PARLE DE MONEY MANAGERS ICI?
495 +%%shows the results for the ZLB period. The $\gamma_m$ coefficients are mostly negative, consistent with the full-sample findings, indicating that macroeconomic surprises during the ZLB period also tend to reduce the bid-ask spread, enhancing market efficiency. The $\theta_m$ coefficients are negative for most commodities, similar to the COVID-19 period, suggesting that financialization continues to improve market efficiency by further reducing the bid-ask spread following macroeconomic announcements.
496 +%
497 +%While the overall patterns in the $\gamma_m$ and $\theta_m$ coefficients are similar to the full sample results, the coefficients tend to be less statistically significant in these sub-periods. This may be due to smaller sample sizes or to unique market conditions during these periods.
498 +%%That being said, the results are consistent with our main findings for bid-ask spreads and suggest there are no meaningful differences in sub-periods.%, the general trend indicates that macroeconomic surprises reduce the bid-ask spread, and financialization enhances this effect, thereby improving market efficiency.
499 +%
500 +%That being said, we find that the beneficial impact of financialization on market efficiency, as measured by a narrower bid-ask spread, is robust across different economic conditions.
501 +%
502 +%%% JE NE CROIS PAS QUE LES PHRASES SUIVANTES AJOUTENT BEAUCOUP; ON SE REPETE.
503 +%%The consistent negative $\theta_m$ coefficients imply that the presence of financial investors, particularly money managers, helps markets process information more efficiently, regardless of the broader economic environment. This supports the argument that financialization contributes to improved market functioning by enhancing the dissemination and incorporation of new information into asset prices. Overall, the sub-period analysis reinforces our full-sample results, highlighting the beneficial role of financialization in commodity markets, particularly in terms of market efficiency as reflected in narrower bid-ask spreads following macroeconomic surprises.
504 +%
505 +%\subsection{Sub-period analysis for money managers and swap dealers}
506 +%
507 +%We discuss in this section a robustness check for the disaggregated analysis using only positions of money managers and swap dealers. In short, these results confirm the baseline finding that money managers bring more stability to markets through a damping effect after a macro surprise, while swap dealers have the opposite impact. The results for money managers are shown in table \ref{tab:macro_fin_mm_fut_returns_covid} (returns equation, COVID-19 period), table \ref{tab:macro_fin_mm_var_covid} (variance equation, COVID-19 period), \ref{tab:macro_fin_mm_fut_returns_zlb} (returns equation, ZLB period), and table \ref{tab:macro_fin_mm_var_zlb} (variance equation, ZLB period). The regression results for swap dealers, in the same order, are presented in tables \ref{tab:macro_fin_sd_fut_returns_covid}, \ref{tab:macro_fin_sd_var_covid}, \ref{tab:macro_fin_sd_fut_returns_zlb}, and \ref{tab:macro_fin_sd_var_zlb}.
508 +%
509 +%%The results during the ZLB sub-period reveal that the $\gamma_m$ coefficients for money managers are generally positive for crude oil and copper and negative for gold, similar to the full-sample findings. This suggests that the presence of money managers mitigates the impact of macroeconomic surprises on commodity futures returns. The $\theta_m$ coefficients are negative, further indicating that financialization by money managers dampens the effects of macro surprises. For swap dealers, the $\gamma_m$ coefficients are positive for most commodities, suggesting that their activities amplify market reactions to macroeconomic announcements. The $\theta_m$ coefficients for swap dealers are positive, supporting the notion that increased trading by swap dealers leads to greater market reactions to surprises.
510 +%
511 +%%During the COVID-19 sub-period, the results are consistent with the ZLB findings. The $\gamma_m$ coefficients for money managers show a similar pattern, with positive values for crude oil and copper and negative values for gold. The $\theta_m$ coefficients remain negative, confirming that financialization by money managers continues to reduce the impact of macroeconomic surprises on returns. For swap dealers, the $\gamma_m$ coefficients are again positive for most commodities, and the $\theta_m$ coefficients are positive, indicating that swap dealers' activities exacerbate market reactions to economic news. The variance equation results during both sub-periods show that the $\phi_m$ coefficients for money managers are negative, indicating that their presence reduces volatility following macroeconomic announcements. In contrast, the $\phi_m$ coefficients for swap dealers are positive, suggesting that their trading activities increase volatility in response to macro surprises.
512 +%
513 +%Overall, the sub-period analysis reinforces our full-sample findings. It highlights how money managers and swap dealers have different impacts on the market's reactions to macroeconomic announcements. While money managers tend to enhance market stability by damping the effects of surprises, swap dealers appear to increase market volatility and amplify the reactions to economic news.
514 +
515 +
516 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
517 +\section{Conclusion} \label{sec:conclusion}
518 +
519 +This paper investigates the impact of speculative trading on the real economy and energy markets through a new angle, namely high-frequency surprises in macroeconomic announcement releases, which allows for better-identified effects. %We study energy commodities as our baseline case, given their economic importance.
520 +We empirically test whether increased speculative trading activity amplifies or dampens the impact of macro surprises on prices and volatility in commodity futures markets.
521 +%ce paragraphe est bon, mais la conslusion est très longue et selon les commentaires on devrait focuser sur les trader types.
522 +%In fact, it is well known that the equity and bond markets react to these surprises. Suppose that a consequence of increased speculative trading intensity is to make energy and commodity futures behave more like financial assets. Then, we should see futures prices display greater reactions to macro surprises. This hypothesis is the basis of our investigation. Moreover, since the increased involvement of financial investors cannot be easily categorized into before and after sub-sample, we measure the intensity of speculative trading by means of a time-varying and commodity-specific proxy.
523 +Our results suggest that increased speculative trading activity has beneficial effects for energy and commodity markets. This is accomplished by reducing volatility and improving price discovery, as indeed price stability and efficient price discovery are crucial for market participants and the broader economy. We find that a greater intensity of speculative trading does not amplify the effects of macro announcement surprises on prices or volatility. On the contrary, an increase in speculative trading in a given commodity has a damping effect: prices and volatility react \emph{less} to macro surprises when speculative trading is higher. This stabilizing effect of speculation is especially valuable to energy markets, where price volatility can have significant economic consequences.
524 +
525 +What is more, our findings are consistent with information diffusion economic arguments. Our analysis of bid-ask spreads in futures contracts further confirms that speculative trading tends to improve market efficiency.
526 +% j'ai ajouté le début du paragraphe pour mettre l'emphase sur le main results
527 +Our results show that these effects are mostly linked to the trading activities of money managers rather than swap dealers. Thus, we contribute to a literature that emphasizes how non-commercial market participants such as money managers are beneficial to commodity markets by supplying liquidity, reducing volatility, and generally improving market efficiency. This finding is particularly relevant for energy markets, which have seen substantial increases in trading volume and complexity. %The results we present are robust to the use of a non-parametric variance estimator, different proxies for speculative trading intensity, and to alternative empirical specifications (e.g., regression specification, high-frequency window, etc.).
528 +
529 +Our findings have important implications for energy market regulation and policy. The damping effect on volatility shocks documented in this paper implies that speculative traders contribute to market stability, which is essential for energy security and economic planning. This stability could also facilitate investment in energy infrastructure and support the ongoing energy transition. By lowering the magnitude of volatility shocks, and thus reducing the real option value of delaying investments, our findings suggest that a greater involvement by speculative traders may also help with sustainability efforts to finance a green energy transition, alongside other instruments such as green bonds and portfolio screens for sustainable investments.
530 +Looking forward, our results suggest several promising avenues for future research in energy markets. First, the role of financial investors (speculators as well as passive investors) in facilitating the energy transition warrants further investigation. Second, the connection between speculative trading and energy market regulation remains an important area for study, given that we find different impacts for money managers and swap dealers. Finally, the impact on energy price discovery of new trading technologies and market participants is an emerging research frontier. These questions are particularly relevant given the increasing importance of energy markets in addressing climate uncertainty and in ensuring economic stability.
531 +
532 +%%%%% BIBLIOGRAPHY HERE %%%%%%%%%%%
533 +\small
534 +\bibliography{master}
535 +%\newpage
536 +
537 +\newpage
538 +%\bibliography{master}
539 +
540 +
541 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
542 +%% TABLES HERE
543 +\include{tables}
544 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
545 +%% FIGURES HERE
546 +\section{Figures}
547 +
548 +
549 +\include{figures}
550 +
551 +%% We do not include the appendices anymore ; but available upon request
552 +%\include{appendix_full} %appendix A for NLS only
553 +
554 +%\include{appendix_covid} %appendix B sub sample covid
555 +%\include{appendix_zlb} % appendix C zlb period sub sample
556 +\end{document}
557 +
558 +
added phd_chap1_20260731/master.bib +2602 −0
@@ -0,0 +1,2602 @@
1 +@article{doi:10.1198/jbes.2009.07205,
2 +author = {S. Borağan Aruoba and Francis X. Diebold and Chiara Scotti},
3 +title = {Real-Time Measurement of Business Conditions},
4 +journal = {Journal of Business \& Economic Statistics},
5 +volume = {27},
6 +number = {4},
7 +pages = {417-427},
8 +year = {2009},
9 +publisher = {Taylor & Francis},
10 +doi = {10.1198/jbes.2009.07205},
11 +
12 +URL = {
13 +
14 + https://doi.org/10.1198/jbes.2009.07205
15 +
16 +
17 +
18 +},
19 +eprint = {
20 +
21 + https://doi.org/10.1198/jbes.2009.07205
22 +
23 +
24 +
25 +}
26 +
27 +}
28 +
29 +
30 +
31 +
32 +@techreport{NBERw19523,
33 + title = "Time Variation in Asset Price Responses to Macro Announcements",
34 + author = "Goldberg, Linda S and Grisse, Christian",
35 + institution = "National Bureau of Economic Research",
36 + type = "Working Paper",
37 + series = "Working Paper Series",
38 + number = "19523",
39 + year = "2013",
40 + month = "October",
41 + doi = {10.3386/w19523},
42 + URL = "http://www.nber.org/papers/w19523",
43 + abstract = {Although the effects of economic news announcements on asset prices are well established, these relationships are unlikely to be stable. This paper documents the time variation in the responses of yield curves and exchange rates using high frequency data from January 2000 through August 2011. Significant time variation in news effects is present for those announcements that have the largest effects on asset prices. The time variation in effects is explained by economic conditions, including the level of policy rates at the time of the release, and risk conditions: government bond yields increase in response to "good news", but less so when risk is elevated. Risk conditions matter since they can capture the effects of uncertainty on the information content of news announcements, the interaction of monetary policy and financial stability objectives of central banks, and the effect of news announcements on the risk premium.},
44 +}
45 +@article{SCOTTI20161,
46 +title = {Surprise and uncertainty indexes: Real-time aggregation of real-activity macro-surprises},
47 +journal = {Journal of Monetary Economics},
48 +volume = {82},
49 +pages = {1-19},
50 +year = {2016},
51 +issn = {0304-3932},
52 +doi = {https://doi.org/10.1016/j.jmoneco.2016.06.002},
53 +url = {https://www.sciencedirect.com/science/article/pii/S0304393216300320},
54 +author = {Chiara Scotti},
55 +keywords = {Business cycle, Dynamic factor model, State space model, Forecasting weights},
56 +abstract = {Two daily, real-time, real-activity indexes are constructed for the United States, euro area, United Kingdom, Canada, and Japan: (i) a surprise index summarizing recent economic data surprises and measuring optimism/pessimism about the state of the economy, and (ii) an uncertainty index measuring uncertainty related to the state of the economy. The surprise index parsimoniously preserves the properties of the underlying series when affecting asset prices. For the United States, the real-activity uncertainty index is compared to other uncertainty proxies to show that, when uncertainty is strictly related to real activity only, it has a potentially milder effect on economic activity.}
57 +}
58 +
59 +@article{10.1257/aer.20190964,
60 +Author = {Känzig, Diego R.},
61 +Title = {The Macroeconomic Effects of Oil Supply News: Evidence from {OPEC} Announcements},
62 +Journal = {American Economic Review},
63 +Volume = {111},
64 +Number = {4},
65 +Year = {2021},
66 +Month = {April},
67 +Pages = {1092-1125},
68 +DOI = {10.1257/aer.20190964},
69 +URL = {https://www.aeaweb.org/articles?id=10.1257/aer.20190964}}
70 +@article{garbade1983price,
71 + title={Price movements and price discovery in futures and cash markets},
72 + author={Garbade, Kenneth D and Silber, William L},
73 + journal={Review of Economics and Statistics},
74 + volume = {65},
75 + number = {2},
76 + pages={289--297},
77 + year={1983},
78 + publisher={JSTOR}
79 +}
80 +@article{andersen1998deutsche,
81 + title={Deutsche mark--dollar volatility: intraday activity patterns, macroeconomic announcements, and longer run dependencies},
82 + author={Andersen, Torben G and Bollerslev, Tim},
83 + journal={Journal of Finance},
84 + volume={53},
85 + number={1},
86 + pages={219--265},
87 + year={1998},
88 + publisher={Wiley Online Library}
89 +}
90 +
91 +@article{andersen2007real,
92 + title={Real-time price discovery in global stock, bond and foreign exchange markets},
93 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X and Vega, Clara},
94 + journal={Journal of International Economics},
95 + volume={73},
96 + number={2},
97 + pages={251--277},
98 + year={2007},
99 + publisher={Elsevier}
100 +}
101 +
102 +@article{andersen2003micro,
103 + title={Micro effects of macro announcements: Real-time price discovery in foreign exchange},
104 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X and Vega, Clara},
105 + journal={American Economic Review},
106 + volume={93},
107 + number={1},
108 + pages={38--62},
109 + year={2003}
110 +}
111 +
112 +@article{basak2016model,
113 + title={A model of financialization of commodities},
114 + author={Basak, Suleyman and Pavlova, Anna},
115 + journal={Journal of Finance},
116 + volume={71},
117 + number={4},
118 + pages={1511--1556},
119 + year={2016},
120 + publisher={Wiley Online Library}
121 +}
122 +
123 +@article{brunetti2016speculators,
124 + title={Speculators, prices, and market volatility},
125 + author={Brunetti, Celso and B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Harris, Jeffrey H},
126 + journal={Journal of Financial and Quantitative Analysis},
127 + volume = {51},
128 + number = {5},
129 + pages={1545--1574},
130 + year={2016},
131 + publisher={JSTOR}
132 +}
133 +
134 +@article{balduzzi2001economic,
135 + title={Economic news and bond prices: Evidence from the US Treasury market},
136 + author={Balduzzi, Pierluigi and Elton, Edwin J and Green, T Clifton},
137 + journal={Journal of Financial and Quantitative analysis},
138 + volume = {36},
139 + number={4},
140 + pages={523--543},
141 + year={2001},
142 + publisher={JSTOR}
143 +}
144 +
145 +@article{beechey2009high,
146 + title={The high-frequency impact of news on long-term yields and forward rates: Is it real?},
147 + author={Beechey, Meredith J and Wright, Jonathan H},
148 + journal={Journal of Monetary Economics},
149 + volume={56},
150 + number={4},
151 + pages={535--544},
152 + year={2009},
153 + publisher={Elsevier}
154 +}
155 +
156 +@article{bernanke2005explains,
157 + title={What explains the stock market's reaction to Federal Reserve policy?},
158 + author={Bernanke, Ben S and Kuttner, Kenneth N},
159 + journal={Journal of Finance},
160 + volume={60},
161 + number={3},
162 + pages={1221--1257},
163 + year={2005},
164 + publisher={Wiley Online Library}
165 +}
166 +
167 +@article{boehm2020us,
168 + title={The {US}, economic news, and the global financial cycle},
169 + author={Boehm, Christoph and Kroner, Niklas},
170 + journal={Economic News, and the Global Financial Cycle (January 10, 2020)},
171 + year={2020}
172 +}
173 +
174 +
175 +@article{bollerslev1986generalized,
176 + title={Generalized autoregressive conditional heteroskedasticity},
177 + author={Bollerslev, Tim},
178 + journal={Journal of Econometrics},
179 + volume={31},
180 + number={3},
181 + pages={307--327},
182 + year={1986},
183 + publisher={Elsevier}
184 +}
185 +
186 +@article{boyd2005stock,
187 + title={The stock market's reaction to unemployment news: Why bad news is usually good for stocks},
188 + author={Boyd, John H and Hu, Jian and Jagannathan, Ravi},
189 + journal={Journal of Finance},
190 + volume={60},
191 + number={2},
192 + pages={649--672},
193 + year={2005},
194 + publisher={Wiley Online Library}
195 +}
196 +
197 +@article{bryan1993consumer,
198 + title={The Consumer Price Index as a measure of inflation},
199 + author={Bryan, Michael F and Cecchetti, Stephen G},
200 + journal={Economic Review-Federal Reserve Bank of Cleveland},
201 + volume={29},
202 + number={4},
203 + pages={15},
204 + year={1993},
205 + publisher={Federal Reserve Bank of Cleveland}
206 +}
207 +
208 +@article{buyukcsahin2014speculators,
209 + title={Speculators, commodities and cross-market linkages},
210 + author={B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Robe, Michel A},
211 + journal={Journal of International Money and Finance},
212 + volume={42},
213 + pages={38--70},
214 + year={2014},
215 + publisher={Elsevier}
216 +}
217 +
218 +@article{cai2001moves,
219 + title={What moves the gold market?},
220 + author={Cai, Jun and Cheung, Yan-Leung and Wong, Michael CS},
221 + journal={Journal of Futures Markets},
222 + volume={21},
223 + number={3},
224 + pages={257--278},
225 + year={2001},
226 + publisher={Wiley Online Library}
227 +}
228 +@article{clark1997us,
229 + title={{US} inflation developments in 1996},
230 + author={Clark, Todd E and others},
231 + journal={Economic Review-Federal Reserve Bank of Kansas City},
232 + volume={82},
233 + pages={11--30},
234 + year={1997},
235 + publisher={Federal Reserve Bank of Kansas City}
236 +}
237 +
238 +
239 +@article{couleau2020corn,
240 + title={Are corn futures prices getting `jumpy'?},
241 + author={Couleau, Anabelle and Serra, Teresa and Garcia, Philip},
242 + journal={American Journal of Agricultural Economics},
243 + volume={102},
244 + number={2},
245 + pages={569--588},
246 + year={2020},
247 + publisher={Wiley Online Library}
248 +}
249 +
250 +@article{creti2013links,
251 + title={On the links between stock and commodity markets' volatility},
252 + author={Creti, Anna and Jo{\"e}ts, Marc and Mignon, Val{\'e}rie},
253 + journal={Energy Economics},
254 + volume={37},
255 + pages={16--28},
256 + year={2013},
257 + publisher={Elsevier}
258 +}
259 +
260 +@article{deaton1992behaviour,
261 + title={On the behaviour of commodity prices},
262 + author={Deaton, Angus and Laroque, Guy},
263 + journal={Review of Economic Studies},
264 + volume={59},
265 + number={1},
266 + pages={1--23},
267 + year={1992},
268 + publisher={Wiley-Blackwell}
269 +}
270 +
271 +@article{engle1982autoregressive,
272 + title={Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation},
273 + author={Engle, Robert F},
274 + journal={Econometrica},
275 + pages={987--1007},
276 + year={1982},
277 + publisher={JSTOR}
278 +}
279 +
280 +@article{ehrmann2004taking,
281 + title={Taking stock: Monetary policy transmission to equity markets},
282 + author={Ehrmann, Michael and Fratzscher, Marcel},
283 + journal={Journal of Money, Credit and Banking},
284 + pages={719--737},
285 + year={2004},
286 + publisher={JSTOR}
287 +}
288 +
289 +@incollection{fama2016commodity,
290 + title={Commodity futures prices: Some evidence on forecast power, premiums, and the theory of storage},
291 + author={Fama, Eugene F and French, Kenneth R},
292 + booktitle={The World Scientific Handbook of Futures Markets},
293 + pages={79--102},
294 + year={2016},
295 + publisher={World Scientific}
296 +}
297 +
298 +@article{fattouh2013role,
299 + title={The role of speculation in oil markets: What have we learned so far?},
300 + author={Fattouh, Bassam and Kilian, Lutz and Mahadeva, Lavan},
301 + journal={The Energy Journal},
302 + volume={34},
303 + number={3},
304 + year={2013},
305 + publisher={International Association for Energy Economics}
306 +}
307 +
308 +@article{flannery2002macroeconomic,
309 + title={Macroeconomic factors do influence aggregate stock returns},
310 + author={Flannery, Mark J and Protopapadakis, Aris A},
311 + journal={Review of Financial Studies},
312 + volume={15},
313 + number={3},
314 + pages={751--782},
315 + year={2002},
316 + publisher={Oxford University Press}
317 +}
318 +
319 +@article{fleming1999price,
320 + title={Price formation and liquidity in the US Treasury market: The response to public information},
321 + author={Fleming, Michael J and Remolona, Eli M},
322 + journal={Journal of Finance},
323 + volume={54},
324 + number={5},
325 + pages={1901--1915},
326 + year={1999},
327 + publisher={Wiley Online Library}
328 +}
329 +
330 +@article{fleming1997moves,
331 + title={What moves the bond market?},
332 + author={Fleming, Michael J and Remolona, Eli M},
333 + journal={Federal Reserve Bank of New York Economic Policy Review},
334 + volume={3},
335 + number={4},
336 + year={1997}
337 +}
338 +
339 +
340 +@article{getz1990barometer,
341 + title={Diffusion indexes: a barometer of the economy},
342 + author={Getz, Patricia M and Ulmer, Mark G},
343 + journal={Bureau of Labor Statistics Monthly Labor Review},
344 + pages={13},
345 + year={1990}
346 +}
347 +
348 +@article{wirl2004impact,
349 + title={The impact of {OPEC} Conference outcomes on world oil prices 1984-2001},
350 + author={Wirl, Franz and Kujundzic, Azra},
351 + journal={The Energy Journal},
352 + volume={25},
353 + number={1},
354 + pages = {45--62},
355 + year={2004},
356 + publisher={International Association for Energy Economics}
357 +}
358 +
359 +@article{gorton2006facts,
360 + title={Facts and fantasies about commodity futures},
361 + author={Gorton, Gary and Rouwenhorst, K Geert},
362 + journal={Financial Analysts Journal},
363 + volume={62},
364 + number={2},
365 + pages={47--68},
366 + year={2006},
367 + publisher={Taylor \& Francis}
368 +}
369 +
370 +@article{glick2012central,
371 + title={Central bank announcements of asset purchases and the impact on global financial and commodity markets},
372 + author={Glick, Reuven and Leduc, Sylvain},
373 + journal={Journal of International Money and Finance},
374 + volume={31},
375 + number={8},
376 + pages={2078--2101},
377 + year={2012},
378 + publisher={Elsevier}
379 +}
380 +
381 +@article{gurkaynak2005sensitivity,
382 + title={The sensitivity of long-term interest rates to economic news: Evidence and implications for macroeconomic models},
383 + author={G{\"u}rkaynak, Refet S and Sack, Brian and Swanson, Eric},
384 + journal={American Economic Review},
385 + volume={95},
386 + number={1},
387 + pages={425--436},
388 + year={2005}
389 +}
390 +
391 +@article{glosten1993relation,
392 + title={On the relation between the expected value and the volatility of the nominal excess return on stocks},
393 + author={Glosten, Lawrence R and Jagannathan, Ravi and Runkle, David E},
394 + journal={Journal of Finance},
395 + volume={48},
396 + number={5},
397 + pages={1779--1801},
398 + year={1993},
399 + publisher={Wiley Online Library}
400 +}
401 +
402 +@article{hakkio1985reaction,
403 + title={The reaction of exchange rates to economic news},
404 + author={Hakkio, Craig S and Pearce, Douglas K},
405 + journal={Economic Inquiry},
406 + volume={23},
407 + number={4},
408 + pages={621--636},
409 + year={1985},
410 + publisher={Wiley Online Library}
411 +}
412 +
413 +@article{hammoudeh2008metal,
414 + title={Metal volatility in presence of oil and interest rate shocks},
415 + author={Hammoudeh, Shawkat and Yuan, Yuan},
416 + journal={Energy Economics},
417 + volume={30},
418 + number={2},
419 + pages={606--620},
420 + year={2008},
421 + publisher={Elsevier}
422 +}
423 +
424 +
425 +@article{hu2017early,
426 + title={Early peek advantage? Efficient price discovery with tiered information disclosure},
427 + author={Hu, Grace Xing and Pan, Jun and Wang, Jiang},
428 + journal={Journal of Financial Economics},
429 + volume={126},
430 + number={2},
431 + pages={399--421},
432 + year={2017},
433 + publisher={Elsevier}
434 +}
435 +
436 +@article{irwin2012financialization,
437 + title={Financialization and structural change in commodity futures markets},
438 + author={Irwin, Scott H and Sanders, Dwight R},
439 + journal={Journal of Agricultural and Applied Economics},
440 + volume={44},
441 + number={1379-2016-113661},
442 + pages={371--396},
443 + year={2012}
444 +}
445 +
446 +@article{joets2017does,
447 + title={Does the volatility of commodity prices reflect macroeconomic uncertainty?},
448 + author={Jo{\"e}ts, Marc and Mignon, Val{\'e}rie and Razafindrabe, Tovonony},
449 + journal={Energy Economics},
450 + volume={68},
451 + pages={313--326},
452 + year={2017},
453 + publisher={Elsevier}
454 +}
455 +
456 +@article{kilian2014role,
457 + title={The role of inventories and speculative trading in the global market for crude oil},
458 + author={Kilian, Lutz and Murphy, Daniel P},
459 + journal={Journal of Applied econometrics},
460 + volume={29},
461 + number={3},
462 + pages={454--478},
463 + year={2014},
464 + publisher={Wiley Online Library}
465 +}
466 +
467 +@article{kurov2019price,
468 + title={Price drift before US macroeconomic news: Private information about public announcements?},
469 + author={Kurov, Alexander and Sancetta, Alessio and Strasser, Georg and Wolfe, Marketa Halova},
470 + journal={Journal of Financial and Quantitative Analysis},
471 + volume={54},
472 + number={1},
473 + pages={449--479},
474 + year={2019},
475 + publisher={Cambridge University Press}
476 +}
477 +
478 +@article{kurov2018monetary,
479 + title={Monetary policy uncertainty and the market reaction to macroeconomic news},
480 + author={Kurov, Alexander and Stan, Raluca},
481 + journal={Journal of Banking \& Finance},
482 + volume={86},
483 + pages={127--142},
484 + year={2018},
485 + publisher={Elsevier}
486 +}
487 +
488 +@article{lee1995oil,
489 + title={Oil shocks and the macroeconomy: the role of price variability},
490 + author={Lee, Kiseok and Ni, Shawn and Ratti, Ronald A},
491 + journal={The Energy Journal},
492 + volume={16},
493 + number={4},
494 + pages = {39--56},
495 + year={1995},
496 + publisher={International Association for Energy Economics}
497 +}
498 +
499 +@article{nelson1991conditional,
500 + title={Conditional heteroskedasticity in asset returns: A new approach},
501 + author={Nelson, Daniel B},
502 + journal={Econometrica},
503 + pages={347--370},
504 + year={1991},
505 + publisher={JSTOR}
506 +}
507 +
508 +@techreport{pearce1984stock,
509 + title={Stock prices and economic news},
510 + author={Pearce, Douglas K and Roley, V Vance},
511 + year={1984},
512 + institution={National Bureau of Economic Research}
513 +}
514 +
515 +@article{singleton2014investor,
516 + title={Investor flows and the 2008 boom/bust in oil prices},
517 + author={Singleton, Kenneth J},
518 + journal={Management Science},
519 + volume={60},
520 + number={2},
521 + pages={300--318},
522 + year={2014},
523 + publisher={INFORMS}
524 +}
525 +
526 +@article{stoll2010commodity,
527 + title={Commodity index investing and commodity futures prices},
528 + author={Stoll, Hans R and Whaley, Robert E},
529 + journal={Journal of Applied Finance (Formerly Financial Practice and Education)},
530 + volume={20},
531 + number={1},
532 + year={2010}
533 +}
534 +
535 +@article{savor2013much,
536 + title={How much do investors care about macroeconomic risk? Evidence from scheduled economic announcements},
537 + author={Savor, Pavel and Wilson, Mungo},
538 + journal={Journal of Financial and Quantitative Analysis},
539 + pages={343--375},
540 + year={2013},
541 + publisher={JSTOR}
542 +}
543 +
544 +@article{scholtus2014speed,
545 + title={Speed, algorithmic trading, and market quality around macroeconomic news announcements},
546 + author={Scholtus, Martin and Van Dijk, Dick and Frijns, Bart},
547 + journal={Journal of Banking \& Finance},
548 + volume={38},
549 + pages={89--105},
550 + year={2014},
551 + publisher={Elsevier}
552 +}
553 +
554 +@article{horan2004implied,
555 + title={Implied volatility of oil futures options surrounding {OPEC} meetings},
556 + author={Horan, Stephen M and Peterson, Jeffrey H and Mahar, James},
557 + journal={The Energy Journal},
558 + volume={25},
559 + number={3},
560 + pages = {103--125},
561 + year={2004},
562 + publisher={International Association for Energy Economics}
563 +}
564 +
565 +@article{tang2012index,
566 + title={Index investment and the financialization of commodities},
567 + author={Tang, Ke and Xiong, Wei},
568 + journal={Financial Analysts Journal},
569 + volume={68},
570 + number={6},
571 + pages={54--74},
572 + year={2012},
573 + publisher={Taylor \& Francis}
574 +}
575 +
576 +@article{vivian2012commodity,
577 + title={Commodity volatility breaks},
578 + author={Vivian, Andrew and Wohar, Mark E},
579 + journal={Journal of International Financial Markets, Institutions and Money},
580 + volume={22},
581 + number={2},
582 + pages={395--422},
583 + year={2012},
584 + publisher={Elsevier}
585 +}
586 +
587 +@article{working1949theory,
588 + title={The theory of price of storage},
589 + author={Working, Holbrook},
590 + journal={American Economic Review},
591 + volume={39},
592 + number={6},
593 + pages={1254--1262},
594 + year={1949},
595 + publisher={JSTOR}
596 +}
597 +
598 +@article{working1960speculation,
599 + title={Speculation on hedging markets},
600 + author={Working, Holbrook},
601 + journal={Food Research Institute Studies},
602 + volume={1},
603 + number={2},
604 + pages={185--220},
605 + year={1960}
606 +}
607 +
608 +@article{goldstein2022commodity,
609 + title={Commodity financialization and information transmission},
610 + author={Goldstein, Itay and Yang, Liyan},
611 + journal={The Journal of Finance},
612 + volume={77},
613 + number={5},
614 + pages={2613--2667},
615 + year={2022},
616 + publisher={Wiley Online Library}
617 +}
618 +
619 +@article{cheng2014financialization,
620 + title={Financialization of commodity markets},
621 + author={Cheng, Ing-Haw and Xiong, Wei},
622 + journal={Annual Review of Financial Economics},
623 + volume={6},
624 + number={1},
625 + pages={419--441},
626 + year={2014},
627 + publisher={Annual Reviews}
628 +}
629 +
630 +@incollection{kaldor1976speculation,
631 + title={Speculation and economic stability},
632 + author={Kaldor, Nicholas},
633 + booktitle={The Economics of Futures Trading},
634 + pages={111--123},
635 + year={1976},
636 + publisher={Springer}
637 +}
638 +
639 +
640 +
641 +@incollection{brennan1976supply,
642 + title={The supply of storage},
643 + author={Brennan, Michael J},
644 + booktitle={The Economics of Futures Trading},
645 + pages={100--107},
646 + year={1976},
647 + publisher={Springer}
648 +}
649 +
650 +@article{keynes1923some,
651 + title={Some aspects of commodity markets},
652 + author={Keynes, John Maynard},
653 + journal={Manchester Guardian Commercial: European Reconstruction Series},
654 + volume={13},
655 + pages={784--786},
656 + year={1923}
657 +}
658 +
659 +@article{hicks1975value,
660 + title={Value and capital: An inquiry into some fundamental principles of economic theory},
661 + author={Hicks, John Richard and others},
662 + journal={OUP Catalogue},
663 + year={1975},
664 + publisher={Oxford University Press}
665 +}
666 +
667 +
668 +@article{brennan1991price,
669 + title={The price of convenience and the valuation of commodity contingent claims,[w:] D. Land, B. Oeksendal},
670 + author={Brennan, MJ},
671 + journal={Stochastic Models and Options Values, Elsevier Science Publications},
672 + year={1991}
673 +}
674 +
675 +@book{pirrong2011commodity,
676 + title={Commodity price dynamics: A structural approach},
677 + author={Pirrong, Craig},
678 + year={2011},
679 + publisher={Cambridge University Press}
680 +}
681 +
682 +@article{kang2020tale,
683 + title={A tale of two premiums: the role of hedgers and speculators in commodity futures markets},
684 + author={Kang, Wenjin and Rouwenhorst, K Geert and Tang, Ke},
685 + journal={Journal of Finance},
686 + volume={75},
687 + number={1},
688 + pages={377--417},
689 + year={2020},
690 + publisher={Wiley Online Library}
691 +}
692 +
693 +@article{boyd2018update,
694 + title={An update on speculation and financialization in commodity markets},
695 + author={Boyd, Naomi E and Harris, Jeffrey H and Li, Bingxin},
696 + journal={Journal of Commodity Markets},
697 + volume={10},
698 + pages={91--104},
699 + year={2018},
700 + publisher={Elsevier}
701 +}
702 +
703 +@article{irwin2011index,
704 + title={Index funds, financialization, and commodity futures markets},
705 + author={Irwin, Scott H and Sanders, Dwight R},
706 + journal={Applied Economic Perspectives and Policy},
707 + volume={33},
708 + number={1},
709 + pages={1--31},
710 + year={2011},
711 + publisher={Oxford University Press}
712 +}
713 +
714 +@article{irwin2012financialization,
715 + title={Financialization and structural change in commodity futures markets},
716 + author={Irwin, Scott H and Sanders, Dwight R},
717 + journal={Journal of Agricultural and Applied Economics},
718 + volume={44},
719 + number={3},
720 + pages={371--396},
721 + year={2012},
722 + publisher={Cambridge University Press}
723 +}
724 +
725 +@article{irwin2012testing,
726 + title={Testing the {Masters} hypothesis in commodity futures markets},
727 + author={Irwin, Scott H and Sanders, Dwight R},
728 + journal={Energy Economics},
729 + volume={34},
730 + number={1},
731 + pages={256--269},
732 + year={2012},
733 + publisher={Elsevier}
734 +}
735 +
736 +
737 +@article{brunetti2014commodity,
738 + title={Commodity index trading and hedging costs},
739 + author={Brunetti, Celso and Reiffen, David},
740 + journal={Journal of Financial Markets},
741 + volume={21},
742 + pages={153--180},
743 + year={2014},
744 + publisher={Elsevier}
745 +}
746 +
747 +@article{hamilton2014risk,
748 + title={Risk premia in crude oil futures prices},
749 + author={Hamilton, James D and Wu, Jing Cynthia},
750 + journal={Journal of International Money and Finance},
751 + volume={42},
752 + pages={9--37},
753 + year={2014},
754 + publisher={Elsevier}
755 +}
756 +
757 +@article{fleming1997moves,
758 + title={What moves the bond market?},
759 + author={Fleming, Michael J and Remolona, Eli M},
760 + journal={Economic Policy Review},
761 + volume={3},
762 + number={4},
763 + year={1997}
764 +}
765 +
766 +@article{kilian2011energy,
767 + title={Do energy prices respond to US macroeconomic news? A test of the hypothesis of predetermined energy prices},
768 + author={Kilian, Lutz and Vega, Clara},
769 + journal={Review of Economics and Statistics},
770 + volume={93},
771 + number={2},
772 + pages={660--671},
773 + year={2011},
774 + publisher={The MIT Press}
775 +}
776 +
777 +@article{frankel1985commodity,
778 + title={Commodity prices, money surprises and fed credibilit},
779 + author={Frankel, Jeffrey A and Hardouvelis, Gikas A},
780 + journal={Journal of Money, Credit and Banking},
781 + volume={17},
782 + number={4},
783 + pages={425--438},
784 + year={1985},
785 + publisher={JSTOR}
786 +}
787 +
788 +
789 +@article{christie2000macroeconomics,
790 + title={Do macroeconomics news releases affect gold and silver prices?},
791 + author={Christie--David, Rohan and Chaudhry, Mukesh and Koch, Timothy W},
792 + journal={Journal of Economics and Business},
793 + volume={52},
794 + number={5},
795 + pages={405--421},
796 + year={2000},
797 + publisher={Elsevier}
798 +}
799 +
800 +@article{cai2001moves,
801 + title={What moves the gold market?},
802 + author={Cai, Jun and Cheung, Yan-Leung and Wong, Michael CS},
803 + journal={Journal of Futures Markets},
804 + volume={21},
805 + number={3},
806 + pages={257--278},
807 + year={2001},
808 + publisher={Wiley Online Library}
809 +}
810 +
811 +
812 +@article{hess2008commodity,
813 + title={How do commodity futures respond to macroeconomic news?},
814 + author={Hess, Dieter and Huang, He and Niessen, Alexandra},
815 + journal={Financial Markets and Portfolio Management},
816 + volume={22},
817 + number={2},
818 + pages={127--146},
819 + year={2008},
820 + publisher={Springer}
821 +}
822 +
823 +
824 +@article{gu2018drives,
825 + title={What drives informed trading before public releases? Evidence from natural gas inventory announcements},
826 + author={Gu, Chen and Kurov, Alexander},
827 + journal={Journal of Futures Markets},
828 + volume={38},
829 + number={9},
830 + pages={1079--1096},
831 + year={2018},
832 + publisher={Wiley Online Library}
833 +}
834 +
835 +@article{shanker2017new,
836 + title={New indices of adequate and excess speculation and their relationship with volatility in the crude oil futures market},
837 + author={Shanker, Latha},
838 + journal={Journal of Commodity Markets},
839 + volume={5},
840 + pages={18--35},
841 + year={2017},
842 + publisher={Elsevier}
843 +}
844 +
845 +@article{hedegaard2011margins,
846 + title={How margins are set and affect asset prices},
847 + author={Hedegaard, Esben},
848 + journal={Job Market Paper},
849 + year={2011}
850 +}
851 +
852 +@article{hammoudeh2008metal,
853 + title={Metal volatility in presence of oil and interest rate shocks},
854 + author={Hammoudeh, Shawkat and Yuan, Yuan},
855 + journal={Energy Economics},
856 + volume={30},
857 + number={2},
858 + pages={606--620},
859 + year={2008},
860 + publisher={Elsevier}
861 +}
862 +
863 +
864 +@article{hirshleifer1990hedging,
865 + title={Hedging pressure and futures price movements in a general equilibrium model},
866 + author={Hirshleifer, David},
867 + journal={Econometrica},
868 + pages={411--428},
869 + year={1990},
870 + publisher={JSTOR}
871 +}
872 +
873 +
874 +@article{hirshleifer1991seasonal,
875 + title={Seasonal patterns of futures hedging and the resolution of output uncertainty},
876 + author={Hirshleifer, David},
877 + journal={Journal of Economic Theory},
878 + volume={53},
879 + number={2},
880 + pages={304--327},
881 + year={1991},
882 + publisher={Elsevier}
883 +}
884 +
885 +
886 +@article{hordahl2020expectations,
887 + title={Expectations and risk premia at 8: 30 am: Deciphering the responses of bond yields to macroeconomic announcements},
888 + author={H{\"o}rdahl, Peter and Remolona, Eli M and Valente, Giorgio},
889 + journal={Journal of Business \& Economic Statistics},
890 + volume={38},
891 + number={1},
892 + pages={27--42},
893 + year={2020},
894 + publisher={Taylor \& Francis}
895 +}
896 +
897 +@article{wu2019does,
898 + title={Does gold or Bitcoin hedge economic policy uncertainty?},
899 + author={Wu, Shan and Tong, Mu and Yang, Zhongyi and Derbali, Abdelkader},
900 + journal={Finance Research Letters},
901 + volume={31},
902 + pages={171--178},
903 + year={2019},
904 + publisher={Elsevier}
905 +}
906 +
907 +@article{baur2010gold,
908 + title={Is gold a hedge or a safe haven? {An} analysis of stocks, bonds and gold},
909 + author={Baur, Dirk G and Lucey, Brian M},
910 + journal={Financial Review},
911 + volume={45},
912 + number={2},
913 + pages={217--229},
914 + year={2010},
915 + publisher={Wiley Online Library}
916 +}
917 +
918 +@article{hollstein2020volatility,
919 + title={Volatility term structures in commodity markets},
920 + author={Hollstein, Fabian and Prokopczuk, Marcel and W{\"u}rsig, Christoph},
921 + journal={Journal of Futures Markets},
922 + volume={40},
923 + number={4},
924 + pages={527--555},
925 + year={2020},
926 + publisher={Wiley Online Library}
927 +}
928 +
929 +@article{ye2021macroeconomic,
930 + title={Macroeconomic forecasts and commodity futures volatility},
931 + author={Ye, Wuyi and Guo, Ranran and Deschamps, Bruno and Jiang, Ying and Liu, Xiaoquan},
932 + journal={Economic Modelling},
933 + volume={94},
934 + pages={981--994},
935 + year={2021},
936 + publisher={Elsevier}
937 +}
938 +
939 +@article{bohl2013does,
940 + title={Does futures speculation destabilize spot prices? New evidence for commodity markets},
941 + author={Bohl, Martin T and Stephan, Patrick M},
942 + journal={Journal of Agricultural and Applied Economics},
943 + volume={45},
944 + number={4},
945 + pages={595--616},
946 + year={2013},
947 + publisher={Cambridge University Press}
948 +}
949 +
950 +@article{bohl2012commodity,
951 + title={Do commodity index traders destabilize agricultural futures prices?},
952 + author={Bohl, Martin T and Javed, Farrukh and Stephan, Patrick M and others},
953 + journal={Diskussionspapier, am},
954 + volume={24},
955 + year={2012},
956 + publisher={Citeseer}
957 +}
958 +
959 +@article{brunetti2009speculation,
960 + title={Is speculation destabilizing?},
961 + author={Brunetti, Celso and Buyuksahin, Bahattin},
962 + journal={SSRN eLibrary},
963 + year={2009},
964 + publisher={Citeseer}
965 +}
966 +
967 +@article{harris2009role,
968 + title={The role of speculators in the crude oil futures market},
969 + author={Harris, Jeffrey H and Buyuksahin, Bahattin},
970 + journal={Available at SSRN 1435042},
971 + year={2009}
972 +}
973 +
974 +@article{irwin2012testing,
975 + title={Testing the Masters Hypothesis in commodity futures markets},
976 + author={Irwin, Scott H and Sanders, Dwight R},
977 + journal={Energy Economics},
978 + volume={34},
979 + number={1},
980 + pages={256--269},
981 + year={2012},
982 + publisher={Elsevier}
983 +}
984 +
985 +@article{sanders2010speculative,
986 + title={A speculative bubble in commodity futures prices? Cross-sectional evidence},
987 + author={Sanders, Dwight R and Irwin, Scott H},
988 + journal={Agricultural Economics},
989 + volume={41},
990 + number={1},
991 + pages={25--32},
992 + year={2010},
993 + publisher={Wiley Online Library}
994 +}
995 +
996 +
997 +@article{irwin2012testing,
998 + title={Testing the Masters Hypothesis in commodity futures markets},
999 + author={Irwin, Scott H and Sanders, Dwight R},
1000 + journal={Energy Economics},
1001 + volume={34},
1002 + number={1},
1003 + pages={256--269},
1004 + year={2012},
1005 + publisher={Elsevier}
1006 +}
1007 +
1008 +@article{kim2015does,
1009 + title={Does futures speculation destabilize commodity markets?},
1010 + author={Kim, Abby},
1011 + journal={Journal of Futures Markets},
1012 + volume={35},
1013 + number={8},
1014 + pages={696--714},
1015 + year={2015},
1016 + publisher={Wiley Online Library}
1017 +}
1018 +
1019 +@article{yang2005futures,
1020 + title={Futures trading activity and commodity cash price volatility},
1021 + author={Yang, Jian and Balyeat, R Brian and Leatham, David J},
1022 + journal={Journal of Business Finance \& Accounting},
1023 + volume={32},
1024 + number={1-2},
1025 + pages={297--323},
1026 + year={2005},
1027 + publisher={Wiley Online Library}
1028 +}
1029 +
1030 +@article{bryant2006causality,
1031 + title={Causality in futures markets},
1032 + author={Bryant, Henry L and Bessler, David A and Haigh, Michael S},
1033 + journal={Journal of Futures Markets},
1034 + volume={26},
1035 + number={11},
1036 + pages={1039--1057},
1037 + year={2006},
1038 + publisher={Wiley Online Library}
1039 +}
1040 +
1041 +@article{masters2009testimony,
1042 + title={Testimony before the commodity futures trading commission},
1043 + author={Masters, Michael W},
1044 + journal={Testimony to the Commodity Futures Trading Commission},
1045 + year={2009}
1046 +}
1047 +
1048 +@article{hu2013noise,
1049 + title={Noise as information for illiquidity},
1050 + author={Hu, Grace Xing and Pan, Jun and Wang, Jiang},
1051 + journal={Journal of Finance},
1052 + volume={68},
1053 + number={6},
1054 + pages={2341--2382},
1055 + year={2013},
1056 + publisher={Wiley Online Library}
1057 +}
1058 +
1059 +@article{hautsch2011impact,
1060 + title={The impact of macroeconomic news on quote adjustments, noise, and informational volatility},
1061 + author={Hautsch, Nikolaus and Hess, Dieter and Veredas, David},
1062 + journal={Journal of Banking \& Finance},
1063 + volume={35},
1064 + number={10},
1065 + pages={2733--2746},
1066 + year={2011},
1067 + publisher={Elsevier}
1068 +}
1069 +
1070 +@article{bernile2016can,
1071 + title={Can information be locked up? Informed trading ahead of macro-news announcements},
1072 + author={Bernile, Gennaro and Hu, Jianfeng and Tang, Yuehua},
1073 + journal={Journal of Financial Economics},
1074 + volume={121},
1075 + number={3},
1076 + pages={496--520},
1077 + year={2016},
1078 + publisher={Elsevier}
1079 +}
1080 +
1081 +@article{alquist2013role,
1082 + title={The role of financial speculation in driving the price of crude oil},
1083 + author={Alquist, Ron and Gervais, Olivier},
1084 + journal={The Energy Journal},
1085 + volume={34},
1086 + number={3},
1087 + pages={35-54},
1088 + year={2013},
1089 + publisher={International Association for Energy Economics}
1090 +}
1091 +
1092 +@incollection{kothari2007econometrics,
1093 + title={Econometrics of event studies},
1094 + author={Kothari, Sagar P and Warner, Jerold B},
1095 + booktitle={Handbook of Empirical Corporate Finance},
1096 + pages={3--36},
1097 + year={2007},
1098 + publisher={Elsevier}
1099 +}
1100 +
1101 +@article{buyukcsahin2010matters,
1102 + title={It matters who trades: Hedge funds, swap dealers, and cross-market linkages},
1103 + author={B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Robe, Michel A},
1104 + year={2010}
1105 +}
1106 +
1107 +@article{kilian2014role,
1108 + title={The role of inventories and speculative trading in the global market for crude oil},
1109 + author={Kilian, Lutz and Murphy, Daniel P},
1110 + journal={Journal of Applied econometrics},
1111 + volume={29},
1112 + number={3},
1113 + pages={454--478},
1114 + year={2014},
1115 + publisher={Wiley Online Library}
1116 +}
1117 +
1118 +@misc{newey1986simple,
1119 + title={A simple, positive semi-definite, heteroskedasticity and autocorrelationconsistent covariance matrix},
1120 + author={Newey, Whitney K and West, Kenneth D},
1121 + year={1986},
1122 + publisher={National Bureau of Economic Research, Cambridge, Mass., USA}
1123 +}
1124 +
1125 +@article{andersen2008realized,
1126 + title={Realized Volatility, Working Paper 2008-14},
1127 + author={Andersen, Torben Gustav and Benzoni, Luca},
1128 + year={2008}
1129 +}
1130 +
1131 +@article{liu2015does,
1132 + title={Does anything beat 5-minute RV? A comparison of realized measures across multiple asset classes},
1133 + author={Liu, Lily Y and Patton, Andrew J and Sheppard, Kevin},
1134 + journal={Journal of Econometrics},
1135 + volume={187},
1136 + number={1},
1137 + pages={293--311},
1138 + year={2015},
1139 + publisher={Elsevier}
1140 +}
1141 +
1142 +@article{andersen2001distribution,
1143 + title={The distribution of realized stock return volatility},
1144 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X and Ebens, Heiko},
1145 + journal={Journal of Financial Economics},
1146 + volume={61},
1147 + number={1},
1148 + pages={43--76},
1149 + year={2001},
1150 + publisher={Elsevier}
1151 +}
1152 +
1153 +@article{barndorff2002estimating,
1154 + title={Estimating quadratic variation using realized variance},
1155 + author={Barndorff-Nielsen, Ole E and Shephard, Neil},
1156 + journal={Journal of Applied Econometrics},
1157 + volume={17},
1158 + number={5},
1159 + pages={457--477},
1160 + year={2002},
1161 + publisher={Wiley Online Library}
1162 +}
1163 +
1164 +
1165 +@article{gospodinov2012effects,
1166 + title={The effects of Federal funds rate surprises on S\&P 500 volatility and volatility risk premium},
1167 + author={Gospodinov, Nikolay and Jamali, Ibrahim},
1168 + journal={Journal of Empirical Finance},
1169 + volume={19},
1170 + number={4},
1171 + pages={497--510},
1172 + year={2012},
1173 + publisher={Elsevier}
1174 +}
1175 +
1176 +@article{chang1997interday,
1177 + title={Interday variations in volume, variance and participation of large speculators},
1178 + author={Chang, Eric C and Pinegar, J Michael and Schachter, Barry},
1179 + journal={Journal of Banking \& Finance},
1180 + volume={21},
1181 + number={6},
1182 + pages={797--810},
1183 + year={1997},
1184 + publisher={Elsevier}
1185 +}
1186 +
1187 +@article{daigler1999impact,
1188 + title={The impact of trader type on the futures volatility-volume relation},
1189 + author={Daigler, Robert T and Wiley, Marilyn K},
1190 + journal={Journal of Finance},
1191 + volume={54},
1192 + number={6},
1193 + pages={2297--2316},
1194 + year={1999},
1195 + publisher={Wiley Online Library}
1196 +}
1197 +
1198 +
1199 +@incollection{irwin2004effect,
1200 +author = {Irwin, Scott H and Holt, Bryce},
1201 +title = {The effect of large hedge fund and {CTA} trading on futures market volatility},
1202 +booktitle = {Commodity trading advisors: Risk, performance analysis and selection},
1203 + publisher={John Wiley and Sons, Inc},
1204 +address = {New York, NY},
1205 +year = {2004}
1206 +}
1207 +
1208 +
1209 +@article{irwin1987note,
1210 + title={A note on the factors affecting technical trading system returns},
1211 + author={Irwin, Scott H and Brorsen, B Wade},
1212 + journal={Journal of Futures Markets},
1213 + volume={7},
1214 + number={5},
1215 + pages={591--595},
1216 + year={1987},
1217 + publisher={Wiley Subscription Services, Inc., A Wiley Company New York}
1218 +}
1219 +
1220 +@article{irwin1999managed,
1221 + title={Managed futures, positive feedback trading, and futures price volatility},
1222 + author={Irwin, Scott H and Yoshimaru, Satoko},
1223 + journal={Journal of Futures Markets},
1224 + volume={19},
1225 + number={7},
1226 + pages={759--776},
1227 + year={1999},
1228 + publisher={Wiley Online Library}
1229 +}
1230 +
1231 +@article{bryant2006causality,
1232 + title={Causality in futures markets},
1233 + author={Bryant, Henry L and Bessler, David A and Haigh, Michael S},
1234 + journal={Journal of Futures Markets},
1235 + volume={26},
1236 + number={11},
1237 + pages={1039--1057},
1238 + year={2006},
1239 + publisher={Wiley Online Library}
1240 +}
1241 +
1242 +@article{haigh2007hedge,
1243 + title={Hedge funds, volatility, and liquidity provision in energy futures markets},
1244 + author={Haigh, Michael S and Hranaiova, Jana and Overdahl, James A},
1245 + journal={Journal of Alternative Investments},
1246 + volume={9},
1247 + number={4},
1248 + pages={10--38},
1249 + year={2007},
1250 + publisher={Institutional Investor Journals Umbrella}
1251 +}
1252 +
1253 +@inproceedings{aulerich2012bubbles,
1254 + title={Bubbles},
1255 + author={Aulerich, Nicole M and Irwin, Scott H and Garcia, Philip},
1256 + booktitle={Food Prices, and Speculation: Evidence from the CFTC’s Daily Large Trader Data Files, Paper prepared for presentation at the NBER Conference on “Economics of Food Price Volatility” in Seattle, WA},
1257 + year={2012}
1258 +}
1259 +
1260 +@article{irwin2009devil,
1261 + title={Devil or angel? The role of speculation in the recent commodity price boom (and bust)},
1262 + author={Irwin, Scott H and Sanders, Dwight R and Merrin, Robert P},
1263 + journal={Journal of Agricultural and Applied Economics},
1264 + volume={41},
1265 + number={2},
1266 + pages={377--391},
1267 + year={2009},
1268 + publisher={Cambridge University Press}
1269 +}
1270 +
1271 +@techreport{cheng2012convective,
1272 + title={Convective risk flows in commodity futures markets},
1273 + author={Cheng, Haw and Kirilenko, Andrei and Xiong, Wei},
1274 + year={2012},
1275 + institution={National Bureau of Economic Research}
1276 +}
1277 +
1278 +@article{cheng2015convective,
1279 + title={Convective risk flows in commodity futures markets},
1280 + author={Cheng, Ing-Haw and Kirilenko, Andrei and Xiong, Wei},
1281 + journal={Review of Finance},
1282 + volume={19},
1283 + number={5},
1284 + pages={1733--1781},
1285 + year={2015},
1286 + publisher={Oxford University Press}
1287 +}
1288 +
1289 +@article{goldstein2014speculation,
1290 + title={Speculation and hedging in segmented markets},
1291 + author={Goldstein, Itay and Li, Yan and Yang, Liyan},
1292 + journal={Review of Financial Studies},
1293 + volume={27},
1294 + number={3},
1295 + pages={881--922},
1296 + year={2014},
1297 + publisher={Oxford University Press}
1298 +}
1299 +
1300 +
1301 +@article{knuth2018breakthroughs,
1302 + title={“Breakthroughs” for a green economy? Financialization and clean energy transition},
1303 + author={Knuth, Sarah},
1304 + journal={Energy Research \& Social Science},
1305 + volume={41},
1306 + pages={220--229},
1307 + year={2018},
1308 + publisher={Elsevier}
1309 +}
1310 +
1311 +@article{kellogg2014effect,
1312 + title={The effect of uncertainty on investment: evidence from Texas oil drilling},
1313 + author={Kellogg, Ryan},
1314 + journal={American Economic Review},
1315 + volume={104},
1316 + number={6},
1317 + pages={1698--1734},
1318 + year={2014}
1319 +}
1320 +
1321 +
1322 +
1323 +@article{almutairi2021value,
1324 + author={Almutairi, Hossa and Pierru, Axel and Smith, James A},
1325 + doi={10.1111/opec.12199},
1326 + title={The Value of OPEC’s Spare Capacity to the Oil Market and Global Economy},
1327 + journal={OPEC Energy Review},
1328 + year={2021}
1329 +}
1330 +
1331 +@article{alquist2013role,
1332 + title = {The Role of Financial Speculation in Driving the Price of Crude Oil},
1333 + author = {Alquist, Ron and Gervais, Olivier},
1334 + journal = {The Energy Journal},
1335 + volume = {34},
1336 + number = {3},
1337 + pages = {35--54},
1338 + year = {2013},
1339 + publisher = {International Association for Energy Economics}
1340 +}
1341 +
1342 +@article{andersen1998deutsche,
1343 + title = {Deutsche Mark--Dollar Volatility: Intraday Activity Patterns, Macroeconomic Announcements, and Longer Run Dependencies},
1344 + author = {Andersen, Torben G. and Bollerslev, Tim},
1345 + journal = {Journal of Finance},
1346 + volume = {53},
1347 + number = {1},
1348 + pages = {219--265},
1349 + year = {1998},
1350 + publisher = {Wiley Online Library}
1351 +}
1352 +
1353 +@article{andersen2001distribution,
1354 + title = {The Distribution of Realized Stock Return Volatility},
1355 + author = {Andersen, Torben G. and Bollerslev, Tim and Diebold, Francis X. and Ebens, Heiko},
1356 + journal = {Journal of Financial Economics},
1357 + volume = {61},
1358 + number = {1},
1359 + pages = {43--76},
1360 + year = {2001},
1361 + publisher = {Elsevier}
1362 +}
1363 +
1364 +@article{andersen2007real,
1365 + title = {Real-Time Price Discovery in Global Stock, Bond and Foreign Exchange Markets},
1366 + author = {Andersen, Torben G. and Bollerslev, Tim and Diebold, Francis X. and Vega, Clara},
1367 + journal = {Journal of International Economics},
1368 + volume = {73},
1369 + number = {2},
1370 + pages = {251--277},
1371 + year = {2007},
1372 + publisher = {Elsevier}
1373 +}
1374 +
1375 +@article{aruoba2009real,
1376 + title = {Real-Time Measurement of Business Conditions},
1377 + author = {Aruoba, S. Borağan and Diebold, Francis X. and Scotti, Chiara},
1378 + journal = {Journal of Business \& Economic Statistics},
1379 + volume = {27},
1380 + number = {4},
1381 + pages = {417-427},
1382 + year = {2009},
1383 + publisher = {Taylor \& Francis},
1384 + doi = {10.1198/jbes.2009.07205},
1385 + url = {https://doi.org/10.1198/jbes.2009.07205}
1386 +}
1387 +
1388 +@inproceedings{aulerich2012bubbles,
1389 + title = {Bubbles},
1390 + author = {Aulerich, Nicole M. and Irwin, Scott H. and Garcia, Philip},
1391 + booktitle = {Food Prices, and Speculation: Evidence from the CFTC’s Daily Large Trader Data Files, Paper Prepared for Presentation at the NBER Conference on “Economics of Food Price Volatility” in Seattle, WA},
1392 + year = {2012}
1393 +}
1394 +
1395 +@article{balduzzi2001economic,
1396 + title = {Economic News and Bond Prices: Evidence from the US Treasury Market},
1397 + author = {Balduzzi, Pierluigi and Elton, Edwin J. and Green, T. Clifton},
1398 + journal = {Journal of Financial and Quantitative Analysis},
1399 + volume = {36},
1400 + number = {4},
1401 + pages = {523--543},
1402 + year = {2001},
1403 + publisher = {JSTOR}
1404 +}
1405 +
1406 +@article{barndorff2002estimating,
1407 + title = {Estimating Quadratic Variation Using Realized Variance},
1408 + author = {Barndorff-Nielsen, Ole E. and Shephard, Neil},
1409 + journal = {Journal of Applied Econometrics},
1410 + volume = {17},
1411 + number = {5},
1412 + pages = {457--477},
1413 + year = {2002},
1414 + publisher = {Wiley Online Library}
1415 +}
1416 +
1417 +@article{basak2016model,
1418 + title = {A Model of Financialization of Commodities},
1419 + author = {Basak, Suleyman and Pavlova, Anna},
1420 + journal = {Journal of Finance},
1421 + volume = {71},
1422 + number = {4},
1423 + pages = {1511--1556},
1424 + year = {2016},
1425 + publisher = {Wiley Online Library}
1426 +}
1427 +
1428 +@article{baur2010gold,
1429 + title = {Is Gold a Hedge or a Safe Haven? {An} Analysis of Stocks, Bonds and Gold},
1430 + author = {Baur, Dirk G. and Lucey, Brian M.},
1431 + journal = {Financial Review},
1432 + volume = {45},
1433 + number = {2},
1434 + pages = {217--229},
1435 + year = {2010},
1436 + publisher = {Wiley Online Library}
1437 +}
1438 +
1439 +@article{bernile2016can,
1440 + title = {Can Information Be Locked Up? Informed Trading Ahead of Macro-News Announcements},
1441 + author = {Bernile, Gennaro and Hu, Jianfeng and Tang, Yuehua},
1442 + journal = {Journal of Financial Economics},
1443 + volume = {121},
1444 + number = {3},
1445 + pages = {496--520},
1446 + year = {2016},
1447 + publisher = {Elsevier}
1448 +}
1449 +
1450 +@book{brennan1991price,
1451 + title = {The Price of Convenience and the Valuation of Commodity Contingent Claims},
1452 + author = {Brennan, Michael J.},
1453 + year = {1991},
1454 + publisher = {Elsevier Science Publications}
1455 +}
1456 +
1457 +@incollection{brennan1976supply,
1458 + title = {The Supply of Storage},
1459 + author = {Brennan, Michael J.},
1460 + booktitle = {The Economics of Futures Trading},
1461 + pages = {100--107},
1462 + year = {1976},
1463 + publisher = {Springer}
1464 +}
1465 +
1466 +@article{bredin2015does,
1467 + title = {Does Gold Glitter in the Long-Run? Gold as a Hedge and Safe Haven Across Time and Investment Horizon},
1468 + author = {Bredin, Don and Conlon, Thomas and Pot{\`\i}, Valerio},
1469 + journal = {International Review of Financial Analysis},
1470 + volume = {41},
1471 + pages = {320--328},
1472 + year = {2015},
1473 + publisher = {Elsevier}
1474 +}
1475 +
1476 +@article{brunetti2009speculation,
1477 + title = {Is Speculation Destabilizing?},
1478 + author = {Brunetti, Celso and B{\"u}y{\"u}k{\c{s}}ahin, Bahattin},
1479 + journal = {SSRN eLibrary},
1480 + year = {2009},
1481 + publisher = {Citeseer}
1482 +}
1483 +
1484 +@article{brunetti2011speculators,
1485 + author = {Brunetti, Celso and B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Harris, Jeffrey},
1486 + title = {Speculators, Prices and Market Volatility},
1487 + journal = {SSRN Electronic Journal},
1488 + year = {2011},
1489 + doi = {10.2139/ssrn.1736737}
1490 +}
1491 +
1492 +@article{brunetti2014commodity,
1493 + title = {Commodity Index Trading and Hedging Costs},
1494 + author = {Brunetti, Celso and Reiffen, David},
1495 + journal = {Journal of Financial Markets},
1496 + volume = {21},
1497 + pages = {153--180},
1498 + year = {2014},
1499 + publisher = {Elsevier}
1500 +}
1501 +
1502 +@article{brunetti2016speculators,
1503 + title = {Speculators, Prices, and Market Volatility},
1504 + author = {Brunetti, Celso and B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Harris, Jeffrey H.},
1505 + journal = {Journal of Financial and Quantitative Analysis},
1506 + volume = {51},
1507 + number = {5},
1508 + pages = {1545--1574},
1509 + year = {2016},
1510 + doi = {10.1017/s0022109016000569},
1511 + publisher = {JSTOR}
1512 +}
1513 +
1514 +@article{bryant2006causality,
1515 + title = {Causality in Futures Markets},
1516 + author = {Bryant, Henry L. and Bessler, David A. and Haigh, Michael S.},
1517 + journal = {Journal of Futures Markets},
1518 + volume = {26},
1519 + number = {11},
1520 + pages = {1039--1057},
1521 + year = {2006},
1522 + publisher = {Wiley Online Library}
1523 +}
1524 +
1525 +@article{buyukcsahin2010matters,
1526 + title = {It Matters Who Trades: Hedge Funds, Swap Dealers, and Cross-Market Linkages},
1527 + author = {B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Robe, Michel A.},
1528 + journal = {SSRN eLibrary},
1529 + year = {2010}
1530 +}
1531 +
1532 +@article{cai2001moves,
1533 + title = {What Moves the Gold Market?},
1534 + author = {Cai, Jun and Cheung, Yan-Leung and Wong, Michael C. S.},
1535 + journal = {Journal of Futures Markets},
1536 + volume = {21},
1537 + number = {3},
1538 + pages = {257--278},
1539 + year = {2001},
1540 + publisher = {Wiley Online Library}
1541 +}
1542 +
1543 +@article{chang1997interday,
1544 + title = {Interday Variations in Volume, Variance and Participation of Large Speculators},
1545 + author = {Chang, Eric C. and Pinegar, J. Michael and Schachter, Barry},
1546 + journal = {Journal of Banking \& Finance},
1547 + volume = {21},
1548 + number = {6},
1549 + pages = {797--810},
1550 + year = {1997},
1551 + publisher = {Elsevier}
1552 +}
1553 +
1554 +@article{cheng2014financialization,
1555 + title = {Financialization of Commodity Markets},
1556 + author = {Cheng, Ing-Haw and Xiong, Wei},
1557 + journal = {Annual Review of Financial Economics},
1558 + volume = {6},
1559 + number = {1},
1560 + pages = {419--441},
1561 + year = {2014},
1562 + publisher = {Annual Reviews}
1563 +}
1564 +
1565 +@techreport{cheng2012convective,
1566 + title = {Convective Risk Flows in Commodity Futures Markets},
1567 + author = {Cheng, Ing-Haw and Kirilenko, Andrei and Xiong, Wei},
1568 + year = {2012},
1569 + institution = {National Bureau of Economic Research}
1570 +}
1571 +
1572 +@article{cheng2015convective,
1573 + title = {Convective Risk Flows in Commodity Futures Markets},
1574 + author = {Cheng, Ing-Haw and Kirilenko, Andrei and Xiong, Wei},
1575 + journal = {Review of Finance},
1576 + volume = {19},
1577 + number = {5},
1578 + pages = {1733--1781},
1579 + year = {2015},
1580 + publisher = {Oxford University Press}
1581 +}
1582 +
1583 +@article{christie2000macroeconomics,
1584 + title = {Do Macroeconomics News Releases Affect Gold and Silver Prices?},
1585 + author = {Christie--David, Rohan and Chaudhry, Mukesh and Koch, Timothy W.},
1586 + journal = {Journal of Economics and Business},
1587 + volume = {52},
1588 + number = {5},
1589 + pages = {405--421},
1590 + year = {2000},
1591 + publisher = {Elsevier}
1592 +}
1593 +
1594 +@article{daigler1999impact,
1595 + title = {The Impact of Trader Type on the Futures Volatility-Volume Relation},
1596 + author = {Daigler, Robert T. and Wiley, Marilyn K.},
1597 + journal = {Journal of Finance},
1598 + volume = {54},
1599 + number = {6},
1600 + pages = {2297--2316},
1601 + year = {1999},
1602 + publisher = {Wiley Online Library}
1603 +}
1604 +
1605 +@article{erb2013golden,
1606 + title = {The Golden Dilemma},
1607 + author = {Erb, Claude B and Harvey, Campbell R},
1608 + journal = {Financial Analysts Journal},
1609 + volume = {69},
1610 + number = {4},
1611 + pages = {10--42},
1612 + year = {2013},
1613 + publisher = {Taylor \& Francis}
1614 +}
1615 +
1616 +@article{fishe2012identifying,
1617 + title = {Identifying Informed Traders in Futures Markets},
1618 + author = {Fishe, Raymond P.H. and Smith, Aaron},
1619 + journal = {Journal of Financial Markets},
1620 + volume = {15},
1621 + number = {3},
1622 + pages = {329--359},
1623 + year = {2012},
1624 + url = {https://files.asmith.ucdavis.edu/2012_JFinM_FS_informed.pdf},
1625 + keywords = {finance},
1626 + abstract = {We use daily positions of futures market participants to identify informed traders. These data contain 8,921 unique traders. We identify between 94 and 230 traders as overnight informed and 91 as intraday informed with little overlap. Floor brokers/traders are over-represented in the overnight informed group. The intraday informed group is dominated by managed money traders/hedge funds and swap dealers, with commercial hedgers under-represented. We find that characteristics such as experience, position size, trading activity, and type of positions held offer significant predictive power for who is informed. An analysis of daily trader profits confirms that we select highly profitable traders.},
1627 + publisher = {North-Holland}
1628 +}
1629 +
1630 +@article{flemming1997moves,
1631 + title = {What Moves the Bond Market?},
1632 + author = {Flemming, Michael J. and Remolona, Eli M.},
1633 + journal = {Economic Policy Review},
1634 + volume = {3},
1635 + number = {4},
1636 + year = {1997}
1637 +}
1638 +
1639 +@article{frankel1985commodity,
1640 + title = {Commodity Prices, Money Surprises and Fed Credibility},
1641 + author = {Frankel, Jeffrey A. and Hardouvelis, Gikas A.},
1642 + journal = {Journal of Money, Credit and Banking},
1643 + volume = {17},
1644 + number = {4},
1645 + pages = {425--438},
1646 + year = {1985},
1647 + publisher = {JSTOR}
1648 +}
1649 +
1650 +@article{garbade1983price,
1651 + title = {Price Movements and Price Discovery in Futures and Cash Markets},
1652 + author = {Garbade, Kenneth D. and Silber, William L.},
1653 + journal = {Review of Economics and Statistics},
1654 + volume = {65},
1655 + number = {2},
1656 + pages = {289--297},
1657 + year = {1983},
1658 + publisher = {JSTOR}
1659 +}
1660 +
1661 +@article{galkin2018opec,
1662 + author = {Galkin, Philipp and Atalla, Tarek and Ren, Zhongyuan},
1663 + doi = {10.30573/ks--2018-dp38},
1664 + title = {An Estimation of the Drivers Behind OPEC’s Quota Decisions},
1665 + journal = {King Abdullah Petroleum Studies and Research Center Discussion Paper},
1666 + year = {2018}
1667 +}
1668 +
1669 +@article{gebarowski2015competition,
1670 + author = {Gębarowski, Robert and Drożdż, Stanisław and Górski, Arkadiusz and Oświęcimka, Paweł},
1671 + title = {Competition of Commodities for the Status of Money in an Agent-Based Model},
1672 + journal = {Acta Physica Polonica A},
1673 + volume = {127},
1674 + number = {3a},
1675 + pages = {A-51--A-54},
1676 + year = {2015},
1677 + doi = {10.12693/aphyspola.127.a-51}
1678 +}
1679 +
1680 +@techreport{goldberg2013time,
1681 + title = {Time Variation in Asset Price Responses to Macro Announcements},
1682 + author = {Goldberg, Linda S. and Grisse, Christian},
1683 + institution = {National Bureau of Economic Research},
1684 + type = {Working Paper},
1685 + series = {Working Paper Series},
1686 + number = {19523},
1687 + year = {2013},
1688 + month = {October},
1689 + doi = {10.3386/w19523},
1690 + url = {http://www.nber.org/papers/w19523}
1691 +}
1692 +
1693 +@article{goldstein2014speculation,
1694 + title = {Speculation and Hedging in Segmented Markets},
1695 + author = {Goldstein, Itay and Li, Yan and Yang, Liyan},
1696 + journal = {Review of Financial Studies},
1697 + volume = {27},
1698 + number = {3},
1699 + pages = {881--922},
1700 + year = {2014},
1701 + publisher = {Oxford University Press}
1702 +}
1703 +
1704 +@article{goldstein2022commodity,
1705 + title = {Commodity Financialization and Information Transmission},
1706 + author = {Goldstein, Itay and Yang, Liyan},
1707 + journal = {The Journal of Finance},
1708 + volume = {77},
1709 + number = {5},
1710 + pages = {2613--2667},
1711 + year = {2022},
1712 + publisher = {Wiley Online Library}
1713 +}
1714 +
1715 +@article{gospodinov2012effects,
1716 + title = {The Effects of Federal Funds Rate Surprises on S\&P 500 Volatility and Volatility Risk Premium},
1717 + author = {Gospodinov, Nikolay and Jamali, Ibrahim},
1718 + journal = {Journal of Empirical Finance},
1719 + volume = {19},
1720 + number = {4},
1721 + pages = {497--510},
1722 + year = {2012},
1723 + publisher = {Elsevier}
1724 +}
1725 +
1726 +@article{graicer2021opec,
1727 + author = {Graicer, Rafael and Nery, Thiago Affonso and Jucá, Michele Nascimento and Junior, Eli Hadad and Bastos, Douglas Dias},
1728 + doi = {10.5902/1983465963575},
1729 + title = {The Impact of the OPEC Announcement on the Oil Companies' Share Prices},
1730 + journal = {Revista De Administração Da Ufsm},
1731 + year = {2021}
1732 +}
1733 +
1734 +@article{grossman1980impossibility,
1735 + title = {On the Impossibility of Informationally Efficient Markets},
1736 + author = {Grossman, Sanford J and Stiglitz, Joseph E},
1737 + journal = {American Economic Review},
1738 + volume = {70},
1739 + number = {3},
1740 + pages = {393--408},
1741 + year = {1980},
1742 + publisher = {JSTOR}
1743 +}
1744 +
1745 +@article{gu2018drives,
1746 + title = {What Drives Informed Trading Before Public Releases? Evidence from Natural Gas Inventory Announcements},
1747 + author = {Gu, Chen and Kurov, Alexander},
1748 + journal = {Journal of Futures Markets},
1749 + volume = {38},
1750 + number = {9},
1751 + pages = {1079--1096},
1752 + year = {2018},
1753 + publisher = {Wiley Online Library}
1754 +}
1755 +
1756 +@article{hamilton2014risk,
1757 + title = {Risk Premia in Crude Oil Futures Prices},
1758 + author = {Hamilton, James D. and Wu, Jing Cynthia},
1759 + journal = {Journal of International Money and Finance},
1760 + volume = {42},
1761 + pages = {9--37},
1762 + year = {2014},
1763 + publisher = {Elsevier}
1764 +}
1765 +
1766 +@article{hautsch2011impact,
1767 + title = {The Impact of Macroeconomic News on Quote Adjustments, Noise, and Informational Volatility},
1768 + author = {Hautsch, Nikolaus and Hess, Dieter and Veredas, David},
1769 + journal = {Journal of Banking \& Finance},
1770 + volume = {35},
1771 + number = {10},
1772 + pages = {2733--2746},
1773 + year = {2011},
1774 + publisher = {Elsevier}
1775 +}
1776 +
1777 +@article{hedge2007funds,
1778 + title = {Hedge Funds, Volatility, and Liquidity Provision in Energy Futures Markets},
1779 + author = {Haigh, Michael S. and Hranaiova, Jana and Overdahl, James A.},
1780 + journal = {Journal of Alternative Investments},
1781 + volume = {9},
1782 + number = {4},
1783 + pages = {10--38},
1784 + year = {2007},
1785 + publisher = {Institutional Investor Journals Umbrella}
1786 +}
1787 +
1788 +@article{hedegaard2011margins,
1789 + title = {How Margins Are Set and Affect Asset Prices},
1790 + author = {Hedegaard, Esben},
1791 + journal = {Job Market Paper},
1792 + year = {2011}
1793 +}
1794 +
1795 +@article{hess2008commodity,
1796 + title = {How Do Commodity Futures Respond to Macroeconomic News?},
1797 + author = {Hess, Dieter and Huang, He and Niessen, Alexandra},
1798 + journal = {Financial Markets and Portfolio Management},
1799 + volume = {22},
1800 + number = {2},
1801 + pages = {127--146},
1802 + year = {2008},
1803 + publisher = {Springer}
1804 +}
1805 +
1806 +@article{hicks1975value,
1807 + title = {Value and Capital: An Inquiry into Some Fundamental Principles of Economic Theory},
1808 + author = {Hicks, John Richard and others},
1809 + journal = {OUP Catalogue},
1810 + year = {1975},
1811 + publisher = {Oxford University Press}
1812 +}
1813 +
1814 +@article{hirshleifer1990hedging,
1815 + title = {Hedging Pressure and Futures Price Movements in a General Equilibrium Model},
1816 + author = {Hirshleifer, David},
1817 + journal = {Econometrica},
1818 + pages = {411--428},
1819 + year = {1990},
1820 + publisher = {JSTOR}
1821 +}
1822 +
1823 +@article{hirshleifer1991seasonal,
1824 + title = {Seasonal Patterns of Futures Hedging and the Resolution of Output Uncertainty},
1825 + author = {Hirshleifer, David},
1826 + journal = {Journal of Economic Theory},
1827 + volume = {53},
1828 + number = {2},
1829 + pages = {304--327},
1830 + year = {1991},
1831 + publisher = {Elsevier}
1832 +}
1833 +
1834 +@article{hollstein2020volatility,
1835 + title = {Volatility Term Structures in Commodity Markets},
1836 + author = {Hollstein, Fabian and Prokopczuk, Marcel and W{\"u}rsig, Christoph},
1837 + journal = {Journal of Futures Markets},
1838 + volume = {40},
1839 + number = {4},
1840 + pages = {527--555},
1841 + year = {2020},
1842 + publisher = {Wiley Online Library}
1843 +}
1844 +
1845 +@article{horan2004implied,
1846 + title = {Implied Volatility of Oil Futures Options Surrounding {OPEC} Meetings},
1847 + author = {Horan, Stephen M. and Peterson, Jeffrey H. and Mahar, James},
1848 + journal = {The Energy Journal},
1849 + volume = {25},
1850 + number = {3},
1851 + pages = {103--125},
1852 + year = {2004},
1853 + publisher = {International Association for Energy Economics}
1854 +}
1855 +
1856 +@article{hordahl2020expectations,
1857 + title = {Expectations and Risk Premia at 8:30 am: Deciphering the Responses of Bond Yields to Macroeconomic Announcements},
1858 + author = {H{\"o}rdahl, Peter and Remolona, Eli M. and Valente, Giorgio},
1859 + journal = {Journal of Business \& Economic Statistics},
1860 + volume = {38},
1861 + number = {1},
1862 + pages = {27--42},
1863 + year = {2020},
1864 + publisher = {Taylor \& Francis}
1865 +}
1866 +
1867 +@article{humphreys2010great,
1868 + title = {The Great Metals Boom: A Retrospective},
1869 + author = {Humphreys, David},
1870 + journal = {Resources Policy},
1871 + volume = {35},
1872 + number = {1},
1873 + pages = {1--13},
1874 + year = {2010},
1875 + publisher = {Elsevier}
1876 +}
1877 +
1878 +@article{hu2013noise,
1879 + title = {Noise as Information for Illiquidity},
1880 + author = {Hu, Grace Xing and Pan, Jun and Wang, Jiang},
1881 + journal = {Journal of Finance},
1882 + volume = {68},
1883 + number = {6},
1884 + pages = {2341--2382},
1885 + year = {2013},
1886 + publisher = {Wiley Online Library}
1887 +}
1888 +
1889 +@article{irwin1987note,
1890 + title = {A Note on the Factors Affecting Technical Trading System Returns},
1891 + author = {Irwin, Scott H. and Brorsen, B. Wade},
1892 + journal = {Journal of Futures Markets},
1893 + volume = {7},
1894 + number = {5},
1895 + pages = {591--595},
1896 + year = {1987},
1897 + publisher = {Wiley Subscription Services, Inc., A Wiley Company New York}
1898 +}
1899 +
1900 +@article{irwin1999managed,
1901 + title = {Managed Futures, Positive Feedback Trading, and Futures Price Volatility},
1902 + author = {Irwin, Scott H. and Yoshimaru, Satoko},
1903 + journal = {Journal of Futures Markets},
1904 + volume = {19},
1905 + number = {7},
1906 + pages = {759--776},
1907 + year = {1999},
1908 + publisher = {Wiley Online Library}
1909 +}
1910 +
1911 +@article{irwin2009devil,
1912 + title = {Devil or Angel? The Role of Speculation in the Recent Commodity Price Boom (and Bust)},
1913 + author = {Irwin, Scott H. and Sanders, Dwight R. and Merrin, Robert P.},
1914 + journal = {Journal of Agricultural and Applied Economics},
1915 + volume = {41},
1916 + number = {2},
1917 + pages = {377--391},
1918 + year = {2009},
1919 + publisher = {Cambridge University Press}
1920 +}
1921 +
1922 +@article{irwin2011index,
1923 + title = {Index Funds, Financialization, and Commodity Futures Markets},
1924 + author = {Irwin, Scott H. and Sanders, Dwight R.},
1925 + journal = {Applied Economic Perspectives and Policy},
1926 + volume = {33},
1927 + number = {1},
1928 + pages = {1--31},
1929 + year = {2011},
1930 + publisher = {Oxford University Press}
1931 +}
1932 +
1933 +@article{irwin2012financialization,
1934 + title = {Financialization and Structural Change in Commodity Futures Markets},
1935 + author = {Irwin, Scott H. and Sanders, Dwight R.},
1936 + journal = {Journal of Agricultural and Applied Economics},
1937 + volume = {44},
1938 + number = {3},
1939 + pages = {371--396},
1940 + year = {2012},
1941 + publisher = {Cambridge University Press}
1942 +}
1943 +
1944 +@article{irwin2012testing,
1945 + title = {Testing the Masters Hypothesis in Commodity Futures Markets},
1946 + author = {Irwin, Scott H. and Sanders, Dwight R.},
1947 + journal = {Energy Economics},
1948 + volume = {34},
1949 + number = {1},
1950 + pages = {256--269},
1951 + year = {2012},
1952 + publisher = {Elsevier}
1953 +}
1954 +
1955 +@incollection{irwin2004effect,
1956 + author = {Irwin, Scott H. and Holt, Bryce},
1957 + title = {The Effect of Large Hedge Fund and {CTA} Trading on Futures Market Volatility},
1958 + booktitle = {Commodity Trading Advisors: Risk, Performance Analysis and Selection},
1959 + publisher = {John Wiley and Sons, Inc},
1960 + address = {New York, NY},
1961 + year = {2004}
1962 +}
1963 +
1964 +@article{kaldor1976speculation,
1965 + title = {Speculation and Economic Stability},
1966 + author = {Kaldor, Nicholas},
1967 + booktitle = {The Economics of Futures Trading},
1968 + pages = {111--123},
1969 + year = {1976},
1970 + publisher = {Springer}
1971 +}
1972 +
1973 +@article{kang2020tale,
1974 + title = {A Tale of Two Premiums: The Role of Hedgers and Speculators in Commodity Futures Markets},
1975 + author = {Kang, Wenjin and Rouwenhorst, K. Geert and Tang, Ke},
1976 + journal = {Journal of Finance},
1977 + volume = {75},
1978 + number = {1},
1979 + pages = {377--417},
1980 + year = {2020},
1981 + publisher = {Wiley Online Library}
1982 +}
1983 +
1984 +@article{kellogg2014effect,
1985 + title = {The Effect of Uncertainty on Investment: Evidence from Texas Oil Drilling},
1986 + author = {Kellogg, Ryan},
1987 + journal = {American Economic Review},
1988 + volume = {104},
1989 + number = {6},
1990 + pages = {1698--1734},
1991 + year = {2014}
1992 +}
1993 +
1994 +@article{keynes1923some,
1995 + title = {Some Aspects of Commodity Markets},
1996 + author = {Keynes, John Maynard},
1997 + journal = {Manchester Guardian Commercial: European Reconstruction Series},
1998 + volume = {13},
1999 + pages = {784--786},
2000 + year = {1923}
2001 +}
2002 +
2003 +@article{kim2015does,
2004 + title = {Does Futures Speculation Destabilize Commodity Markets?},
2005 + author = {Kim, Abby},
2006 + journal = {Journal of Futures Markets},
2007 + volume = {35},
2008 + number = {8},
2009 + pages = {696--714},
2010 + year = {2015},
2011 + publisher = {Wiley Online Library}
2012 +}
2013 +
2014 +@article{kilian2011energy,
2015 + title = {Do Energy Prices Respond to US Macroeconomic News? A Test of the Hypothesis of Predetermined Energy Prices},
2016 + author = {Kilian, Lutz and Vega, Clara},
2017 + journal = {Review of Economics and Statistics},
2018 + volume = {93},
2019 + number = {2},
2020 + pages = {660--671},
2021 + year = {2011},
2022 + publisher = {The MIT Press}
2023 +}
2024 +
2025 +@article{kilian2014role,
2026 + title = {The Role of Inventories and Speculative Trading in the Global Market for Crude Oil},
2027 + author = {Kilian, Lutz and Murphy, Daniel P.},
2028 + journal = {Journal of Applied Econometrics},
2029 + volume = {29},
2030 + number = {3},
2031 + pages = {454--478},
2032 + year = {2014},
2033 + publisher = {Wiley Online Library}
2034 +}
2035 +
2036 +@article{knuth2018breakthroughs,
2037 + title = {“Breakthroughs” for a Green Economy? Financialization and Clean Energy Transition},
2038 + author = {Knuth, Sarah},
2039 + journal = {Energy Research \& Social Science},
2040 + volume = {41},
2041 + pages = {220--229},
2042 + year = {2018},
2043 + publisher = {Elsevier}
2044 +}
2045 +
2046 +@article{kothari2007econometrics,
2047 + title = {Econometrics of Event Studies},
2048 + author = {Kothari, Sagar P. and Warner, Jerold B.},
2049 + booktitle = {Handbook of Empirical Corporate Finance},
2050 + pages = {3--36},
2051 + year = {2007},
2052 + publisher = {Elsevier}
2053 +}
2054 +
2055 +
2056 +
2057 +@article{kyle1985continuous,
2058 + title = {Continuous Auctions and Insider Trading},
2059 + author = {Kyle, Albert S},
2060 + journal = {Econometrica},
2061 + volume = {53},
2062 + number = {6},
2063 + pages = {1315--1335},
2064 + year = {1985},
2065 + publisher = {JSTOR}
2066 +}
2067 +
2068 +@article{lagi2011food,
2069 + author = {Lagi, Marco and Bertrand, Karla and Bar‐Yam, Yaneer},
2070 + title = {The Food Crises and Political Instability in North Africa and the Middle East},
2071 + journal = {SSRN Electronic Journal},
2072 + year = {2011},
2073 + doi = {10.2139/ssrn.1910031}
2074 +}
2075 +
2076 +@article{lin2010opec,
2077 + author = {Lin, Sharon Xiaowen and Tamvakis, Michael},
2078 + doi = {10.1016/j.enpol.2009.10.053},
2079 + title = {OPEC Announcements and Their Effects on Crude Oil Prices},
2080 + journal = {Energy Policy},
2081 + volume = {38},
2082 + number = {2},
2083 + pages = {1010--1016},
2084 + year = {2010}
2085 +}
2086 +
2087 +@article{liu2015does,
2088 + title = {Does Anything Beat 5-Minute RV? A Comparison of Realized Measures Across Multiple Asset Classes},
2089 + author = {Liu, Lily Y. and Patton, Andrew J. and Sheppard, Kevin},
2090 + journal = {Journal of Econometrics},
2091 + volume = {187},
2092 + number = {1},
2093 + pages = {293--311},
2094 + year = {2015},
2095 + publisher = {Elsevier}
2096 +}
2097 +
2098 +@article{lucey2015precious,
2099 + title = {What Precious Metals Act as Safe Havens, and When? Some {US} Evidence},
2100 + author = {Lucey, Brian M and Li, Sile},
2101 + journal = {Applied Economics Letters},
2102 + volume = {22},
2103 + number = {1},
2104 + pages = {35--45},
2105 + year = {2015},
2106 + publisher = {Taylor \& Francis}
2107 +}
2108 +
2109 +@article{masters2009testimony,
2110 + title = {Testimony Before the Commodity Futures Trading Commission},
2111 + author = {Masters, Michael W.},
2112 + journal = {Testimony to the Commodity Futures Trading Commission},
2113 + year = {2009}
2114 +}
2115 +
2116 +@misc{newey1986simple,
2117 + title = {A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix},
2118 + author = {Newey, Whitney K. and West, Kenneth D.},
2119 + year = {1986},
2120 + publisher = {National Bureau of Economic Research, Cambridge, MA, USA}
2121 +}
2122 +
2123 +@article{pal2023agricultural,
2124 + author = {Pal, Dipanwita},
2125 + title = {Do Agricultural Prices Respond to Interest on Reserves?},
2126 + journal = {The Economics and Finance Letters},
2127 + volume = {10},
2128 + number = {1},
2129 + pages = {94--101},
2130 + year = {2023},
2131 + doi = {10.18488/29.v10i1.3311}
2132 +}
2133 +
2134 +@article{park2019speculators,
2135 + author = {Park, Jinwook},
2136 + title = {Effect of Speculators’ Position Changes on the LME Futures Market},
2137 + journal = {International Journal of Financial Studies},
2138 + volume = {7},
2139 + number = {2},
2140 + pages = {32},
2141 + year = {2019},
2142 + doi = {10.3390/ijfs7020032}
2143 +}
2144 +
2145 +@article{pescatori2022opec,
2146 + author = {Pescatori, Andrea and Nazer, Yousef},
2147 + doi = {10.5089/9798400219788.001},
2148 + title = {OPEC and the Oil Market},
2149 + journal = {IMF Working Paper},
2150 + year = {2022}
2151 +}
2152 +
2153 +@book{pirrong2011commodity,
2154 + title = {Commodity Price Dynamics: A Structural Approach},
2155 + author = {Pirrong, Craig},
2156 + year = {2011},
2157 + publisher = {Cambridge University Press}
2158 +}
2159 +
2160 +@article{rakowski2021twitter,
2161 + title = {Twitter Activity, Investor Attention, and the Diffusion of Information},
2162 + author = {Rakowski, David and Shirley, Sara E and Stark, Jeffrey R},
2163 + journal = {Financial Management},
2164 + volume = {50},
2165 + number = {1},
2166 + pages = {3--46},
2167 + year = {2021},
2168 + publisher = {Wiley Online Library}
2169 +}
2170 +
2171 +@article{rlmishra2021financialization,
2172 + author = {Mishra, Ramesh and Mishra, Alok},
2173 + title = {Financialization of Indian Agricultural Commodities: The Case of Index Investments},
2174 + journal = {International Journal of Social Economics},
2175 + volume = {49},
2176 + number = {1},
2177 + pages = {73--96},
2178 + year = {2021},
2179 + doi = {10.1108/ijse-05-2021-0254}
2180 +}
2181 +
2182 +@article{sanders2016bubbles,
2183 + title = {Bubbles, Froth and Facts: Another Look at the Masters Hypothesis in Commodity Futures Markets},
2184 + author = {Sanders, Dwight and Irwin, Scott},
2185 + journal = {Journal of Agricultural Economics},
2186 + volume = {68},
2187 + number = {2},
2188 + pages = {345--365},
2189 + year = {2016},
2190 + doi = {10.1111/1477-9552.12191}
2191 +}
2192 +
2193 +@article{scotti2016surprise,
2194 + title = {Surprise and Uncertainty Indexes: Real-Time Aggregation of Real-Activity Macro-Surprises},
2195 + journal = {Journal of Monetary Economics},
2196 + volume = {82},
2197 + pages = {1--19},
2198 + year = {2016},
2199 + issn = {0304-3932},
2200 + doi = {https://doi.org/10.1016/j.jmoneco.2016.06.002},
2201 + url = {https://www.sciencedirect.com/science/article/pii/S0304393216300320},
2202 + author = {Scotti, Chiara},
2203 + keywords = {Business cycle, Dynamic factor model, State space model, Forecasting weights},
2204 + abstract = {Two daily, real-time, real-activity indexes are constructed for the United States, euro area, United Kingdom, Canada, and Japan: (i) a surprise index summarizing recent economic data surprises and measuring optimism/pessimism about the state of the economy, and (ii) an uncertainty index measuring uncertainty related to the state of the economy. The surprise index parsimoniously preserves the properties of the underlying series when affecting asset prices. For the United States, the real-activity uncertainty index is compared to other uncertainty proxies to show that, when uncertainty is strictly related to real activity only, it has a potentially milder effect on economic activity.}
2205 +}
2206 +
2207 +@article{shanker2017new,
2208 + title = {New Indices of Adequate and Excess Speculation and Their Relationship with Volatility in the Crude Oil Futures Market},
2209 + author = {Shanker, Latha},
2210 + journal = {Journal of Commodity Markets},
2211 + volume = {5},
2212 + pages = {18--35},
2213 + year = {2017},
2214 + publisher = {Elsevier}
2215 +}
2216 +
2217 +@article{singleton2014investor,
2218 + author = {Singleton, Kenneth},
2219 + title = {Investor Flows and the 2008 Boom/Bust in Oil Prices},
2220 + journal = {Management Science},
2221 + volume = {60},
2222 + number = {2},
2223 + pages = {300--318},
2224 + year = {2014},
2225 + doi = {10.1287/mnsc.2013.1756}
2226 +}
2227 +
2228 +@article{soana2020index,
2229 + author = {Soana, Maria and Verga, Giovanna and Volpi, Maurizio},
2230 + title = {Did Index Trader and Swap Dealer Activity Produce a Bubble in the Agricultural Commodity Market?},
2231 + journal = {African Journal of Business Management},
2232 + volume = {14},
2233 + number = {1},
2234 + pages = {9--24},
2235 + year = {2020},
2236 + doi = {10.5897/ajbm2019.8877}
2237 +}
2238 +
2239 +@article{tang2012index,
2240 + title = {Index Investment and the Financialization of Commodities},
2241 + author = {Tang, Ke and Xiong, Wei},
2242 + journal = {Financial Analysts Journal},
2243 + volume = {68},
2244 + number = {6},
2245 + pages = {54--74},
2246 + year = {2012},
2247 + publisher = {Taylor \& Francis}
2248 +}
2249 +
2250 +@article{vivian2012commodity,
2251 + title = {Commodity Volatility Breaks},
2252 + author = {Vivian, Andrew and Wohar, Mark E.},
2253 + journal = {Journal of International Financial Markets, Institutions and Money},
2254 + volume = {22},
2255 + number = {2},
2256 + pages = {395--422},
2257 + year = {2012},
2258 + publisher = {Elsevier}
2259 +}
2260 +
2261 +@article{working1949theory,
2262 + title = {The Theory of Price of Storage},
2263 + author = {Working, Holbrook},
2264 + journal = {American Economic Review},
2265 + volume = {39},
2266 + number = {6},
2267 + pages = {1254--1262},
2268 + year = {1949},
2269 + publisher = {JSTOR}
2270 +}
2271 +
2272 +@article{working1960speculation,
2273 + title = {Speculation on Hedging Markets},
2274 + author = {Working, Holbrook},
2275 + journal = {Food Research Institute Studies},
2276 + volume = {1},
2277 + number = {2},
2278 + pages = {185--220},
2279 + year = {1960}
2280 +}
2281 +
2282 +@article{wu2019does,
2283 + title = {Does Gold or Bitcoin Hedge Economic Policy Uncertainty?},
2284 + author = {Wu, Shan and Tong, Mu and Yang, Zhongyi and Derbali, Abdelkader},
2285 + journal = {Finance Research Letters},
2286 + volume = {31},
2287 + pages = {171--178},
2288 + year = {2019},
2289 + publisher = {Elsevier}
2290 +}
2291 +
2292 +@article{yang2005futures,
2293 + title = {Futures Trading Activity and Commodity Cash Price Volatility},
2294 + author = {Yang, Jian and Balyeat, R. Brian and Leatham, David J.},
2295 + journal = {Journal of Business Finance \& Accounting},
2296 + volume = {32},
2297 + number = {1-2},
2298 + pages = {297--323},
2299 + year = {2005},
2300 + publisher = {Wiley Online Library}
2301 +}
2302 +
2303 +@article{ye2021macroeconomic,
2304 + title = {Macroeconomic Forecasts and Commodity Futures Volatility},
2305 + author = {Ye, Wuyi and Guo, Ranran and Deschamps, Bruno and Jiang, Ying and Liu, Xiaoquan},
2306 + journal = {Economic Modelling},
2307 + volume = {94},
2308 + pages = {981--994},
2309 + year = {2021},
2310 + publisher = {Elsevier}
2311 +}
2312 +
2313 +@article{zhang2018informed,
2314 + title = {Informed Options Trading Prior to Dividend Change Announcements},
2315 + author = {Zhang, Jun},
2316 + journal = {Financial Management},
2317 + volume = {47},
2318 + number = {1},
2319 + pages = {81--103},
2320 + year = {2018},
2321 + publisher = {Wiley Online Library}
2322 +}
2323 +
2324 +@article{zhang2022hedging,
2325 + author = {Zhang, Tao},
2326 + title = {Hedging Pressure and Liquidity Provision in Commodity Options Markets},
2327 + journal = {Journal of Futures Markets},
2328 + volume = {42},
2329 + number = {7},
2330 + pages = {1212--1233},
2331 + year = {2022},
2332 + doi = {10.1002/fut.22327}
2333 +}
2334 +
2335 +@article{zorn1983sovereignty,
2336 + author = {Zorn, Stephen A},
2337 + doi = {10.1111/j.1477-8947.1983.tb00276.x},
2338 + title = {Permanent Sovereignty Over Natural Resources},
2339 + journal = {Natural Resources Forum},
2340 + volume = {7},
2341 + number = {4},
2342 + pages = {321--328},
2343 + year = {1983}
2344 +}
2345 +@article{roll1984,
2346 + author = {Roll, R.},
2347 + title = {A Simple Implicit Measure of the Effective Bid-Ask Spread},
2348 + journal = {The Journal of Finance},
2349 + volume = {39},
2350 + number = {4},
2351 + pages = {1127--1139},
2352 + year = {1984},
2353 +}
2354 +
2355 +
2356 +@article{chordia2008liquidity,
2357 + author = {Chordia, T. and Roll, R. and Subrahmanyam, A.},
2358 + title = {Liquidity and market efficiency},
2359 + journal = {Journal of Financial Economics},
2360 + volume = {87},
2361 + number = {2},
2362 + pages = {249--268},
2363 + year = {2008},
2364 + issn = {0304-405X},
2365 + doi = {10.1016/j.jfineco.2007.03.005},
2366 + url = {https://www.sciencedirect.com/science/article/pii/S0304405X07001833},
2367 + keywords = {Liquidity, Market efficiency, Order flow},
2368 + abstract = {Short-horizon return predictability from order flows is an inverse indicator of market efficiency. We find that such predictability is diminished when bid-ask spreads are narrower, and has declined over time with the minimum tick size. Variance ratio tests suggest that prices were closer to random walk benchmarks in the more liquid decimal regime than in other ones. These findings indicate that liquidity stimulates arbitrage activity, which, in turn, enhances market efficiency. Further, as the tick size decreased, open-close/close-open return variance ratios increased, while return autocorrelations decreased. This suggests an increased incorporation of private information into prices during more liquid regimes.}
2369 +}
2370 +
2371 +@inproceedings{boons2014price,
2372 + author = {Martijn Boons and Frans A. de Roon and Marta Szymanowska},
2373 + title = {The Price of Commodity Risk in Stock and Futures Markets},
2374 + booktitle = {AFA 2012 Chicago Meetings Paper},
2375 + year = {2014},
2376 + note = {Available at SSRN: \url{https://ssrn.com/abstract=1785728} or \url{http://dx.doi.org/10.2139/ssrn.1785728}}
2377 +
2378 +@article{da2024financialization,
2379 + title={Financialization and commodity markets serial dependence},
2380 + author={Da, Zhi and Tang, Ke and Tao, Yubo and Yang, Liyan},
2381 + journal={Management Science},
2382 + volume={70},
2383 + number={4},
2384 + pages={2122--2143},
2385 + year={2024},
2386 + publisher={INFORMS}
2387 +}
2388 +
2389 +@article{kang2023financialization,
2390 + title={Financialization of commodity markets ten years later},
2391 + author={Kang, Wenjin and Tang, Ke and Wang, Ningli},
2392 + journal={Journal of Commodity Markets},
2393 + volume={30},
2394 + pages={100313},
2395 + year={2023},
2396 + publisher={Elsevier}
2397 +}
2398 +
2399 +@article{ready2022order,
2400 + title={Order flows and financial investor impacts in commodity futures markets},
2401 + author={Ready, Mark J and Ready, Robert C},
2402 + journal={The Review of Financial Studies},
2403 + volume={35},
2404 + number={10},
2405 + pages={4712--4755},
2406 + year={2022},
2407 + publisher={Oxford University Press}
2408 +}
2409 +
2410 +@article{cheng2014financialization,
2411 + title={Financialization of commodity markets},
2412 + author={Cheng, Ing-Haw and Xiong, Wei},
2413 + journal={Annu. Rev. Financ. Econ.},
2414 + volume={6},
2415 + number={1},
2416 + pages={419--441},
2417 + year={2014},
2418 + publisher={Annual Reviews}
2419 +}
2420 +
2421 +@article{shleifer1990noise,
2422 + title={The noise trader approach to finance},
2423 + author={Shleifer, Andrei and Summers, Lawrence H},
2424 + journal={Journal of Economic perspectives},
2425 + volume={4},
2426 + number={2},
2427 + pages={19--33},
2428 + year={1990},
2429 + publisher={American Economic Association}
2430 +}
2431 +
2432 +@article{erb2013golden,
2433 + title={The golden dilemma},
2434 + author={Erb, Claude B and Harvey, Campbell R},
2435 + journal={Financial Analysts Journal},
2436 + volume={69},
2437 + number={4},
2438 + pages={10--42},
2439 + year={2013},
2440 + publisher={Taylor \& Francis}
2441 +}
2442 +
2443 +@article{connolly2022stock,
2444 + title={Stock returns and inflation shocks in weaker economic times},
2445 + author={Connolly, Robert A and Stivers, Chris and Sun, Licheng},
2446 + journal={Financial Management},
2447 + volume={51},
2448 + number={3},
2449 + pages={827--867},
2450 + year={2022},
2451 + publisher={Wiley Online Library}
2452 +}
2453 +
2454 +@article{zhu2022oil,
2455 + title={Oil price shocks and stock market anomalies},
2456 + author={Zhu, Zhaobo and Sun, Licheng and Tu, Jun and Ji, Qiang},
2457 + journal={Financial Management},
2458 + volume={51},
2459 + number={2},
2460 + pages={573--612},
2461 + year={2022},
2462 + publisher={Wiley Online Library}
2463 +}
2464 +
2465 +
2466 +@article{karali2014macro,
2467 + title={Macro determinants of volatility and volatility spillover in energy markets},
2468 + author={Karali, Berna and Ramirez, Octavio A},
2469 + journal={Energy Economics},
2470 + volume={46},
2471 + pages={413--421},
2472 + year={2014},
2473 + publisher={Elsevier},
2474 + doi={10.1016/j.eneco.2014.06.004}
2475 +}
2476 +
2477 +@article{cao2024us,
2478 + title={U.S. monetary policy: The pushing hands of crude oil price?},
2479 + author={Cao, Fangzhi and Su, Chi-Wei and Sun, Dian and Qin, Meng and Umar, Muhammad},
2480 + journal={Energy Economics},
2481 + volume={134},
2482 + pages={107555},
2483 + year={2024},
2484 + publisher={Elsevier},
2485 + doi={10.1016/j.eneco.2024.107555}
2486 +}
2487 +
2488 +@article{kang2020economic,
2489 + title={Economic determinants of oil futures volatility: A term structure perspective},
2490 + author={Kang, Boda and Nikitopoulos, Christina Sklibosios and Prokopczuk, Marcel},
2491 + journal={Energy Economics},
2492 + volume={88},
2493 + pages={104743},
2494 + year={2020},
2495 + publisher={Elsevier},
2496 + doi={10.1016/j.eneco.2020.104743}
2497 +}
2498 +
2499 +@article{kilian2011energy,
2500 + title={Do energy prices respond to US macroeconomic news? {A} test of the hypothesis of predetermined energy prices},
2501 + author={Kilian, Lutz and Vega, Clara},
2502 + journal={Review of Economics and Statistics},
2503 + volume={93},
2504 + number={2},
2505 + pages={660--671},
2506 + year={2011},
2507 + publisher={The MIT Press}
2508 +}
2509 +
2510 +
2511 +@article{kilian2014role,
2512 + title={The role of inventories and speculative trading in the global market for crude oil},
2513 + author={Kilian, Lutz and Murphy, Daniel P},
2514 + journal={Journal of Applied Econometrics},
2515 + volume={29},
2516 + number={3},
2517 + pages={454--478},
2518 + year={2014},
2519 + publisher={Wiley Online Library}
2520 +}
2521 +
2522 +@article{fattouh2013role,
2523 + title={The role of speculation in oil markets: What have we learned so far?},
2524 + author={Fattouh, Bassam and Kilian, Lutz and Mahadeva, Lavan},
2525 + journal={The Energy Journal},
2526 + volume={34},
2527 + number={3},
2528 + pages={7--33},
2529 + year={2013},
2530 + publisher={SAGE Publications Sage CA: Los Angeles, CA}
2531 +}
2532 +
2533 +@article{baumeister2014oil,
2534 + title={Do oil price increases cause higher food prices?},
2535 + author={Baumeister, Christiane and Kilian, Lutz},
2536 + journal={Economic Policy},
2537 + volume={29},
2538 + number={80},
2539 + pages={691--747},
2540 + year={2014},
2541 + publisher={Oxford University Press}
2542 +}
2543 +
2544 +@article{henderson2015new,
2545 + title={New evidence on the financialization of commodity markets},
2546 + author={Henderson, Brian J and Pearson, Neil D and Wang, Li},
2547 + journal={Review of Financial Studies},
2548 + volume={28},
2549 + number={5},
2550 + pages={1285--1311},
2551 + year={2015},
2552 + publisher={Oxford University Press}
2553 +}
2554 +
2555 +@article{domanski2007financial,
2556 + title={Financial investors and commodity markets},
2557 + author={Domanski, Dietrich and Heath, Alexandra},
2558 + journal={BIS Quarterly Review},
2559 + volume={3},
2560 + number={1},
2561 + pages={53--67},
2562 + year={2007}
2563 +}
2564 +
2565 +@article{newey1994automatic,
2566 + title={Automatic lag selection in covariance matrix estimation},
2567 + author={Newey, Whitney K and West, Kenneth D},
2568 + journal={Review of Economic Studies},
2569 + volume={61},
2570 + number={4},
2571 + pages={631--653},
2572 + year={1994},
2573 + publisher={Wiley-Blackwell}
2574 +}
2575 +
2576 +@article{ghysels2004midas,
2577 + title={The MIDAS touch: Mixed data sampling regression models},
2578 + author={Ghysels, Eric and Santa-Clara, Pedro and Valkanov, Rossen},
2579 + year={2004}
2580 +}
2581 +
2582 +@article{buyukcsahin2011speculators,
2583 + title={Do speculators drive crude oil futures prices?},
2584 + author={B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Harris, Jeffrey H},
2585 + journal={The Energy Journal},
2586 + volume={32},
2587 + number={2},
2588 + pages={167--202},
2589 + year={2011},
2590 + publisher={SAGE Publications Sage CA: Los Angeles, CA}
2591 +}
2592 +
2593 +@article{alquist2013role,
2594 + title={The role of financial speculation in driving the price of crude oil},
2595 + author={Alquist, Ron and Gervais, Olivier},
2596 + journal={The Energy Journal},
2597 + volume={34},
2598 + number={3},
2599 + pages={35--54},
2600 + year={2013},
2601 + publisher={SAGE Publications Sage CA: Los Angeles, CA}
2602 +}
\ No newline at end of file
added phd_chap1_20260731/pi_cl.png +0 −0

Binary file not shown.

added phd_chap1_20260731/pi_gc.png +0 −0

Binary file not shown.

added phd_chap1_20260731/pi_hg.png +0 −0

Binary file not shown.

added phd_chap1_20260731/pi_ng.png +0 −0

Binary file not shown.

added phd_chap1_20260731/pi_pa.png +0 −0

Binary file not shown.

added phd_chap1_20260731/pi_si.png +0 −0

Binary file not shown.

added phd_chap1_20260731/tables.tex +656 −0
@@ -0,0 +1,656 @@
1 +\section{Tables}
2 +
3 +
4 +
5 +
6 +\begin{landscape}
7 +\begin{table}[]
8 +\caption{Summary of the literature: Effect of non-commercial trading activity on commodity futures volatility}
9 +\label{tab:fin}
10 +\centering
11 +\begin{tabular}{@{}lll@{}}
12 +\toprule
13 +\textbf{References} & \textbf{Proxy used for financialization or speculation} & \textbf{Impact on volatility} \\ \midrule
14 +\citet{chang1997interday} & CFTC’s definition of speculators & \multirow{4}{*}{\textbf{Increase}} \\
15 +\citet{daigler1999impact} & CFTC’s definition of speculators & \\
16 +\citet{irwin2004effect} & Set speculators & \\
17 +\citet{tang2012index} & Commodity index trader (CIT) positions & \\ \midrule
18 +\citet{irwin1987note} & Amount of money invested in traded futures funds & \multirow{4}{*}{\textbf{No change}} \\
19 +\citet{irwin1999managed} & Trading volume of large-commodity pool operators & \\
20 +\citet{bryant2006causality} & CFTC’s definition of speculators & \\
21 +\citet{haigh2007hedge} & Number and positions of commodity pool operators and hedge funds & \\ \midrule
22 +\citet{brunetti2016speculators} & Net positions of hedge funds and floor brokers & \multirow{2}{*}{\textbf{Decrease}} \\
23 +\citet{aulerich2012bubbles} & Commodity index trader (CIT) positions & \\ \bottomrule
24 +\end{tabular}
25 +\begin{tablenotes}
26 +\singlespacing
27 +\footnotesize
28 +This table summarizes the findings of a range of studies on the effect of financialization and speculation on commodity futures volatility. The impact on volatility is categorized as Positive, Neutral, or Negative based on the results reported by each study. The proxies used for financialization or speculation include definitions and positions from the CFTC, set speculators, commodity index trader (CIT) positions, trading volume, and net positions of hedge funds and floor brokers.
29 +\end{tablenotes}
30 +\end{table}
31 +\end{landscape}
32 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
33 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
34 +\begin{table}[]
35 +\begin{center}
36 +\caption{List of macroeconomic announcements in our sample}
37 +\label{tab:stat1}
38 +\begin{tabular}{@{}lcccc@{}}
39 +\toprule
40 +\multicolumn{1}{c}{\textbf{Announcement}} & \textbf{Frequency} & \textbf{Source*} & \textbf{Unit} & \textbf{Time} \\ \midrule
41 +\textbf{GDP advance} & Quarterly & BEA & \% & 8:30 \\
42 +\textbf{GDP preliminary} & Quarterly & BEA & \% & 8:30 \\
43 +\textbf{GDP final} & Quarterly & BEA & \% & 8:30 \\
44 +\textbf{Personal income} & Monthly & BEA & \% & 8:30 \\
45 +\textbf{ADP employment} & Monthly & ADP & Number of jobs & 8:15 \\
46 +\textbf{Initial jobless claims} & Weekly & ETA & Number of claims & 8:30 \\
47 +\textbf{Non-farm employment} & Monthly & BLS & Number of jobs & 8:30 \\
48 +\textbf{Factory orders} & Monthly & BC & \% & 10:00 \\
49 +\textbf{Industrial production} & Monthly & FRB & \% & 9:15 \\
50 +\textbf{Construction spending} & Monthly & BC & \% & 10:00 \\
51 +\textbf{Durable goods orders} & Monthly & BC & \% & 8:30 \\
52 +\textbf{Advance retail sales} & Monthly & BC & \% & 8:30 \\
53 +\textbf{Consumer credit} & Monthly & FRB & USD & 15:00 \\
54 +\textbf{Personal consumption} & Monthly & BEA & \% & 8:30 \\
55 +\textbf{Building permits} & Monthly & BC & Number of permits & 8:30 \\
56 +\textbf{Existing home sales} & Monthly & NAR & Number of homes & 10:00 \\
57 +\textbf{Housing starts} & Monthly & BC & Number of homes & 8:30 \\
58 +\textbf{New home sales} & Monthly & BC & Number of homes & 10:00 \\
59 +\textbf{Pending home sales} & Monthly & NAR & \% & 10:00 \\
60 +\textbf{Trade balance} & Monthly & BEA & USD & 8:30 \\
61 +\textbf{Consumer price index} & Monthly & BLS & \% & 8:30 \\
62 +\textbf{Producer price index} & Monthly & BLS & \% & 8:30 \\
63 +\textbf{CB Consumer confidence index} & Monthly & CB & Index & 10:00 \\
64 +\textbf{UM Consumer sentiment} & Monthly & TR/UM & Index & 9:55 \\
65 +\textbf{Weekly Crude Oil Stock} & Weekly & EIA & number of barrels & 10:30 \\
66 +\textbf{Weekly Natural Gas Stock} & Weekly & EIA & number of cubic feet & 11:00 \\\bottomrule
67 +\end{tabular}
68 +\end{center}
69 +\begin{tablenotes}
70 + \singlespacing
71 + \footnotesize
72 +This table shows the category, frequency, source, unit of measure, and release time for each macroeconomic announcements. *(Automatic Data Processing, Inc. (ADP), Bureau of the Census (BC), Bureau of Economic Analysis (BEA), Bureau of Labor Statistics (BLS), Conference Board (CB), Employment and Training Administration (ETA), Federal Reserve Board (FRB), Institute for Supply Management (ISM), National Association of Realtors (NAR), Thomson Reuters/University of Michigan (TR/UM), and U.S. Department of the Treasury (USDT).)
73 +\end{tablenotes}
74 +\end{table}
75 +
76 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
77 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
78 +\begin{landscape}
79 +\begin{table}[]
80 +\begin{center}
81 +\caption{Descriptive statistics for the standardized surprise calculated for each of the macroeconomic announcements}
82 +\label{tab:stat2}
83 +\begin{tabular}{@{}lccccccc@{}}
84 +\toprule
85 +\multicolumn{1}{c}{\textbf{Announcements}} & \textbf{Nb. obs.} & \textbf{Min.} & \textbf{1st Qu.} & \textbf{Med.} & \textbf{Mean} & \textbf{3rd Qu.} & \textbf{Max.} \\ \midrule
86 +\textbf{Initial jobless claims} & 825 & -3.407 & -0.0720 & -0.007 & 0.068 & 0.065 & 22.672 \\
87 +\textbf{ADP Employment} & 202 & -2.751 & -0.0640 & 0.008 & 0.046 & 0.078 & 12.880 \\
88 +\textbf{CB Consumer} & 201 & -2.635 & -0.4638 & 0.093 & 0.102 & 0.872 & 2.412 \\
89 +\textbf{Advance retail sales} & 202 & -4.028 & -0.3661 & -0.092 & 0.023 & 0.183 & 8.879 \\
90 +\textbf{Building permit} & 198 & -2.375 & -0.5356 & 0.025 & 0.082 & 0.627 & 3.205 \\
91 +\textbf{Construction spending} & 202 & -3.054 & -0.5912 & -0.099 & -0.130 & 0.493 & 4.335 \\
92 +\textbf{Consumer\_credit} & 202 & -2.055 & -0.5217 & 0.104 & 0.061 & 0.619 & 3.131 \\
93 +\textbf{Consumer price index} & 201 & -3.483 & -0.6966 & 0.000 & -0.035 & 0.697 & 4.180 \\
94 +\textbf{Durable goods orders} & 193 & -2.702 & -0.5757 & 0.022 & 0.037 & 0.531 & 6.688 \\
95 +\textbf{Existing home sales} & 202 & -4.729 & -0.4627 & 0.000 & -0.067 & 0.488 & 2.467 \\
96 +\textbf{Factory orders} & 202 & -3.040 & -0.3378 & 0.000 & 0.081 & 0.507 & 2.534 \\
97 +\textbf{GDP} & 185 & -2.589 & -0.3698 & 0.000 & -0.026 & 0.370 & 2.958 \\
98 +\textbf{Housing starts} & 199 & -2.285 & -0.6178 & 0.000 & 0.042 & 0.624 & 3.401 \\
99 +\textbf{Industrial production} & 385 & -4.773 & -0.5727 & 0.000 & -0.080 & 0.573 & 2.291 \\
100 +\textbf{Michigan Sentiment Index} & 202 & -3.922 & -0.4100 & 0.036 & -0.049 & 0.463 & 3.244 \\
101 +\textbf{New home sales} & 202 & -3.062 & -0.3466 & 0.116 & 0.118 & 0.631 & 3.562 \\
102 +\textbf{Non-farm employment} & 201 & -0.892 & -0.0564 & 0.005 & 0.078 & 0.071 & 13.169 \\
103 +\textbf{Pending home sales} & 202 & -2.949 & -0.4244 & 0.022 & 0.075 & 0.581 & 5.674 \\
104 +\textbf{Personal consumption} & 201 & -3.666 & -0.3666 & 0.000 & -0.026 & 0.367 & 2.566 \\
105 +\textbf{Personal income} & 201 & -1.079 & -0.0771 & 0.000 & 0.095 & 0.077 & 13.108 \\
106 +\textbf{Producer price index} & 188 & -3.168 & -0.5760 & 0.000 & 0.083 & 0.864 & 2.880 \\
107 +\textbf{Trade balance} & 202 & -1.831 & -0.1801 & -0.018 & 0.013 & 0.207 & 2.359 \\ \bottomrule
108 +\end{tabular}
109 +\end{center}
110 +\begin{tablenotes}
111 + \singlespacing
112 + \footnotesize
113 +This table presents descriptive statistics for the standardized surprise $(A_{kt}-E_{kt})/\sigma_{kt}$ for each of the macroeconomic announcements. The column (Nb. Observations) shows the number of individual surprises that can be calculated over the whole analysis period. The columns (Min.), (1st Qu.), (Median), (Mean), (3rd Qu.) and (Max) present respectively the minimum value, the first quartile, the median, the mean, the third quartile and the maximum value for the standardized surprise of each macroeconomic announcement
114 +\end{tablenotes}
115 +\end{table}
116 +\end{landscape}
117 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
118 +
119 +%%%%%%%%%%%%%%%%%%%%%%%%%%%
120 +
121 +
122 +
123 +
124 +
125 +\begin{landscape}
126 +\begin{table}[]
127 +\begin{center}
128 +\caption{Descriptive statistics: 5-minute intraday futures price returns}
129 +\label{tab:stat4}
130 +\begin{tabular}{@{}lllllll@{}}
131 +\toprule
132 +\textbf{Commodity Futures} & \textbf{Min (\%)} & \textbf{1st Qu. (\%)} & \textbf{Med. (\%)} & \textbf{Mean (\%)} & \textbf{3rd Qu. (\%)} & \textbf{Max (\%)} \\ \midrule
133 +\textbf{Crude Oil (CL=F)} & -33.91 & -0.0441 & 0.000 & 0.000 & 0.0447 & 41.64 \\
134 +\textbf{Gold (GC=F)} & -2.782 & -0.0241 & 0.000 & 0.0001 & 0.0244 & 3.064 \\
135 +\textbf{Copper (HG=F)} & -4.534 & -0.0363 & 0.000 & -0.0001 & 0.0365 & 8.877 \\
136 +\textbf{Natural Gas (NG=F)} & -6.735 & -0.0528 & 0.000 & -0.0003 & 0.0532 & 15.62 \\
137 +\textbf{Palladium (PA=F)} & -13.350 & -0.034 & 0.000 & 0.0001 & 0.0348 & 9.467 \\
138 +\textbf{Silver (SI=F)} & -7.504 & -0.0394 & 0.000 & 0.0001 & 0.0415 & 4.242 \\ \bottomrule
139 +\end{tabular}
140 +\end{center}
141 +\begin{tablenotes}
142 + \singlespacing
143 + \footnotesize
144 +Shows descriptive statistics of the 5-minute intraday returns, for each commodity futures. The columns (Min.), (1st Qu.), (Median), (Mean), (3rd Qu.) and (Max) present respectively the minimum value, the first quartile, the median, the mean, the third quartile and the maximum value for the 5 minute intraday returns.
145 +\end{tablenotes}
146 +\end{table}
147 +\end{landscape}
148 +
149 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
150 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
151 +
152 +\begin{table}[ht]
153 +\centering
154 +\caption{Descriptive statistics: Computed value of the NLS proxy (respectively: full sample, money managers-only sample, swap dealers-only sample)}
155 +\label{tab:stat5}
156 +\small
157 +\begin{tabular}{lcccccc}
158 +\toprule
159 + & \textbf{CL} & \textbf{GC} & \textbf{HG} & \textbf{SI} & \textbf{PA} & \textbf{NG} \\
160 +\midrule
161 + & & & \textbf{$NLS$} & & & \\
162 +\midrule
163 +Min. & -0.1667 & -0.4505 & -0.3238 & -0.1364 & -0.6246 & -0.2745 \\
164 +1st Qu. & 0.0358 & 0.2285 & -0.1067 & 0.1547 & 0.3181 & -0.1731 \\
165 +Median & 0.0642 & 0.2397 & 0.0235 & 0.2163 & 0.3056 & -0.0563 \\
166 +Mean & 0.0774 & 0.1780 & 0.0278 & 0.2123 & 0.2685 & -0.0579 \\
167 +3rd Qu. & 0.1834 & 0.4012 & 0.1417 & 0.3334 & 0.5652 & -0.0321 \\
168 +Max. & 0.2941 & 0.5269 & 0.4413 & 0.5748 & 0.7343 & 0.0794 \\
169 +\midrule
170 + & & & \textbf{$NLS_{MM$}} & & & \\
171 +\midrule
172 +Min. & -0.0370 & -0.2353 & -0.2727 & -0.2303 & -0.6005 & -0.2471 \\
173 +1st Qu. & 0.0358 & 0.2177 & 0.0238 & 0.1550 & 0.3919 & -0.0345 \\
174 +Median & 0.1059 & 0.2177 & 0.0238 & 0.1550 & 0.3919 & -0.0345 \\
175 +Mean & 0.1011 & 0.2125 & 0.0350 & 0.1425 & 0.3294 & -0.0371 \\
176 +3rd Qu. & 0.1834 & 0.4012 & 0.1417 & 0.3334 & 0.5652 & -0.0321 \\
177 +Max. & 0.2051 & 0.4563 & 0.3923 & 0.4477 & 0.7330 & 0.1867 \\
178 +\midrule
179 + & & & \textbf{$NLS_{SD}$} & & & \\
180 +\midrule
181 +Min. & -0.2645 & -0.4092 & -0.2263 & -0.2263 & -0.3640 & -0.0689 \\
182 +1st Qu. & -0.1119 & -0.1457 & 0.0238 & -0.0255 & -0.0251 & 0.1091 \\
183 +Median & -0.1119 & -0.1457 & 0.2398 & -0.0255 & -0.0251 & 0.1091 \\
184 +Mean & -0.0897 & -0.1612 & 0.2395 & -0.0276 & 0.0086 & 0.1087 \\
185 +3rd Qu. & 0.1847 & 0.1290 & 0.3923 & 0.3334 & 0.5652 & 0.2743 \\
186 +Max. & 0.1847 & 0.1290 & 0.3923 & 0.3334 & 0.5652 & 0.2743 \\
187 +\bottomrule
188 +\end{tabular}
189 +\begin{tablenotes}
190 +\singlespacing
191 +\footnotesize
192 +This table provides descriptive statistics of the NLS proxy for speculative trading intensity for each commodity futures contract in our sample. The lines (Min.), (1st Qu.), (Median), (Mean), (3rd Qu.) and (Max) present respectively the minimum value, the first quartile, the median, the mean, the third quartile, and the maximum value. CL: crude oil, GC: gold, HG: high-grade copper, SI: silver, PA: palladium, NG: natural gas.
193 +\end{tablenotes}
194 +\end{table}
195 +
196 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
197 +
198 +\begin{sidewaystable}
199 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (NLS) on Futures Returns - Full Sample}
200 +\label{tab:macro_fin_nls_fut_returns_full}
201 +\centering
202 +\resizebox{\linewidth}{!}{%
203 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
204 +%\begin{tabular}{@{}lllllllllllll@{}}
205 +\toprule
206 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
207 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
208 +
209 +\textbf{Initial jobless claims} & -0.19763*** & 0.87888*** & 0.70022*** & -1.33955*** & -0.44930*** & 1.98801*** & -0.00740 & -0.01540 & 0.02913** & -0.22057*** & -0.00266 & -0.10938 \\
210 +\textbf{ADP Employment} & 0.34291*** & -1.06756** & -1.18741*** & 2.52498*** & 0.58107*** & -2.36989*** & -0.02591*** & 0.16900 & 0.00801 & -0.14170 & -0.11464*** & -2.79764*** \\
211 +\textbf{CB Consumer} & 0.11672*** & -0.46895*** & -0.05938*** & 0.07030* & 0.07957*** & -0.15159** & 0.00049 & 0.00567 & -0.06841*** & 0.11631** & 0.02933 & -0.01440 \\
212 +\textbf{Advance retail sales} & 0.19991*** & -0.74542*** & -0.22861*** & 0.44747*** & 0.10409*** & -0.42536** & 0.01333 & -0.08638 & -0.01767 & 0.00664 & -0.02725 & -0.47149 \\
213 +\textbf{Building permit} & -0.00037 & 0.03995 & -0.02651** & 0.03697 & 0.06270*** & -0.15653** & 0.00698 & -0.00358 & -0.00233 & -0.03748 & 0.04272* & 0.36164** \\
214 +\textbf{Construction spending} & 0.02190 & -0.14609 & -0.04079** & 0.09211** & 0.04044* & -0.14376* & -0.00070 & -0.18868*** & 0.00951 & -0.00257 & -0.02886 & -0.17899 \\
215 +\textbf{Consumer credit} & -0.01646 & 0.08344 & -0.01359* & 0.02955 & -0.01554* & 0.06217 & 0.00217 & 0.02437 & 0.01087 & -0.01043 & -0.01175 & -0.11173 \\
216 +\textbf{Consumer price index} & 0.08391*** & -0.29308** & -0.17224*** & 0.31022*** & 0.26865*** & -0.75945*** & -0.03002*** & -0.02265 & -0.14233*** & 0.23064*** & 0.01302 & 0.15012 \\
217 +\textbf{Durable goods orders} & 0.15300*** & -0.73816*** & -0.07265*** & 0.14703*** & 0.06435*** & -0.14930* & -0.01052* & 0.00024 & -0.05354*** & 0.14525*** & -0.00328 & -0.09972 \\
218 +\textbf{Existing home sales} & 0.08722*** & -0.59238*** & -0.02000 & 0.05026 & -0.02244 & 0.10822 & -0.02238*** & -0.06905 & 0.01177 & -0.04488 & 0.04083 & 0.21897 \\
219 +\textbf{Factory orders} & -0.00112 & 0.01036 & -0.02949 & 0.00285 & -0.05976** & 0.14591 & 0.00357 & 0.02488 & -0.01887 & 0.00026 & 0.10729*** & 0.68503*** \\
220 +\textbf{Gross domestic product} & 0.04571* & -0.18390 & -0.16085*** & 0.24165*** & 0.16953*** & -0.33056*** & 0.00920 & -0.00478 & -0.05405*** & 0.00789 & -0.03601 & -0.33730* \\
221 +\textbf{Housing starts} & 0.03224 & -0.12690 & -0.06263*** & 0.11383*** & 0.07118*** & -0.14876** & 0.00503 & 0.00563 & 0.00159 & -0.03648* & 0.02340 & 0.38343** \\
222 +\textbf{Industrial production} & 0.02075 & -0.15376 & -0.00085 & -0.07097 & -0.01332 & -0.11571 & -0.00947 & -0.00797 & -0.02227 & 0.02109 & -0.02279 & -0.02294 \\
223 +\textbf{New home sales} & 0.09853*** & -0.43638** & -0.06940*** & 0.13595*** & -0.03309* & -0.04095 & -0.02368*** & 0.08652* & -0.00796 & 0.02235 & -0.00278 & -0.08427 \\
224 +\textbf{Non-farm employment} & 1.40422*** & -5.11125*** & -3.19414*** & 6.81185*** & 1.38897*** & -6.04367*** & -0.02702*** & 0.62417** & 0.00013 & 0.15340 & -0.21810*** & -6.29434*** \\
225 +\textbf{Pending home sales} & 0.07333*** & -0.34943** & -0.01840 & 0.00310 & -0.01292 & -0.02301 & -0.01712** & -0.11126 & -0.00877 & 0.03099 & -0.02417 & -0.39913 \\
226 +\textbf{Personal consumption} & -0.01478 & 0.06799 & -0.03594** & 0.06306* & 0.00926 & -0.05828 & 0.00274 & 0.08690 & 0.01300 & -0.05027 & 0.00370 & 0.36897 \\
227 +\textbf{Personal income} & 0.00873 & -0.13628 & -0.07190 & 0.15000 & 0.25364*** & -1.05220** & -0.01709* & 0.27874** & -0.01271 & -0.02240 & -0.02828 & -0.82024 \\
228 +\textbf{Producer price index} & 0.03513* & -0.23561** & -0.07698*** & 0.15026*** & -0.01476 & -0.04153 & -0.00480 & -0.01293 & -0.02035 & 0.04451 & -0.04862* & -0.22346 \\
229 +\textbf{Trade balance} & 0.00320 & -0.05635 & -0.05207** & 0.11911** & -0.01111 & 0.04144 & -0.00354 & -0.13232 & -0.03871 & 0.07731 & -0.04667 & -0.45064 \\
230 +\textbf{Crude Oil Weekly inventory} & -0.07659*** & -0.39633*** & & & & & & & & & & \\
231 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.36472*** & -0.07672 \\ \midrule
232 +
233 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
234 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{0.198} & \multicolumn{2}{c}{0.204} & \multicolumn{2}{c}{0.101} & \multicolumn{2}{c}{0.0290} & \multicolumn{2}{c}{0.0246} & \multicolumn{2}{c}{0.118} \\ \bottomrule
235 +\end{tabular}
236 +}
237 +\begin{tablenotes}
238 +\singlespacing
239 +\footnotesize
240 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and speculative trading intensity variable $X_{t}=NLS_t$. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in the futures return when an announcement has just occurred. The $\theta_m$ coefficients capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of speculative trading intensity.
241 +\end{tablenotes}
242 +\end{sidewaystable}
243 +
244 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
245 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
246 +
247 +\begin{sidewaystable}
248 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (NLS) on Futures Conditional Variance - Full Sample}
249 +\label{tab:macro_fin_nls_var_full}
250 +\centering
251 +\resizebox{\linewidth}{!}{%
252 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
253 +%\begin{tabular}{@{}lllllllllllll@{}}
254 +\toprule
255 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
256 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
257 +
258 +\textbf{Initial jobless claims} & 0.10116*** & -0.43618*** & 0.04743*** & 0.01989 & 0.11651*** & -0.18846*** & 0.02474*** & -0.05555*** & 0.11786*** & -0.11355*** & -0.01167 & -0.02107 \\
259 +\textbf{ADP Employment} & 0.02792 & -0.20043 & 0.06377*** & -0.08427*** & 0.05327*** & -0.03006 & 0.02831*** & 0.03382 & -0.02858* & 0.01538 & 0.01600 & 0.05689 \\
260 +\textbf{CB Consumer} & 0.10356*** & -0.41440*** & 0.03139*** & 0.01209 & 0.04540*** & 0.02788 & 0.02838*** & -0.10290** & 0.06091*** & -0.09895*** & -0.00770 & -0.01276 \\
261 +\textbf{Advance retail sales} & 0.16868*** & -0.86165*** & 0.09604*** & -0.05612* & 0.12379*** & -0.05358 & 0.02819*** & -0.15572*** & 0.11351*** & -0.16926*** & 0.05769*** & 0.25105* \\
262 +\textbf{Building permit} & 0.10470*** & -0.54035*** & 0.04592*** & -0.06034* & 0.06133*** & -0.19301*** & 0.00484 & -0.07408* & 0.07672*** & -0.06975*** & 0.02190 & -0.02627 \\
263 +\textbf{Construction spending} & 0.15338*** & -0.62724*** & 0.09518*** & -0.11889*** & 0.08606*** & -0.09793 & 0.04108*** & -0.18959*** & 0.04626*** & 0.01330 & -0.01904 & -0.32089** \\
264 +\textbf{Consumer credit} & 0.04798* & -0.26103* & 0.00778 & -0.00318 & 0.01661 & 0.00378 & 0.00280 & 0.03858 & -0.00633 & 0.03412 & 0.00152 & 0.01392 \\
265 +\textbf{Consumer price index} & 0.07237** & -0.06680 & 0.13915*** & -0.07513* & 0.26627*** & -0.52898*** & 0.09389*** & -0.37452*** & 0.18224*** & -0.27404*** & 0.02706 & 0.03922 \\
266 +\textbf{Durable goods orders} & 0.09826*** & -0.41079*** & 0.02310** & 0.01561 & 0.04156*** & 0.02205 & 0.00773 & -0.02411 & 0.06781*** & -0.03730 & -0.02225 & -0.25576* \\
267 +\textbf{Existing home sales} & 0.04757* & -0.05948 & 0.01879* & 0.04102 & 0.03838*** & -0.02961 & 0.02231*** & -0.13249*** & 0.04815*** & -0.06749** & 0.02201 & -0.11251 \\
268 +\textbf{Factory orders} & 0.08762*** & -0.31063** & -0.00678 & 0.16162*** & 0.01842 & 0.15052** & 0.02641*** & -0.13159*** & 0.03966*** & -0.04030 & -0.03041 & -0.69494*** \\
269 +\textbf{Gross domestic product} & 0.11632*** & -0.61837*** & 0.04659*** & 0.05303 & 0.09959*** & -0.06323 & 0.02464*** & 0.01653 & 0.11025*** & -0.12093*** & -0.05116** & -0.30947** \\
270 +\textbf{Housing starts} & 0.11313*** & -0.60745*** & 0.03660*** & -0.03112 & 0.06213*** & -0.19012*** & 0.00848 & -0.09479** & 0.05963*** & -0.04924 & 0.01829 & 0.04047 \\
271 +\textbf{Industrial production} & 0.09578*** & -0.63833*** & 0.01928* & 0.00308 & 0.02059 & -0.00706 & 0.00182 & -0.04794 & 0.01964 & -0.07145** & -0.03281 & -0.18705 \\
272 +\textbf{New home sales} & 0.10149*** & -0.52688*** & 0.04068*** & 0.01515 & 0.05778*** & 0.02800 & 0.02485*** & -0.00096 & 0.06768*** & -0.11522*** & -0.02516 & -0.22929* \\
273 +\textbf{Non-farm employment} & 0.36475*** & -1.20583*** & 0.23592*** & 0.04682 & 0.38126*** & -0.14107** & 0.12881*** & -0.15602*** & 0.17336*** & -0.05048 & 0.02807 & -0.35383** \\
274 +\textbf{Pending home sales} & 0.10920*** & -0.49393*** & 0.01999* & -0.01624 & 0.02548* & 0.02096 & 0.02833*** & -0.00831 & 0.05870*** & -0.08511*** & -0.01374 & -0.30813** \\
275 +\textbf{Personal consumption} & 0.00638 & 0.02122 & 0.06203*** & -0.08145** & 0.10177*** & -0.26781*** & 0.00662 & -0.00781 & 0.11081*** & -0.14798*** & 0.02410 & -0.37604** \\
276 +\textbf{Personal income} & 0.00650 & 0.04657 & 0.09149*** & -0.15016*** & 0.12450*** & -0.34101*** & 0.01322** & -0.02338 & 0.09417*** & -0.10866*** & 0.00283 & -0.43167*** \\
277 +\textbf{Producer price index} & 0.09735*** & -0.34959** & 0.01720 & 0.08462** & 0.09045*** & -0.09926 & 0.01106* & -0.05096 & 0.11987*** & -0.06696** & 0.00810 & 0.00477 \\
278 +\textbf{Trade balance} & 0.07844*** & -0.12829 & -0.00650 & 0.20643*** & 0.00273 & 0.39295*** & 0.00919 & 0.02977 & 0.10323*** & -0.08686*** & 0.01821 & 0.11158 \\
279 +\textbf{Crude Oil Weekly inventory} & 0.00456 & 0.62373*** & & & & & & & & & & \\
280 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.53064*** & -0.68665*** \\ \midrule
281 +
282 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
283 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{7.065} & \multicolumn{2}{c}{8.064} & \multicolumn{2}{c}{7.396} & \multicolumn{2}{c}{7.057} & \multicolumn{2}{c}{7.629} & \multicolumn{2}{c}{13.60} \\ \bottomrule
284 +\end{tabular}
285 +}
286 +\begin{tablenotes}
287 +\singlespacing
288 +\footnotesize
289 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the speculative trading intensity variable $NLS_t$. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2007-04-01 to 2024-02-11. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of speculative trading intensity.
290 +\end{tablenotes}
291 +\end{sidewaystable}
292 +
293 +
294 +
295 +
296 +
297 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
298 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
299 +\begin{sidewaystable}
300 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (NLS) on Futures Price Bid-Ask Spreads - Full Sample}
301 +\label{tab:return-fin-full}
302 +\centering
303 +\resizebox{\linewidth}{!}{%
304 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
305 +%\begin{tabular}{@{}lllllllllllll@{}}
306 +\toprule
307 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Palladium}} \\ \midrule
308 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule
309 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
310 +\textbf{Initial jobless claims} & -2.7096*** & 13.6978 & 5.4253 & -10.3796*** & 5.8502 & -26.3969*** & -0.2493 & -4.5426*** & 0.0148 & 0.8567 & -0.1912 & 0.5368 \\
311 +\textbf{ADP Employment} & 3.6287 & -10.9417*** & -10.5416*** & 22.3535 & -7.3992*** & 31.3360 & -0.6515 & -13.8892*** & 0.2498 & 2.0429 & 0.0364 & -0.4670\\
312 +\textbf{CB Consumer} & 1.4188 & -6.2213*** & -0.8502 & 1.4474 & -1.1319 & 3.0482 & 0.5121 & 0.3016 & 0.0264 & 0.2419 & -0.7750 & 1.9280 \\
313 +\textbf{Advance retail sales} & 2.0886 & -7.0546*** & -2.2792 & 4.4308 & -0.8497 & 2.8479 & -0.2272 & -2.0227 & 0.3505 & -3.2196*** & -0.4188 & 1.5128 \\
314 +\textbf{Building permit} & -0.1970 & 1.7823 & -0.4372 & 0.6514 & -0.7742 & 2.5610 & 0.3011 & 3.7647 & 0.0299 & 0.3677 & -0.1989 & -0.2028 \\
315 +\textbf{Construction spending} & 0.1768 & -1.0346 & -0.0637 & 0.0761 & -0.2650 & 0.8472 & -0.1867 & -2.6962*** & 0.0609 & -2.4199*** & -0.1259 & 0.2982 \\
316 +\textbf{Consumer credit} & 0.3706 & -1.8987 & -0.0269 & 0.0457 & -0.0288 & 0.1356 & -0.0409 & 0.0473 & 0.0549 & 0.3589 & -0.0071 & 0.0208 \\
317 +\textbf{Consumer price index} & -0.8292 & 1.3319 & -2.2389 & 4.8692 & -3.0533 & 9.4014 & -0.0417 & 1.9823 & -0.5620 & 1.0961 & -1.8277 & 3.2679 \\
318 +\textbf{Durable goods orders} & 1.3175 & -7.3678*** & -0.9559 & 2.3132 & -1.3099 & 5.5532 & -0.1211 & -1.4967 & 0.2027 & -0.1838 & -0.3876 & 1.2514 \\
319 +\textbf{Existing home sales} & 1.0542 & -5.3658*** & -0.3977 & 1.0670 & -0.1607 & 1.4199 & 1.0635 & 5.2952 & 0.4218 & -1.8117 & 0.0162 & -0.0411 \\
320 +\textbf{Factory orders} & -0.3378 & 1.5881 & -0.8512 & 1.3429 & -0.7687 & 1.8326 & 1.5490 & 11.3206 & 0.0137 & 1.0877 & -0.0331 & -0.3245 \\
321 +\textbf{GDP} & 0.3581 & -1.5995 & -1.8905 & 3.7376 & -2.2481 & 6.3831 & -0.3939 & -3.6342*** & 0.1213 & -0.7319 & -0.5648 & 0.3738 \\
322 +\textbf{Housing starts} & 0.1898 & -1.2327 & -0.9901 & 1.8960 & -0.9695 & 2.5715 & 0.5413 & 6.5382 & 0.0832 & 0.5107 & 0.0362 & -0.5628 \\
323 +\textbf{Industrial production} & 0.1195 & -2.0990 & -0.1954 & -0.4140 & -0.0606 & -1.9941 & -0.2034 & -1.1923 & -0.0584 & -0.4020 & -0.1814 & 0.2817 \\
324 +\textbf{Manufacturing capacity} & -0.1028 & 2.2785 & -0.6122 & 0.9215 & -0.8221 & 1.9754 & -0.3658 & -2.8980*** & -0.0373 & 0.6966 & -0.1408 & 0.5487 \\
325 +\textbf{New home sales} & 1.6308 & -8.3434*** & -0.9063 & 1.6516 & -0.3849 & -0.4070 & -0.2523 & -0.8854 & 0.3688 & -0.4561 & -0.0098 & 0.0570 \\
326 +\textbf{Non-farm payroll} & 10.7879 & -38.8878*** & -29.7105*** & 63.3421 & -10.2858*** & 43.9569 & -2.4909 & -71.4713*** & 0.2717 & 8.5221 & -0.0000 & 1.4673 \\
327 +\textbf{Pending home sales} & 0.8251 & -2.5452 & -0.4223 & 0.5242 & -0.1941 & -0.1573 & -0.5748 & -9.0134*** & 0.1567 & -2.3599*** & -0.3537 & 0.2819 \\
328 +\textbf{Personal consumption} & -0.2150 & -0.0394 & -0.8262 & 1.6288 & -0.0344 & 0.1194 & -0.1174 & 3.9321 & 0.0919 & 2.0435 & -0.2523 & 0.5773 \\
329 +\textbf{Personal income} & -0.0876 & -0.8200 & -1.5351 & 3.2898 & -4.4002*** & 17.6153 & -0.4605 & -12.8607*** & -0.2459 & -3.8381*** & -0.1619 & 0.2354 \\
330 +\textbf{Producer price index} & 0.7142 & -4.4244*** & -0.7227 & 1.4344 & 0.2196 & -1.8938 & -0.5144 & -2.8027*** & 0.0608 & -0.1209 & -0.7231 & 1.4795 \\
331 +\textbf{Treasury balance} & 0.0459 & -1.6195 & -0.8212 & 1.9771 & -0.8034 & 2.9669 & -0.1261 & -0.7379 & -0.0332 & -1.5029 & -0.3137 & 0.8416 \\
332 +\midrule
333 +\multicolumn{13}{c}{\textbf{Announcements specific to commodity markets}} \\ \midrule
334 +\textbf{Weekly crude oil stock} & -0.9773 & -3.4730*** & & & & & & & & & & \\
335 +\textbf{Natural Gas Weekly inventory} & & & & & & & -1.7473 & -7.0273*** & & & & \\
336 +\midrule
337 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{1,101,836} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{749,168} \\
338 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{1.98} & \multicolumn{2}{c}{1.77} & \multicolumn{2}{c}{0.56} & \multicolumn{2}{c}{1.85} & \multicolumn{2}{c}{1.28} & \multicolumn{2}{c}{1.51} \\
339 +\bottomrule
340 +\end{tabular}
341 +}
342 +\begin{tablenotes}
343 + \singlespacing
344 + \footnotesize
345 + This table presents estimates of the equation $R_{SPREAD_{t}}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+\delta X_{t,i}+\sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{SPREAD_{t-\tau}}^{t}+\epsilon_t$, analyzing the effects of speculative trading intensity and macroeconomic announcements on the bid-ask spread using the speculative trading intensity variable $NLS_t$. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the speculative trading intensity variable $NLS_t$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and speculative trading intensity.
346 +\end{tablenotes}
347 +\end{sidewaystable}
348 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
349 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
350 +\setcounter{table}{8}
351 +\renewcommand\thetable{9(a)}
352 +\begin{sidewaystable}
353 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Money Managers only) on Futures Returns}
354 +\label{tab:macro_fin_mm_fut_returns_full}
355 +\centering
356 +\resizebox{\linewidth}{!}{%
357 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
358 +%\begin{tabular}{@{}lllllllllllll@{}}
359 +\toprule
360 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
361 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
362 +
363 +\textbf{Initial jobless claims} & -0.02956 & 0.01952 & 0.30064*** & -0.96225*** & -0.02630 & 0.33366 & -0.00912 & -0.02954 & 0.04621** & -0.27526*** & 0.00955 & 0.06074 \\
364 +\textbf{ADP Employment} & 0.29189** & -1.23728* & -0.06824 & 0.15437 & 0.22281***& -0.97429** & -0.02291***& 0.14952 & 0.01012 & -0.11814 & -0.02204 & -0.49202 \\
365 +\textbf{CB Consumer} & 0.11505*** & -0.65428** & -0.04763***& 0.05599 & 0.06964***& -0.13979** & 0.00036 & 0.00188 & -0.06685***& 0.11545** & 0.03748** & 0.12201 \\
366 +\textbf{Advance retail sales} & 0.18018*** & -0.95056***& 0.00625 & -0.22754*** & 0.09507***& -0.52284***& 0.01327 & -0.06932 & -0.02139 & 0.02073 & -0.00518 & -0.24418 \\
367 +\textbf{Building permit} & -0.03075 & 0.30773* & -0.01409** & -0.00246 & 0.04883***& -0.12171** & 0.00701 & 0.00070 & -0.00304 & -0.03782 & 0.00348 & 0.08695 \\
368 +\textbf{Construction spending} & 0.04684 & -0.41529 & -0.04794***& 0.17091*** & 0.03370** & -0.15213* & 0.00487 & -0.12621** & 0.00961 & -0.00307 & -0.01120 & -0.09293 \\
369 +\textbf{Consumer credit} & -0.00932 & 0.05736 & -0.01044** & 0.04101* & -0.01023 & 0.05293 & 0.00149 & 0.02019 & 0.01095 & -0.01252 & -0.00177 & -0.04766 \\
370 +\textbf{Consumer price index} & 0.08568*** & -0.42002** & -0.15785***& 0.47194*** & 0.21663***& -0.72688***& -0.02944***& -0.01458 & -0.13577***& 0.22337*** & -0.00076 & 0.07776 \\
371 +\textbf{Durable goods orders} & 0.09316*** & -0.69602***& -0.05849***& 0.15716*** & 0.05268***& -0.11874 & -0.01489** & -0.05119 & -0.06161***& 0.17723*** & 0.00976 & -0.02018 \\
372 +\textbf{Existing home sales} & 0.09536*** & -0.80184***& -0.00676 & 0.01184 & -0.01163 & 0.07853 & -0.02082** & -0.01303 & 0.01267 & -0.05028 & 0.01320 & 0.05330 \\
373 +\textbf{Factory orders} & -0.02429 & 0.21872 & -0.04389***& 0.06866 & 0.05024** & -0.14478 & 0.00322 & 0.01249 & -0.01854 & -0.00031 & 0.04152** & 0.38741** \\
374 +\textbf{Gross domestic product} & 0.04408* & -0.25171 & -0.08308***& 0.02306 & 0.12802***& -0.18740** & 0.00637 & 0.02704 & -0.05987***& 0.02416 & -0.00388 & -0.28244* \\
375 +\textbf{Housing starts} & -0.01672 & 0.19837 & -0.02757***& 0.00801 & 0.04577***& -0.03650 & 0.00348 & 0.02420 & 0.00214 & -0.04053* & -0.01298 & 0.14226 \\
376 +\textbf{Industrial production} & -0.00324 & -0.03528 & -0.02254** & -0.02049 & -0.02215 & -0.10057 & -0.00675 & -0.03402 & -0.03174 & 0.04511 & -0.01321 & 0.23336 \\
377 +\textbf{New home sales} & 0.03959 & -0.20140 & -0.03508***& 0.05297 & 0.03238** & 0.06133 & -0.02410***& -0.04529 & -0.01226 & 0.03240 & 0.00382 & -0.18259 \\
378 +\textbf{Non-farm employment} & 0.79766*** & -4.07313***& -0.51670***& 2.14516*** & 0.77114***& -4.43004***& -0.01496* & 0.61309*** & 0.02467 & -0.16206 & -0.00450 & -0.95575 \\
379 +\textbf{Pending home sales} & 0.05641 & -0.31219 & -0.01885* & 0.00739 & -0.02053 & 0.01626 & -0.01554* & -0.03397 & -0.01511 & 0.05025 & 0.01033 & -0.26951 \\
380 +\textbf{Personal consumption} & 0.09082** & -0.64077***& -0.01948** & 0.04335 & 0.00547 & -0.06372 & -0.00024 & 0.08100 & 0.01162 & -0.04886 & -0.02503 & 0.23224 \\
381 +\textbf{Personal income} & 0.05847 & -0.50101 & 0.01713 & -0.09655 & 0.14463** & -0.83394** & -0.00826 & -0.20026 & -0.01654 & 0.00889 & 0.00405 & -1.05049* \\
382 +\textbf{Producer price index} & 0.03892* & -0.32583* & -0.05523***& 0.13297*** & -0.00600 & -0.12185 & -0.00347 & -0.03082 & -0.02220 & 0.05468 & -0.02632 & -0.11214 \\
383 +\textbf{Trade balance} & 0.00732 & -0.10801 & -0.03076** & 0.08094* & -0.00911 & 0.03998 & 0.00382 & -0.11800 & -0.02494 & 0.06033 & -0.00010 & -0.26851 \\
384 +\textbf{Crude Oil Weekly inventory} & -0.07233*** & -0.61043***& & & & & & & & & & \\
385 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.35459*** & -0.12230 \\ \midrule
386 +
387 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
388 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{0.181} & \multicolumn{2}{c}{0.151} & \multicolumn{2}{c}{0.0923} & \multicolumn{2}{c}{0.0284} & \multicolumn{2}{c}{0.0235} & \multicolumn{2}{c}{0.113} \\ \bottomrule
389 +\end{tabular}
390 +}
391 +\begin{tablenotes}
392 +\singlespacing
393 +\footnotesize
394 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and the money manager-only speculative trading intensity variable $NLS_{t,MM}$, calculated with the money manager positions. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of speculative trading intensity.
395 +\end{tablenotes}
396 +\end{sidewaystable}
397 +
398 +
399 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
400 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
401 +
402 +\renewcommand\thetable{9(b)}
403 +\begin{sidewaystable}
404 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Swap Dealers only) on Futures Returns}
405 +\label{tab:macro_fin_sd_fut_returns_full}
406 +\centering
407 +\resizebox{\linewidth}{!}{%
408 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
409 +%\begin{tabular}{@{}lllllllllllll@{}}
410 +\toprule
411 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
412 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
413 +
414 +\textbf{Initial jobless claims} & -0.18601*** & -0.76285*** & 0.36069*** & 1.02251*** & 0.01407 & -0.51053 & 0.08238** & 0.44801** & 0.00571 & -0.03460 & -0.00709 & 0.23566 \\
415 +\textbf{ADP Employment} & 0.42554*** & 1.77009*** & -0.73172***& -1.92332*** & 0.14323*** & 1.90790*** & 0.10514 & -0.41655 & -0.08432* & -0.37900* & -0.07015***& 1.98134*** \\
416 +\textbf{CB Consumer} & 0.08606*** & 0.38665*** & -0.02682***& 0.06149 & 0.04702*** & 0.04493 & -0.05678***& -0.25240***& -0.02421** & -0.13691** & 0.02902 & 0.01534 \\
417 +\textbf{Advance retail sales} & 0.18764*** & 0.83656*** & -0.22074***& -0.62235*** & 0.02334* & 0.32997 & -0.05224** & -0.34386** & -0.02373 & 0.03729 & -0.01204 & 0.18256 \\
418 +\textbf{Building permit} & 0.01182 & 0.03350 & -0.02794***& -0.06415** & 0.03852*** & 0.19589* & 0.00878 & -0.00836 & -0.01490** & 0.06228 & 0.06712** & -0.57136***\\
419 +\textbf{Construction spending} & 0.00921 & 0.10371 & -0.01531* & -0.03974 & 0.01559 & 0.15444 & -0.04579** & -0.19927** & 0.01295 & 0.06733 & -0.00588 & -0.03512 \\
420 +\textbf{Consumer credit} & -0.00057 & 0.00723 & -0.00297 & 0.00152 & -0.00353 & 0.00472 & 0.00671 & -0.01905 & 0.00686 & 0.02748 & -0.02010 & 0.18485 \\
421 +\textbf{Consumer price index} & -0.05142*** & -0.12282 & -0.04196***& -0.11608** & 0.10927*** & 0.50120*** & -0.07969***& -0.20611** & -0.05426***& -0.28370*** & 0.04152 & -0.40373 \\
422 +\textbf{Durable goods orders} & 0.08229*** & 0.40797*** & -0.02824***& -0.02785 & 0.02337** & 0.27759*** & -0.00067 & 0.04659 & 0.00927 & -0.09008** & 0.01138 & -0.01219 \\
423 +\textbf{Existing home sales} & 0.02521** & 0.28450*** & -0.00669 & -0.02050 & -0.00093 & 0.07534 & -0.06009***& -0.33603***& -0.00636 & 0.05054 & 0.05433 & -0.33599 \\
424 +\textbf{Factory orders} & -0.00312 & -0.03814 & -0.00410 & 0.14500** & 0.03244*** & -0.19383 & 0.01462 & -0.04673 & -0.01972 & -0.03200 & 0.09608*** & -0.62225** \\
425 +\textbf{Gross domestic product} & 0.03848** & 0.17661** & -0.09123***& -0.06216 & 0.09968*** & -0.15355 & -0.02494 & 0.14602** & -0.05188***& -0.00867 & -0.03176 & 0.30154 \\
426 +\textbf{Housing starts} & 0.03038** & 0.13425* & -0.05461***& -0.14857*** & 0.05147*** & 0.21799** & -0.01782 & 0.11965* & -0.01043 & 0.02270 & 0.01927 & -0.32116 \\
427 +\textbf{Industrial production} & 0.02354 & 0.19850* & -0.00350 & 0.10713** & 0.03408*** & 0.14657 & -0.02953 & 0.09807 & -0.01335 & -0.01264 & -0.02947 & 0.09028 \\
428 +\textbf{New home sales} & 0.07156*** & 0.34303*** & -0.04011***& -0.08154* & 0.03797*** & 0.09004 & -0.03774** & -0.28795***& 0.00009 & -0.02615 & 0.01621 & -0.11025 \\
429 +\textbf{Non-farm employment} & 1.81964*** & 8.77192*** & -3.07930***& -8.25792*** & 0.22593*** & 3.88087*** & -0.48535***& -2.66148***& -0.14565* & -0.68396* & -0.06665** & 2.42302** \\
430 +\textbf{Pending home sales} & 0.04227*** & 0.27467*** & -0.01662 & 0.00210 & -0.01694 & 0.07454 & -0.03279 & 0.19949** & 0.00391 & -0.02496 & -0.00429 & 0.25203 \\
431 +\textbf{Personal consumption} & -0.00158 & -0.00543 & -0.01737* & -0.02729 & 0.00538 & 0.32148** & -0.00779 & 0.03095 & -0.00577 & 0.07145 & 0.00273 & -0.38325 \\
432 +\textbf{Personal income} & -0.02221 & 0.00621 & -0.05774* & -0.15405 & 0.08167*** & 1.82948*** & -0.09051 & 0.43742 & -0.02297 & 0.03880 & -0.00815 & 0.53016 \\
433 +\textbf{Producer price index} & 0.00818 & 0.07471 & -0.02420***& -0.00196 & 0.02453*** & 0.08862 & -0.03150 & 0.11098 & -0.00586 & -0.03594 & -0.03630 & 0.15142 \\
434 +\textbf{Trade balance} & -0.00089 & 0.03751 & -0.01618 & -0.06246 & -0.01214 & -0.23857 & -0.04471 & 0.14389 & -0.01350 & -0.22026* & 0.00043 & 0.12505 \\
435 +\textbf{Crude Oil Weekly inventory} & -0.14367*** & 0.03576 & & & & & & & & & & \\
436 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.47323***& 0.90803*** \\ \midrule
437 +
438 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
439 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{0.196} & \multicolumn{2}{c}{0.206} & \multicolumn{2}{c}{0.0809} & \multicolumn{2}{c}{0.0337} & \multicolumn{2}{c}{0.0195} & \multicolumn{2}{c}{0.123} \\ \bottomrule
440 +\end{tabular}
441 +}
442 +\begin{tablenotes}
443 +\singlespacing
444 +\footnotesize
445 +This table presents estimates of eq. \ref{eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and the swap dealer-specific speculative trading intensity variable $NLS_{t,SD}$, calculated with the swap dealer positions. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of speculative trading intensity.
446 +\end{tablenotes}
447 +\end{sidewaystable}
448 +
449 +
450 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
451 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
452 +\setcounter{table}{9}
453 +\renewcommand\thetable{10(a)}
454 +\begin{sidewaystable}
455 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Money Managers only) on Futures Conditional Variance}
456 +\label{tab:macro_fin_mm_var_full}
457 +\centering
458 +\resizebox{\linewidth}{!}{%
459 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
460 +%\begin{tabular}{@{}lllllllllllll@{}}
461 +\toprule
462 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
463 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
464 +
465 +\textbf{Initial jobless claims} & 0.08555*** & -0.47011*** & 0.05181*** & 0.01050 & 0.10409*** & -0.17922*** & 0.02685*** & -0.04784*** & 0.12581*** & -0.13984*** & -0.01102** & -0.04846 \\
466 +\textbf{ADP Employment} & 0.00118 & -0.03201 & 0.04021*** & -0.02019 & 0.05068*** & -0.02457 & 0.02543*** & -0.06310* & -0.02832* & 0.01564 & 0.01156 & 0.05686 \\
467 +\textbf{CB Consumer} & 0.11197*** & -0.66649*** & 0.03980*** & -0.01979 & 0.04828*** & 0.01872 & 0.03063*** & -0.04699 & 0.06245*** & -0.10441*** & -0.00788 & -0.04287 \\
468 +\textbf{Advance retail sales} & 0.14174*** & -0.96296*** & 0.08087*** & -0.01399 & 0.12479*** & -0.08493 & 0.03330*** & -0.11206*** & 0.12530*** & -0.20972*** & 0.03867*** & 0.27501** \\
469 +\textbf{Building permit} & 0.10251*** & -0.75666*** & 0.02103*** & 0.02773 & 0.04613*** & -0.16591*** & 0.00635 & -0.03797 & 0.08523*** & -0.10006*** & 0.02777** & 0.07714 \\
470 +\textbf{Construction spending} & 0.14202*** & -0.78417*** & 0.06648*** & -0.04959* & 0.07206*** & -0.04082 & 0.04891*** & -0.17441*** & 0.05608*** & -0.01314 & 0.00738 & -0.30331*** \\
471 +\textbf{Consumer credit} & 0.03931 & -0.28727 & 0.00369 & 0.01455 & 0.00970 & 0.05397 & 0.00119 & 0.03617 & -0.00823 & 0.03933 & 0.00029 & 0.00914 \\
472 +\textbf{Consumer price index} & 0.05875* & 0.03855 & 0.15937*** & -0.21038***& 0.23368*** & -0.51978*** & 0.10425*** & -0.21797*** & 0.18249*** & -0.28192*** & 0.02276 & 0.01481 \\
473 +\textbf{Durable goods orders} & 0.05652** & -0.21010 & 0.02385*** & 0.02141 & 0.04912*** & -0.02468 & 0.00903 & -0.03221 & 0.07695*** & -0.06578* & 0.00235 & -0.15833 \\
474 +\textbf{Existing home sales} & 0.11456*** & -0.69145*** & 0.02386*** & 0.03978 & 0.04052*** & -0.05757 & 0.02726*** & -0.11502*** & 0.05097*** & -0.07660** & 0.03224*** & -0.08380 \\
475 +\textbf{Factory orders} & 0.16239*** & -1.11410*** & 0.03236*** & -0.07028** & 0.02909*** & -0.13809** & 0.03208*** & -0.13161*** & 0.04159*** & -0.04631 & 0.03273*** & -0.46006*** \\
476 +\textbf{Gross domestic product} & 0.08125*** & -0.53798** & 0.05453*** & 0.04432 & 0.09089*** & -0.02946 & 0.02457*** & 0.00785 & 0.12198*** & -0.15871*** & -0.02217* & -0.24014** \\
477 +\textbf{Housing starts} & 0.11207*** & -0.85733*** & 0.02082*** & 0.02774 & 0.04880*** & -0.17498*** & 0.01059* & -0.05686 & 0.06737*** & -0.07388** & 0.01719 & 0.09484 \\
478 +\textbf{Industrial production} & 0.12021*** & -1.14614*** & 0.01389** & 0.03136 & 0.03199*** & -0.09631 & 0.00429 & -0.06702* & 0.02154 & -0.07965** & -0.01891 & -0.21217* \\
479 +\textbf{New home sales} & 0.07830*** & -0.53409** & 0.04935*** & -0.01730 & 0.06779*** & -0.03253 & 0.02495*** & -0.00382 & 0.06925*** & -0.12089*** & -0.00394 & -0.15829 \\
480 +\textbf{Non-farm employment} & 0.30168*** & -1.11309*** & 0.19942*** & -0.24293***& 0.34791*** & 0.03828 & 0.13642*** & -0.17279*** & 0.17368*** & -0.05267 & 0.05711*** & -0.32357*** \\
481 +\textbf{Pending home sales} & 0.12382*** & -0.82687*** & 0.02174*** & -0.03272 & 0.02609** & 0.02470 & 0.02837*** & 0.00193 & 0.06077*** & -0.09181*** & 0.01005 & -0.33617*** \\
482 +\textbf{Personal consumption} & 0.03205 & -0.20100 & 0.06603*** & -0.14244***& 0.08686*** & -0.27403*** & 0.00835 & -0.03042 & 0.12020*** & -0.17807*** & 0.06250*** & -0.19183 \\
483 +\textbf{Personal income} & 0.03464 & -0.18771 & 0.06633*** & -0.11505***& 0.09950*** & -0.30582*** & 0.01641** & -0.05830 & 0.10262*** & -0.13314*** & 0.04515*** & -0.21897* \\
484 +\textbf{Producer price index} & 0.12303*** & -0.72893*** & 0.03575*** & 0.04448 & 0.08899*** & -0.12987** & 0.01285** & -0.03913 & 0.12713*** & -0.09331** & 0.00855 & 0.02351 \\
485 +\textbf{Trade balance} & 0.10961*** & -0.46684* & 0.01732** & -0.20946***& 0.03656*** & -0.31451*** & 0.00883 & 0.00687 & 0.10595*** & -0.09640*** & 0.00668 & 0.03687 \\
486 +\textbf{Crude Oil Weekly inventory} & 0.05271* & 0.55262** & & & & & & & & & & \\
487 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.59765*** & -0.38025*** \\ \midrule
488 +
489 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
490 +\textbf{$R^2$} & \multicolumn{2}{c}{7.869} & \multicolumn{2}{c}{8.237} & \multicolumn{2}{c}{7.398} & \multicolumn{2}{c}{7.119} & \multicolumn{2}{c}{7.50} & \multicolumn{2}{c}{13.643} \\ \bottomrule
491 +\end{tabular}
492 +}
493 +\begin{tablenotes}
494 +\singlespacing
495 +\footnotesize
496 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the speculative trading intensity variable $NLS_{t,MM}$, calculated with the money manager positions. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2007-04-01 to 2024-02-11. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of speculative trading intensity.
497 +\end{tablenotes}
498 +\end{sidewaystable}
499 +
500 +
501 +
502 +
503 +
504 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
505 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
506 +
507 +\setcounter{table}{9}
508 +\renewcommand\thetable{10(b)}
509 +\begin{sidewaystable}
510 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Swap Dealers only) on Futures Conditional Variance }
511 +\label{tab:macro_fin_sd_var_full}
512 +\centering
513 +\resizebox{\linewidth}{!}{%
514 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
515 +%\begin{tabular}{@{}lllllllllllll@{}}
516 +\toprule
517 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
518 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
519 +
520 +\textbf{Initial jobless claims} & 0.05822*** & 0.23343*** & 0.03845*** & 0.09404*** & 0.08327*** & 0.26382*** & 0.01604** & 0.17118*** & 0.07776*** & 0.21244*** & -0.02694** & 0.14805* \\
521 +\textbf{ADP Employment} & 0.01013 & 0.12754 & 0.03466*** & -0.00761 & 0.04490*** & -0.11581 & -0.00362 & 0.13183** & -0.02324** & -0.06171 & 0.01926 & -0.08546 \\
522 +\textbf{CB Consumer} & 0.06461*** & 0.24214*** & 0.02474*** & 0.06479** & 0.05249*** & 0.09368 & 0.06041***& 0.37330*** & 0.02534** & 0.16265*** & 0.00279 & -0.07654 \\
523 +\textbf{Advance retail sales} & 0.09553*** & 0.54512*** & 0.07545*** & -0.01412 & 0.11206*** & -0.05012 & 0.06957***& 0.41281*** & 0.05474*** & 0.28103*** & 0.04847** & -0.17485 \\
524 +\textbf{Building permit} & 0.04924*** & 0.24970*** & 0.03390*** & 0.04372 & 0.02824*** & 0.23562*** & 0.06755***& 0.32009*** & 0.05311*** & 0.13539*** & -0.01950 & 0.37466** \\
525 +\textbf{Construction spending} & 0.08554*** & 0.28910*** & 0.05709*** & 0.00763 & 0.06924*** & 0.19964** & 0.07736***& 0.49540*** & 0.05126*** & 0.03006 & 0.00924 & 0.08375 \\
526 +\textbf{Consumer credit} & 0.02409* & 0.15378* & 0.00977 & 0.01706 & 0.01612** & -0.06572 & 0.01001 & -0.03267 & 0.00416 & -0.08387 & -0.00125 & 0.00774 \\
527 +\textbf{Consumer price index} & 0.03913** & 0.23500** & 0.08146*** & 0.20522*** & 0.16575*** & 0.29275*** & 0.05399** & 0.17749** & 0.08766*** & 0.44880*** & -0.02157 & 0.35991* \\
528 +\textbf{Durable goods orders} & 0.06383*** & 0.28226*** & 0.02366*** & -0.02745 & 0.04815*** & 0.19932** & 0.03407** & 0.17902*** & 0.05428*** & 0.10969** & -0.04285** & 0.43043***\\
529 +\textbf{Existing home sales} & 0.03173** & -0.05715 & 0.03054*** & -0.00958 & 0.03451*** & 0.13272 & -0.02703 & 0.21180*** & 0.02430** & 0.16807*** & 0.03280 & 0.02110 \\
530 +\textbf{Factory orders} & 0.06008*** & 0.19180** & 0.01488** & 0.19247*** & 0.04458*** & 0.23291*** & -0.02611 & 0.21673*** & 0.02510** & 0.08760 & -0.02845 & 0.67174***\\
531 +\textbf{Gross domestic product} & 0.05612*** & 0.34455*** & 0.02232*** & 0.24770*** & 0.08973*** & 0.16653* & -0.00972 & 0.13666* & 0.06717*** & 0.24490*** & -0.05860** & 0.36187** \\
532 +\textbf{Housing starts} & 0.04932*** & 0.28860*** & 0.02698*** & 0.00304 & 0.02806*** & 0.19249** & 0.06481***& 0.31655*** & 0.04260*** & 0.09632* & -0.01498 & 0.24401 \\
533 +\textbf{Industrial production} & 0.03118** & 0.31942*** & 0.03257*** & 0.07688** & 0.01915*** & 0.02450 & -0.01125 & 0.05783 & -0.00415 & 0.13347** & -0.02739 & 0.14104 \\
534 +\textbf{New home sales} & 0.04779*** & 0.27850*** & 0.03898*** & -0.04063 & 0.06594*** & 0.18551** & -0.02060 & 0.19254*** & 0.02760*** & 0.20740*** & -0.02547 & 0.23051 \\
535 +\textbf{Non-farm employment} & 0.26824*** & 0.84150*** & 0.24170*** & -0.05819* & 0.36074*** & 0.48117*** & 0.01816 & 0.47516*** & 0.15490*** & 0.18734*** & 0.06170*** & 0.06634 \\
536 +\textbf{Pending home sales} & 0.06020*** & 0.26392*** & 0.01544** & 0.00401 & 0.03015*** & 0.03614 & 0.03653** & 0.26870*** & 0.02869*** & 0.15020*** & 0.01378 & 0.07586 \\
537 +\textbf{Personal consumption} & 0.00517 & -0.04382 & 0.03137*** & -0.01852 & 0.05089*** & 0.26220*** & 0.01457 & -0.04014 & 0.05675*** & 0.31626*** & -0.04473* & 0.91871***\\
538 +\textbf{Personal income} & 0.00784 & -0.05332 & 0.06246*** & 0.12366*** & 0.06217*** & 0.45491*** & 0.01766 & -0.02283 & 0.05419*** & 0.23258*** & -0.06483***& 0.99017***\\
539 +\textbf{Producer price index} & 0.06156*** & 0.16722* & 0.03764*** & -0.04623 & 0.07112*** & 0.05176 & -0.02949 & 0.16895** & 0.09461*** & 0.18704*** & 0.02017 & -0.10796 \\
540 +\textbf{Trade balance} & 0.07471*** & 0.15035* & 0.03543*** & 0.15511*** & 0.06670*** & 0.75961*** & 0.01410 & -0.02210 & 0.07275*** & 0.12279** & 0.01722 & -0.10359 \\
541 +\textbf{Crude Oil Weekly inventory} & 0.11644*** & 0.00487 & & & & & & & & & & \\
542 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.48444*** & 1.02519***\\ \midrule
543 +
544 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
545 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{7.01} & \multicolumn{2}{c}{7.716} & \multicolumn{2}{c}{8.023} & \multicolumn{2}{c}{7.975} & \multicolumn{2}{c}{7.732} & \multicolumn{2}{c}{13.66} \\ \bottomrule
546 +\end{tabular}
547 +}
548 +
549 +\begin{tablenotes}
550 +\singlespacing
551 +\footnotesize
552 +This table presents estimates of eq. \ref{eqn:VarianceEqn} using the speculative trading intensity variable $NLS_{t,SD}$, calculated with the swap dealer positions. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2007-04-01 to 2024-02-11. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of speculative trading intensity.
553 +\end{tablenotes}
554 +\end{sidewaystable}
555 +
556 +
557 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
558 +
559 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
560 +\setcounter{table}{10}
561 +\renewcommand\thetable{11(a)} \begin{sidewaystable}
562 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Money Managers only) on Futures Price Bid-Ask Spreads}
563 +\label{tab:macro_fin_mm_fut_spread_full}
564 +\centering
565 +\resizebox{\linewidth}{!}{%
566 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
567 +%\begin{tabular}{@{}lllllllllllll@{}}
568 +\toprule
569 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
570 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
571 +
572 +\textbf{Initial jobless claims} & -2.46565** & 9.75050 & 0.22202 & -6.10718 & -0.41391 & -2.99267 & -1.52013** & -5.80854 & -7.73754* & 10.17312 & -2.06909** & 2.89216 \\
573 +\textbf{ADP Employment} & -1.10495 & 4.87377 & -0.03119 & 0.31326 & -0.44597 & -0.60609 & -0.46357 & 0.04112 & 11.32547 & -20.08663 & -1.56664 & 9.51315 \\
574 +\textbf{CB Consumer} & -0.88291 & -7.59047 & -1.07264** & -21.88274***& -0.60931 & -6.32195 & -2.26961* & -5.00564 & -4.89368 & 6.32172 & -6.10967***& 29.78721 \\
575 +\textbf{Advance retail sales} & -1.75295 & 6.53863 & -1.84813 & 4.41188 & -1.50715* & -4.13973 & -1.11161 & -5.34223 & -8.47285 & 16.81488 & -2.52082 & -23.72607 \\
576 +\textbf{Building permit} & 0.04804 & -5.08377 & 0.12315 & -2.62534 & -0.43797 & -3.66026 & -0.85890 & -0.39189 & -16.80677** & -9.47013 & -2.48078 & -3.85074 \\
577 +\textbf{Construction spending} & 0.71484 & -21.50516 & 0.41072 & -19.63649** & -2.03618** & -6.82194 & -0.78189 & 1.91535 & -3.65053 & 14.68763 & -6.18198***& 14.39348 \\
578 +\textbf{Consumer credit} & -0.39090 & -2.14483 & -0.23236 & 0.92328 & -0.63429 & -0.72773 & 0.72304 & 4.26292 & 2.89371 & 0.82550 & -2.08503 & -10.93259 \\
579 +\textbf{Consumer price index} & -1.79444 & 13.11087 & -0.32839 & -2.43101 & -0.59389 & -0.57386 & -1.52632 & -7.46228 & -28.00983** & 33.08946 & -2.16619 & 17.81488 \\
580 +\textbf{Durable goods orders} & -2.20101 & 5.98876 & 0.23721 & -7.01173 & -1.02867 & -5.37622 & -2.46602* & -19.42954** & -19.84697** & -52.26667***& -3.50474 & 22.77345 \\
581 +\textbf{Existing home sales} & -0.49431 & -6.37862 & 0.15230 & -2.61367 & -0.83207 & -5.40951 & -1.45784 & -2.82054 & -5.47968 & 6.15830 & -5.71583***& 23.65943 \\
582 +\textbf{Factory orders} & -2.69030 & 15.96650 & -0.51700 & -6.28637 & -1.62052* & -9.04768 & -1.96193 & -1.64059 & -0.55217 & 28.54841 & -4.09505** & 17.20082 \\
583 +\textbf{Gross domestic product} & -3.81281 & 16.72542 & -1.27400 & -5.90060 & -1.18407 & -6.39056 & -1.55842 & -17.41209* & -10.42842 & 13.46385 & -2.79300 & 6.61661 \\
584 +\textbf{Housing starts} & -0.26839 & -3.67286 & 0.02239 & -2.47882 & -0.32593 & -3.62060 & -0.12310 & -1.59499 & -14.93911* & -11.91340 & -2.48037 & -3.10561 \\
585 +\textbf{Industrial production} & 0.99908 & -10.14181 & -0.03976 & 8.87670 & -0.26830 & -0.37440 & 0.52140 & 0.47759 & 7.05215 & -13.19391 & -3.89105* & 10.56246 \\
586 +\textbf{New home sales} & 0.51742 & -13.67619 & -0.31951 & -4.68388 & -0.54238 & -2.95472 & -0.87358 & -12.35468 & -8.91154 & 6.44362 & -4.41648** & 11.79880 \\
587 +\textbf{Non-farm employment} & -1.43941 & 1.49780 & -0.83880 & -7.00960 & -3.09250***& -24.19358***& -0.26216 & -5.05771 & -19.80034** & 22.46053 & -2.06886 & 19.68747 \\
588 +\textbf{Pending home sales} & -1.22436 & 4.32860 & 0.23288 & -3.30352 & -0.65997 & -2.14049 & -2.38112* & -5.64287 & 4.60257 & -14.70125 & -5.08845***& 10.99935 \\
589 +\textbf{Personal consumption} & -1.58014 & 8.27394 & -0.38176 & 2.58232 & 0.28583 & -1.30713 & -0.71712 & -10.63446 & -14.71023* & -0.85336 & -2.60387 & 16.06990 \\
590 +\textbf{Personal income} & -1.83332 & 4.18659 & 0.03450 & -0.93297 & -0.54847 & -3.16160 & -1.78660 & -3.12819 & -18.44907** & -7.56018 & -3.68559 & 15.01741 \\
591 +\textbf{Producer price index} & 1.23242 & -25.45917 & 0.17337 & -5.42396 & -1.58493 & -8.72610 & 0.46325 & -4.84467 & -3.85292 & -6.53376 & -0.31231 & 7.49785 \\
592 +\textbf{Trade balance} & 0.77994 & -13.30115 & -0.00374 & 0.05686 & 0.13732 & -10.60780 & -0.44476 & -5.85479 & -11.44187 & 39.18528 & -0.22393 & 5.23703 \\
593 +\textbf{Crude Oil Weekly inventory} & 0.49358 & -12.74294 & & & & & & & & & & \\
594 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -7.59576***& -19.17343* \\ \midrule
595 +
596 +\textbf{Observations} & \multicolumn{2}{c}{1,041,497} & \multicolumn{2}{c}{1,022,592} & \multicolumn{2}{c}{1,023,897} & \multicolumn{2}{c}{1,022,006} & \multicolumn{2}{c}{683,875} & \multicolumn{2}{c}{995,919} \\
597 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{0.0572} & \multicolumn{2}{c}{0.115} & \multicolumn{2}{c}{0.12} & \multicolumn{2}{c}{0.118} & \multicolumn{2}{c}{0.034} & \multicolumn{2}{c}{0.0942} \\ \bottomrule
598 +\end{tabular}
599 +}
600 +\begin{tablenotes}
601 + \singlespacing
602 + \footnotesize
603 + This table presents estimates of the equation $R_{\text{SPREAD}}_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{\text{SPREAD}}_{t-\tau}^{t}+\epsilon_{t}$, analyzing the effects of speculative trading intensity and macroeconomic announcements on the bid-ask spread using the speculative trading intensity variable $NLS_{t,MM}$, calculated with the money manager positions. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the speculative trading intensity variable $NLS_{t,MM}$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and speculative trading intensity. The $\beta$ coefficient represents the effect of the lagged bid-ask spread.
604 +\end{tablenotes}
605 +
606 +\end{sidewaystable}
607 +
608 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
609 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
610 +\setcounter{table}{10}
611 +\renewcommand\thetable{11(b)}
612 +\begin{sidewaystable}
613 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Swap Dealers only) on Futures Price Bid-Ask Spreads}
614 +\label{tab:macro_fin_sd_fut_spread_full}
615 +\centering
616 +\resizebox{\linewidth}{!}{%
617 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
618 +%\begin{tabular}{@{}lllllllllllll@{}}
619 +\toprule
620 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
621 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
622 +
623 +\textbf{Initial jobless claims} & -2.76158*** & 8.99275* & -2.06041 & -5.31071 & -0.63502 & -2.29125 & 1.03040 & -14.26906 & -4.55195 & -14.41688 & -0.23562 & -20.00611 \\
624 +\textbf{ADP Employment} & -1.53692 & -6.19939 & 1.56579 & 4.65731 & -1.00409 & -6.69557 & -0.02006 & -3.16419 & 4.42780 & 8.98661 & 1.48355 & -32.33990 \\
625 +\textbf{CB Consumer} & -2.10022 & -2.32545 & -8.62678** & -22.90121 & -1.03518 & -2.63864 & 1.08755 & -19.13223 & -2.84047 & -14.63955 & 1.02725 & 76.98272*** \\
626 +\textbf{Advance retail sales} & -0.39770 & 4.05260 & -4.03721 & -14.26039 & -2.88363** & 16.59467* & -1.58737 & 4.11864 & -3.70598 & -15.43327 & -3.91203 & 24.52836 \\
627 +\textbf{Building permit} & -0.47588 & 0.18138 & -1.72855 & -6.97984 & -0.98877 & -5.01949 & -2.65095 & 12.39262 & -20.04496***& 17.71476 & -1.26962 & -10.40198 \\
628 +\textbf{Construction spending} & -2.17096 & -3.44095 & -3.30488 & 5.82834 & -3.51547** & -15.86227 & 1.24637 & -12.79038 & 1.03029 & -16.52376 & -2.31365 & -39.77206 \\
629 +\textbf{Consumer credit} & -1.20889 & -3.80213 & -4.20991 & -19.23425 & -1.53730 & -10.76448 & -2.64266 & 18.44068 & 3.18871 & 5.41810 & -0.78792 & -12.38910 \\
630 +\textbf{Consumer price index} & -0.73150 & -2.25239 & -3.39799 & -10.56428 & -1.30151 & -8.59060 & 0.41782 & -9.96604 & -16.99259** & -50.19107 & 1.32430 & -41.28821 \\
631 +\textbf{Durable goods orders} & -1.77394 & -1.82834 & -2.58254 & -8.90295 & -1.86417 & -8.85290 & -1.15772 & -7.40161 & -3.72137 & 56.42319* & -1.13899 & -26.70623 \\
632 +\textbf{Existing home sales} & -3.86759** & 17.19577* & -1.30608 & -5.41876 & -1.44836 & -5.31788 & 1.86846 & -18.96790 & -3.19545 & -16.48154 & -0.81606 & -51.32055* \\
633 +\textbf{Factory orders} & -1.79369 & -5.86609 & -7.11295* & -20.33911 & -3.07893** & -13.68582 & 6.84761** & 52.38087*** & 9.27291 & -49.32580 & 1.51334 & 62.45159** \\
634 +\textbf{Gross domestic product} & -3.16022* & -8.26499 & -2.81458 & -1.40512 & -2.08926 & -8.87196 & -1.76528 & -1.47472 & -5.93161 & -23.15106 & -0.13820 & -30.61457 \\
635 +\textbf{Housing starts} & -0.58545 & 0.77527 & -2.08499 & -8.11537 & -0.88969 & -4.66326 & -3.39511 & 20.13366 & -18.99183***& 23.19105 & -0.84224 & -15.89329 \\
636 +\textbf{Industrial production} & -0.29436 & -1.05059 & 5.50709 & 13.45477 & -0.04290 & 1.71610 & -2.71137 & 19.60256 & 2.81663 & 10.15067 & -0.86143 & -33.18089 \\
637 +\textbf{New home sales} & -2.88999 & -11.62913 & -0.67553 & 2.29246 & -1.33152 & -7.42828 & 0.00223 & -5.27181 & -6.63642 & -19.19409 & 1.59693 & 66.71395** \\
638 +\textbf{Non-farm employment} & -2.46100 & -7.53641 & -3.23439 & -3.80813 & -5.00635*** & 12.32649 & 0.73351 & -7.12626 & -13.54711* & -21.50040 & 2.51278 & 55.39346* \\
639 +\textbf{Pending home sales} & -2.61983 & -11.86787 & 1.00528 & 1.47548 & -1.26851 & -4.63955 & 0.42822 & -15.44296 & -0.19188 & 5.13867 & -2.25446 & -28.30922 \\
640 +\textbf{Personal consumption} & -1.67718 & -6.90320 & -9.79488** & 50.08330** & 0.15770 & 0.20581 & 0.92260 & -10.78664 & -14.99575* & 0.79995 & -0.64875 & -19.87802 \\
641 +\textbf{Personal income} & -4.57034** & -20.24571 & -4.80584 & -25.18481 & -1.16029 & -5.97089 & 1.28405 & -17.01679 & -21.24693***& 17.20807 & -0.65659 & -31.76262 \\
642 +\textbf{Producer price index} & -1.37042 & 1.60079 & -4.05858 & -13.59722 & -3.14320** & -14.84227 & -2.95390 & 17.63070 & -5.66177 & 0.27909 & 0.31519 & -8.87114 \\
643 +\textbf{Trade balance} & -1.05000 & -2.68201 & 0.70212 & 3.23724 & -0.23813 & 6.58851 & 2.77586 & -18.99333 & -0.38237 & -34.84505 & -0.46430 & 2.48295 \\
644 +\textbf{Crude Oil Weekly inventory} & 0.05943 & 7.79488 & & & & & & & & & & \\
645 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -1.31214 & 67.43257*** \\ \midrule
646 +
647 +\textbf{Observations} & \multicolumn{2}{c}{1,041,497} & \multicolumn{2}{c}{1,022,592} & \multicolumn{2}{c}{1,023,897} & \multicolumn{2}{c}{1,022,006} & \multicolumn{2}{c}{683,875} & \multicolumn{2}{c}{995,919} \\
648 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{0.071} & \multicolumn{2}{c}{0.063} & \multicolumn{2}{c}{0.20} & \multicolumn{2}{c}{0.29} & \multicolumn{2}{c}{0.031} & \multicolumn{2}{c}{0.48} \\ \bottomrule
649 +\end{tabular}
650 +}
651 +\begin{tablenotes}
652 + \singlespacing
653 + \footnotesize
654 + This table presents estimates of the equation $R_{\text{SPREAD}, t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{\text{SPREAD}, t-\tau}^{t}+\epsilon_{t}$, analyzing the effects of financialization and macroeconomic announcements on the bid-ask spread using the speculative trading intensity variable $NLS_{t,SD}$, calculated with the swap dealer positions. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the speculative trading intensity variable $NLS_{t,SD}$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and speculative trading intensity. The $\beta$ coefficient represents the effect of the lagged bid-ask spread.
655 +\end{tablenotes}
656 +\end{sidewaystable}
added phd_chap2_20260731/figures/figures.tex +77 −0
@@ -0,0 +1,77 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Figures}
4 +
5 +
6 +\begin{landscape}
7 +\begin{figure}[t]
8 +\includegraphics[height=0.78\textheight,keepaspectratio]{oil_rv.pdf}
9 +\centering
10 +\caption{Realized volatility of the Crude Oil ETF and its iNAV. The figure plots daily realized volatility (sum of squared intraday log returns, $\times 10^{-3}$) for the USO ETF (blue) and its iNAV, proxied by the CL futures contract (red), from 2008-03-27 to 2025-08-01. Panels A, B, and C use returns sampled at the 1-minute, 5-minute, and 30-minute frequencies, respectively. For readability, each series is winsorized at its 99th percentile within each panel.}
11 +\label{fig:rv_uso}
12 +\end{figure}
13 +\end{landscape}
14 +
15 +\begin{landscape}
16 +\begin{figure}[t]
17 +\includegraphics[height=0.78\textheight,keepaspectratio]{gold_rv.pdf}
18 +\centering
19 +\caption{Realized volatility of the Gold ETF and its iNAV. The figure plots daily realized volatility (sum of squared intraday log returns, $\times 10^{-3}$) for the GLD ETF (blue) and its iNAV, proxied by the GC futures contract (red), from 2008-03-27 to 2025-08-01. Panels A, B, and C use returns sampled at the 1-minute, 5-minute, and 30-minute frequencies, respectively. For readability, each series is winsorized at its 99th percentile within each panel.}
20 +\label{fig:rv_gld}
21 +\end{figure}
22 +\end{landscape}
23 +
24 +\begin{landscape}
25 +\begin{figure}[t]
26 +\includegraphics[height=0.78\textheight,keepaspectratio]{silver_rv.pdf}
27 +\centering
28 +\caption{Realized volatility of the Silver ETF and its iNAV. The figure plots daily realized volatility (sum of squared intraday log returns, $\times 10^{-3}$) for the SLV ETF (blue) and its iNAV, proxied by the SI futures contract (red), from 2008-03-27 to 2025-08-01. Panels A, B, and C use returns sampled at the 1-minute, 5-minute, and 30-minute frequencies, respectively. For readability, each series is winsorized at its 99th percentile within each panel.}
29 +\label{fig:rv_slv}
30 +\end{figure}
31 +\end{landscape}
32 +
33 +\begin{landscape}
34 +\begin{figure}[t]
35 +\includegraphics[height=0.78\textheight,keepaspectratio]{gas_rv.pdf}
36 +\centering
37 +\caption{Realized volatility of the Natural Gas ETF and its iNAV. The figure plots daily realized volatility (sum of squared intraday log returns, $\times 10^{-3}$) for the UNG ETF (blue) and its iNAV, proxied by the NG futures contract (red), from 2008-03-27 to 2025-08-01. Panels A, B, and C use returns sampled at the 1-minute, 5-minute, and 30-minute frequencies, respectively. For readability, each series is winsorized at its 99th percentile within each panel.}
38 +\label{fig:rv_ung}
39 +\end{figure}
40 +\end{landscape}
41 +
42 +
43 +\begin{landscape}
44 +\begin{figure}[t]
45 +\includegraphics[width=16cm,trim=0 0 0 22,clip]{oil__uso_cl__irf.pdf}
46 +\centering
47 +\caption{Impulse Response Functions (IRF) for USO ETF and Crude Oil (CL) iNAV Volatility. Bayesian VAR impulse responses showing fast shock absorption with systematic asymmetry: iNAV shocks cause large and persistent responses in ETF volatility while ETF shocks generate smaller and more transitory effects on iNAV, validating bidirectional but asymmetric transmission in crude oil markets.}
48 +\label{fig:irf1}
49 +\end{figure}
50 +\end{landscape}
51 +
52 +\begin{landscape}
53 +\begin{figure}[t]
54 +\includegraphics[width=16cm,trim=0 0 0 22,clip]{gold__gld_gc__irf.pdf}
55 +\centering
56 +\caption{Impulse Response Functions (IRF) for GLD ETF and Gold (GC) iNAV Volatility. Bayesian VAR impulse responses exhibiting the most pronounced asymmetries: iNAV shocks create large and persistent responses in ETF volatility that decay smoothly, while ETF shocks generate negligible responses in iNAV, providing strong dynamic evidence for unidirectional transmission in gold markets.}
57 +\label{fig:irf2}
58 +\end{figure}
59 +\end{landscape}
60 +
61 +\begin{landscape}
62 +\begin{figure}[t]
63 +\includegraphics[width=16cm,trim=0 0 0 22,clip]{silver__slv_si__irf.pdf}
64 +\centering
65 +\caption{Impulse Response Functions (IRF) for SLV ETF and Silver (SI) iNAV Volatility. Bayesian VAR impulse responses preserving the gold market's asymmetric structure but with qualitatively richer adjustment dynamics and greater variability, reflecting silver's dual function as both an industrial and a precious metal.}
66 +\label{fig:irf3}
67 +\end{figure}
68 +\end{landscape}
69 +
70 +\begin{landscape}
71 +\begin{figure}[t]
72 +\includegraphics[width=16cm,trim=0 0 0 22,clip]{gas__ung_ng__irf.pdf}
73 +\centering
74 +\caption{Impulse Response Functions (IRF) for UNG ETF and Natural Gas (NG) iNAV Volatility. Bayesian VAR impulse responses exhibiting the most elaborate structures with large responses in both directions, delayed peaks, and periodic behavior, substantiating bidirectional transmission involving numerous channels and horizons in natural gas markets.}
75 +\label{fig:irf4}
76 +\end{figure}
77 +\end{landscape}
added phd_chap2_20260731/figures/gas__ung_ng__irf.pdf +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/gas_rv.pdf +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/gas_rv.png +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/gold__gld_gc__irf.pdf +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/gold_rv.pdf +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/gold_rv.png +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/oil__uso_cl__irf.pdf +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/oil_rv.pdf +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/oil_rv.png +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/silver__slv_si__irf.pdf +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/silver_rv.pdf +0 −0

Binary file not shown.

added phd_chap2_20260731/figures/silver_rv.png +0 −0

Binary file not shown.

added phd_chap2_20260731/main.bbl +228 −0
@@ -0,0 +1,228 @@
1 +\begin{thebibliography}{}
2 +
3 +\bibitem[\protect\citeauthoryear{Ackert and Tian}{Ackert and
4 + Tian}{2000}]{ackert2000arbitrage}
5 +Ackert, L.~F. and Y.~S. Tian (2000).
6 +\newblock Arbitrage and valuation in the market for {S}tandard \& {P}oor's
7 + depositary receipts.
8 +\newblock {\em Financial Management\/}~{\em 29\/}(3), 71--87.
9 +
10 +\bibitem[\protect\citeauthoryear{Andersen, Bollerslev, and Diebold}{Andersen
11 + et~al.}{2007}]{andersen2007roughing}
12 +Andersen, T.~G., T.~Bollerslev, and F.~X. Diebold (2007).
13 +\newblock Roughing it up: Including jump components in the measurement,
14 + modeling, and forecasting of return volatility.
15 +\newblock {\em The Review of Economics and Statistics\/}~{\em 89\/}(4),
16 + 701--720.
17 +
18 +\bibitem[\protect\citeauthoryear{Andersen, Bollerslev, Diebold, and
19 + Ebens}{Andersen et~al.}{2001}]{andersen2001distribution}
20 +Andersen, T.~G., T.~Bollerslev, F.~X. Diebold, and H.~Ebens (2001).
21 +\newblock The distribution of realized stock return volatility.
22 +\newblock {\em Journal of Financial Economics\/}~{\em 61\/}(1), 43--76.
23 +
24 +\bibitem[\protect\citeauthoryear{Bai and Perron}{Bai and
25 + Perron}{2003}]{bai2003computation}
26 +Bai, J. and P.~Perron (2003).
27 +\newblock Computation and analysis of multiple structural change models.
28 +\newblock {\em Journal of Applied Econometrics\/}~{\em 18\/}(1), 1--22.
29 +
30 +\bibitem[\protect\citeauthoryear{Barndorff-Nielsen, Hansen, Lunde, and
31 + Shephard}{Barndorff-Nielsen et~al.}{2009}]{barndorff2009realized}
32 +Barndorff-Nielsen, O.~E., P.~R. Hansen, A.~Lunde, and N.~Shephard (2009).
33 +\newblock Realized kernels in practice: Trades and quotes.
34 +\newblock {\em The Econometrics Journal\/}~{\em 12\/}(3), C1--C32.
35 +
36 +\bibitem[\protect\citeauthoryear{Barndorff-Nielsen and
37 + Shephard}{Barndorff-Nielsen and Shephard}{2002}]{barndorff2002econometric}
38 +Barndorff-Nielsen, O.~E. and N.~Shephard (2002).
39 +\newblock Econometric analysis of realized volatility and its use in estimating
40 + stochastic volatility models.
41 +\newblock {\em Journal of the Royal Statistical Society: Series B\/}~{\em
42 + 64\/}(2), 253--280.
43 +
44 +\bibitem[\protect\citeauthoryear{Barndorff-Nielsen and
45 + Shephard}{Barndorff-Nielsen and Shephard}{2004}]{barndorff2004power}
46 +Barndorff-Nielsen, O.~E. and N.~Shephard (2004).
47 +\newblock Power and bipower variation with stochastic volatility and jumps.
48 +\newblock {\em Journal of Financial Econometrics\/}~{\em 2\/}(1), 1--37.
49 +
50 +\bibitem[\protect\citeauthoryear{Basak and Pavlova}{Basak and
51 + Pavlova}{2016}]{basak2016model}
52 +Basak, S. and A.~Pavlova (2016).
53 +\newblock A model of financialization of commodities.
54 +\newblock {\em The Journal of Finance\/}~{\em 71\/}(4), 1511--1556.
55 +
56 +\bibitem[\protect\citeauthoryear{Ben-David, Franzoni, and Moussawi}{Ben-David
57 + et~al.}{2018}]{ben2018etfs}
58 +Ben-David, I., F.~Franzoni, and R.~Moussawi (2018).
59 +\newblock Do {ETFs} increase volatility?
60 +\newblock {\em The Journal of Finance\/}~{\em 73\/}(6), 2471--2535.
61 +
62 +\bibitem[\protect\citeauthoryear{Brownlees and Gallo}{Brownlees and
63 + Gallo}{2006}]{brownlees2014practical}
64 +Brownlees, C.~T. and G.~M. Gallo (2006).
65 +\newblock Financial econometric analysis at ultra-high frequency: Data handling
66 + concerns.
67 +\newblock {\em Computational Statistics \& Data Analysis\/}~{\em 51\/}(4),
68 + 2232--2245.
69 +
70 +\bibitem[\protect\citeauthoryear{Buyuksahin and Robe}{Buyuksahin and
71 + Robe}{2014}]{buyuksahin2014speculation}
72 +Buyuksahin, B. and M.~A. Robe (2014).
73 +\newblock Speculation, commodities and cross-market linkages.
74 +\newblock {\em Journal of International Money and Finance\/}~{\em 42}, 38--70.
75 +
76 +\bibitem[\protect\citeauthoryear{Carriero, Kapetanios, and Marcellino}{Carriero
77 + et~al.}{2009}]{carriero2015forecasting}
78 +Carriero, A., G.~Kapetanios, and M.~Marcellino (2009).
79 +\newblock Forecasting exchange rates with a large {B}ayesian {VAR}.
80 +\newblock {\em International Journal of Forecasting\/}~{\em 25\/}(2), 400--417.
81 +
82 +\bibitem[\protect\citeauthoryear{Corsi}{Corsi}{2009}]{corsi2009simple}
83 +Corsi, F. (2009).
84 +\newblock A simple approximate long-memory model of realized volatility.
85 +\newblock {\em Journal of Financial Econometrics\/}~{\em 7\/}(2), 174--196.
86 +
87 +\bibitem[\protect\citeauthoryear{Da and Shive}{Da and
88 + Shive}{2018}]{da2018exchange}
89 +Da, Z. and S.~Shive (2018).
90 +\newblock Exchange traded funds and asset return correlations.
91 +\newblock {\em European Financial Management\/}~{\em 24\/}(1), 136--168.
92 +
93 +\bibitem[\protect\citeauthoryear{Dannhauser}{Dannhauser}{2017}]{dannhauser2017effect}
94 +Dannhauser, C.~D. (2017).
95 +\newblock The impact of innovation: Evidence from corporate bond
96 + exchange-traded funds.
97 +\newblock {\em Journal of Financial Economics\/}~{\em 125\/}(3), 537--560.
98 +
99 +\bibitem[\protect\citeauthoryear{Glosten, Nallareddy, and Zou}{Glosten
100 + et~al.}{2021}]{glosten2021etf}
101 +Glosten, L., S.~Nallareddy, and Y.~Zou (2021).
102 +\newblock {ETF} activity and informational efficiency of underlying securities.
103 +\newblock {\em Management Science\/}~{\em 67\/}(1), 22--47.
104 +
105 +\bibitem[\protect\citeauthoryear{Gorton and Rouwenhorst}{Gorton and
106 + Rouwenhorst}{2006}]{gorton2006facts}
107 +Gorton, G. and K.~G. Rouwenhorst (2006).
108 +\newblock Facts and fantasies about commodity futures.
109 +\newblock {\em Financial Analysts Journal\/}~{\em 62\/}(2), 47--68.
110 +
111 +\bibitem[\protect\citeauthoryear{Gromb and Vayanos}{Gromb and
112 + Vayanos}{2010}]{gromb2010limits}
113 +Gromb, D. and D.~Vayanos (2010).
114 +\newblock Limits of arbitrage: The state of the theory.
115 +\newblock {\em Annual Review of Financial Economics\/}~{\em 2}, 251--275.
116 +
117 +\bibitem[\protect\citeauthoryear{Hansen and Lunde}{Hansen and
118 + Lunde}{2005}]{hansen2005realized}
119 +Hansen, P.~R. and A.~Lunde (2005).
120 +\newblock A realized variance for the whole day based on intermittent
121 + high-frequency data.
122 +\newblock {\em Journal of Financial Econometrics\/}~{\em 3\/}(4), 525--554.
123 +
124 +\bibitem[\protect\citeauthoryear{Hasbrouck}{Hasbrouck}{2003}]{hasbrouck2003intraday}
125 +Hasbrouck, J. (2003).
126 +\newblock Intraday price formation in {US} equity index markets.
127 +\newblock {\em The Journal of Finance\/}~{\em 58\/}(6), 2375--2400.
128 +
129 +\bibitem[\protect\citeauthoryear{Hendershott and Riordan}{Hendershott and
130 + Riordan}{2013}]{hendershott2013relationship}
131 +Hendershott, T. and R.~Riordan (2013).
132 +\newblock Algorithmic trading and the market for liquidity.
133 +\newblock {\em Journal of Financial and Quantitative Analysis\/}~{\em 48\/}(4),
134 + 1001--1024.
135 +
136 +\bibitem[\protect\citeauthoryear{Huang and Tauchen}{Huang and
137 + Tauchen}{2005}]{huang2005using}
138 +Huang, X. and G.~Tauchen (2005).
139 +\newblock The relative contribution of jumps to total price variance.
140 +\newblock {\em Journal of Financial Econometrics\/}~{\em 3\/}(4), 456--499.
141 +
142 +\bibitem[\protect\citeauthoryear{Israeli, Lee, and Sridharan}{Israeli
143 + et~al.}{2017}]{israeli2017etf}
144 +Israeli, D., C.~M. Lee, and S.~A. Sridharan (2017).
145 +\newblock Is there a dark side to exchange traded funds? {An} information
146 + perspective.
147 +\newblock {\em Review of Accounting Studies\/}~{\em 22\/}(3), 1048--1083.
148 +
149 +\bibitem[\protect\citeauthoryear{Koop}{Koop}{2013}]{koop2011forecasting}
150 +Koop, G. (2013).
151 +\newblock Forecasting with medium and large {B}ayesian {VARs}.
152 +\newblock {\em Journal of Applied Econometrics\/}~{\em 28\/}(2), 177--203.
153 +
154 +\bibitem[\protect\citeauthoryear{Litterman}{Litterman}{1986}]{litterman1986forecasting}
155 +Litterman, R.~B. (1986).
156 +\newblock Forecasting with {B}ayesian vector autoregressions—five years of
157 + experience.
158 +\newblock {\em Journal of Business \& Economic Statistics\/}~{\em 4\/}(1),
159 + 25--38.
160 +
161 +\bibitem[\protect\citeauthoryear{Liu, Patton, and Sheppard}{Liu
162 + et~al.}{2015}]{liu2015does}
163 +Liu, L.~Y., A.~J. Patton, and K.~Sheppard (2015).
164 +\newblock Does anything beat 5-minute {RV}? {A} comparison of realized measures
165 + across multiple asset classes.
166 +\newblock {\em Journal of Econometrics\/}~{\em 187\/}(1), 293--311.
167 +
168 +\bibitem[\protect\citeauthoryear{Madhavan}{Madhavan}{2012}]{madhavan2012exchange}
169 +Madhavan, A. (2012).
170 +\newblock Exchange-traded funds, market structure, and the flash crash.
171 +\newblock {\em Financial Analysts Journal\/}~{\em 68\/}(4), 20--35.
172 +
173 +\bibitem[\protect\citeauthoryear{Müller, Dacorogna, Davé, Olsen, Pictet, and
174 + von Weizsäcker}{Müller et~al.}{1997}]{muller1997volatilities}
175 +Müller, U.~A., M.~M. Dacorogna, R.~D. Davé, R.~B. Olsen, O.~V. Pictet, and
176 + J.~E. von Weizsäcker (1997).
177 +\newblock Volatilities of different time resolutions—analyzing the dynamics
178 + of market components.
179 +\newblock {\em Journal of Empirical Finance\/}~{\em 4\/}(2-3), 213--239.
180 +
181 +\bibitem[\protect\citeauthoryear{O'Hara and Zhou}{O'Hara and
182 + Zhou}{2021}]{ohara2021etf}
183 +O'Hara, M. and X.~A. Zhou (2021).
184 +\newblock Anatomy of a liquidity crisis: Corporate bonds in the {COVID}-19
185 + crisis.
186 +\newblock {\em Journal of Financial Economics\/}~{\em 142\/}(1), 46--68.
187 +
188 +\bibitem[\protect\citeauthoryear{Pan and Zeng}{Pan and Zeng}{2016}]{pan2016etf}
189 +Pan, K. and Y.~Zeng (2016).
190 +\newblock {ETF} arbitrage under liquidity mismatch.
191 +\newblock {\em Journal of Financial Economics\/}~{\em 120\/}(3), 617--635.
192 +
193 +\bibitem[\protect\citeauthoryear{Petäjistö}{Petäjistö}{2017}]{petajisto2017inefficiencies}
194 +Petäjistö, A. (2017).
195 +\newblock Inefficiencies in the pricing of exchange-traded funds.
196 +\newblock {\em Financial Analysts Journal\/}~{\em 73\/}(1), 24--54.
197 +
198 +\bibitem[\protect\citeauthoryear{Pontiff}{Pontiff}{1996}]{pontiff1996costly}
199 +Pontiff, J. (1996).
200 +\newblock Costly arbitrage: Evidence from closed-end funds.
201 +\newblock {\em The Quarterly Journal of Economics\/}~{\em 111\/}(4),
202 + 1135--1151.
203 +
204 +\bibitem[\protect\citeauthoryear{Richie, Daigler, and Gleason}{Richie
205 + et~al.}{2008}]{richie2008examination}
206 +Richie, N., R.~T. Daigler, and K.~C. Gleason (2008).
207 +\newblock The limits to stock index arbitrage: Examining {S\&P} 500 futures and
208 + {SPDRs}.
209 +\newblock {\em Journal of Futures Markets\/}~{\em 28\/}(12), 1182--1205.
210 +
211 +\bibitem[\protect\citeauthoryear{Sims and Zha}{Sims and
212 + Zha}{1999}]{sims1999error}
213 +Sims, C.~A. and T.~Zha (1999).
214 +\newblock Error bands for impulse responses.
215 +\newblock {\em Econometrica\/}~{\em 67\/}(5), 1113--1155.
216 +
217 +\bibitem[\protect\citeauthoryear{Staer}{Staer}{2017}]{staer2017asset}
218 +Staer, A. (2017).
219 +\newblock Asset management via {ETFs}.
220 +\newblock {\em The Review of Financial Studies\/}~{\em 30\/}(9), 3225--3264.
221 +
222 +\bibitem[\protect\citeauthoryear{Todorov}{Todorov}{2024}]{todorov2021etf}
223 +Todorov, K. (2024).
224 +\newblock When passive funds affect prices: Evidence from volatility and
225 + commodity {ETFs}.
226 +\newblock {\em Review of Finance\/}~{\em 28\/}(3), 831--863.
227 +
228 +\end{thebibliography}
added phd_chap2_20260731/main.pdf +0 −0

Binary file not shown.

added phd_chap2_20260731/main.tex +105 −0
@@ -0,0 +1,105 @@
1 +% For submission to the Journal of Futures Markets (JFM) — version with authors
2 +
3 +\documentclass[12pt]{article}
4 +
5 +\usepackage{amssymb,amsmath,amsfonts,eurosym,geometry,ulem,graphicx,caption,color,setspace,comment,footmisc,pdflscape,subfigure,array,hyperref}
6 +\graphicspath{{figures/}}
7 +\usepackage{rotating}
8 +\usepackage{multirow}
9 +\usepackage{tabularx}
10 +\usepackage{booktabs}
11 +\usepackage{lscape}
12 +\usepackage[dvipsnames]{xcolor}
13 +\definecolor{mypink1}{rgb}{0.858, 0.188, 0.478}
14 +\usepackage{tablefootnote}
15 +\usepackage{ragged2e}
16 +\usepackage{nicematrix}
17 +\usepackage{enumitem}
18 +\usepackage{wasysym}
19 +\usepackage{framed}
20 +\usepackage{pgfgantt}
21 +\usepackage{tikz}
22 +\usetikzlibrary{shapes.geometric, arrows.meta, positioning, fit, backgrounds}
23 +\usepackage{subfloat}
24 +\usepackage{blindtext}
25 +\usepackage{colortbl}
26 +\usepackage{palatino}
27 +\usepackage{mathpazo}
28 +\usepackage[flushleft]{threeparttable}
29 +\usepackage[margin=15pt,font=small,labelfont={bf,sf}]{caption}
30 +\usepackage[semibold]{sourcesanspro}
31 +\usepackage{sectsty}
32 +\allsectionsfont{\sffamily}
33 +\usepackage{xcolor}
34 +\hypersetup{colorlinks,
35 +citecolor=blue,
36 +linkcolor=magenta
37 +}
38 +
39 +\usepackage{environ,atbegshi}
40 +\makeatletter
41 +\newif\ifAtEnd@DontShowMe
42 +\newcommand*\contents@DontShowMe{}
43 +
44 +\normalem
45 +
46 +\doublespacing
47 +\newtheorem{theorem}{Theorem}
48 +\newtheorem{corollary}[theorem]{Corollary}
49 +\newtheorem{proposition}{Proposition}
50 +\newenvironment{proof}[1][Proof]{\noindent\textbf{#1.} }{\ \rule{0.5em}{0.5em}}
51 +
52 +\newtheorem{hyp}{Hypothesis}
53 +\newtheorem{subhyp}{Hypothesis}[hyp]
54 +\renewcommand{\thesubhyp}{\thehyp\alph{subhyp}}
55 +
56 +\usepackage{natbib}
57 +\bibliographystyle{chicago}
58 +\setcitestyle{authoryear,open={(},close={)}}
59 +\geometry{left=1.0in,right=1.0in,top=1.0in,bottom=1.0in}
60 +\title{\textbf{Seeing Through the ETF: Indicative NAV and Commodity Volatility Transmission }
61 +}
62 +
63 +\author{Simon-Pierre Boucher\thanks{PhD student in finance, Universit\'e Laval, Quebec City QC Canada G1V 0A6, email: \href{mailto:simon-pierre.boucher.1@ulaval.ca}{simon-pierre.boucher.1@ulaval.ca}}
64 +\and Marie-H\'el\`ene Gagnon\thanks{Professor of Finance and Research Fellow, CRREP, Universit\'e Laval, email: \href{mailto:marie-helene.gagnon@fsa.ulaval.ca}{marie-helene.gagnon@fsa.ulaval.ca}}
65 +\and Gabriel J. Power\thanks{IG Wealth Management Chairholder, Professor of Finance and Research Fellow, CRREP and CRIB, Universit\'e Laval, email: \href{mailto:gabriel.power@fsa.ulaval.ca}{gabriel.power@fsa.ulaval.ca}}}
66 +
67 +\date{\today}
68 +\begin{document}
69 +\begin{titlepage}
70 +\maketitle
71 +
72 +\begin{abstract}
73 +\singlespacing
74 +\noindent
75 +This paper builds a novel dataset of indicative Net Asset Value (iNAV) observations for commodity ETFs and measures volatility transmission between ETFs and their underlying baskets using high-frequency realized variance. Applying this design to crude oil, gold, silver, and natural gas, we show that the iNAV provides a sharper image of the volatility relationship. Decomposing realized variance into continuous and jump components, we establish that transmission runs primarily through jumps rather than diffusion, a channel that standard volatility connectedness measures may obscure. Sampling frequency also matters: 1-minute data reveal daily volatility transmission up to twice as large as 30-minute estimates. Finally, transmission differs across commodity types. For gold and silver, it is unidirectional from iNAV to ETF, consistent with passive arbitrage. For energy ETFs, which are more liquid and hold futures as underlying assets, transmission is bidirectional and asymmetric.
76 +
77 +
78 +\vspace{0.2in}
79 +\noindent\textbf{Keywords:} ETF, volatility, transmission, commodity markets, futures, intraday, realized variance, jump, HAR, arbitrage, microstructure.\\
80 +\textbf{JEL Classification:} G12, G13, G14, C32
81 +
82 +\bigskip
83 +\end{abstract}
84 +\doublespacing
85 +
86 +\setcounter{page}{0}
87 +\thispagestyle{empty}
88 +\end{titlepage}
89 +\pagebreak \newpage
90 +
91 +
92 +\input{sections/01_introduction}
93 +\input{sections/03_data}
94 +\input{sections/04_methods}
95 +\input{sections/05_results}
96 +\input{sections/06_conclusion}
97 +
98 +\bibliography{master}
99 +\doublespacing
100 +
101 +\input{tables/tables}
102 +\input{figures/figures}
103 +
104 +
105 +\end{document}
\ No newline at end of file
added phd_chap2_20260731/master.bib +770 −0
@@ -0,0 +1,770 @@
1 +@article{petajisto2017inefficiencies,
2 + title={Inefficiencies in the pricing of exchange-traded funds},
3 + author={Petäjistö, Antti},
4 + journal={Financial Analysts Journal},
5 + volume={73},
6 + number={1},
7 + pages={24--54},
8 + year={2017},
9 + publisher={CFA Institute}
10 +}
11 +
12 +
13 +@article{gorton2006facts,
14 + title={Facts and fantasies about commodity futures},
15 + author={Gorton, Gary and Rouwenhorst, K Geert},
16 + journal={Financial Analysts Journal},
17 + volume={62},
18 + number={2},
19 + pages={47--68},
20 + year={2006},
21 + publisher={CFA Institute}
22 +}
23 +
24 +
25 +@article{ackert2000arbitrage,
26 + title={Arbitrage and valuation in the market for {S}tandard \& {P}oor's depositary receipts},
27 + author={Ackert, Lucy F and Tian, Yisong S},
28 + journal={Financial Management},
29 + volume={29},
30 + number={3},
31 + pages={71--87},
32 + year={2000},
33 + publisher={Wiley}
34 +}
35 +
36 +
37 +@article{madhavan2012exchange,
38 + title={Exchange-traded funds, market structure, and the flash crash},
39 + author={Madhavan, Ananth},
40 + journal={Financial Analysts Journal},
41 + volume={68},
42 + number={4},
43 + pages={20--35},
44 + year={2012},
45 + publisher={CFA Institute}
46 +}
47 +
48 +
49 +@article{ben2018etfs,
50 + title={Do {ETFs} increase volatility?},
51 + author={Ben-David, Itzhak and Franzoni, Francesco and Moussawi, Rabih},
52 + journal={The Journal of Finance},
53 + volume={73},
54 + number={6},
55 + pages={2471--2535},
56 + year={2018},
57 + publisher={Wiley}
58 +}
59 +
60 +
61 +@article{israeli2017etf,
62 + title={Is there a dark side to exchange traded funds? {An} information perspective},
63 + author={Israeli, Doron and Lee, Charles MC and Sridharan, Suhas A},
64 + journal={Review of Accounting Studies},
65 + volume={22},
66 + number={3},
67 + pages={1048--1083},
68 + year={2017},
69 + publisher={Springer}
70 +}
71 +
72 +
73 +@article{todorov2021etf,
74 + title={When passive funds affect prices: Evidence from volatility and commodity {ETFs}},
75 + author={Todorov, Karamfil},
76 + journal={Review of Finance},
77 + year={2024},
78 + volume={28},
79 + number={3},
80 + pages={831--863},
81 + publisher={Oxford University Press}
82 +}
83 +
84 +
85 +@article{corsi2009simple,
86 + title={A simple approximate long-memory model of realized volatility},
87 + author={Corsi, Fulvio},
88 + journal={Journal of Financial Econometrics},
89 + volume={7},
90 + number={2},
91 + pages={174--196},
92 + year={2009},
93 + publisher={Oxford University Press}
94 +}
95 +
96 +
97 +@article{koop2011forecasting,
98 + title={Forecasting with medium and large {B}ayesian {VARs}},
99 + author={Koop, Gary},
100 + journal={Journal of Applied Econometrics},
101 + volume={28},
102 + number={2},
103 + pages={177--203},
104 + year={2013},
105 + publisher={Wiley}
106 +}
107 +
108 +
109 +@article{andersen2001distribution,
110 + title={The distribution of realized stock return volatility},
111 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X and Ebens, Heiko},
112 + journal={Journal of Financial Economics},
113 + volume={61},
114 + number={1},
115 + pages={43--76},
116 + year={2001},
117 + publisher={Elsevier}
118 +}
119 +
120 +
121 +@article{barndorff2004power,
122 + title={Power and bipower variation with stochastic volatility and jumps},
123 + author={Barndorff-Nielsen, Ole E and Shephard, Neil},
124 + journal={Journal of Financial Econometrics},
125 + volume={2},
126 + number={1},
127 + pages={1--37},
128 + year={2004},
129 + publisher={Oxford University Press}
130 +}
131 +
132 +
133 +@article{pan2016etf,
134 + title={{ETF} arbitrage under liquidity mismatch},
135 + author={Pan, Kevin and Zeng, Yao},
136 + journal={Journal of Financial Economics},
137 + volume={120},
138 + number={3},
139 + pages={617--635},
140 + year={2016},
141 + publisher={Elsevier}
142 +}
143 +
144 +
145 +@article{fang2023etf,
146 + title={{ETF} arbitrage, non-fundamental demand, and return predictability},
147 + author={Fang, Lily H and Jiang, Joel},
148 + journal={Review of Financial Studies},
149 + volume={36},
150 + number={4},
151 + pages={1433--1470},
152 + year={2023},
153 + publisher={Oxford University Press}
154 +}
155 +
156 +
157 +@article{lettau2018exchange,
158 + title={Exchange-traded funds 101 for economists},
159 + author={Lettau, Martin and Madhavan, Ananth},
160 + journal={Journal of Economic Perspectives},
161 + volume={32},
162 + number={1},
163 + pages={135--154},
164 + year={2018},
165 + publisher={American Economic Association}
166 +}
167 +
168 +
169 +@article{glosten2021etf,
170 + title={{ETF} activity and informational efficiency of underlying securities},
171 + author={Glosten, Lawrence and Nallareddy, Suresh and Zou, Yuan},
172 + journal={Management Science},
173 + volume={67},
174 + number={1},
175 + pages={22--47},
176 + year={2021},
177 + publisher={INFORMS}
178 +}
179 +
180 +
181 +@article{da2018exchange,
182 + title={Exchange traded funds and asset return correlations},
183 + author={Da, Zhi and Shive, Sophie},
184 + journal={European Financial Management},
185 + volume={24},
186 + number={1},
187 + pages={136--168},
188 + year={2018},
189 + publisher={Wiley}
190 +}
191 +
192 +
193 +@article{hamm2011liquidity,
194 + title={Liquidity and firm value: Evidence from {ETF} holdings},
195 + author={Hamm, Sophia JW},
196 + journal={Journal of Corporate Finance},
197 + volume={20},
198 + pages={1--15},
199 + year={2014},
200 + publisher={Elsevier}
201 +}
202 +
203 +
204 +@article{krause2014exchange,
205 + title={Exchange-traded funds, liquidity and volatility},
206 + author={Krause, Tanja and Ehsani, Shahram and Lien, Donald},
207 + journal={Applied Financial Economics},
208 + volume={24},
209 + number={24},
210 + pages={1617--1630},
211 + year={2014},
212 + publisher={Taylor \& Francis}
213 +}
214 +
215 +
216 +@article{ohara2021etf,
217 + title={Anatomy of a liquidity crisis: Corporate bonds in the {COVID}-19 crisis},
218 + author={O'Hara, Maureen and Zhou, Xing Alex},
219 + journal={Journal of Financial Economics},
220 + volume={142},
221 + number={1},
222 + pages={46--68},
223 + year={2021},
224 + publisher={Elsevier}
225 +}
226 +
227 +
228 +@article{barndorff2002econometric,
229 + title={Econometric analysis of realized volatility and its use in estimating stochastic volatility models},
230 + author={Barndorff-Nielsen, Ole E and Shephard, Neil},
231 + journal={Journal of the Royal Statistical Society: Series B},
232 + volume={64},
233 + number={2},
234 + pages={253--280},
235 + year={2002},
236 + publisher={Wiley}
237 +}
238 +
239 +
240 +@article{buckle2018price,
241 + title={Price discovery in the {ETF} market: the role of market makers},
242 + author={Buckle, David J and Curran, Edward and McGroarty, Frank},
243 + journal={Journal of Banking \& Finance},
244 + volume={93},
245 + pages={32--45},
246 + year={2018},
247 + publisher={Elsevier}
248 +}
249 +
250 +
251 +@article{hamm2014liquidity,
252 + title={Liquidity and firm value: Evidence from {ETF} holdings},
253 + author={Hamm, Sophia JW},
254 + journal={Journal of Corporate Finance},
255 + volume={29},
256 + pages={1--15},
257 + year={2014},
258 + publisher={Elsevier}
259 +}
260 +
261 +
262 +@article{agarwal2018etfs,
263 + title={Do {ETFs} increase the commonality in liquidity of underlying stocks?},
264 + author={Agarwal, Vikas and Hanouna, Paul and Moussawi, Rabih and Stahel, Christof W},
265 + journal={Journal of Financial Economics},
266 + volume={127},
267 + number={1},
268 + pages={67--90},
269 + year={2018},
270 + publisher={Elsevier}
271 +}
272 +
273 +
274 +@article{basak2016model,
275 + title={A model of financialization of commodities},
276 + author={Basak, Suleyman and Pavlova, Anna},
277 + journal={The Journal of Finance},
278 + volume={71},
279 + number={4},
280 + pages={1511--1556},
281 + year={2016},
282 + publisher={Wiley}
283 +}
284 +
285 +
286 +@article{buyuksahin2014speculation,
287 + title={Speculation, commodities and cross-market linkages},
288 + author={Buyuksahin, Bahattin and Robe, Michel A},
289 + journal={Journal of International Money and Finance},
290 + volume={42},
291 + pages={38--70},
292 + year={2014},
293 + publisher={Elsevier}
294 +}
295 +
296 +
297 +@article{andersen2007roughing,
298 + title={Roughing it up: Including jump components in the measurement, modeling, and forecasting of return volatility},
299 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X},
300 + journal={The Review of Economics and Statistics},
301 + volume={89},
302 + number={4},
303 + pages={701--720},
304 + year={2007},
305 + publisher={MIT Press}
306 +}
307 +
308 +
309 +@article{corsi2010threshold,
310 + title={Threshold bipower variation and the impact of jumps on volatility forecasting},
311 + author={Corsi, Fulvio and Pirino, Davide and Reno, Roberto},
312 + journal={Journal of Econometrics},
313 + volume={159},
314 + number={2},
315 + pages={276--288},
316 + year={2010},
317 + publisher={Elsevier}
318 +}
319 +
320 +
321 +@article{bubak2011volatility,
322 + title={Volatility transmission in emerging European foreign exchange markets},
323 + author={Bub{\'a}k, Vít and Kočenda, Evžen and Žikeš, Filip},
324 + journal={Journal of Banking \& Finance},
325 + volume={35},
326 + number={11},
327 + pages={2829--2841},
328 + year={2011},
329 + publisher={Elsevier}
330 +}
331 +
332 +
333 +@article{carriero2015forecasting,
334 + title={Forecasting exchange rates with a large {B}ayesian {VAR}},
335 + author={Carriero, Andrea and Kapetanios, George and Marcellino, Massimiliano},
336 + journal={International Journal of Forecasting},
337 + volume={25},
338 + number={2},
339 + pages={400--417},
340 + year={2009},
341 + publisher={Elsevier}
342 +}
343 +
344 +
345 +@article{diebold2012measuring,
346 + title={Better to give than to receive: Predictive directional measurement of volatility spillovers},
347 + author={Diebold, Francis X and Yilmaz, Kamil},
348 + journal={International Journal of Forecasting},
349 + volume={28},
350 + number={1},
351 + pages={57--66},
352 + year={2012},
353 + publisher={Elsevier}
354 +}
355 +
356 +
357 +@article{barunik2018measuring,
358 + title={Measuring the frequency dynamics of financial connectedness and systemic risk},
359 + author={Baruník, Jozef and Křehlík, Tomáš},
360 + journal={Journal of Financial Econometrics},
361 + volume={16},
362 + number={2},
363 + pages={271--296},
364 + year={2018},
365 + publisher={Oxford University Press}
366 +}
367 +
368 +
369 +@article{guo2015leveraged,
370 + title={Leveraged {ETF} performance during periods of market turmoil},
371 + author={Guo, Hui and Leung, Tim},
372 + journal={The Journal of Derivatives},
373 + volume={22},
374 + number={3},
375 + pages={61--74},
376 + year={2015},
377 + publisher={Institutional Investor Journals}
378 +}
379 +
380 +
381 +@article{ozdurak2020price,
382 + title={Price discovery in crude oil markets: Intraday volatility interactions between crude oil futures and energy exchange traded funds},
383 + author={Ozdurak, Caner and Ulusoy, Veysel},
384 + journal={International Journal of Energy Economics and Policy},
385 + volume={10},
386 + number={3},
387 + pages={402--413},
388 + year={2020}
389 +}
390 +
391 +
392 +@article{pontiff1996costly,
393 + title={Costly arbitrage: Evidence from closed-end funds},
394 + author={Pontiff, Jeffrey},
395 + journal={The Quarterly Journal of Economics},
396 + volume={111},
397 + number={4},
398 + pages={1135--1151},
399 + year={1996},
400 + publisher={Oxford University Press}
401 +}
402 +
403 +
404 +@article{gromb2010limits,
405 + title={Limits of arbitrage: The state of the theory},
406 + author={Gromb, Denis and Vayanos, Dimitri},
407 + journal={Annual Review of Financial Economics},
408 + volume={2},
409 + pages={251--275},
410 + year={2010},
411 + publisher={Annual Reviews}
412 +}
413 +
414 +
415 +@article{hendershott2013relationship,
416 + title={Algorithmic trading and the market for liquidity},
417 + author={Hendershott, Terrence and Riordan, Ryan},
418 + journal={Journal of Financial and Quantitative Analysis},
419 + volume={48},
420 + number={4},
421 + pages={1001--1024},
422 + year={2013},
423 + publisher={Cambridge University Press}
424 +}
425 +
426 +
427 +@article{malamud2016portfolio,
428 + title={Portfolio delegation and market efficiency},
429 + author={Malamud, Semyon},
430 + journal={The Journal of Finance},
431 + volume={71},
432 + number={6},
433 + pages={2529--2574},
434 + year={2016},
435 + publisher={Wiley}
436 +}
437 +
438 +
439 +@article{ben2017etfs,
440 + title={Do {ETFs} increase volatility?},
441 + author={Ben-David, Itzhak and Franzoni, Francesco and Moussawi, Rabih},
442 + journal={The Journal of Finance},
443 + volume={73},
444 + number={6},
445 + pages={2471--2535},
446 + year={2018},
447 + publisher={Wiley}
448 +}
449 +
450 +
451 +@article{shin2013basis,
452 + title={Basis and trading volume in international {ETF} markets},
453 + author={Shin, Sujeong and Soydemir, Gokce},
454 + journal={Journal of Futures Markets},
455 + volume={33},
456 + number={10},
457 + pages={967--992},
458 + year={2013},
459 + publisher={Wiley}
460 +}
461 +
462 +
463 +@article{dannhauser2017effect,
464 + title={The impact of innovation: Evidence from corporate bond exchange-traded funds},
465 + author={Dannhauser, Caitlin D},
466 + journal={Journal of Financial Economics},
467 + volume={125},
468 + number={3},
469 + pages={537--560},
470 + year={2017},
471 + publisher={Elsevier}
472 +}
473 +
474 +
475 +@article{staer2017asset,
476 + title={Asset management via {ETFs}},
477 + author={Staer, Alexander},
478 + journal={The Review of Financial Studies},
479 + volume={30},
480 + number={9},
481 + pages={3225--3264},
482 + year={2017},
483 + publisher={Oxford University Press}
484 +}
485 +
486 +
487 +@article{richie2008examination,
488 + title={The limits to stock index arbitrage: Examining {S\&P} 500 futures and {SPDRs}},
489 + author={Richie, Nivine and Daigler, Robert T and Gleason, Kimberly C},
490 + journal={Journal of Futures Markets},
491 + volume={28},
492 + number={12},
493 + pages={1182--1205},
494 + year={2008},
495 + publisher={Wiley}
496 +}
497 +
498 +
499 +@article{hasbrouck2003intraday,
500 + title={Intraday price formation in {US} equity index markets},
501 + author={Hasbrouck, Joel},
502 + journal={The Journal of Finance},
503 + volume={58},
504 + number={6},
505 + pages={2375--2400},
506 + year={2003},
507 + publisher={Wiley}
508 +}
509 +
510 +
511 +@article{broman2016liquidity,
512 + title={Liquidity, style investing and excess comovement of exchange-traded fund returns},
513 + author={Broman, Markus S},
514 + journal={Journal of Financial Markets},
515 + volume={30},
516 + pages={27--53},
517 + year={2016},
518 + publisher={Elsevier}
519 +}
520 +
521 +
522 +@article{huang2005using,
523 + title={The relative contribution of jumps to total price variance},
524 + author={Huang, Xin and Tauchen, George},
525 + journal={Journal of Financial Econometrics},
526 + volume={3},
527 + number={4},
528 + pages={456--499},
529 + year={2005},
530 + publisher={Oxford University Press}
531 +}
532 +
533 +
534 +@article{hansen2005realized,
535 + title={A realized variance for the whole day based on intermittent high-frequency data},
536 + author={Hansen, Peter Reinhard and Lunde, Asger},
537 + journal={Journal of Financial Econometrics},
538 + volume={3},
539 + number={4},
540 + pages={525--554},
541 + year={2005},
542 + publisher={Oxford University Press}
543 +}
544 +
545 +
546 +@article{liu2015does,
547 + title={Does anything beat 5-minute {RV}? {A} comparison of realized measures across multiple asset classes},
548 + author={Liu, Lily Y and Patton, Andrew J and Sheppard, Kevin},
549 + journal={Journal of Econometrics},
550 + volume={187},
551 + number={1},
552 + pages={293--311},
553 + year={2015},
554 + publisher={Elsevier}
555 +}
556 +
557 +
558 +@article{clark2008forecasting,
559 + title={Forecasting {US} inflation using {B}ayesian factor augmented autoregressive models},
560 + author={Clark, Todd E and McCracken, Michael W},
561 + journal={The Review of Economics and Statistics},
562 + volume={90},
563 + number={4},
564 + pages={665--678},
565 + year={2008},
566 + publisher={MIT Press}
567 +}
568 +
569 +
570 +@article{primiceri2005time,
571 + title={Time varying structural vector autoregressions and monetary policy},
572 + author={Primiceri, Giorgio E},
573 + journal={The Review of Economic Studies},
574 + volume={72},
575 + number={3},
576 + pages={821--852},
577 + year={2005},
578 + publisher={Oxford University Press}
579 +}
580 +
581 +
582 +@article{buyuksahin2010fundamentals,
583 + title={Fundamentals, trader activity and derivative pricing},
584 + author={Buyuksahin, Bahattin and Robe, Michel A},
585 + journal={Journal of Futures Markets},
586 + volume={30},
587 + number={4},
588 + pages={317--341},
589 + year={2010},
590 + publisher={Wiley}
591 +}
592 +
593 +
594 +@article{singleton2014investor,
595 + title={Investor flows and the 2008 boom/bust in oil prices},
596 + author={Singleton, Kenneth J},
597 + journal={Management Science},
598 + volume={60},
599 + number={2},
600 + pages={300--318},
601 + year={2014},
602 + publisher={INFORMS}
603 +}
604 +
605 +
606 +@article{dannhauser2020etf,
607 + title={{ETF} arbitrage during market turmoil},
608 + author={Dannhauser, Caitlin D},
609 + journal={Journal of Financial Economics},
610 + volume={138},
611 + number={2},
612 + pages={355--374},
613 + year={2020},
614 + publisher={Elsevier}
615 +}
616 +
617 +
618 +@article{pan2017market,
619 + title={Market disruptions and destabilizing speculation},
620 + author={Pan, Kevin and Zeng, Yao},
621 + journal={Journal of Financial Economics},
622 + volume={126},
623 + number={1},
624 + pages={450--471},
625 + year={2017},
626 + publisher={Elsevier}
627 +}
628 +
629 +
630 +@article{lynch2021market,
631 + title={Market maker of last resort: The {F}ederal {R}eserve's role in the {COVID}-19 bond market crisis},
632 + author={Lynch, David P},
633 + journal={Journal of Financial Economics},
634 + volume={142},
635 + number={2},
636 + pages={588--605},
637 + year={2021},
638 + publisher={Elsevier}
639 +}
640 +
641 +
642 +@article{barndorff2009realized,
643 + title={Realized kernels in practice: Trades and quotes},
644 + author={Barndorff-Nielsen, Ole E and Hansen, Peter Reinhard and Lunde, Asger and Shephard, Neil},
645 + journal={The Econometrics Journal},
646 + volume={12},
647 + number={3},
648 + pages={C1--C32},
649 + year={2009},
650 + publisher={Oxford University Press}
651 +}
652 +
653 +
654 +@article{brownlees2014practical,
655 + title={Financial econometric analysis at ultra-high frequency: Data handling concerns},
656 + author={Brownlees, Christian T and Gallo, Giampiero M},
657 + journal={Computational Statistics \& Data Analysis},
658 + volume={51},
659 + number={4},
660 + pages={2232--2245},
661 + year={2006},
662 + publisher={Elsevier}
663 +}
664 +
665 +
666 +@article{lee1991inferring,
667 + title={Inferring trade direction from intraday data},
668 + author={Lee, Charles MC and Ready, Mark J},
669 + journal={The Journal of Finance},
670 + volume={46},
671 + number={2},
672 + pages={733--746},
673 + year={1991},
674 + publisher={Wiley}
675 +}
676 +
677 +
678 +@article{krehbiel2019real,
679 + title={Real-time price discovery via verbal communication: Method and application},
680 + author={Krehbiel, Tim and Adkins, Lee C},
681 + journal={Journal of Futures Markets},
682 + volume={39},
683 + number={5},
684 + pages={596--613},
685 + year={2019},
686 + publisher={Wiley}
687 +}
688 +
689 +
690 +@article{bai2003computation,
691 + title={Computation and analysis of multiple structural change models},
692 + author={Bai, Jushan and Perron, Pierre},
693 + journal={Journal of Applied Econometrics},
694 + volume={18},
695 + number={1},
696 + pages={1--22},
697 + year={2003},
698 + publisher={Wiley}
699 +}
700 +
701 +
702 +@article{muller1997volatilities,
703 + title={Volatilities of different time resolutions—analyzing the dynamics of market components},
704 + author={Müller, Ulrich A and Dacorogna, Michel M and Davé, Rakhal D and Olsen, Richard B and Pictet, Olivier V and von Weizsäcker, Jacob E},
705 + journal={Journal of Empirical Finance},
706 + volume={4},
707 + number={2-3},
708 + pages={213--239},
709 + year={1997},
710 + publisher={Elsevier}
711 +}
712 +
713 +
714 +@article{litterman1986forecasting,
715 + title={Forecasting with {B}ayesian vector autoregressions—five years of experience},
716 + author={Litterman, Robert B},
717 + journal={Journal of Business \& Economic Statistics},
718 + volume={4},
719 + number={1},
720 + pages={25--38},
721 + year={1986},
722 + publisher={Taylor \& Francis}
723 +}
724 +
725 +
726 +@article{sims1999error,
727 + title={Error bands for impulse responses},
728 + author={Sims, Christopher A and Zha, Tao},
729 + journal={Econometrica},
730 + volume={67},
731 + number={5},
732 + pages={1113--1155},
733 + year={1999},
734 + publisher={Wiley}
735 +}
736 +
737 +
738 +@article{koop2013forecasting,
739 + title={Forecasting with medium and large {B}ayesian {VARs}},
740 + author={Koop, Gary},
741 + journal={Journal of Applied Econometrics},
742 + volume={28},
743 + number={2},
744 + pages={177--203},
745 + year={2013},
746 + publisher={Wiley}
747 +}
748 +
749 +
750 +@article{giannone2015prior,
751 + title={Prior selection for vector autoregressions},
752 + author={Giannone, Domenico and Lenza, Michele and Primiceri, Giorgio E},
753 + journal={The Review of Economics and Statistics},
754 + volume={97},
755 + number={2},
756 + pages={436--451},
757 + year={2015},
758 + publisher={MIT Press}
759 +}
760 +
761 +
762 +@article{geweke1992evaluating,
763 + title={Evaluating the accuracy of sampling-based approaches to the calculation of posterior moments},
764 + author={Geweke, John},
765 + journal={Bayesian Statistics},
766 + volume={4},
767 + pages={169--193},
768 + year={1992},
769 + publisher={Oxford University Press}
770 +}
added phd_chap2_20260731/sections/01_introduction.tex +22 −0
@@ -0,0 +1,22 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Introduction}
4 +
5 +The exchange-traded fund (ETF) market has grown rapidly over the past two decades, with global assets under management (AUM) exceeding \$23 trillion as of May 2026. The growth in the ETF market has also altered the structure of underlying asset markets \citep{petajisto2017inefficiencies}. ETF ownership has been found to increase non-fundamental volatility in underlying equities \citep{ben2018etfs}, weaken the link between prices and fundamentals \citep{israeli2017etf}, and strengthen correlations during periods of market stress \citep{da2018exchange}. The number and economic size of commodity ETFs, such as the SPDR Gold Trust (GLD), has rapidly grown in recent years. These products offer exposure to commodity markets without the need for direct futures trading or physical storage \citep{gorton2006facts}, thus increasing the participation of financial investors, creating new economic links between commodity and equity prices, and establishing the ETF as an alternative venue for price discovery \citep{basak2016model, buyuksahin2014speculation}.
6 +
7 +Commodities offer a particularly interesting setting to study ETF pricing and volatility transmission. Exchange-traded funds are usually studied as near-transparent wrappers around their holdings. For example, VOO and IVV replicate the U.S. S\&P~500 market index and hold the stocks that comprise this index according to their weights. For a typical equity ETF this framing is close to exact. The underlying securities trade on the same exchanges during the same hours, and creation-redemption arbitrage pins the fund's price to its net asset value quickly and in essentially one direction. Thus, for an equity ETF, asking whether the ETF leads the basket would be similar to asking whether a shadow leads the object that casts it. In contrast, commodity ETFs break this identity in a way that is not feasible in other major ETF classes. This is because their underlying may not be the physical good, but could be instead a derivative claim. For instance, if it is a futures position in crude oil or natural gas, there are economic implications for cost-of-carry, roll, and the shape of the term structure. In the case of a stored physical claim in gold and silver there is lease-and-storage economics to consider.
8 +
9 +Thus, for commodity ETFs, the arbitrage relationship is not as simple as it is for equities or bonds, and volatility transmission becomes a genuine economic question rather than a purely mechanical one. To the point of this paper, it is important to investigate the fund's indicative net asset value (iNAV), which is the real-time fair value of that basket. Commodity ETFs also play a role in the financialization debate \citep{basak2016model}, as they are popular retail-accessible instruments. It is important for investors, traders, hedgers and policymakers to understand how volatility travels between these vehicles and their underlyings. The direction of this volatility flow is not clear a priori. Indeed, commodity ETFs differ along another significant dimension, namely that unlike equities, in commodities the ETF may well be more liquid and more continuously accessible than its underlying. This feature provides a clear reason why the ETF might contribute to price discovery rather than simply inherit it. Whether it does, however, remains an empirical question.
10 +
11 +
12 +The theoretical foundation of ETF pricing rests on arbitrage. Authorized participants (APs) maintain price alignment through creation and redemption, buying undervalued ETFs while selling their underlying constituents, or vice versa \citep{ackert2000arbitrage}. Classic arbitrage theory predicts that volatility transmission would be unidirectional, flowing from the underlying constituents (NAV) to the ETF: when underlying volatility rises, arbitrageurs trade more actively to maintain alignment, which transmits volatility to the ETF. Recent evidence challenges this prediction, documenting bidirectional transmission in which ETF trading influences underlying asset volatility \citep{ben2018etfs, da2018exchange}. This outcome arises because ETFs often trade more frequently and with smaller spreads than their constituents, making the ETF a primary venue for price formation \citep{glosten2021etf, pan2016etf}. The relative magnitude and direction of transmission therefore reveal which market dominates information discovery and whether arbitrage functions efficiently. These questions have direct implications for hedging (directional dependencies) and for regulation, e.g., as feedback effects may amplify volatility during crises \citep{madhavan2012exchange, petajisto2017inefficiencies}.
13 +
14 +This paper studies how volatility is transmitted between commodity ETFs and their underlying assets using high-frequency realized variance over 2010--2023. We focus on four major single-commodity ETFs spanning two market structures: physically-backed precious metals (SPDR Gold Trust, GLD; iShares Silver Trust, SLV) and futures-based energy funds (United States Oil Fund, USO; United States Natural Gas Fund, UNG). Our central question is whether transmission is unidirectional or bidirectional, and how its direction, strength, and time horizon vary with commodity type, sampling frequency, and the continuous versus jump nature of volatility. We study volatility transmission rather than price discovery because it reveals how risk, not just information, propagates across linked markets and whether ETF arbitrage stabilizes or amplifies it---a question central to risk management and systemic-risk regulation.
15 +
16 +The empirical literature on ETF volatility transmission is incomplete in three respects. First, it focuses on equity ETFs using daily data \citep{ben2018etfs, israeli2017etf}, leaving the intraday dynamics that govern arbitrage largely unexplored. Indeed, as arbitrage operates continuously through the day, daily aggregation may obscure rapid transmission. Second, the literature has not examined how spillovers differ across commodity types, even though precious metals trade in liquid global markets with physical arbitrage while energy commodities rely on futures with rollover costs and storage constraints---features that should generate systematically different transmission. Third, it has not distinguished the roles of underlying assets versus ETFs in driving volatility, leaving open the questionwhether ETFs passively follow their constituents or actively feed back into them.
17 +
18 +We address these gaps by constructing minute-by-minute indicative Net Asset Value (iNAV) series for the four ETFs over thirteen years spanning the European sovereign debt crisis, the 2014--2016 commodity collapse, the COVID-19 pandemic, and the subsequent inflation surge. We combine Heterogeneous Autoregressive (HAR) models that capture the long memory of realized volatility across daily, weekly, and monthly horizons \citep{corsi2009simple} with Bayesian Vector Autoregression (BVAR) models to accommodate time-varying dependence while avoiding overfitting \citep{koop2011forecasting}. We measure realized variance from high-frequency returns \citep{andersen2001distribution} and decompose it into continuous and jump components \citep{barndorff2004power}. We separate jumps because commodity prices respond to discrete shocks---geopolitical events, supply disruptions---that may transmit through different channels than smooth price movements, with distinct consequences for tail-risk hedging. On this basis, we test four hypotheses, stated formally in Section~\ref{sec:methodolog}: (i) whether transmission is uni- or bidirectional and varies by commodity type; (ii) whether high-frequency sampling reveals dynamics hidden in daily data; (iii) whether jumps or the continuous component dominate transmission; and (iv) whether transmission is stable over time.
19 +
20 +Our results reveal substantial heterogeneity across commodity types and sampling frequencies. For precious metals (GLD, SLV), transmission is strongly unidirectional from iNAV to ETF, with spillover coefficients ranging from 0.42 (silver) to 0.63 (gold) at one-minute frequency and negligible reverse effects. Energy ETFs (USO, UNG) show bidirectional transmission, with the iNAV-to-ETF direction dominant for crude oil (about four to one) and the two directions comparable for natural gas. Sampling frequency matters: the daily iNAV-to-ETF spillover is up to roughly twice as large in one-minute as in thirty-minute data. Jump components dominate continuous transmission, especially for precious metals. The BVAR analysis confirms these patterns through impulse responses and variance decompositions in which underlying volatility rivals or exceeds ETF self-persistence in explaining ETF volatility.
21 +
22 +The paper makes two contributions. Methodologically, we construct the first comprehensive high-frequency iNAV series for commodity ETFs over an extended period, enabling precise measurement of intraday arbitrage relationships that are unobservable with daily NAV data. Empirically, we document systematic differences in transmission across commodity categories and sampling frequencies, showing that market structure shapes information transmission and that empirical inference is affected by temporal aggregation. Thus, we extend the literature on ETF volatility transmission from equity to commodity markets \citep{ben2018etfs, israeli2017etf}. We further connect it to the microstructure literature \citep{hasbrouck2003intraday, richie2008examination}, and we contribute to research on commodity-ETF financialization \citep{todorov2021etf, buyuksahin2014speculation} by showing that the effects vary in systematic ways between physically-backed and futures-based ETFs. The remainder of the paper is as follows. Section~\ref{sec:data} describes the data and the construction of high-frequency iNAV and realized-volatility series. Section~\ref{sec:methodolog} presents the econometric framework and states the hypotheses. Section~\ref{sec:results} reports the empirical results. Section~\ref{sec:conclusion} concludes.
added phd_chap2_20260731/sections/03_data.tex +76 −0
@@ -0,0 +1,76 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Data and Sample Construction} \label{sec:data}
4 +
5 +Our data processing approach improves on what has been previously used in the ETF volatility literature, which typically relies on daily data and may not detect high-frequency transmission channels \citep{ben2018etfs, israeli2017etf}. By constructing minute-by-minute price and indicative NAV series for four commodity ETFs over thirteen years, we capture volatility dynamics at horizons not previously examined and compare volatility transmission across heterogeneous commodity market structures.
6 +
7 +\subsection{Sample Selection and Data Sources}
8 +
9 +We study four large single-commodity ETFs, each among the top funds in its category ranked by assets under management:
10 +
11 +\begin{itemize}
12 +\item \textbf{SPDR Gold Trust (GLD)} and \textbf{iShares Silver Trust (SLV)} are physically-backed precious-metals ETFs holding bullion in trust (over \$50 billion and \$10 billion in assets, respectively). Physical backing minimizes tracking error but creates arbitrage frictions tied to delivery, storage, and insurance.
13 +\item \textbf{United States Oil Fund (USO)} and \textbf{United States Natural Gas Fund (UNG)} are futures-based energy ETFs tracking WTI crude oil and Henry Hub natural gas through NYMEX futures. Continuous rolling of expiring contracts generates tracking errors related to contango and backwardation \citep{todorov2021etf}, and natural gas adds extreme seasonality and storage constraints.
14 +\end{itemize}
15 +
16 +We select these ETFs for three reasons: high liquidity (top-five funds by assets in each category), continuous tradability over the full sample (January 2010 to January 2023) and-, most importantly, contrasting tracking mechanisms. Physically-backed ETFs (GLD, SLV) settle through physical delivery against London Bullion Market Association (LBMA) spot prices, with no rollover costs but slower arbitrage. Futures-based ETFs (USO, UNG) are arbitraged electronically against NYMEX futures, with faster settlement but involving roll costs, contango/backwardation patterns, and basis risk. Since these market structures create different frictions, transaction costs and lags, we expect them to generate different patterns of volatility transmission. We use single-commodity rather than index ETFs to avoid cross-commodity correlation effects.
17 +
18 +Our primary data source is the Bloomberg terminal, which provides tick-by-tick trade prices, bid--ask quotes, and volume with millisecond timestamps for the ETFs and their underlying assets. We supplement this with futures contract price data from the Chicago Mercantile Exchange (CME) and spot price data from the LBMA. We begin the sample in 2010 for three reasons: By this date, the ETFs (launched 2004--2007) had plausibly reached the trading volume and market-making infrastructure needed for reliable high-frequency data and meaningful arbitrage. Moreover, starting in 2010 avoids the atypical microstructure effects during certain periods of the 2008--2009 crisis. Lastly, consistent tick-level data became available across all instruments in our sample. The resulting thirteen-year period spans the 2010--2012 European sovereign debt crisis, the 2014--2016 commodity collapse, the 2020 COVID-19 pandemic, and the 2021--2022 inflation surge, covering diverse volatility regimes.
19 +
20 +\subsection{High-Frequency Data Construction and Cleaning}
21 +
22 +High-frequency data require careful filtering to remove microstructure noise. Starting from raw tick data over regular U.S. market hours (9:30 AM--4:00 PM EST), we apply standard procedures adapted to the ETF market structure \citep{barndorff2009realized}. First, we remove outliers using the \citet{brownlees2014practical} method, deleting observations more than 10 standard deviations from a rolling 20-minute median (about 0.03\% of observations). Second, we apply the duration filter suggested by \citet{hansen2005realized}, excluding trades separated by more than 30 minutes, which typically indicates closures or technical failures (less than 0.1\% of observations). We then build synchronized price series at 1-, 5-, and 30-minute frequencies using previous-tick interpolation \citep{andersen2001distribution}. The final cleaned dataset contains roughly 45 million price observations. Data loss is minimal and concentrated in thinly traded periods.
23 +
24 +\subsection{Indicative Net Asset Value (iNAV) Construction}
25 +
26 +Official NAV is published only at the end of each trading day \citep{petajisto2017inefficiencies}. This variable cannot capture the intraday arbitrage activities that actually drive ETF pricing. We therefore construct an indicative NAV (iNAV) at a high frequency, which provides a real-time estimate of fundamental value from current underlying prices. This approach lets us measure arbitrage and transmission at the frequency where authorized participants make creation/redemption decisions, and let us separate fundamental ETF volatility (changes in underlying value) from non-fundamental volatility (liquidity shocks, inventory effects, or temporary arbitrage breakdowns).
27 +
28 +For physically-backed ETFs (GLD, SLV), iNAV follows a composition-based approach:
29 +
30 +\begin{equation}
31 +\text{iNAV}_{t} = \frac{1}{N_t} \left[ \text{Cash}_t + \sum_{i} \left( P_{it} \cdot f_{it} \cdot q_{it} \cdot c_{it} \right) \right]
32 +\end{equation}
33 +
34 +where $N_t$ is the number of outstanding ETF shares, $\text{Cash}_t$ is the fund's cash holdings, $P_{it}$ is the price of underlying asset $i$ in local currency, $f_{it}$ the currency conversion factor, $q_{it}$ the quantity held, and $c_{it}$ an adjustment for accrued interest, dividends, or other cash flows. We use LBMA gold and silver prices converted to U.S. dollars at real-time exchange rates. Physical holdings are updated daily with creation/redemption activity, while intraday changes reflect only price movements.
35 +
36 +For futures-based ETFs (USO, UNG), iNAV follows a futures-position model:
37 +
38 +\begin{equation}
39 +\text{iNAV}_{t} = \frac{1}{N_t} \left[ \text{Cash}_t + \sum_{j} \left( F_{jt} \cdot cc_{jt} \cdot q_{jt} \cdot m_{jt} \right) \right] \cdot FX_t
40 +\end{equation}
41 +
42 +where $F_{jt}$ is the price of futures contract $j$, $cc_{jt}$ the contract conversion factor, $q_{jt}$ the number of contracts held, $m_{jt}$ the contract multiplier, and $FX_t$ any currency conversion. We use real-time NYMEX WTI crude oil and Henry Hub natural gas futures, accounting for the funds' actual contract positions and monthly roll schedules.
43 +
44 +A technical consideration is the mismatch between ETF trading hours and underlying market hours: gold and silver trade nearly around the clock in London and Asia, while energy futures have defined sessions. To handle this issue, we weight each market session by its share of price discovery and carry forward the most recent adjusted prices when an underlying market is closed. We validate the construction against published end-of-day NAV. We find that our iNAV has a correlation greater than 0.999 with the official NAV, and mean absolute deviations of less than 5 basis points.
45 +
46 +\subsection{Realized Variance Construction and Jump Detection}
47 +
48 +We measure volatility using realized variance, i.e., the sum of squared intraday returns \citep{andersen2001distribution}. For asset $i$ on day $t$:
49 +
50 +\begin{equation}
51 +RV_{i,t} = \sum_{j=1}^{M} r_{i,t,j}^2
52 +\end{equation}
53 +
54 +where $r_{i,t,j} = \log(P_{i,t,j}) - \log(P_{i,t,j-1})$ and $M$ is the number of intraday returns, giving $M=390$ (1-minute), $M=78$ (5-minute), and $M=13$ (30-minute). Under standard conditions, realized variance converges to integrated variance as sampling intensifies \citep{barndorff2002econometric}. The choice of frequency trades statistical efficiency against microstructure bias from bid--ask bounce \citep{hansen2005realized, liu2015does}, so we use three frequencies to assess robustness.
55 +
56 +To separate continuous movements from discrete jumps, we use the bipower variation \citep{barndorff2004power}:
57 +
58 +\begin{equation}
59 +BV_{i,t} = \mu_1^{-2} \sum_{j=2}^{M} |r_{i,t,j}| \cdot |r_{i,t,j-1}|
60 +\end{equation}
61 +
62 +where $\mu_1 = \sqrt{2/\pi} \approx 0.798$. Bipower variation consistently estimates integrated variance even in the presence of jumps. The jump component is
63 +
64 +\begin{equation}
65 +J_{i,t} = \max(RV_{i,t} - BV_{i,t}, 0)
66 +\end{equation}
67 +
68 +and the continuous component is $C_{i,t} = RV_{i,t} - J_{i,t}$ \citep{huang2005using}. This continuous component is reported as quadratic power variation (QPV) in the estimation tables. Separating the two matters because jumps---from supply disruptions or geopolitical events---may transmit across markets differently than smooth volatility.
69 +
70 +\subsection{Descriptive Statistics and Stylized Facts}
71 +
72 +Tables \ref{tab:desc_stats_5min}, \ref{tab:desc_stats_1min}, and \ref{tab:desc_stats_30min} report descriptive statistics for realized variance, bipower variation, and jumps across the three frequencies. Several stylized facts emerge. Realized variance is heterogeneous across commodities and between ETFs and their iNAVs. Natural gas and crude oil have the highest average volatility: the 5-minute mean realized variance is 0.104\%, and 0.081\% for the iNAV. For crude oil, the iNAV (0.081\%) is more volatile than the ETF value (0.066\%), as expected from the damping effect of arbitrage. The heaviest distributional tails occur in crude oil and gold: For crude oil, the iNAV realized variance reaches a maximum above 40\%. For gold, the ETF series shows occasional extreme outliers (maximum near 79\%), reflecting occasional disruptions in physical gold arbitrage adjustments. Precious metals are otherwise less volatile on average than energy, though for gold the ETF mean volatility (0.034\%) exceeds its iNAV mean (0.012\%). Jump activity is most pronounced for crude oil, consistent with this commodity's sensitivity to geopolitical and supply shocks, and it is weakest for precious metals, for which prices move more smoothly.
73 +
74 +Figures \ref{fig:rv_uso}, \ref{fig:rv_gld}, \ref{fig:rv_slv}, and \ref{fig:rv_ung} show the evolution of realized variance over time. All series show strong volatility clustering, which is most pronounced during the 2014--2016 oil collapse, the 2016 Brexit referendum (precious metals), and the 2020 pandemic (for all commodities). ETF--iNAV synchronization is high for precious metals but more variable for energy commodities, where periods of close co-movement alternate with divergence. As these relationships are time-varying, a simple correlation analysis would fail to detect economically significant dynamics, which motivates the modeling framework in Section~\ref{sec:methodolog}.
75 +
76 +Finally, all realized-variance series are stationary in both levels and logs (augmented Dickey--Fuller tests), and we validate the GLD iNAV against the NYSE indicative optimized portfolio value (IOPV), obtaining correlations above 0.995. Tests for structural breaks and the full set of robustness checks are reported with the econometric framework in Section~\ref{sec:methodolog}.
added phd_chap2_20260731/sections/04_methods.tex +103 −0
@@ -0,0 +1,103 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Econometric Methodology} \label{sec:methodolog}
4 +
5 +We analyze volatility transmission between commodity ETFs and their iNAVs with two complementary frameworks. Heterogeneous Autoregressive (HAR) models \citep{corsi2009simple} capture the long memory of realized volatility across daily, weekly, and monthly horizons and, in cross-market form, yield interpretable spillover coefficients. Bayesian Vector Autoregression (BVAR) models \citep{koop2011forecasting, carriero2015forecasting} treat ETF and iNAV volatility as jointly endogenous, using Minnesota-prior shrinkage to control parameter proliferation and providing impulse responses and variance decompositions with proper uncertainty bands. We favor realized-variance models over GARCH because realized variance is a nearly model-free volatility estimate that forecasts better \citep{andersen2001distribution}, and HAR over fractionally integrated (ARFIMA) specifications because it captures long memory parsimoniously through a horizon cascade \citep{corsi2009simple}.
6 +
7 +\subsection{Theoretical Framework for Volatility Transmission}
8 +
9 +Under frictionless arbitrage, ETF prices equal their NAV and volatility transmission is instantaneous and bidirectional \citep{ackert2000arbitrage, petajisto2017inefficiencies}. Real-world frictions---transaction costs, inventory and funding constraints---can instead create asymmetric transmission that varies across horizons and market conditions. We study volatility transmission rather than price discovery because it reveals how risk propagates and whether arbitrage stabilizes or amplifies fluctuations, with direct implications for hedging effectiveness and systemic risk. Examining multiple horizons follows the heterogeneous market hypothesis \citep{muller1997volatilities}, under which day traders, institutions, and longer-horizon participants generate volatility that persists over their characteristic time scales. Finally, the contrast between physically-backed ETFs (GLD, SLV), whose arbitrage requires physical delivery, and futures-based ETFs (USO, UNG), which face roll costs and basis risk \citep{todorov2021etf}, motivates our cross-sectional comparison.
10 +
11 +\subsection{Heterogeneous Autoregressive (HAR) Models}
12 +
13 +The HAR model \citep{corsi2009simple} is the dominant framework for realized-volatility dynamics \citep{andersen2007roughing}. The baseline specification is
14 +
15 +\begin{equation} \label{eq:har_baseline}
16 +\log(RV_{i,t}) = \beta_0 + \beta_1 \log(RV_{i,t-1}) + \beta_2 \log(\overline{RV}_{i,t-5:t-1}) + \beta_3 \log(\overline{RV}_{i,t-22:t-1}) + \varepsilon_{i,t}
17 +\end{equation}
18 +
19 +where $\overline{RV}_{i,t-h:t-1} = \frac{1}{h}\sum_{j=1}^{h} RV_{i,t-j}$ averages realized variance over the previous $h$ days, capturing daily, weekly ($h=5$), and monthly ($h=22$) persistence. The log transform keeps fitted volatility positive and stabilizes the residual variance.
20 +
21 +To measure transmission, we add cross-market terms. For iNAV volatility,
22 +
23 +\begin{align} \label{eq:har_x_nav}
24 +\log(RV_{t,\text{NAV}}) &= \beta_0 + \beta_1 \log(RV_{t-1,\text{NAV}}) + \beta_2 \log(\overline{RV}_{t-5:t-1,\text{NAV}}) + \beta_3 \log(\overline{RV}_{t-22:t-1,\text{NAV}}) \\
25 +&\quad + \alpha_1 \log(RV_{t-1,\text{ETF}}) + \varepsilon_{t,\text{NAV}} \nonumber
26 +\end{align}
27 +
28 +and symmetrically for ETF volatility,
29 +
30 +\begin{align} \label{eq:har_x_etf}
31 +\log(RV_{t,\text{ETF}}) &= \gamma_0 + \gamma_1 \log(RV_{t-1,\text{ETF}}) + \gamma_2 \log(\overline{RV}_{t-5:t-1,\text{ETF}}) + \gamma_3 \log(\overline{RV}_{t-22:t-1,\text{ETF}}) \\
32 +&\quad + \delta_1 \log(RV_{t-1,\text{NAV}}) + \varepsilon_{t,\text{ETF}} \nonumber
33 +\end{align}
34 +
35 +Each equation includes the own daily, weekly, and monthly terms and a single cross-market term at the daily lag, following the parsimony of the HAR cascade. The coefficient $\alpha_1$ measures the daily ETF-to-iNAV spillover and $\delta_1$ the daily iNAV-to-ETF spillover; the own weekly ($\beta_2,\gamma_2$) and monthly ($\beta_3,\gamma_3$) terms capture longer-horizon volatility persistence.
36 +
37 +To separate the transmission of continuous and jump volatility, we estimate a HAR-CJ-X specification using the components $C$ and $J$ defined in Section~\ref{sec:data}:
38 +
39 +\begin{align} \label{eq:har_cj_x_nav}
40 +\log(RV_{t,\text{NAV}}) &= \beta_0 + \beta_1 \log(C_{t-1,\text{NAV}}) + \beta_2 \log(\overline{C}_{t-5:t-1,\text{NAV}}) + \beta_3 \log(\overline{C}_{t-22:t-1,\text{NAV}}) \\
41 +&\quad + \beta_4 \log(1+J_{t-1,\text{NAV}}) + \beta_5 \log(1+\overline{J}_{t-5:t-1,\text{NAV}}) + \beta_6 \log(1+\overline{J}_{t-22:t-1,\text{NAV}}) \nonumber \\
42 +&\quad + \alpha_1 \log(C_{t-1,\text{ETF}}) + \alpha_2 \log(1+J_{t-1,\text{ETF}}) + \varepsilon_{t,\text{NAV}} \nonumber
43 +\end{align}
44 +
45 +with an analogous ETF equation. The own continuous and jump components enter at all three horizons through $\beta_1$--$\beta_3$ and $\beta_4$--$\beta_6$; the daily cross-market continuous and jump spillovers are $\alpha_1$ and $\alpha_2$. The transform $\log(1+J_{t})$ keeps jump terms defined when $J_{t}=0$.
46 +
47 +\subsection{Bayesian Vector Autoregression (BVAR)}
48 +
49 +The HAR cascade imposes a fixed horizon structure. As a flexible complement, we estimate a VAR treating ETF and iNAV volatility as jointly endogenous:
50 +
51 +\begin{equation} \label{eq:var_baseline}
52 +\mathbf{y}_t = \mathbf{c} + \sum_{k=1}^{p} \mathbf{A}_k \mathbf{y}_{t-k} + \mathbf{u}_t
53 +\end{equation}
54 +
55 +where $\mathbf{y}_t = [\log(RV_{t,\text{ETF}}), \log(RV_{t,\text{NAV}})]'$ and $\mathbf{u}_t \sim \mathcal{N}(\mathbf{0}, \mathbf{\Sigma})$. Information criteria (BIC) select $p=2$. Because unrestricted VARs over-parameterize, we apply the Minnesota prior \citep{litterman1986forecasting}:
56 +
57 +\begin{align} \label{eq:minnesota_prior}
58 +\beta_{ij}^{(k)} &\sim \mathcal{N}(0, \lambda_1^2 \cdot k^{-\lambda_3}) \quad \text{for } i \neq j \\
59 +\beta_{ii}^{(1)} &\sim \mathcal{N}(1, \lambda_1^2) \nonumber \\
60 +\beta_{ii}^{(k)} &\sim \mathcal{N}(0, \lambda_1^2 \cdot k^{-\lambda_3}) \quad \text{for } k > 1 \nonumber
61 +\end{align}
62 +
63 +where $\beta_{ij}^{(k)}$ is the coefficient on variable $j$ at lag $k$ in equation $i$; $\lambda_1$ controls overall tightness, $\lambda_2$ cross-variable shrinkage, and $\lambda_3$ lag decay. We set $\lambda_1=0.2$, $\lambda_2=0.5$, $\lambda_3=2$ and vary them in sensitivity analysis. The prior encodes that own lags matter more than others, recent lags more than distant ones, and coefficients are not extreme---assumptions well suited to persistent, mean-reverting volatility.
64 +
65 +We estimate by Gibbs sampling, alternating between the coefficients (multivariate normal given $\mathbf{\Sigma}$) and the covariance (inverse-Wishart given the coefficients):
66 +
67 +\begin{equation} \label{eq:coefficient_posterior}
68 +\boldsymbol{\beta} | \mathbf{\Sigma}, \mathbf{Y} \sim \mathcal{N}(\hat{\boldsymbol{\beta}}, \mathbf{\Sigma} \otimes (\mathbf{X}'\mathbf{X} + \mathbf{V}_0^{-1})^{-1}), \qquad
69 +\hat{\boldsymbol{\beta}} = (\mathbf{X}'\mathbf{X} + \mathbf{V}_0^{-1})^{-1}(\mathbf{X}'\text{vec}(\mathbf{Y}) + \mathbf{V}_0^{-1}\boldsymbol{\beta}_0)
70 +\end{equation}
71 +
72 +\begin{equation} \label{eq:covariance_posterior}
73 +\mathbf{\Sigma} | \boldsymbol{\beta}, \mathbf{Y} \sim \text{IW}(\mathbf{S} + \mathbf{S}_0, T + \nu_0)
74 +\end{equation}
75 +
76 +where $\mathbf{Y}$ and $\mathbf{X}$ are the stacked data, $\mathbf{V}_0$, $\mathbf{S}_0$, and $\nu_0$ are prior parameters, $\mathbf{S}$ the residual sum of squares, and $T$ the sample size. We run 50{,}000 iterations, discard 10{,}000 as burn-in, and thin every tenth draw; convergence is assessed with trace plots and the Geweke test.
77 +
78 +From the posterior we compute orthogonalized impulse responses and forecast error variance decompositions:
79 +
80 +\begin{equation} \label{eq:impulse_response}
81 +\text{IRF}(h) = \mathbf{C}_h \mathbf{P}, \qquad
82 +\text{FEVD}_{i,j}(h) = \frac{\sum_{k=0}^{h-1} [\mathbf{C}_k \mathbf{P}]_{i,j}^2}{\sum_{k=0}^{h-1} [\mathbf{C}_k \mathbf{\Sigma} \mathbf{C}_k']_{i,i}}
83 +\end{equation}
84 +
85 +where $\mathbf{C}_h$ is the $h$-step moving-average matrix and $\mathbf{P}$ the Cholesky factor of $\mathbf{\Sigma}$. We order iNAV before ETF, so iNAV innovations may affect ETF volatility contemporaneously but not the reverse, reflecting that underlying price movements lead ETF adjustments through arbitrage. The confidence bands follow \citet{sims1999error}. A high $\text{FEVD}_{\text{ETF},\text{NAV}}(h)$ indicates that iNAV innovations explain ETF volatility, and conversely for $\text{FEVD}_{\text{NAV},\text{ETF}}(h)$.
86 +
87 +\subsection{Hypotheses and Testing}
88 +
89 +We test four hypotheses, mapping each to coefficients in the equations above.
90 +
91 +\textbf{H1 (Direction and cross-commodity heterogeneity).} Theory predicts unidirectional iNAV-to-ETF transmission, but a more liquid ETF can reverse the flow \citep{glosten2021etf}. We expect physically-backed precious metals to show unidirectional iNAV-to-ETF transmission and futures-based energy ETFs more balanced bidirectional transmission. In equations~\eqref{eq:har_x_nav}--\eqref{eq:har_x_etf}, ETF-to-iNAV transmission tests $H_0:\alpha_1=0$ against $H_a:\alpha_1\neq 0$, and iNAV-to-ETF transmission tests $H_0:\delta_1=0$ against $H_a:\delta_1\neq 0$. Transmission is unidirectional when only $\delta_1$ is significant and bidirectional when both $\alpha_1$ and $\delta_1$ are significant. Heterogeneity means these outcomes differ across commodities.
92 +
93 +\textbf{H2 (Frequency dependence).} If arbitrage occurs within minutes, daily aggregation would understate transmission. Using the daily spillover coefficients $\delta_1$ and $\alpha_1$ estimated at the 1-, 5-, and 30-minute frequencies, we test $H_0: \delta_1^{(1\text{m})}=\delta_1^{(5\text{m})}=\delta_1^{(30\text{m})}$ against $H_a:$ the coefficients differ across frequencies.
94 +
95 +\textbf{H3 (Jumps versus continuous component).} Tail risk from discrete shocks is harder to hedge than smooth volatility, which is why we decompose volatility into continuous and jump parts. In the HAR-CJ-X model~\eqref{eq:har_cj_x_nav}, we compare the daily cross-market jump and continuous spillovers, testing $H_0:\alpha_2=\alpha_1$ (jump and continuous transmission equal) against $H_a:\alpha_2>\alpha_1$ (jump transmission dominates), and analogously for the ETF equation.
96 +
97 +\textbf{H4 (Stability over time).} Over a data sample spanning several crises, patterns of transmission may evolve. We re-estimate the BVAR over sub-periods (2010--2014, 2015--2019, 2020--2023) and apply the structural-break test of \citet{bai2003computation}, testing $H_0:$ transmission coefficients are constant across regimes against $H_a:$ they change.
98 +
99 +\subsection{Estimation, Model Selection, and Robustness}
100 +
101 +We select lag lengths and compare nested models with the AIC and BIC (the DIC for Bayesian models, which accounts for shrinkage), and evaluate out-of-sample forecasts with rolling windows (a 1{,}000-day estimation window re-estimated every 250 days) at the 1-, 5-, and 22-day horizons using RMSE and MAE. Residual diagnostics include the Ljung--Box (serial correlation), Breusch--Pagan (heteroskedasticity), and Jarque--Bera (normality) tests, plus posterior predictive checks for the BVAR. We also gauge economic magnitude through the effect of one-standard-deviation shocks on forecast volatility.
102 +
103 +We assess robustness along three dimensions, consolidating the checks noted in Section~\ref{sec:data}. First, \emph{specifications}: alternative HAR lag structures, BVAR lags from one to four, level and square-root (rather than log) transforms, and time-varying-parameter versions. Second, \emph{sample stability}: the sub-period and regime estimation underlying H4, structural-break tests \citep{bai2003computation}---which flag the 2014 oil collapse, the 2016 Brexit referendum, and the 2020 pandemic---and bootstrap inference. Our transmission results hold across regimes, indicating they capture general mechanisms rather than period-specific effects. Third, \emph{data construction}: additional 15- and 60-minute frequencies, alternative estimators (truncated realized variance, realized kernels, range-based measures), and alternative iNAV constructions (currency conversion, cash treatment), including validation against the GLD IOPV. Our main findings are robust throughout.
added phd_chap2_20260731/sections/05_results.tex +59 −0
@@ -0,0 +1,59 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Empirical Results} \label{sec:results}
4 +
5 +This section reports the results of our empirical analysis of volatility transmission between commodity ETFs and their underlying assets. We organize the presentation of results around the four hypotheses stated in Section~\ref{sec:methodolog}, analyzing them in order: namely, the direction of transmission and its variation across commodity types (H1), the role of sampling frequency (H2), the relative importance of jump and continuous components (H3), and the stability of transmission over time (H4).
6 +
7 +\subsection{Direction of Transmission and Commodity-Specific Asymmetries}
8 +
9 +Our first hypothesis concerns whether transmission between an ETF and its iNAV is unidirectional or bidirectional, and whether this varies across commodities (H1). As discussed in Section~\ref{sec:methodolog}, theory predicts a flow only from iNAV to ETF, but a more liquid ETF can reverse the direction of the flow. In the case of commodities, the storability and settlement mechanism is also expected to matter. We therefore expect physically-backed precious metals to show unidirectional iNAV-to-ETF transmission, and futures-based energy ETFs more balanced bidirectional transmission.
10 +
11 +The HAR-X estimates in Tables~\ref{tab:HAR_5min} through~\ref{tab:HAR_30min} reject the null of uniform bidirectional transmission across all commodities, extending the heterogeneity documented by \citet{gorton2006facts} and \citet{buyuksahin2014speculation} to the ETF setting and consistent with \citet{basak2016model} on how financialization effects vary across commodity types.
12 +
13 +The results for precious metals show a strongly unidirectional transmission from iNAV to ETF, with little evidence of reverse transmission. Consider the iNAV-to-ETF coefficient, which is the coefficient for lagged NAV volatility in the ETF equation. For gold at a 1-minute frequency, this coefficient is 0.632 (Table~\ref{tab:HAR_1min}), which is among the largest across commodities, while the ETF-to-iNAV effect (i.e., lagged ETF volatility in the NAV equation) is not statistically different from zero ($-0.001$, $p=0.96$). The results for silver show a similar but weaker pattern: the NAV-to-ETF coefficient is 0.418 while the reverse effect is 0.004 and not significant. On the other hand, our results for energy commodities show bidirectional transmission. For crude oil at a 1-minute frequency, the iNAV-to-ETF coefficient is 0.378 and the ETF-to-iNAV coefficient is 0.089. Both are significant at the 1\% level, and the coefficient for iNAV-to-ETF is roughly four times larger. The results for natural gas are the most directionally balanced: the effects in both directions are significant and of similar magnitude (iNAV-to-ETF 0.079; ETF-to-iNAV 0.127). This contrast between the strongly unidirectional results for metals and the bidirectional results for energy supports our hypothesis H1. Indeed, the evidence confirms that physical versus futures-based arbitrage, settlement mechanism, and liquidity shape the transmission of volatility between the commodity underlying and the ETF.
14 +
15 +These patterns are confirmed by the results for the formal tests described in Section~\ref{sec:methodolog}. The ETF-to-iNAV restriction $H_0:\alpha_1=0$ in equation~\eqref{eq:har_x_nav} is not rejected for gold or silver---their reverse coefficients are insignificant at every frequency---but is rejected at the 1\% level for crude oil and natural gas. The iNAV-to-ETF restriction $H_0:\delta_1=0$ in equation~\eqref{eq:har_x_etf} is rejected at the 1\% level for all four commodities. Transmission is therefore unidirectional (iNAV-to-ETF) for precious metals and bidirectional for energy. Moreover, the iNAV-to-ETF channel dominates for crude oil (with a magnitude of about four to one), while for natural gas the effects for the two directions are comparable.
16 +
17 +The economic interpretation is consistent across models. In precious metals, arbitrage requires physical delivery against LBMA bullion, which is costly and slow. Authorized participants readily create or redeem ETF shares in response to underlying price moves, but cannot easily push ETF-specific shocks back into the tightly arbitraged spot market, so volatility flows essentially one way. The reverse coefficients for both metals hover near zero at all frequencies, confirming that ETF activity does not transmit volatility back to the spot market. In energy markets, by contrast, both the ETF and the underlying futures settle electronically and trade with comparable liquidity, so shocks propagate in both directions, even though fundamental supply-and-demand information still enters first through the futures-based iNAV.
18 +
19 +Within precious metals, gold shows stronger unidirectional effects than silver at every frequency. The iNAV-to-ETF coefficients are 0.632 at a 1-minute and 0.387 at a 30-minute frequency for gold, versus 0.418 to 0.277 for silver. These results are consistent with the hypothesis that gold is a financial store of value, while silver carries additional industrial demand-related volatility. The reverse (ETF-to-iNAV) coefficients for both metals stay close to zero across frequencies: gold between $-0.015$ and 0.021, silver between $-0.034$ and 0.004, and neither is economically meaningful. In the category of energy commodities, crude oil shows a clear iNAV-to-ETF dominance across frequencies. However, the effects are more balanced for natural gas. For iNAV-to-ETF and ETF-to-iNAV, respectively, the coefficients are 0.105 and 0.126 at the 5-minute, and 0.081 and 0.074 at the 30-minute frequency. These results reflect the illiquidity, storage limits, and contango that impede arbitrage in the ETF and futures markets.
20 +
21 +As the HAR-X specification includes cross-market terms only at the daily lag, directional transmission is identified at the daily horizon. The weekly and monthly coefficients measure each series' own persistence rather than spillovers. This own-persistence is high and, for precious metals, stable across horizons and frequencies: for instance, gold's weekly own-volatility coefficient stays near 0.33, confirming the presence of long memory in realized volatility.
22 +
23 +\subsection{The Relevance of Using High-Frequency Data}
24 +
25 +Our second hypothesis (H2) is that higher-frequency sampling reveals transmission that is obscured in daily data. Tables~\ref{tab:HAR_5min}, \ref{tab:HAR_1min}, and~\ref{tab:HAR_30min} report the HAR-X estimates at 5-, 1-, and 30-minute frequencies. The daily iNAV-to-ETF transmission is markedly larger at a finer sampling frequency. For crude oil, it is measured at 0.378 at a 1-minute, 0.311 at a 5-minute, and 0.20 at a 30-minute frequency. The 1-minute estimate represents a 22\% increase over the 5-minute estimate and nearly double the 30-minute estimate. Thus, much of the same-day arbitrage transmission occurs within minutes, and measurements of this mechanism would be understated at a coarser sampling, as is traditionally used in the literature.
26 +
27 +The sensitivity of estimates according to frequency also varies by commodity. The iNAV-to-ETF transmission decreases substantially from 1-minute to 30-minute sampling in the case of crude oil (0.378 to 0.20, i.e., a 47\% drop), gold (0.632 to 0.387, or 39\% smaller), and silver (0.418 to 0.277, or 34\% smaller). It is, however, essentially flat for natural gas (0.079 versus 0.081). The economic interpretation is that transmission in the actively and continuously arbitraged crude oil and bullion markets clears within minutes, while for natural gas this adjusts over longer horizons because storage and pipeline constraints slow down the arbitrage activities that would otherwise help equalize ETF and underlying volatility.
28 +
29 +The reverse channel, ETF-to-iNAV transmission, is empirically weaker. Moreover, unlike the iNAV-to-ETF channel, the estimated effects do not strengthen monotonically with the sampling frequency. For crude oil, the daily ETF-to-iNAV coefficient is 0.089 at a 1-minute, 0.111 at a 5-minute, and 0.103 at a 30-minute frequency, peaking at the intermediate horizon rather than rising with the frequency. The contrast between the strongly frequency-dependent iNAV-to-ETF channel and this flatter reverse channel reinforces the directional dominance finding which we describe under hypothesis H1.
30 +
31 +Overall, we reject the null hypothesis of frequency-invariant transmission. A Wald test of $H_0:\delta_1^{(1\text{m})}=\delta_1^{(5\text{m})}=\delta_1^{(30\text{m})}$ on the daily iNAV-to-ETF coefficient rejects equality at the 1\% level for crude oil, gold, and silver; for natural gas, whose spillover is small and flat across frequencies, the difference is not significant. Daily data thus understate short-horizon transmission by up to roughly a factor of two for the most affected commodities, which would lead to incorrect conclusions about arbitrage effectiveness, addressing the sampling-frequency question raised by \citet{hansen2005realized}. For practitioners, this finding supports the relevance of high-frequency market monitoring to detect transmission patterns that are understated in daily analysis.
32 +
33 +\subsection{Jump Components and Discontinuous Volatility Transmission}
34 +
35 +Our third hypothesis (H3) concerns whether transmission is driven by the continuous (diffusion) component or by jumps. Tables~\ref{tab:HAR_CJ_5min}, \ref{tab:HAR_CJ_1min}, and~\ref{tab:HAR_CJ_30min} report the HAR-CJ-X model estimates. Across the four commodities, the daily cross-market jump coefficient dwarfs the continuous one: discrete price moves transmit volatility, while smooth price moves essentially do not. In the dominant iNAV-to-ETF direction at a 1-minute frequency, the jump transmission (lagged NAV jump in the ETF equation) is large and positive---0.892 for gold, 0.578 for silver, 0.432 for crude oil, 0.09 for natural gas---while the corresponding continuous transmission is near zero or negative for every commodity (gold $-0.123$, silver $-0.065$, crude oil $-0.019$, natural gas $-0.002$). The effect is strongest for precious metals, where the jump component is the largest in the sample.
36 +
37 +The same pattern holds in the reverse (ETF-to-iNAV) direction but with smaller economic magnitudes: the daily jump coefficient is 0.096 for crude oil and 0.128 for natural gas, while the continuous coefficients are near $-0.01$. Natural gas shows the weakest iNAV-to-ETF jump coefficient (0.090), which is consistent with natural gas volatility spikes arising from idiosyncratic, localized events---hurricanes, pipeline failures, extreme weather---that do not propagate systematically. In contrast, there is a larger value for the \emph{own} continuous effect: i.e., the weekly own-continuous coefficient reaches 0.116 at 1-minute frequency. This result suggests a separate, within-market phenomenon rather than a cross-market transmission pattern. Testing $H_0:\alpha_2=\alpha_1$ (daily jump versus continuous effect) against $H_a:\alpha_2>\alpha_1$, we find that the dominance of the jump component is confirmed at the 1\% level for all four commodities. This result supports H3: transmission mainly occurs through discrete jumps rather than continuous diffusion, which has implications for tail-risk hedging. Indeed, jump-driven volatility is harder to hedge with strategies that are usually built for the assumption of continuous price processes and volatility diffusions.
38 +
39 +\subsection{Bayesian VAR Analysis and Stability Over Time}
40 +
41 +Our fourth hypothesis (H4) concerns whether transmission is stable over time. The BVAR results shown in Tables~\ref{tab:VAR_USO}--\ref{tab:VAR_UNG} characterize the joint dynamics of ETF and iNAV volatility and, through sub-period estimation, the stability of the coefficient estimates. Since the BVAR treats the two volatilities as endogenous, it further provides a check on the HAR-X spillover coefficients without imposing a cascade structure. Across the four commodities, the results for the BVAR model confirm the asymmetries documented using the HAR-X model. Moreover, the cross-market coefficients remain stable across the 2010--2014, 2015--2019, and 2020--2023 sub-periods. Thus, we do not find evidence that the transmission mechanisms meaningfully change over time. The structural break test due to \citet{bai2003computation} indicates breaks in the volatility time series during the 2014 crude oil price collapse, the 2016 Brexit referendum, and the 2020 Covid-19 pandemic. However, reestimating the BVAR model for each regime period does not change the sign and relative magnitude of the cross-market coefficients. Therefore, we fail to reject the null hypothesis (H4) of a volatility transmission model that is stable over the full sample period.
42 +
43 +For crude oil (Table~\ref{tab:VAR_USO}), both volatilities are strongly persistent (iNAV first-lag coefficient 0.5704, 95\% credible interval [0.5237, 0.6171]; second lag 0.2606 [0.2206, 0.3005]). The ETF has only a weak effect on the iNAV (first lag 0.0735 [0.0343, 0.1135]), whereas the iNAV strongly drives the ETF: its first-lag effect (0.2861 [0.2315, 0.3433]) is comparable to ETF self-persistence (0.2941 [0.2460, 0.3418]) and its second-lag effect remains economically large (0.1934 [0.1443, 0.2410]). That the iNAV's first-lag effect rivals the ETF's own self-persistence is striking: for crude oil, underlying volatility is about as important as the ETF's recent volatility in explaining today's ETF volatility---the BVAR counterpart of the strong iNAV-to-ETF spillover found in the HAR-X estimates.
44 +
45 +Gold (Table~\ref{tab:VAR_GLD}) shows the most asymmetric configuration. The ETF has essentially no effect on the iNAV (first lag $-0.0101$ [$-0.0547$, 0.0336], with only a marginal second-lag effect of 0.0383 [0.0001, 0.0759]), while the iNAV dominates the ETF: its first-lag effect (0.3487 [0.2848, 0.4126]) exceeds ETF self-persistence (0.1549 [0.1004, 0.2083]) by more than twofold, with a large second lag (0.2758 [0.2221, 0.3307]). Silver (Table~\ref{tab:VAR_SLV}) is intermediate: iNAV self-persistence is the highest in the sample (first lag 0.6192 [0.5631, 0.6738]), the ETF-to-iNAV effect remains negligible ($-0.0446$ [$-0.0932$, 0.0051], turning weakly positive but still economically trivial at the second lag, 0.0584 [0.0176, 0.0996]), and the iNAV-to-ETF effect is strong (first lag 0.3760 [0.3102, 0.4404] versus ETF self-persistence 0.1790 [0.1226, 0.2362]) and remains economically large at the second lag (0.2016 [0.1471, 0.2563]). Natural gas (Table~\ref{tab:VAR_UNG}) shows the clearest bidirectionality. Its iNAV is moderately persistent (first lag 0.3919 [0.3488, 0.4366], second lag 0.2313 [0.1915, 0.2715]) and, unlike the other commodities, the ETF affects the iNAV through both lags, with a larger second-lag effect (0.0799 [0.0381, 0.1214] then 0.1567 [0.1189, 0.1938]), indicating that ETF activity feeds back to the underlying market through delayed channels tied to natural-gas storage operations. The forecast error variance decompositions reinforce this interpretation. Indeed, iNAV innovations explain a large and increasing share of ETF volatility forecast errors at longer horizons, while the share of iNAV forecast-error variance attributable to ETF innovations stays small for every commodity and is near zero for gold and silver. The impulse responses examined below confirm the same asymmetry.
46 +
47 +\subsection{Graphical Evidence on Volatility Patterns and Dynamic Responses}
48 +
49 +The realized volatility series shown in Figures~\ref{fig:rv_uso}--\ref{fig:rv_ung} corroborate the findings reported above. For crude oil (Figure~\ref{fig:rv_uso}), ETF and iNAV volatility co-move closely with synchronized peaks during stress periods, consistent with bidirectional transmission. For gold (Figure~\ref{fig:rv_gld}), iNAV volatility consistently precedes ETF volatility in periods of significant spikes, which is the visual counterpart of unidirectional iNAV-to-ETF transmission. Silver (Figure~\ref{fig:rv_slv}) shows the same pattern, but with occasional divergence, while natural gas (Figure~\ref{fig:rv_ung}) shows the most frequent ETF--iNAV divergences.
50 +
51 +The impulse responses shown in Figures~\ref{fig:irf1}--\ref{fig:irf4} tell the same story but add evidence on how the shocks evolve over time. For crude oil (Figure~\ref{fig:irf1}), iNAV shocks produce large, persistent responses in ETF volatility while ETF shocks produce small, transitory responses in the iNAV. Gold (Figure~\ref{fig:irf2}) shows the most pronounced asymmetry, with negligible responses of the iNAV to ETF shocks. Silver (Figure~\ref{fig:irf3}) is similar but noisier, while natural gas (Figure~\ref{fig:irf4}) shows sizable responses in both directions with delayed peaks, confirming bidirectional transmission.
52 +
53 +\subsection{Summary of the Evidence for the Hypotheses}
54 +
55 +Taken together, the estimates from the HAR-X, HAR-CJ-X and BVAR models support our four research hypotheses, and the formal tests defined in Section~\ref{sec:methodolog} reject each null hypothesis. We can summarize as follows.
56 +
57 +\textbf{H1.} We find that transmission is heterogeneous: the reverse (ETF-to-iNAV) coefficient is not significant for gold and silver but it is significant for crude oil and natural gas, while the forward (iNAV-to-ETF) coefficient is significant everywhere. Transmission is therefore unidirectional from iNAV to ETF for precious metals (a magnitude about four to one greater) and bidirectional for energy commodities (the coefficients are similar in magnitude). Thus, we reject the null hypothesis of uniform bidirectional transmission and we extend the evidence from \citet{gorton2006facts} to setting of commodity volatility. \textbf{H2.} Transmission is frequency-dependent: the daily iNAV-to-ETF spillover is up to roughly twice as large at the 1-minute as the 30-minute frequency. It is flat only for natural gas. These results confirm that daily data understate these volatility dynamics \citep{hansen2005realized, liu2015does}. \textbf{H3.} The daily cross-market jump component of transmission dwarfs the continuous part for all four commodities---most sharply for precious metals (i.e., the jump coefficient for gold iNAV-to-ETF jump is 0.892 versus $-0.123$ for the continuous component)---so transmission occurs mainly through discrete jumps. \textbf{H4.} The cross-market coefficients are stable across sub-periods and are robust to astructural-break test, so the documented mechanisms hold in general and are not specific to sub-periods.
58 +
59 +These findings build around a single mechanism, namely that the market structure and the arbitrage mechanism that links an ETF to its underlying asset---physical delivery for precious metals versus electronic futures settlement for energy---strongly shapes the characteristics of volatility transmission (e.g., direction, magnitude, horizon, and importance of jump vs diffusion parts), consistent with the arguments of \citet{petajisto2017inefficiencies} and \citet{basak2016model}. Finally, we explainthe implications for investors, market makers, and regulators in the conclusion.
added phd_chap2_20260731/sections/06_conclusion.tex +14 −0
@@ -0,0 +1,14 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Conclusion} \label{sec:conclusion}
4 +
5 +Using high-frequency realized variance and a combination of HAR and Bayesian VAR models, this paper examines volatility transmission between four commodity ETFs and their underlying assets over the period 2010--2023. We find that volatility transmission is unidirectional from iNAV to ETF for physically-backed precious metals (GLD, SLV), while it is bidirectional (though still dominated by iNAV-to-ETF effects) for futures-based energy commodity funds (USO, UNG). Short-horizon transmission is far stronger in high-frequency data than in daily data, while longer-horizon effects are frequency-invariant; jump components transmit more strongly than continuous ones across all commodities; and these patterns are stable across sub-periods.
6 +
7 +These results show how arbitrage operates in practice and how its effectiveness depends on market structure, extending the limits-to-arbitrage frameworks of \citet{ackert2000arbitrage}, \citet{pontiff1996costly}, and \citet{gromb2010limits}. The unidirectional transmission which we document for precious metals is consistent with physical-delivery frictions that limit ETF activity from influencing underlying prices. In contrast, the more balanced volatility transmission in energy commodity markets fits the concept of futures-based arbitrage with electronic settlement, as argued by \citet{basak2016model}. The frequency-specific results suggest that arbitrage operates primarily through high-frequency channels. The evidence of longer-horizon persistence reflects separate fundamental forces, building on the intraday microstructure analysis of \citet{hasbrouck2003intraday} and \citet{richie2008examination}. The dominance of jumps in the transmission of volatility, which we identify using the bipower variation decomposition due to \citet{barndorff2004power} and \citet{huang2005using}, indicates that models used for ETF pricing should pay separate attention to the continuous and discontinuous components of volatility.
8 +
9 +In terms of methodology, our high-frequency iNAV series is the first that are built specifically for commodity ETFs over an extended sample period. These new series allow us to measure intraday arbitrage relationships that are unobservable when one uses end-of-day NAV \citep{petajisto2017inefficiencies}. Moreover, our comparison of HAR and BVAR models, together with a thorough analysis by sampling frequency, shows that temporal aggregation strongly affects the reliability of the empirical conclusions that can be drawn \citep{corsi2009simple, koop2011forecasting, andersen2007roughing}.
10 +
11 +The findings in this paper carry practical implications for different market participants and extend the risk management discussion of \citet{madhavan2012exchange} and \citet{staer2017asset}. For investors and risk managers, we show that precious metals ETF volatility can be forecast from underlying volatility alone, while energy ETF models must account for bidirectional feedback. Both benefit from high-frequency information for short horizons. For market makers and authorized participants, arbitrage in precious metals flows mainly from underlying markets to ETFs, while energy markets contain more balanced bidirectional opportunities \citep{hendershott2013relationship}. For regulators, the asymmetries that we document suggest monitoring that is tailored to the commodity type: precious metals markets show little ETF feedback and thus have a lower destabilization risk, while energy markets exhibit stronger bidirectional links warranting closer monitoring during stress periods \citep{ohara2021etf, dannhauser2017effect}. The strength of the jump transmission further suggests that stress testing should account for discontinuous shock scenarios.
12 +
13 +
14 +Finally, for future work, we note that the high-frequency iNAV methodology shown in this paper could be applied to equity, international, and fixed-income ETFs where similar arbitrage mechanisms operate. Future research could also investigate the relationship between jumps, which we find drive volatility transmission, to specific news events, order-flow imbalances, and market-maker inventory constraints.
added phd_chap2_20260731/tables/tables.tex +678 −0
@@ -0,0 +1,678 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Tables}
4 +
5 +\begin{table}[htbp]
6 +\centering
7 +\caption{Descriptive Statistics for Realized Volatility, Quadratic Power Variation, and Jump Variables}
8 +\label{tab:desc_stats_5min}
9 +\begin{threeparttable}
10 +\begin{tabular}{@{}lcccccc@{}}
11 +\toprule
12 +\multicolumn{1}{l}{Commodity} & Variable & Obs & Mean & Std. Dev. & Min & Max \\
13 +\midrule
14 +\multicolumn{7}{l}{\textit{Panel A: Realized Volatility}} \\
15 +\addlinespace[0.1cm]
16 +Crude Oil & $RV_{t,NAV}$ & 3,935 & 0.081 & 0.717 & 0.002 & 42.082 \\
17 + & $RV_{t,ETF}$ & 3,935 & 0.066 & 0.204 & 0.001 & 10.335 \\
18 +\addlinespace[0.05cm]
19 +Gold & $RV_{t,NAV}$ & 3,935 & 0.012 & 0.017 & 0.001 & 0.327 \\
20 + & $RV_{t,ETF}$ & 3,935 & 0.034 & 1.259 & 0.001 & 78.988 \\
21 +\addlinespace[0.05cm]
22 +Silver & $RV_{t,NAV}$ & 3,935 & 0.043 & 0.063 & 0.002 & 1.120 \\
23 + & $RV_{t,ETF}$ & 3,935 & 0.042 & 0.067 & 0.004 & 1.286 \\
24 +\addlinespace[0.05cm]
25 +Natural Gas & $RV_{t,NAV}$ & 3,935 & 0.104 & 0.173 & 0.010 & 6.635 \\
26 + & $RV_{t,ETF}$ & 3,935 & 0.103 & 0.116 & 0.004 & 1.651 \\
27 +\addlinespace[0.2cm]
28 +
29 +\multicolumn{7}{l}{\textit{Panel B: Quadratic Power Variation}} \\
30 +\addlinespace[0.1cm]
31 +Crude Oil & $QPV_{t,NAV}$ & 3,935 & 0.110 & 6.797 & 0.000 & 426.363 \\
32 + & $QPV_{t,ETF}$ & 3,935 & 0.003 & 0.121 & 0.000 & 7.295 \\
33 +\addlinespace[0.05cm]
34 +Gold & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.007 \\
35 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.027 \\
36 +\addlinespace[0.05cm]
37 +Silver & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.003 & 0.000 & 0.139 \\
38 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.043 \\
39 +\addlinespace[0.05cm]
40 +Natural Gas & $QPV_{t,NAV}$ & 3,935 & 0.003 & 0.152 & 0.000 & 9.555 \\
41 + & $QPV_{t,ETF}$ & 3,935 & 0.001 & 0.030 & 0.000 & 1.881 \\
42 +\addlinespace[0.2cm]
43 +
44 +\multicolumn{7}{l}{\textit{Panel C: Jump Component}} \\
45 +\addlinespace[0.1cm]
46 +Crude Oil & $J_{t,NAV}$ & 3,935 & $-$0.028 & 6.132 & 0.000 & 8.941 \\
47 + & $J_{t,ETF}$ & 3,935 & 0.063 & 0.124 & 0.000 & 3.040 \\
48 +\addlinespace[0.05cm]
49 +Gold & $J_{t,NAV}$ & 3,935 & 0.012 & 0.017 & 0.001 & 0.324 \\
50 + & $J_{t,ETF}$ & 3,935 & 0.034 & 1.259 & 0.001 & 78.981 \\
51 +\addlinespace[0.05cm]
52 +Silver & $J_{t,NAV}$ & 3,935 & 0.042 & 0.061 & 0.002 & 1.094 \\
53 + & $J_{t,ETF}$ & 3,935 & 0.042 & 0.066 & 0.004 & 1.243 \\
54 +\addlinespace[0.05cm]
55 +Natural Gas & $J_{t,NAV}$ & 3,935 & 0.101 & 0.209 & 0.000 & 6.634 \\
56 + & $J_{t,ETF}$ & 3,935 & 0.102 & 0.113 & 0.000 & 1.651 \\
57 +\bottomrule
58 +\end{tabular}
59 +\begin{tablenotes}
60 +\small
61 +\item This table presents descriptive statistics for realized volatility (RV), quadratic power variation (QPV), and jump component (J) variables constructed using 5-minute price data. All values are expressed in percentages. NAV refers to net asset value prices, and ETF refers to exchange-traded fund prices. The sample period includes 3,935 daily observations for each commodity.
62 +\end{tablenotes}
63 +\end{threeparttable}
64 +\end{table}
65 +
66 +
67 +\begin{table}[htbp]
68 +\centering
69 +\caption{Descriptive Statistics for Realized Volatility, Quadratic Power Variation, and Jump Variables (1-minute data)}
70 +\label{tab:desc_stats_1min}
71 +\begin{threeparttable}
72 +\begin{tabular}{@{}lcccccc@{}}
73 +\toprule
74 +\multicolumn{1}{l}{Commodity} & Variable & Obs & Mean & Std. Dev. & Min & Max \\
75 +\midrule
76 +\multicolumn{7}{l}{\textit{Panel A: Realized Volatility}} \\
77 +\addlinespace[0.1cm]
78 +Crude Oil & $RV_{t,NAV}$ & 3,935 & 0.083 & 0.959 & 0.001 & 58.167 \\
79 + & $RV_{t,ETF}$ & 3,935 & 0.058 & 0.250 & 0.001 & 14.051 \\
80 +\addlinespace[0.05cm]
81 +Gold & $RV_{t,NAV}$ & 3,935 & 0.012 & 0.018 & 0.001 & 0.302 \\
82 + & $RV_{t,ETF}$ & 3,935 & 0.031 & 1.259 & 0.001 & 78.986 \\
83 +\addlinespace[0.05cm]
84 +Silver & $RV_{t,NAV}$ & 3,935 & 0.040 & 0.064 & 0.001 & 1.347 \\
85 + & $RV_{t,ETF}$ & 3,935 & 0.036 & 0.061 & 0.002 & 1.367 \\
86 +\addlinespace[0.05cm]
87 +Natural Gas & $RV_{t,NAV}$ & 3,935 & 0.099 & 0.174 & 0.007 & 6.699 \\
88 + & $RV_{t,ETF}$ & 3,935 & 0.086 & 0.096 & 0.004 & 1.634 \\
89 +\addlinespace[0.2cm]
90 +
91 +\multicolumn{7}{l}{\textit{Panel B: Quadratic Power Variation}} \\
92 +\addlinespace[0.1cm]
93 +Crude Oil & $QPV_{t,NAV}$ & 3,935 & 0.019 & 1.152 & 0.000 & 72.288 \\
94 + & $QPV_{t,ETF}$ & 3,935 & 0.004 & 0.191 & 0.000 & 11.936 \\
95 +\addlinespace[0.05cm]
96 +Gold & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.004 \\
97 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.016 \\
98 +\addlinespace[0.05cm]
99 +Silver & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.060 \\
100 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.025 \\
101 +\addlinespace[0.05cm]
102 +Natural Gas & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.004 & 0.000 & 0.232 \\
103 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.039 \\
104 +\addlinespace[0.2cm]
105 +
106 +\multicolumn{7}{l}{\textit{Panel C: Jump Component}} \\
107 +\addlinespace[0.1cm]
108 +Crude Oil & $J_{t,NAV}$ & 3,935 & 0.064 & 0.330 & 0.000 & 8.933 \\
109 + & $J_{t,ETF}$ & 3,935 & 0.055 & 0.112 & 0.001 & 3.172 \\
110 +\addlinespace[0.05cm]
111 +Gold & $J_{t,NAV}$ & 3,935 & 0.012 & 0.018 & 0.001 & 0.300 \\
112 + & $J_{t,ETF}$ & 3,935 & 0.031 & 1.259 & 0.001 & 78.969 \\
113 +\addlinespace[0.05cm]
114 +Silver & $J_{t,NAV}$ & 3,935 & 0.040 & 0.063 & 0.001 & 1.328 \\
115 + & $J_{t,ETF}$ & 3,935 & 0.036 & 0.060 & 0.002 & 1.344 \\
116 +\addlinespace[0.05cm]
117 +Natural Gas & $J_{t,NAV}$ & 3,935 & 0.098 & 0.173 & 0.007 & 6.697 \\
118 + & $J_{t,ETF}$ & 3,935 & 0.086 & 0.095 & 0.004 & 1.634 \\
119 +\bottomrule
120 +\end{tabular}
121 +\begin{tablenotes}
122 +\small
123 +\item This table presents descriptive statistics for realized volatility (RV), quadratic power variation (QPV), and jump component (J) variables constructed using 1-minute price data. All values are expressed in percentages. NAV refers to net asset value prices, and ETF refers to exchange-traded fund prices. The sample period includes 3,935 daily observations for each commodity.
124 +\end{tablenotes}
125 +\end{threeparttable}
126 +\end{table}
127 +
128 +\begin{table}[htbp]
129 +\centering
130 +\caption{Descriptive Statistics for Realized Volatility, Quadratic Power Variation, and Jump Variables (30-minute data)}
131 +\label{tab:desc_stats_30min}
132 +\begin{threeparttable}
133 +\begin{tabular}{@{}lcccccc@{}}
134 +\toprule
135 +\multicolumn{1}{l}{Commodity} & Variable & Obs & Mean & Std. Dev. & Min & Max \\
136 +\midrule
137 +\multicolumn{7}{l}{\textit{Panel A: Realized Volatility}} \\
138 +\addlinespace[0.1cm]
139 +Crude Oil & $RV_{t,NAV}$ & 3,935 & 0.073 & 0.459 & 0.001 & 23.817 \\
140 + & $RV_{t,ETF}$ & 3,935 & 0.052 & 0.176 & 0.000 & 8.995 \\
141 +\addlinespace[0.05cm]
142 +Gold & $RV_{t,NAV}$ & 3,935 & 0.011 & 0.019 & 0.001 & 0.407 \\
143 + & $RV_{t,ETF}$ & 3,935 & 0.030 & 1.257 & 0.000 & 78.854 \\
144 +\addlinespace[0.05cm]
145 +Silver & $RV_{t,NAV}$ & 3,935 & 0.039 & 0.069 & 0.001 & 1.701 \\
146 + & $RV_{t,ETF}$ & 3,935 & 0.033 & 0.060 & 0.001 & 1.618 \\
147 +\addlinespace[0.05cm]
148 +Natural Gas & $RV_{t,NAV}$ & 3,935 & 0.093 & 0.176 & 0.004 & 6.829 \\
149 + & $RV_{t,ETF}$ & 3,935 & 0.077 & 0.093 & 0.002 & 1.668 \\
150 +\addlinespace[0.2cm]
151 +
152 +\multicolumn{7}{l}{\textit{Panel B: Quadratic Power Variation}} \\
153 +\addlinespace[0.1cm]
154 +Crude Oil & $QPV_{t,NAV}$ & 3,935 & 0.002 & 0.067 & 0.000 & 4.036 \\
155 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.013 & 0.000 & 0.831 \\
156 +\addlinespace[0.05cm]
157 +Gold & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.002 \\
158 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.002 \\
159 +\addlinespace[0.05cm]
160 +Silver & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.050 \\
161 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.037 \\
162 +\addlinespace[0.05cm]
163 +Natural Gas & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.044 \\
164 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.035 \\
165 +\addlinespace[0.2cm]
166 +
167 +\multicolumn{7}{l}{\textit{Panel C: Jump Component}} \\
168 +\addlinespace[0.1cm]
169 +Crude Oil & $J_{t,NAV}$ & 3,935 & 0.072 & 0.401 & 0.001 & 19.781 \\
170 + & $J_{t,ETF}$ & 3,935 & 0.051 & 0.165 & 0.000 & 8.164 \\
171 +\addlinespace[0.05cm]
172 +Gold & $J_{t,NAV}$ & 3,935 & 0.011 & 0.018 & 0.001 & 0.404 \\
173 + & $J_{t,ETF}$ & 3,935 & 0.030 & 1.257 & 0.000 & 78.852 \\
174 +\addlinespace[0.05cm]
175 +Silver & $J_{t,NAV}$ & 3,935 & 0.038 & 0.068 & 0.001 & 1.665 \\
176 + & $J_{t,ETF}$ & 3,935 & 0.033 & 0.059 & 0.001 & 1.583 \\
177 +\addlinespace[0.05cm]
178 +Natural Gas & $J_{t,NAV}$ & 3,935 & 0.093 & 0.175 & 0.004 & 6.823 \\
179 + & $J_{t,ETF}$ & 3,935 & 0.077 & 0.093 & 0.002 & 1.633 \\
180 +\bottomrule
181 +\end{tabular}
182 +\begin{tablenotes}
183 +\small
184 +\item This table presents descriptive statistics for realized volatility (RV), quadratic power variation (QPV), and jump component (J) variables constructed using 30-minute price data. All values are expressed in percentages. NAV refers to net asset value prices, and ETF refers to exchange-traded fund prices. The sample period includes 3,935 daily observations for each commodity.
185 +\end{tablenotes}
186 +\end{threeparttable}
187 +\end{table}
188 +
189 +
190 +\begin{landscape}
191 +\begin{table}[htbp]
192 +\centering
193 +\caption{HAR-X Model Estimates with 5-minute Realized Variance}
194 +\label{tab:HAR_5min}
195 +\begin{threeparttable}
196 +\footnotesize
197 +\begin{tabular}{@{}lcccccccc@{}}
198 +\toprule
199 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
200 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
201 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
202 +\midrule
203 +$RV_{t-1,NAV}$ & 0.288*** & 0.311*** & 0.253*** & 0.515*** & 0.345*** & 0.383*** & 0.092** & 0.105*** \\
204 + & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) \\
205 +\addlinespace[0.1cm]
206 +$\overline{RV}_{t-5,NAV}$ & 0.408*** & & 0.338*** & & 0.318*** & & 0.393*** & \\
207 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
208 +\addlinespace[0.1cm]
209 +$\overline{RV}_{t-22,NAV}$ & 0.144*** & & 0.321*** & & 0.271*** & & 0.313*** & \\
210 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
211 +\addlinespace[0.1cm]
212 +$RV_{t-1,ETF}$ & 0.111*** & 0.098*** & 0.021 & $-$0.084* & $-$0.009 & $-$0.015 & 0.126*** & 0.084*** \\
213 + & (0.000) & (0.000) & (0.158) & (0.062) & (0.612) & (0.464) & (0.000) & (0.000) \\
214 +\addlinespace[0.1cm]
215 +$\overline{RV}_{t-5,ETF}$ & & 0.297*** & & 0.270*** & & 0.209*** & & 0.431*** \\
216 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
217 +\addlinespace[0.1cm]
218 +$\overline{RV}_{t-22,ETF}$ & & 0.227*** & & 0.164*** & & 0.342*** & & 0.332*** \\
219 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
220 +\midrule
221 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
222 +\bottomrule
223 +\end{tabular}
224 +\begin{tablenotes}
225 +\small
226 +\item This table presents estimation results for the HAR-X model using 5-minute realized variance data. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $RV_{t-1}$ represents the lagged daily realized variance, $\overline{RV}_{t-5}$ is the average of the past 5 days' realized variances, and $\overline{RV}_{t-22}$ is the average of the past 22 days' realized variances.
227 +\end{tablenotes}
228 +\end{threeparttable}
229 +\end{table}
230 +\end{landscape}
231 +
232 +
233 +\begin{landscape}
234 +\begin{table}[htbp]
235 +\centering
236 +\caption{HAR-X Model Estimates with 1-minute Realized Variance}
237 +\label{tab:HAR_1min}
238 +\begin{threeparttable}
239 +\footnotesize
240 +\begin{tabular}{@{}lcccccccc@{}}
241 +\toprule
242 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
243 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
244 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
245 +\midrule
246 +$RV_{t-1,NAV}$ & 0.371*** & 0.378*** & 0.331*** & 0.632*** & 0.404*** & 0.418*** & 0.121*** & 0.079*** \\
247 + & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) \\
248 +\addlinespace[0.1cm]
249 +$\overline{RV}_{t-5,NAV}$ & 0.387*** & & 0.329*** & & 0.297*** & & 0.386*** & \\
250 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
251 +\addlinespace[0.1cm]
252 +$\overline{RV}_{t-22,NAV}$ & 0.109*** & & 0.280*** & & 0.226*** & & 0.285*** & \\
253 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
254 +\addlinespace[0.1cm]
255 +$RV_{t-1,ETF}$ & 0.089*** & 0.097*** & $-$0.001 & $-$0.059* & 0.004 & 0.015 & 0.127*** & 0.124*** \\
256 + & (0.000) & (0.000) & (0.961) & (0.065) & (0.804) & (0.410) & (0.000) & (0.000) \\
257 +\addlinespace[0.1cm]
258 +$\overline{RV}_{t-5,ETF}$ & & 0.272*** & & 0.196*** & & 0.206*** & & 0.438*** \\
259 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
260 +\addlinespace[0.1cm]
261 +$\overline{RV}_{t-22,ETF}$ & & 0.200*** & & 0.113*** & & 0.276*** & & 0.318*** \\
262 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
263 +\midrule
264 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
265 +\bottomrule
266 +\end{tabular}
267 +\begin{tablenotes}
268 +\small
269 +\item This table presents estimation results for the HAR-X model using 1-minute realized variance data. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $RV_{t-1}$ represents the lagged daily realized variance, $\overline{RV}_{t-5}$ is the average of the past 5 days' realized variances, and $\overline{RV}_{t-22}$ is the average of the past 22 days' realized variances.
270 +\end{tablenotes}
271 +\end{threeparttable}
272 +\end{table}
273 +\end{landscape}
274 +
275 +
276 +\begin{landscape}
277 +\begin{table}[htbp]
278 +\centering
279 +\caption{HAR-X Model Estimates with 30-minute Realized Variance}
280 +\label{tab:HAR_30min}
281 +\begin{threeparttable}
282 +\footnotesize
283 +\begin{tabular}{@{}lcccccccc@{}}
284 +\toprule
285 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
286 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
287 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
288 +\midrule
289 +$RV_{t-1,NAV}$ & 0.133*** & 0.200*** & 0.150*** & 0.387*** & 0.226*** & 0.277*** & 0.043 & 0.081*** \\
290 + & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.121) & (0.000) \\
291 +\addlinespace[0.1cm]
292 +$\overline{RV}_{t-5,NAV}$ & 0.461*** & & 0.335*** & & 0.319*** & & 0.399*** & \\
293 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
294 +\addlinespace[0.1cm]
295 +$\overline{RV}_{t-22,NAV}$ & 0.243*** & & 0.445*** & & 0.394*** & & 0.393*** & \\
296 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
297 +\addlinespace[0.1cm]
298 +$RV_{t-1,ETF}$ & 0.103*** & 0.049 & $-$0.015 & $-$0.128** & $-$0.034 & $-$0.075** & 0.074** & 0.012 \\
299 + & (0.000) & (0.149) & (0.476) & (0.019) & (0.140) & (0.013) & (0.030) & (0.662) \\
300 +\addlinespace[0.1cm]
301 +$\overline{RV}_{t-5,ETF}$ & & 0.363*** & & 0.332*** & & 0.238*** & & 0.405*** \\
302 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
303 +\addlinespace[0.1cm]
304 +$\overline{RV}_{t-22,ETF}$ & & 0.319*** & & 0.212*** & & 0.460*** & & 0.447*** \\
305 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
306 +\midrule
307 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
308 +\bottomrule
309 +\end{tabular}
310 +\begin{tablenotes}
311 +\small
312 +\item This table presents estimation results for the HAR-X model using 30-minute realized variance data. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $RV_{t-1}$ represents the lagged daily realized variance, $\overline{RV}_{t-5}$ is the average of the past 5 days' realized variances, and $\overline{RV}_{t-22}$ is the average of the past 22 days' realized variances.
313 +\end{tablenotes}
314 +\end{threeparttable}
315 +\end{table}
316 +\end{landscape}
317 +
318 +
319 +\begin{landscape}
320 +\begin{table}[htbp]
321 +\centering
322 +\caption{HAR-CJ-X Model Estimates with 5-minute Realized Variance}
323 +\label{tab:HAR_CJ_5min}
324 +\begin{threeparttable}
325 +\footnotesize
326 +\begin{tabular}{@{}lcccccccc@{}}
327 +\toprule
328 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
329 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
330 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
331 +\midrule
332 +\multicolumn{9}{l}{\textit{Panel A: Quadratic Power Variation}} \\
333 +\addlinespace[0.1cm]
334 +$QPV_{t-1,NAV}$ & 0.049** & 0.008 & 0.026 & $-$0.001 & $-$0.015 & $-$0.026 & 0.065*** & 0.016 \\
335 + & (0.025) & (0.661) & (0.239) & (0.957) & (0.467) & (0.273) & (0.001) & (0.400) \\
336 +\addlinespace[0.05cm]
337 +$\overline{QPV}_{t-5,NAV}$ & 0.032 & & $-$0.027 & & $-$0.030 & & 0.110*** & \\
338 + & (0.167) & & (0.317) & & (0.236) & & (0.000) & \\
339 +\addlinespace[0.05cm]
340 +$\overline{QPV}_{t-22,NAV}$ & $-$0.046*** & & $-$0.021 & & $-$0.005 & & 0.044** & \\
341 + & (0.007) & & (0.341) & & (0.793) & & (0.018) & \\
342 +\addlinespace[0.05cm]
343 +$QPV_{t-1,ETF}$ & 0.001 & 0.043** & $-$0.015 & $-$0.030 & 0.006 & 0.010 & 0.011 & 0.028* \\
344 + & (0.969) & (0.038) & (0.545) & (0.248) & (0.779) & (0.686) & (0.620) & (0.088) \\
345 +\addlinespace[0.05cm]
346 +$\overline{QPV}_{t-5,ETF}$ & & 0.017 & & $-$0.008 & & $-$0.003 & & 0.062*** \\
347 + & & (0.454) & & (0.737) & & (0.876) & & (0.001) \\
348 +\addlinespace[0.05cm]
349 +$\overline{QPV}_{t-22,ETF}$ & & $-$0.040** & & 0.133*** & & 0.006 & & 0.038 \\
350 + & & (0.043) & & (0.000) & & (0.763) & & (0.137) \\
351 +\addlinespace[0.2cm]
352 +
353 +\multicolumn{9}{l}{\textit{Panel B: Jump Component}} \\
354 +\addlinespace[0.1cm]
355 +$J_{t-1,NAV}$ & 0.207*** & 0.280*** & 0.212*** & 0.491*** & 0.375*** & 0.431*** & $-$0.003 & 0.057* \\
356 + & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.923) & (0.082) \\
357 +\addlinespace[0.05cm]
358 +$\overline{J}_{t-5,NAV}$ & 0.338*** & & 0.407*** & & 0.389*** & & 0.220*** & \\
359 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
360 +\addlinespace[0.05cm]
361 +$\overline{J}_{t-22,NAV}$ & 0.267*** & & 0.354*** & & 0.272*** & & 0.213*** & \\
362 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
363 +\addlinespace[0.05cm]
364 +$J_{t-1,ETF}$ & 0.094*** & 0.019 & 0.041 & $-$0.023 & $-$0.016 & $-$0.030 & 0.044 & 0.039 \\
365 + & (0.000) & (0.473) & (0.156) & (0.497) & (0.527) & (0.291) & (0.190) & (0.195) \\
366 +\addlinespace[0.05cm]
367 +$\overline{J}_{t-5,ETF}$ & & 0.264*** & & 0.246*** & & 0.214*** & & 0.307*** \\
368 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
369 +\addlinespace[0.05cm]
370 +$\overline{J}_{t-22,ETF}$ & & 0.329*** & & $-$0.014 & & 0.327*** & & 0.250*** \\
371 + & & (0.000) & & (0.631) & & (0.000) & & (0.000) \\
372 +\midrule
373 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
374 +\bottomrule
375 +\end{tabular}
376 +\begin{tablenotes}
377 +\small
378 +\item This table presents estimation results for the HAR-CJ-X model using 5-minute realized variance data. The model incorporates both continuous (quadratic power variation, QPV) and jump (J) components. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $QPV_{t-1}$ and $J_{t-1}$ represent the lagged daily components, while $\overline{QPV}_{t-5}$, $\overline{J}_{t-5}$, $\overline{QPV}_{t-22}$, and $\overline{J}_{t-22}$ are the corresponding weekly and monthly averages.
379 +\end{tablenotes}
380 +\end{threeparttable}
381 +\end{table}
382 +\end{landscape}
383 +
384 +\begin{landscape}
385 +\begin{table}[htbp]
386 +\centering
387 +\caption{HAR-CJ-X Model Estimates with 1-minute Realized Variance}
388 +\label{tab:HAR_CJ_1min}
389 +\begin{threeparttable}
390 +\footnotesize
391 +\begin{tabular}{@{}lcccccccc@{}}
392 +\toprule
393 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
394 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
395 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
396 +\midrule
397 +\multicolumn{9}{l}{\textit{Panel A: Quadratic Power Variation}} \\
398 +\addlinespace[0.1cm]
399 +$QPV_{t-1,NAV}$ & 0.036 & $-$0.019 & $-$0.033* & $-$0.123*** & $-$0.041** & $-$0.065*** & 0.008 & $-$0.002 \\
400 + & (0.246) & (0.525) & (0.096) & (0.000) & (0.039) & (0.002) & (0.773) & (0.920) \\
401 +\addlinespace[0.05cm]
402 +$\overline{QPV}_{t-5,NAV}$ & $-$0.021 & & $-$0.076*** & & $-$0.068*** & & 0.116*** & \\
403 + & (0.444) & & (0.004) & & (0.009) & & (0.000) & \\
404 +\addlinespace[0.05cm]
405 +$\overline{QPV}_{t-22,NAV}$ & $-$0.031 & & $-$0.001 & & 0.009 & & $-$0.040** & \\
406 + & (0.309) & & (0.976) & & (0.716) & & (0.046) & \\
407 +\addlinespace[0.05cm]
408 +$QPV_{t-1,ETF}$ & $-$0.011 & 0.009 & $-$0.020* & $-$0.032* & $-$0.011 & $-$0.014 & $-$0.011 & $-$0.021* \\
409 + & (0.665) & (0.698) & (0.062) & (0.089) & (0.617) & (0.514) & (0.671) & (0.089) \\
410 +\addlinespace[0.05cm]
411 +$\overline{QPV}_{t-5,ETF}$ & & $-$0.014 & & 0.016 & & 0.003 & & 0.016 \\
412 + & & (0.591) & & (0.566) & & (0.895) & & (0.554) \\
413 +\addlinespace[0.05cm]
414 +$\overline{QPV}_{t-22,ETF}$ & & $-$0.024** & & 0.060*** & & $-$0.017 & & $-$0.014 \\
415 + & & (0.036) & & (0.003) & & (0.463) & & (0.617) \\
416 +\addlinespace[0.2cm]
417 +
418 +\multicolumn{9}{l}{\textit{Panel B: Jump Component}} \\
419 +\addlinespace[0.1cm]
420 +$J_{t-1,NAV}$ & 0.306*** & 0.432*** & 0.426*** & 0.892*** & 0.510*** & 0.578*** & 0.092* & 0.090*** \\
421 + & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.075) & (0.000) \\
422 +\addlinespace[0.05cm]
423 +$\overline{J}_{t-5,NAV}$ & 0.435*** & & 0.509*** & & 0.468*** & & 0.219*** & \\
424 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
425 +\addlinespace[0.05cm]
426 +$\overline{J}_{t-22,NAV}$ & 0.190*** & & 0.234*** & & 0.161*** & & 0.346*** & \\
427 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
428 +\addlinespace[0.05cm]
429 +$J_{t-1,ETF}$ & 0.096*** & 0.077** & 0.031 & 0.011 & 0.028 & 0.051 & 0.128*** & 0.166*** \\
430 + & (0.000) & (0.019) & (0.290) & (0.689) & (0.313) & (0.106) & (0.000) & (0.000) \\
431 +\addlinespace[0.05cm]
432 +$\overline{J}_{t-5,ETF}$ & & 0.300*** & & 0.132** & & 0.184*** & & 0.401*** \\
433 + & & (0.000) & & (0.023) & & (0.000) & & (0.000) \\
434 +\addlinespace[0.05cm]
435 +$\overline{J}_{t-22,ETF}$ & & 0.250*** & & 0.028 & & 0.295*** & & 0.350*** \\
436 + & & (0.000) & & (0.455) & & (0.000) & & (0.000) \\
437 +\midrule
438 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
439 +\bottomrule
440 +\end{tabular}
441 +\begin{tablenotes}
442 +\small
443 +\item This table presents estimation results for the HAR-CJ-X model using 1-minute realized variance data. The model incorporates both continuous (quadratic power variation, QPV) and jump (J) components. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $QPV_{t-1}$ and $J_{t-1}$ represent the lagged daily components, while $\overline{QPV}_{t-5}$, $\overline{J}_{t-5}$, $\overline{QPV}_{t-22}$, and $\overline{J}_{t-22}$ are the corresponding weekly and monthly averages.
444 +\end{tablenotes}
445 +\end{threeparttable}
446 +\end{table}
447 +\end{landscape}
448 +
449 +
450 +\begin{landscape}
451 +\begin{table}[htbp]
452 +\centering
453 +\caption{HAR-CJ-X Model Estimates with 30-minute Realized Variance}
454 +\label{tab:HAR_CJ_30min}
455 +\begin{threeparttable}
456 +\footnotesize
457 +\begin{tabular}{@{}lcccccccc@{}}
458 +\toprule
459 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
460 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
461 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
462 +\midrule
463 +\multicolumn{9}{l}{\textit{Panel A: Quadratic Power Variation}} \\
464 +\addlinespace[0.1cm]
465 +$QPV_{t-1,NAV}$ & 0.019 & $-$0.026 & 0.058*** & 0.067*** & 0.013 & 0.033 & 0.057*** & 0.027 \\
466 + & (0.535) & (0.421) & (0.005) & (0.003) & (0.596) & (0.235) & (0.008) & (0.278) \\
467 +\addlinespace[0.05cm]
468 +$\overline{QPV}_{t-5,NAV}$ & 0.029 & & $-$0.051** & & $-$0.008 & & 0.114*** & \\
469 + & (0.375) & & (0.044) & & (0.779) & & (0.000) & \\
470 +\addlinespace[0.05cm]
471 +$\overline{QPV}_{t-22,NAV}$ & $-$0.001 & & 0.001 & & $-$0.045* & & 0.156*** & \\
472 + & (0.975) & & (0.977) & & (0.098) & & (0.000) & \\
473 +\addlinespace[0.05cm]
474 +$QPV_{t-1,ETF}$ & $-$0.010 & 0.015 & $-$0.018 & $-$0.027 & $-$0.006 & 0.001 & $-$0.014 & 0.001 \\
475 + & (0.683) & (0.562) & (0.489) & (0.313) & (0.799) & (0.984) & (0.575) & (0.972) \\
476 +\addlinespace[0.05cm]
477 +$\overline{QPV}_{t-5,ETF}$ & & 0.055** & & $-$0.017 & & $-$0.001 & & 0.079*** \\
478 + & & (0.033) & & (0.517) & & (0.966) & & (0.001) \\
479 +\addlinespace[0.05cm]
480 +$\overline{QPV}_{t-22,ETF}$ & & 0.016 & & 0.234*** & & $-$0.035 & & 0.168*** \\
481 + & & (0.517) & & (0.000) & & (0.191) & & (0.000) \\
482 +\addlinespace[0.2cm]
483 +
484 +\multicolumn{9}{l}{\textit{Panel B: Jump Component}} \\
485 +\addlinespace[0.1cm]
486 +$J_{t-1,NAV}$ & 0.110** & 0.221*** & 0.054 & 0.206*** & 0.201*** & 0.216*** & $-$0.006 & 0.027 \\
487 + & (0.019) & (0.000) & (0.190) & (0.000) & (0.000) & (0.000) & (0.870) & (0.381) \\
488 +\addlinespace[0.05cm]
489 +$\overline{J}_{t-5,NAV}$ & 0.403*** & & 0.436*** & & 0.339*** & & 0.204*** & \\
490 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
491 +\addlinespace[0.05cm]
492 +$\overline{J}_{t-22,NAV}$ & 0.248*** & & 0.440*** & & 0.489*** & & 0.089 & \\
493 + & (0.000) & & (0.000) & & (0.000) & & (0.123) & \\
494 +\addlinespace[0.05cm]
495 +$J_{t-1,ETF}$ & 0.110*** & 0.043 & 0.009 & $-$0.059 & $-$0.023 & $-$0.080* & 0.016 & 0.007 \\
496 + & (0.000) & (0.181) & (0.772) & (0.183) & (0.428) & (0.077) & (0.608) & (0.823) \\
497 +\addlinespace[0.05cm]
498 +$\overline{J}_{t-5,ETF}$ & & 0.259*** & & 0.274*** & & 0.241*** & & 0.243*** \\
499 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
500 +\addlinespace[0.05cm]
501 +$\overline{J}_{t-22,ETF}$ & & 0.283*** & & $-$0.082* & & 0.532*** & & 0.088 \\
502 + & & (0.000) & & (0.058) & & (0.000) & & (0.224) \\
503 +\midrule
504 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
505 +\bottomrule
506 +\end{tabular}
507 +\begin{tablenotes}
508 +\small
509 +\item This table presents estimation results for the HAR-CJ-X model using 30-minute realized variance data. The model incorporates both continuous (quadratic power variation, QPV) and jump (J) components. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $QPV_{t-1}$ and $J_{t-1}$ represent the lagged daily components, while $\overline{QPV}_{t-5}$, $\overline{J}_{t-5}$, $\overline{QPV}_{t-22}$, and $\overline{J}_{t-22}$ are the corresponding weekly and monthly averages.
510 +\end{tablenotes}
511 +\end{threeparttable}
512 +\end{table}
513 +\end{landscape}
514 +
515 +
516 +\begin{landscape}
517 +\begin{table}[htbp]
518 +\centering
519 +\caption{Bayesian Vector Autoregression Results: USO ETF and Net Asset Value}
520 +\label{tab:VAR_USO}
521 +\begin{threeparttable}
522 +\footnotesize
523 +\begin{tabular}{@{}lccccccc@{}}
524 +\toprule
525 + & \multicolumn{6}{c}{\textbf{Panel A: $\log(RV_{t,NAV})$}} \\
526 +\cmidrule(lr){2-7}
527 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
528 +\cmidrule(lr){6-7}
529 + & & & & & Lower & Upper \\
530 +\midrule
531 +$\log(RV_{t-1,NAV})$ & 0.570 & 0.024 & 0.000 & 0.570 & 0.524 & 0.617 \\
532 +$\log(RV_{t-2,NAV})$ & 0.261 & 0.021 & 0.000 & 0.261 & 0.221 & 0.300 \\
533 +$\log(RV_{t-1,ETF})$ & 0.074 & 0.020 & 0.000 & 0.074 & 0.034 & 0.113 \\
534 +$\log(RV_{t-2,ETF})$ & 0.016 & 0.018 & 0.000 & 0.016 & $-$0.019 & 0.051 \\
535 +\addlinespace[0.3cm]
536 +
537 + & \multicolumn{6}{c}{\textbf{Panel B: $\log(RV_{t,ETF})$}} \\
538 +\cmidrule(lr){2-7}
539 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
540 +\cmidrule(lr){6-7}
541 + & & & & & Lower & Upper \\
542 +\midrule
543 +$\log(RV_{t-1,NAV})$ & 0.286 & 0.028 & 0.000 & 0.286 & 0.231 & 0.343 \\
544 +$\log(RV_{t-2,NAV})$ & 0.193 & 0.025 & 0.000 & 0.193 & 0.144 & 0.241 \\
545 +$\log(RV_{t-1,ETF})$ & 0.294 & 0.024 & 0.000 & 0.294 & 0.246 & 0.342 \\
546 +$\log(RV_{t-2,ETF})$ & 0.106 & 0.022 & 0.000 & 0.106 & 0.064 & 0.148 \\
547 +\bottomrule
548 +\end{tabular}
549 +\begin{tablenotes}
550 +\small
551 +\item This table presents posterior statistics from a Bayesian Vector Autoregression (VAR) model analyzing the relationship between the USO ETF and its Net Asset Value (NAV). Panel A shows results for the NAV equation, while Panel B shows results for the ETF equation. The model includes two lags of both log realized variances. Mean represents the posterior mean, Std. Dev. is the posterior standard deviation, MCSE is the Monte Carlo standard error, Median is the posterior median, and the 95\% credible interval provides the range of plausible parameter values. All realized variance measures are constructed using 5-minute price data.
552 +\end{tablenotes}
553 +\end{threeparttable}
554 +\end{table}
555 +\end{landscape}
556 +
557 +\begin{landscape}
558 +\begin{table}[htbp]
559 +\centering
560 +\caption{Bayesian Vector Autoregression Results: GLD ETF and Net Asset Value}
561 +\label{tab:VAR_GLD}
562 +\begin{threeparttable}
563 +\footnotesize
564 +\begin{tabular}{@{}lccccccc@{}}
565 +\toprule
566 + & \multicolumn{6}{c}{\textbf{Panel A: $\log(RV_{t,NAV})$}} \\
567 +\cmidrule(lr){2-7}
568 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
569 +\cmidrule(lr){6-7}
570 + & & & & & Lower & Upper \\
571 +\midrule
572 +$\log(RV_{t-1,NAV})$ & 0.551 & 0.027 & 0.000 & 0.551 & 0.498 & 0.603 \\
573 +$\log(RV_{t-2,NAV})$ & 0.248 & 0.023 & 0.000 & 0.248 & 0.203 & 0.293 \\
574 +$\log(RV_{t-1,ETF})$ & $-$0.010 & 0.023 & 0.000 & $-$0.010 & $-$0.055 & 0.034 \\
575 +$\log(RV_{t-2,ETF})$ & 0.038 & 0.019 & 0.000 & 0.038 & 0.000 & 0.076 \\
576 +\addlinespace[0.3cm]
577 +
578 + & \multicolumn{6}{c}{\textbf{Panel B: $\log(RV_{t,ETF})$}} \\
579 +\cmidrule(lr){2-7}
580 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
581 +\cmidrule(lr){6-7}
582 + & & & & & Lower & Upper \\
583 +\midrule
584 +$\log(RV_{t-1,NAV})$ & 0.349 & 0.033 & 0.000 & 0.348 & 0.285 & 0.413 \\
585 +$\log(RV_{t-2,NAV})$ & 0.276 & 0.028 & 0.000 & 0.276 & 0.222 & 0.331 \\
586 +$\log(RV_{t-1,ETF})$ & 0.155 & 0.028 & 0.000 & 0.155 & 0.100 & 0.208 \\
587 +$\log(RV_{t-2,ETF})$ & 0.048 & 0.023 & 0.000 & 0.048 & 0.003 & 0.095 \\
588 +\bottomrule
589 +\end{tabular}
590 +\begin{tablenotes}
591 +\small
592 +\item This table presents posterior statistics from a Bayesian Vector Autoregression (VAR) model analyzing the relationship between the GLD ETF and its Net Asset Value (NAV). Panel A shows results for the NAV equation, while Panel B shows results for the ETF equation. The model includes two lags of both log realized variances. Mean represents the posterior mean, Std. Dev. is the posterior standard deviation, MCSE is the Monte Carlo standard error, Median is the posterior median, and the 95\% credible interval provides the range of plausible parameter values. All realized variance measures are constructed using 5-minute price data.
593 +\end{tablenotes}
594 +\end{threeparttable}
595 +\end{table}
596 +\end{landscape}
597 +
598 +\begin{landscape}
599 +\begin{table}[htbp]
600 +\centering
601 +\caption{Bayesian Vector Autoregression Results: SLV ETF and Net Asset Value}
602 +\label{tab:VAR_SLV}
603 +\begin{threeparttable}
604 +\footnotesize
605 +\begin{tabular}{@{}lccccccc@{}}
606 +\toprule
607 + & \multicolumn{6}{c}{\textbf{Panel A: $\log(RV_{t,NAV})$}} \\
608 +\cmidrule(lr){2-7}
609 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
610 +\cmidrule(lr){6-7}
611 + & & & & & Lower & Upper \\
612 +\midrule
613 +$\log(RV_{t-1,NAV})$ & 0.619 & 0.028 & 0.000 & 0.619 & 0.563 & 0.674 \\
614 +$\log(RV_{t-2,NAV})$ & 0.214 & 0.024 & 0.000 & 0.214 & 0.168 & 0.261 \\
615 +$\log(RV_{t-1,ETF})$ & $-$0.045 & 0.025 & 0.000 & $-$0.045 & $-$0.093 & 0.005 \\
616 +$\log(RV_{t-2,ETF})$ & 0.058 & 0.021 & 0.000 & 0.058 & 0.018 & 0.100 \\
617 +\addlinespace[0.3cm]
618 +
619 + & \multicolumn{6}{c}{\textbf{Panel B: $\log(RV_{t,ETF})$}} \\
620 +\cmidrule(lr){2-7}
621 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
622 +\cmidrule(lr){6-7}
623 + & & & & & Lower & Upper \\
624 +\midrule
625 +$\log(RV_{t-1,NAV})$ & 0.376 & 0.033 & 0.000 & 0.377 & 0.310 & 0.440 \\
626 +$\log(RV_{t-2,NAV})$ & 0.202 & 0.028 & 0.000 & 0.202 & 0.147 & 0.256 \\
627 +$\log(RV_{t-1,ETF})$ & 0.179 & 0.029 & 0.000 & 0.179 & 0.123 & 0.236 \\
628 +$\log(RV_{t-2,ETF})$ & 0.078 & 0.024 & 0.000 & 0.078 & 0.031 & 0.126 \\
629 +\bottomrule
630 +\end{tabular}
631 +\begin{tablenotes}
632 +\small
633 +\item This table presents posterior statistics from a Bayesian Vector Autoregression (VAR) model analyzing the relationship between the SLV ETF and its Net Asset Value (NAV). Panel A shows results for the NAV equation, while Panel B shows results for the ETF equation. The model includes two lags of both log realized variances. Mean represents the posterior mean, Std. Dev. is the posterior standard deviation, MCSE is the Monte Carlo standard error, Median is the posterior median, and the 95\% credible interval provides the range of plausible parameter values. All realized variance measures are constructed using 5-minute price data.
634 +\end{tablenotes}
635 +\end{threeparttable}
636 +\end{table}
637 +\end{landscape}
638 +
639 +\begin{landscape}
640 +\begin{table}[htbp]
641 +\centering
642 +\caption{Bayesian Vector Autoregression Results: UNG ETF and Net Asset Value}
643 +\label{tab:VAR_UNG}
644 +\begin{threeparttable}
645 +\footnotesize
646 +\begin{tabular}{@{}lccccccc@{}}
647 +\toprule
648 + & \multicolumn{6}{c}{\textbf{Panel A: $\log(RV_{t,NAV})$}} \\
649 +\cmidrule(lr){2-7}
650 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
651 +\cmidrule(lr){6-7}
652 + & & & & & Lower & Upper \\
653 +\midrule
654 +$\log(RV_{t-1,NAV})$ & 0.392 & 0.022 & 0.000 & 0.392 & 0.349 & 0.437 \\
655 +$\log(RV_{t-2,NAV})$ & 0.231 & 0.020 & 0.000 & 0.231 & 0.192 & 0.272 \\
656 +$\log(RV_{t-1,ETF})$ & 0.080 & 0.021 & 0.000 & 0.080 & 0.038 & 0.121 \\
657 +$\log(RV_{t-2,ETF})$ & 0.157 & 0.019 & 0.000 & 0.157 & 0.119 & 0.194 \\
658 +\addlinespace[0.3cm]
659 +
660 + & \multicolumn{6}{c}{\textbf{Panel B: $\log(RV_{t,ETF})$}} \\
661 +\cmidrule(lr){2-7}
662 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
663 +\cmidrule(lr){6-7}
664 + & & & & & Lower & Upper \\
665 +\midrule
666 +$\log(RV_{t-1,NAV})$ & 0.130 & 0.024 & 0.000 & 0.130 & 0.083 & 0.177 \\
667 +$\log(RV_{t-2,NAV})$ & 0.165 & 0.022 & 0.000 & 0.165 & 0.124 & 0.208 \\
668 +$\log(RV_{t-1,ETF})$ & 0.339 & 0.023 & 0.000 & 0.339 & 0.294 & 0.383 \\
669 +$\log(RV_{t-2,ETF})$ & 0.240 & 0.020 & 0.000 & 0.240 & 0.200 & 0.280 \\
670 +\bottomrule
671 +\end{tabular}
672 +\begin{tablenotes}
673 +\small
674 +\item This table presents posterior statistics from a Bayesian Vector Autoregression (VAR) model analyzing the relationship between the UNG ETF and its Net Asset Value (NAV). Panel A shows results for the NAV equation, while Panel B shows results for the ETF equation. The model includes two lags of both log realized variances. Mean represents the posterior mean, Std. Dev. is the posterior standard deviation, MCSE is the Monte Carlo standard error, Median is the posterior median, and the 95\% credible interval provides the range of plausible parameter values. All realized variance measures are constructed using 5-minute price data.
675 +\end{tablenotes}
676 +\end{threeparttable}
677 +\end{table}
678 +\end{landscape}
\ No newline at end of file
added these-ulaval/.latexmkrc +6 −0
@@ -0,0 +1,6 @@
1 +# Configuration latexmk — thèse ulthese (pdfLaTeX + BibTeX)
2 +$pdf_mode = 1; # pdflatex
3 +$bibtex_use = 2; # exécuter bibtex au besoin
4 +$pdflatex = 'pdflatex -interaction=nonstopmode -halt-on-error -file-line-error %O %S';
5 +@default_files = ('main.tex');
6 +$clean_ext = 'bbl blg run.xml synctex.gz';
added these-ulaval/INVENTAIRE.md +314 −0
@@ -0,0 +1,314 @@
1 +# INVENTAIRE — Thèse par articles (ULaval, FSA, dépôt 2026)
2 +
3 +Analyse exhaustive du repo `phd_thesis/` réalisée avant construction du document
4 +`these-ulaval/`. **Aucun fichier source des chapitres n'a été modifié** ; toutes
5 +les adaptations sont faites sur des copies dans `these-ulaval/`.
6 +
7 +Titre de la thèse : *Three Essays on High-Frequency Return and Volatility Dynamics
8 +in Commodities and Financial Futures Markets* — Simon-Pierre Boucher.
9 +
10 +## Cartographie du repo
11 +
12 +| Chapitre | Dossier source | Fichier maître | Statut |
13 +|---|---|---|---|
14 +| 1 | `phd_chap1_20260731/` | `main.tex` (monolithique, 558 l.) | Version révisée pour *The Energy Journal* |
15 +| 2 | `phd_chap2_20260731/` | `main.tex` + `sections/` | Version pour soumission au *Journal of Futures Markets* |
16 +| 3 | `PHD_chapitre3_theses_20260731/` | `chapitre3.tex` + `sections/` | Manuscrit daté 2026-03-06 |
17 +
18 +Une seule version de chaque chapitre dans le repo (suffixe commun `20260731`) —
19 +aucune ambiguïté de version.
20 +
21 +---
22 +
23 +## Chapitre 1 — Speculative trading in energy markets: Evidence from macroeconomic surprises
24 +
25 +- **Sources** : `phd_chap1_20260731/` — `main.tex` (texte + préambule + 8 équations numérotées),
26 + `tables.tex` (14 flottants, dont 9 sidewaystable), `figures.tex` (6 figures),
27 + `appendix_full.tex` (2 tableaux), `appendix_covid.tex` (11), `appendix_zlb.tex` (11), `master.bib`.
28 +- **Auteurs** : Simon-Pierre Boucher, Marie-Hélène Gagnon, Gabriel J. Power.
29 +- **Classe/pile** : `article` 11pt, natbib + `\bibliographystyle{chicago}`, ~45 packages
30 + (beaucoup inutilisés), geometry 1in, doublespacing, palatino/mathpazo/sourcesanspro.
31 +- **Structure** : Introduction / Background / Data / Econometric framework and methods /
32 + Results / Discussion and implications / Conclusion (+ sections Tables, Figures en fin d'article).
33 +- **Comptes (actifs)** : 14 tableaux (corps) + 24 en appendices (non inclus dans la version
34 + article — `\include` commentés, « available upon request ») ; 6 figures ; 8 équations numérotées.
35 +- **Figures** : `pi_cl.png, pi_gc.png, pi_hg.png, pi_si.png, pi_pa.png, pi_ng.png` (toutes présentes).
36 + Non référencées : `FIG_MSCT.pdf`, `FIG_NLS.pdf`, `FIG_WT.pdf` (proxies abandonnés, équations commentées).
37 +- **master.bib** : 249 entrées, ~60 réellement citées ; clés hétérogènes (auteur-année-mot,
38 + DOI bruts, sigles NBER) ; ~75 clés dupliquées EN INTERNE (même clé répétée).
39 +- **Contenu scientifique** : effet de l'intensité de la spéculation (proxy NLS construit sur le
40 + COT désagrégé de la CFTC : money managers vs swap dealers) sur la réaction des futures
41 + (rendements WLS-EWMA, variance conditionnelle GARCH, spreads bid-ask) aux surprises
42 + d'annonces macro standardisées (Balduzzi et al. 2001). Données 5 minutes, avril 2007 –
43 + février 2024, CL, NG vs GC, SI, HG, PA. Résultat : effet AMORTISSEUR de la spéculation
44 + (prix, volatilité, spreads), porté par les money managers ; plus fort pour l'énergie
45 + procyclique que pour l'or (valeur refuge).
46 +
47 +### Problèmes détectés (chapitre 1)
48 +
49 +1. 🔴 `figures.tex` : 6 accolades fermantes orphelines (l. 4, 12, 20, 28, 36, 44) — ne compile
50 + pas tel quel. Corrigé mécaniquement dans la copie.
51 +2. 🔴 `\newenvironment{proof}` — collision avec `memoir` (ulthese) qui définit déjà `proof`.
52 + (Environnement de toute façon inutilisé dans le corps.)
53 +3. 🔴 `sectsty`/`\allsectionsfont` et `tocloft` — incompatibles avec `memoir`.
54 +4. 🔴 `subfigure` + `subfloat` + `subcaption` chargés ensemble — mutuellement incompatibles.
55 +5. 🔴 `\include{tables}`/`\include{figures}` — `\include` non imbricable → convertis en `\input`.
56 +6. 🔴 `tables.tex` : `\setcounter{table}{8}`… + `\renewcommand\thetable{9(a)}`…`11(b)`
57 + jamais restaurés → casseraient la numérotation des chapitres suivants de la thèse.
58 +7. 🟡 Citation `Roll1984` (main.tex:327) : correspond en fait à l'entrée `roll1984` du
59 + `master.bib` (BibTeX est insensible à la casse des clés) — aucune entrée manquante.
60 + En revanche, l'entrée `boons2014price` de la source est non fermée (accolade finale
61 + manquante) : réinsérée complète dans le bib consolidé.
62 +8. 🟠 `geometry`, `titlepage`/`\maketitle`, `\pagenumbering`, `\setcounter{page}{0}`,
63 + `\doublespacing`, `\hypersetup` couleurs — à neutraliser (gérés par ulthese).
64 +9. 🟠 `hyperref` chargé trop tôt ; doublons `color`+`xcolor`, `lscape`+`pdflscape` ;
65 + `apalike` (package) + natbib + chicago contradictoires ; ~17 packages jamais utilisés
66 + (`pgfgantt`, `blindtext`, `nicematrix`, `wasysym`, `eurosym`, `siunitx` sans usage, etc.).
67 +10. 🟡 Labels dupliqués entre `tables.tex` et `appendix_full.tex`
68 + (`tab:macro_fin_nls_fut_returns_full`, `tab:macro_fin_nls_var_full`) — pertinent si
69 + l'annexe A est réintégrée.
70 +11. 🟡 Incohérence de contenu : abstract « 26 announcements » vs section Data « 22 » (les
71 + équations somment sur 22). **Non modifié** (fond scientifique) — à arbitrer par l'auteur.
72 +12. 🟡 Labels avec espaces (`eq:Model 1`, `eq:Model 2`, `eqn:auxiliary 2`) ; commentaires de
73 + travail (MH/SP) dans le source ; guillemets ASCII.
74 +
75 +---
76 +
77 +## Chapitre 2 — Seeing Through the ETF: Indicative NAV and Commodity Volatility Transmission
78 +
79 +- **Sources** : `phd_chap2_20260731/` — `main.tex` (préambule + page titre + abstract),
80 + `sections/01_introduction, 03_data, 04_methods, 05_results, 06_conclusion` (pas de fichier
81 + `02_*` : revue de littérature fondue dans l'introduction), `tables/tables.tex`,
82 + `figures/figures.tex`, `master.bib`, `main.pdf` (compilé).
83 +- **Auteurs** : Simon-Pierre Boucher, Marie-Hélène Gagnon, Gabriel J. Power.
84 +- **Classe/pile** : `article` 12pt, natbib + chicago, geometry 1in, sectsty/`\allsectionsfont`,
85 + palatino+mathpazo+sourcesanspro, `caption` chargé deux fois (option clash latent),
86 + `\makeatletter` jamais refermé, tikz/pgfgantt/blindtext... chargés sans usage.
87 +- **Structure** : Introduction / Data and Sample Construction (5 ss.) / Econometric
88 + Methodology (5 ss.) / Empirical Results (6 ss.) / Conclusion / Tables / Figures.
89 +- **Comptes** : 13 tableaux (tous `threeparttable`, les nos 4-13 en `landscape`) ; 8 figures
90 + (toutes en `landscape`) ; 14 équations numérotées. Aucune annexe.
91 +- **Figures** : 8 PDF tous présents (`oil_rv, gold_rv, silver_rv, gas_rv` + 4 IRF) ;
92 + 4 PNG doublons non référencés.
93 +- **master.bib** : 65 entrées (100 % `@article`), clés `auteur+année+mot` ; 36 citées,
94 + 29 non citées ; aucune citation orpheline.
95 +- **Contenu scientifique** : construction inédite d'une iNAV à la minute pour 4 ETF
96 + mono-commodité (GLD, SLV, USO, UNG), 2010–2023 (≈45 M d'observations tick nettoyées) ;
97 + RV décomposée continu/sauts (BNS) ; HAR-X / HAR-CJ-X à 1, 5, 30 min + BVAR Minnesota
98 + (Gibbs). Résultats : transmission unidirectionnelle iNAV→ETF pour les métaux précieux,
99 + bidirectionnelle et asymétrique pour l'énergie ; passe par les **sauts** plutôt que la
100 + diffusion ; ampleur jusqu'à 2× plus grande à 1 min qu'à 30 min ; stable en sous-périodes.
101 +
102 +### Problèmes détectés (chapitre 2)
103 +
104 +1. 🔴 `sectsty` + `\allsectionsfont{\sffamily}` — incompatible memoir. Non repris.
105 +2. 🔴 `caption` chargé 2× (l.5 sans option, l.29 avec options) — option clash latent.
106 +3. 🔴 `subfigure` + `subfloat` + `subcaption` — conflit de packages.
107 +4. 🔴 `\newenvironment{proof}` — collision memoir/amsthm (env. inutilisé).
108 +5. 🔴 `\makeatletter` sans `\makeatother` (l.40-42, vestige `\ifAtEnd@DontShowMe`) — non repris.
109 +6. 🟠 `geometry`, `titlepage`/`\maketitle`, `\doublespacing`, `\hypersetup` couleurs,
110 + triple package de polices — neutralisés (gérés par ulthese).
111 +7. 🟠 `\graphicspath{{figures/}}` — remplacé par le `\graphicspath` global de la thèse.
112 +8. 🟡 `\section{Tables}`/`\section{Figures}` numérotées (vs `\section*` au chap. 3) —
113 + harmonisé : sections numérotées dans les trois chapitres.
114 +9. 🟡 **Incohérence de contenu** : le texte annonce « January 2010 to January 2023 » mais les
115 + 8 légendes de figures indiquent 2008-03-27 → 2025-08-01. **Non modifié** — à arbitrer.
116 +
117 +---
118 +
119 +## Chapitre 3 — Returns and Volatility Around FOMC Announcements: A High-Frequency Analysis of Policy Tone and Novelty
120 +
121 +- **Sources** : `PHD_chapitre3_theses_20260731/` — `chapitre3.tex`,
122 + `sections/01_introduction … 05_conclusion`, `tables/` (tables.tex + 23 fichiers de
123 + tableaux), `figures/figures.tex`, `appendices/appendix_proofs.tex`,
124 + `appendices/appendix_additional.tex`, `master.bib`, `chapitre3.pdf` (compilé sans warning).
125 +- **Auteurs** : Simon-Pierre Boucher, Marie-Hélène Gagnon, Gabriel J. Power. Date 2026-03-06.
126 +- **Classe/pile** : `article` 11pt letterpaper, préambule propre (inputenc/fontenc/babel,
127 + amsthm, natbib + **apalike**, geometry 1in, float `[H]`).
128 +- **Structure** : Introduction / Data (4 ss.) / Methodology (3 ss., 11 définitions amsthm,
129 + 1 proposition, 1 assumption) / Results (4 ss.) / Conclusion (5 ss.) / Tables / Figures /
130 + Annexe A (Mathematical Proofs : 3 théorèmes + 3 preuves, 10 éq.) / Annexe B (Additional
131 + Tables and Figures : 8 tableaux + 10 figures).
132 +- **Comptes** : 24 tableaux (1 corps + 15 via tables.tex + 8 en annexe B ; les 23 fichiers de
133 + `tables/` sont tous inclus, aucun orphelin) ; 19 figures (9 corps + 10 annexe, 21
134 + `\includegraphics` dont 2 paires de sous-figures) ; 45 équations numérotées.
135 +- **Figures** : 17 PDF + 5 PNG (diagnostics NLP en bitmap — qualité d'impression à vérifier),
136 + tous présents, tous référencés.
137 +- **master.bib** : 45 entrées, toutes citées, zéro orpheline ; clés `auteur+année` (convention
138 + différente des chap. 1-2, mais **aucune collision**).
139 +- **Contenu scientifique** : décomposition des communiqués FOMC en **ton** (hawkish/dovish) et
140 + **nouveauté** informationnelle via un ensemble bi-modèle (MiniLM+BERT, TSDAE+MNRL, références
141 + choisies par ACP) ; 217 communiqués (2000–2025), 148 événements FOMC (2008–2025), données
142 + 1 min sur 7 futures (ES, VX, ZN, ZF, DX, CL, GC) ; panels à la minute, régressions
143 + événementielles, projections locales de Jordà, placebos, 5 méthodes d'inférence. Résultats :
144 + le ton prédit les rendements directionnels (ES ≈ +12 pb à 2 h pour un choc dovish 1 σ), la
145 + nouveauté prédit la volatilité (interaction stance×novelty sur VIX, t = −5,06) ; stance
146 + significatif sur log(RV) pour 6/7 contrats ; placebos nuls.
147 +
148 +### Problèmes détectés (chapitre 3)
149 +
150 +1. 🔴 `\appendix` (niveau article : renumérote les sections en A, B) — converti en annexes de
151 + thèse (chapitres d'annexe memoir), voir plan.
152 +2. 🔴 `apalike` vs `chicago` (chap. 1-2) — style unique retenu : chicago (auteur-année).
153 +3. 🟠 `titlepage` avec `\footnotemark[1..3]`/`\footnotetext` manuels — non repris.
154 +4. 🟠 `geometry`, `\doublespacing`, `\hypersetup`, `babel[english]` principal — neutralisés.
155 +5. 🟠 Chemins `figures/...` en dur (21 `\includegraphics`) — adaptés.
156 +6. 🟡 `\addcontentsline` manuels pour Tables/Figures — remplacés par des sections numérotées.
157 +7. 🟡 Sans correction, les tableaux/figures de l'annexe B continueraient la numérotation du
158 + corps — réglé nativement par memoir (annexes = chapitres A/B → A.1, B.1…).
159 +
160 +---
161 +
162 +## Conflits inter-chapitres (détectés AVANT construction)
163 +
164 +| Conflit | Détail | Résolution |
165 +|---|---|---|
166 +| **Styles de citation** | chap. 1-2 : natbib + `chicago` ; chap. 3 : natbib + `apalike`. | Style unique `chicago` (auteur-année, dominant 2/3, norme finance). Différences purement typographiques en bibliographie. |
167 +| **Packages incompatibles memoir** | `geometry` (3×), `sectsty` (1-2), `tocloft` (1), `caption` double (2), `subfigure`+`subfloat` (1-2). | Aucun repris ; préambule minimal reconstruit (`preamble.tex`). |
168 +| **Environnements définis 2×** | `theorem`, `proposition` définis dans les 3 préambules ; `proof` redéfini manuellement (chap. 1-2) en conflit avec amsthm (chap. 3) et memoir. | Définis UNE fois dans `preamble.tex` via amsthm, numérotés par chapitre (`[chapter]`). `proof` = celui d'amsthm. Les envs `hyp`/`subhyp`/`corollary` (chap. 1-2, inutilisés dans les corps) définis aussi, par prudence. |
169 +| **Clés BibTeX** | ~75 clés dupliquées à l'intérieur du bib du chap. 1 ; 5 clés partagées chap. 1/chap. 2 avec contenus divergents (`andersen2001distribution`, `basak2016model`, `gorton2006facts`, `liu2015does`, `singleton2014investor`) ; 18 titres identiques sous clés différentes (versions/coquilles : `fleming1997moves`/`flemming1997moves`, `hu2013noise`/`hu2019`…). Zéro collision chap. 2/chap. 3. | Fusion en `bib/these.bib` : une seule entrée par clé (variante la plus complète) ; clés distinctes conservées telles quelles (aucune citation modifiée) ; doublons de titres documentés — seuls comptent ceux effectivement cités, vérifiés à la compilation. Entrée manquante `Roll1984` ajoutée. |
170 +| **Labels dupliqués entre chapitres** | `sec:data` (2 et 3), `sec:results` (2 et 3), `sec:conclusion` (1, 2, 3), etc. → `multiply defined labels` et renvois faux. | Préfixage mécanique systématique par chapitre dans les copies : `\label{X}`→`\label{chN:X}`, idem `\ref`/`\eqref`/`\pageref`. Les renvois sont tous intra-chapitre (vérifié). |
171 +| **Notation mathématique incohérente** (documentée, NON modifiée) | (i) La surprise standardisée est $S_{kt}$ au chap. 1 et MPS/surprise sémantique au chap. 3 ; (ii) la variance réalisée est $RV$ aux chap. 2-3 mais la volatilité du chap. 1 passe par une variance conditionnelle GARCH $h_t$ ; (iii) $\theta_m$ (chap. 1, texte) vs $\phi_m$ (chap. 1, discussion des tableaux) pour le même coefficient d'interaction ; (iv) indices de temps $t$ à 5 min (chap. 1), 1/5/30 min (chap. 2), 1 min (chap. 3). | Aucune modification (fond scientifique). Signalé pour l'introduction générale, qui définit les objets en termes génériques. |
172 +| **Incohérences internes de contenu** | chap. 1 : « 26 » (abstract) vs « 22 » (Data) annonces ; chap. 2 : période 2010–2023 (texte) vs 2008–2025 (légendes de figures). | **Non modifié** — liste transmise à l'auteur dans le rapport final. |
173 +| **Chemins d'images** | chap. 1 : noms nus ; chap. 2 : `\graphicspath{{figures/}}` + noms nus ; chap. 3 : préfixe `figures/` en dur. | Figures copiées dans `chapitreN/figures/` ; `\graphicspath` global `{{chapitre1/figures/}{chapitre2/figures/}{chapitre3/figures/}}` ; préfixe `figures/` retiré des copies du chap. 3. |
174 +| **`\include` imbriqués** (chap. 1) | `\include{tables}` dans un fichier lui-même inclus → erreur. | Tous les `\include` internes convertis en `\input`. |
175 +| **Numérotation manuelle des tableaux** (chap. 1) | `\setcounter{table}{8}` + `\thetable → 9(a)…11(b)` jamais restaurés. | Supprimés des copies : numérotation continue par chapitre (1.1–1.14). Toutes les références passent par `\ref` (vérifié) — aucun renvoi cassé. |
176 +
177 +## Plan de résolution / construction (exécuté en phase 2)
178 +
179 +1. `preamble.tex` minimal : amsmath/amssymb/amsthm, booktabs, multirow, threeparttable,
180 + rotating, pdflscape, float, subcaption, théorèmes uniques `[chapter]`, `\graphicspath`,
181 + métadonnées hyperref. Rien d'autre (natbib, babel, hyperref, graphicx, xcolor : chargés
182 + par la classe).
183 +2. `main.tex` = gabarit officiel `gabarit-doctorat-articles.tex` adapté (PhD, french+english,
184 + frontispice, liminaires, intro, 3 chapitres, conclusion, annexes A-B, bibliographie unique).
185 +3. Chapitres : wrapper `\chapter` + Résumé FR (≤150 mots, brouillon) + Abstract EN (texte
186 + original de l'article) + corps copié-adapté (préfixage des labels, retrait des éléments de
187 + préambule/page titre, conversions décrites ci-dessus). **Aucun changement au texte
188 + scientifique, aux équations, aux tableaux ni aux figures.**
189 +4. Annexes de thèse A et B = annexes du chap. 3 (`\section`→`\chapter`, `\subsection`→
190 + `\section`), titres suffixés « (Chapter 3) ». Annexes désactivées du chap. 1 : non
191 + intégrées (décision de l'auteur dans la source : « available upon request ») — copies
192 + conservées nulle part ; réactivables depuis le repo source si souhaité.
193 +5. `bib/these.bib` consolidé + `\bibliographystyle{chicago}`.
194 +6. Compilation `latexmk` (pdfLaTeX) jusqu'à zéro erreur / zéro référence indéfinie /
195 + zéro citation manquante ; vérification des comptes (sections, tableaux, figures,
196 + équations) contre les sources.
197 +
198 +## Journal des modifications mécaniques (copies uniquement — sources intactes)
199 +
200 +Aucun texte scientifique, aucune équation, aucun résultat, aucun tableau ni aucune figure
201 +n'a été modifié, aux exceptions près listées ci-dessous, toutes requises pour que le
202 +document compile (les sources des chap. 1 n'avaient jamais été compilées telles quelles).
203 +
204 +### Restructuration (tous chapitres)
205 +
206 +1. Corps des articles copiés dans `chapitreN/` ; préambules, pages de titre d'articles,
207 + `\maketitle`, abstracts de page titre, `\pagenumbering`, `\doublespacing`, `\geometry`,
208 + `\hypersetup` locaux : non repris (gérés par la classe/`preamble.tex`).
209 +2. Préfixage systématique des étiquettes : `\label{X}`→`\label{chN:X}`, idem `\ref`/`\eqref`
210 + (35 labels ch1, 34 ch2, 67 ch3) — résout les collisions `sec:data`, `sec:results`,
211 + `sec:conclusion` entre chapitres. Tous les renvois sont intra-chapitre (vérifié).
212 +3. Figures copiées dans `chapitreN/figures/` ; `\graphicspath` global dans `preamble.tex` ;
213 + préfixe `figures/` retiré des `\includegraphics` et `\input{tables/…}` →
214 + `\input{chapitre3/tables/…}` (chap. 3).
215 +4. Wrappers `chapitreN.tex` : `\chapter{titre original}` + section Résumé (français,
216 + brouillon) + section Abstract (texte original de l'article, conservé intégralement,
217 + avec Keywords/JEL) — exigence FESP pour les thèses par articles.
218 +5. Sections « Tables »/« Figures » de fin d'article : harmonisées en `\section` numérotées
219 + dans les trois chapitres (chap. 2 les avait numérotées, chap. 3 en `\section*` +
220 + `\addcontentsline` manuels, retirés).
221 +
222 +### Chapitre 1 (`phd_chap1_20260731`)
223 +
224 +6. `figures.tex` : suppression de 6 accolades fermantes orphelines (l. 4, 12, 20, 28, 36,
225 + 44 de la source) — **bug bloquant de la source** (`Too many }'s`).
226 +7. `figures.tex` : `\includegraphics[scale=0.4]` → `[width=\textwidth]` (6×) — les images
227 + débordaient de ~13,7 mm dans les marges FESP (30 mm), plus étroites que celles de
228 + l'article (1 po).
229 +8. `tables.tex` : suppression des hacks `\setcounter{table}{8|9|10}` et
230 + `\renewcommand\thetable{9(a)…11(b)}` (jamais restaurés dans la source) → numérotation
231 + memoir continue 1.1–1.14. Toutes les références passent par `\ref` (vérifié — aucun
232 + « Table 9(a) » codé en dur).
233 +9. `tables.tex` : `\begin{tablenotes}` → `\begin{tablenotes}\item[]` (14×) — les notes de
234 + tableaux étaient utilisées sans `\item` et hors `threeparttable` : erreur LaTeX
235 + (« perhaps a missing \item ») ; la source ne compilait pas.
236 +10. `tables.tex` : correction de 2 coquilles mathématiques bloquantes de la source :
237 + `$NLS_{MM$}` → `$NLS_{MM}$` (idem `SD`) et `R_{\text{SPREAD}}_{t}` →
238 + `{R_{\text{SPREAD}}}_{t}` (2×, erreur « Double subscript »).
239 +11. `main.tex` l. 90–531 extraits vers `corps.tex` ; `\include{tables|figures}` →
240 + `\input` (l'`\include` n'est pas imbricable) ; `\small` avant bibliographie non repris.
241 +12. Annexes A/B/C de l'article (full-sample NLS, COVID, ZLB) : **non intégrées** — elles
242 + sont désactivées dans la source (« We do not include the appendices anymore ; but
243 + available upon request »). Réactivables sur demande.
244 +
245 +### Chapitre 2 (`phd_chap2_20260731`)
246 +
247 +13. Aucune modification au-delà des points 1–5 (sections copiées telles quelles).
248 +
249 +### Chapitre 3 (`PHD_chapitre3_theses_20260731`)
250 +
251 +14. Annexes de l'article converties en annexes de thèse : `\section` → `\chapter` (titres
252 + suffixés « (Chapter 3) »), `\subsection` → `\section`. Numérotation memoir automatique
253 + (A.1…, B.1…) pour sections, tableaux et figures d'annexe.
254 +15. Canonicalisation de 4 clés de citation dupliquées (même référence citée sous deux
255 + clés, ce qui doublait les entrées en bibliographie « 2003a/2003b ») :
256 + `andersen2003`→`andersen2003micro` (3×), `andersen2007`→`andersen2007real` (1×),
257 + `balduzzi2001`→`balduzzi2001economic` (1×), `barndorff2002`→`barndorff2002econometric`
258 + (1×) — remplacements dans les copies du chap. 3 uniquement.
259 +
260 +### Préambule harmonisé (`preamble.tex`)
261 +
262 +16. Environnements de théorèmes définis une seule fois via `amsthm`, numérotés par chapitre
263 + (la numérotation des Definitions/Theorems du chap. 3 passe de « Definition 1 » à
264 + « Definition 3.1 » — changement de forme uniquement).
265 +17. `siunitx` repris avec le `\sisetup` d'origine du chap. 1 (colonnes `S` de ses
266 + tableaux) ; `\singlespacing` mappé sur `\SingleSpacing` (memoir) ; packages en conflit
267 + avec memoir non repris (`geometry`, `sectsty`, `tocloft`, `subfigure`, `subfloat`,
268 + `caption` en double, `setspace`, `color`, `lscape`, polices `palatino`/`mathpazo`/
269 + `sourcesanspro`, et ~17 packages jamais utilisés).
270 +
271 +### Bibliographie (`bib/these.bib`)
272 +
273 +18. Fusion des trois `master.bib` : 359 entrées sources → 271 clés uniques ; 4 doublons
274 + exacts supprimés ; 85 clés répétées avec variantes divergentes résolues (variante la
275 + plus complète conservée).
276 +19. L'entrée `boons2014price` de la source (chap. 1) était **non fermée** (accolade finale
277 + manquante — bug de la source) : réinsérée complète dans le bib consolidé.
278 +20. `Roll1984` (cité au chap. 1) correspond à l'entrée `roll1984` de la source — BibTeX
279 + est insensible à la casse des clés : aucune entrée manquante finalement.
280 +21. Doublons « même référence, clés différentes » NON fusionnés lorsqu'une seule clé est
281 + citée (l'entrée non citée n'apparaît pas en bibliographie) ou lorsque les années
282 + diffèrent (versions différentes d'un même travail — à arbitrer par l'auteur) :
283 + `brunetti2011speculators`/`brunetti2016speculators`, `cheng2012convective`/
284 + `cheng2015convective`, `hu2013noise`/`hu2019`, `ben2017etfs`/`ben2018etfs`,
285 + `koop2011forecasting`/`koop2013forecasting`, `hamm2011liquidity`/`hamm2014liquidity`,
286 + `fleming1997moves`/`flemming1997moves`, etc.
287 +
288 +## Vérification finale (comptes source ↔ thèse)
289 +
290 +| Chapitre | Tableaux | Figures | Équations (equation+align) | Écart |
291 +|---|---|---|---|---|
292 +| 1 | 14 (5 table + 9 sidewaystable) | 6 | 8 | **0** |
293 +| 2 | 13 | 8 | 14 | **0** |
294 +| 3 (avec annexes) | 24 | 19 | 44 | **0** |
295 +
296 +Sections/sous-sections : identiques, aux transformations près documentées (annexes du
297 +chap. 3 promues : 2 sections → 2 chapitres d'annexe, 10 sous-sections → 10 sections ;
298 +ajout des sections Résumé/Abstract exigées par la FESP et des sections Tables/Figures).
299 +
300 +Compilation finale : `latexmk` (pdfLaTeX), **0 erreur, 0 référence indéfinie, 0 citation
301 +manquante, 0 label multiplié, 0 flottant perdu, 0 overfull sévère (>20 pt)** ; test réussi
302 +depuis une copie fraîche du dossier seul ; PDF de 188 pages avec métadonnées
303 +(titre/auteur).
304 +
305 +## Incohérences de CONTENU laissées à l'arbitrage de l'auteur (non modifiées)
306 +
307 +- Chap. 1 : « 26 macroeconomic announcement releases » (abstract) vs « 22 announcements »
308 + (section Data ; les équations somment sur 22).
309 +- Chap. 2 : période « January 2010 to January 2023 » (texte) vs « 2008-03-27 to
310 + 2025-08-01 » (légendes des 8 figures).
311 +- Chap. 1 : commentaires de travail (MH/SP) toujours présents dans les sources copiées
312 + (inoffensifs à la compilation) ; guillemets ASCII `"commodity beta"` (l. 5 de
313 + `corps.tex`) rendus typographiquement faux.
314 +- Doublons bibliographiques « versions différentes » du point 21.
added these-ulaval/Makefile +18 −0
@@ -0,0 +1,18 @@
1 +# Compilation de la thèse (classe ulthese, pdfLaTeX + BibTeX via latexmk)
2 +#
3 +# make : compile main.pdf (toutes les passes nécessaires)
4 +# make clean : supprime les fichiers auxiliaires
5 +# make cleanall : supprime aussi le PDF
6 +
7 +.PHONY: all clean cleanall
8 +
9 +all:
10 + latexmk main.tex
11 +
12 +clean:
13 + latexmk -c
14 + find . -name '*.aux' -delete
15 +
16 +cleanall:
17 + latexmk -C
18 + find . -name '*.aux' -delete
added these-ulaval/NOTES_FORMAT.md +61 −0
@@ -0,0 +1,61 @@
1 +# NOTES_FORMAT — Exigences FESP/ULaval et décisions de mise en forme
2 +
3 +Thèse : *Three Essays on High-Frequency Return and Volatility Dynamics in Commodities
4 +and Financial Futures Markets* — Simon-Pierre Boucher, dépôt 2026.
5 +
6 +Sources officielles consultées (juillet 2026) :
7 +
8 +- Tableau synthèse des règles de présentation matérielle (FESP, 2026-07-15) :
9 + <https://www.fesp.ulaval.ca/system/files/public/memoires-theses/synthese_regles_de_presentation_materielle_memoire_these.pdf>
10 +- Aide-mémoire pour le mémoire ou la thèse par articles (FESP, 2025-09-17) :
11 + <https://www.fesp.ulaval.ca/system/files/public/memoires-theses/aide_memoire_pour_la_redaction_avec_articles.pdf>
12 +- Règles de présentation détaillées : <https://www.fesp.ulaval.ca/memoires-et-theses/regles-de-presentation-detaillees>
13 +- Contenu des sections : <https://www.fesp.ulaval.ca/memoires-et-theses/contenu-des-sections>
14 +- Outils de synthèse et langue de rédaction : <https://www.fesp.ulaval.ca/memoires-et-theses/outils-synthese-et-langue-de-redaction>
15 +- Guide de la classe `ulthese` v7.0 (FESP, incluse dans TeX Live 2026, doc locale
16 + `/usr/local/texlive/2026/texmf-dist/doc/latex/ulthese/ulthese.pdf`)
17 +- Programme : <https://www.ulaval.ca/etudes/programmes/doctorat-en-sciences-de-ladministration-finance-et-assurance>
18 +
19 +| # | Exigence (source) | Comment elle est respectée ici |
20 +|---|---|---|
21 +| 1 | **Classe officielle** : la FESP fournit et recommande la classe LaTeX `ulthese` (basée sur `memoir`). | `main.tex` utilise `\documentclass[PhD,11pt,french,english]{ulthese}` (dernière langue = anglais, langue principale de la thèse). Classe installée (TeX Live 2026, v7.0). |
22 +| 2 | **Page de titre** : « une page de titre officielle … sera ajoutée à votre manuscrit par nos systèmes au moment du dépôt final » ; le guide ulthese : « La thèse ou le mémoire déposé à la FESP ne doit **pas** comporter de page de titre. » La classe produit une page **frontispice** obligatoire (`\frontispice`) avec titre, auteur, programme, direction/codirection. | Pas de page de titre manuelle. Frontispice ulthese avec `\titre`, `\auteur{Simon-Pierre Boucher}`, `\programme{Doctorat en sciences de l'administration -- finance et assurance}`, `\direction{Marie-Hélène Gagnon, directrice de recherche}`, `\codirection{Gabriel Power, codirecteur de recherche}`. Le bloc de page titre fourni par l'auteur n'est **pas** utilisé (interdit au dépôt) ; il est conservé en commentaire dans `main.tex` pour référence. |
23 +| 3 | **Comité de thèse / jury** : n'apparaît **nulle part** dans le manuscrit (ni page titre, ni page du jury ; « aucune signature ni information personnelle »). L'examinateur externe (terminologie FESP : « examinateur externe » / « membre externe ») figure au dossier administratif seulement. | Les membres du comité (M.-C. Beaulieu, A. Tremblay-Simard, V. Sokolovski — examinateur externe, University of Alberta) ne sont pas imprimés dans la thèse. Ils sont listés en commentaire dans `main.tex` et peuvent être remerciés dans `remerciements.tex` (placeholder prévu). **Décision documentée ici comme demandé.** |
24 +| 4 | **Intitulé du programme** : seul intitulé officiel trouvé au répertoire ULaval : « Doctorat en sciences de l'administration - finance et assurance », grade Philosophiæ doctor (Ph. D.), FSA. Il n'existe pas de concentration « finance » seule. | `\programme{Doctorat en sciences de l'administration -- finance et assurance}`. ⚠️ **À confirmer avec la direction de programme** (l'intitulé demandé était « … – finance »). |
25 +| 5 | **Langue** : rédaction en anglais permise avec autorisation préalable (formulaire FES-101). Titre dans la langue de rédaction ; autres éléments de la page frontispice en français. | Thèse en anglais ; babel `[french,english]`, anglais actif par défaut ; pages liminaires françaises composées en français via `\selectlanguage`/`otherlanguage`. ⚠️ **Vérifier que le FES-101 est au dossier.** |
26 +| 6 | **Résumé français obligatoire**, max **700 mots**, = page ii ; résumé anglais recommandé, page distincte. | `pages-liminaires/resume.tex` (brouillon < 700 mots, marqué `% BROUILLON À RÉVISER PAR L'AUTEUR`) puis `abstract.tex`. Ordre : résumé FR → abstract EN. |
27 +| 7 | **Ordre des parties** (Tableau synthèse) : page de titre (ajoutée par la FESP) ; résumés ; table des matières ; listes (tableaux, figures, abréviations) ; dédicace/épigraphe (facultatif) ; remerciements ; avant-propos ; introduction (= page 1, non numérotée comme chapitre) ; chapitres ; conclusion (non numérotée) ; bibliographie ; annexes. | `main.tex` suit exactement cet ordre (gabarit officiel `gabarit-doctorat-articles.tex`). Introduction et conclusion en `\chapter*` avec ajout manuel à la TdM. Annexes : placées **avant** la bibliographie, comme dans le gabarit officiel `gabarit-doctorat-articles.tex` (la règle FESP liste « bibliographie puis annexes » mais précise que « selon les disciplines, les annexes peuvent être placées avant la bibliographie » — le gabarit ulthese retient annexes → bibliographie ; décision documentée). |
28 +| 8 | **Avant-propos obligatoire**, deux parties : (1) déclaration sur l'utilisation de l'IA générative (obligatoire pour tous depuis le 12 janvier 2026) ; (2) pour chaque article : état de préparation / date de soumission-acceptation-publication, modifications vs version publiée, statut d'auteur (premier auteur…), rôle exact de l'étudiant et des coauteurs. Position : après les remerciements, avant l'introduction. | `pages-liminaires/avant-propos.tex` structuré en ces deux parties, avec placeholders `[À COMPLÉTER]` pour les statuts de soumission (info absente du repo) et un texte pré-rempli pour les coauteurs (Gagnon, Power) et le rôle de l'auteur. |
29 +| 9 | **Chapitres-articles** : chaque article = un chapitre avec titre original, résumé français ≤ 150 mots (obligatoire), résumé dans la langue de rédaction (l'anglais), apparaissant dans la TdM. | Chaque `chapitreN.tex` commence par `\section{Résumé}` (français, traduit de l'abstract, marqué brouillon) et `\section{Abstract}` (abstract original de l'article), conformément au gabarit `chapitre1-articles.tex` d'ulthese. |
30 +| 10 | **Bibliographie unique** : depuis l'automne 2023, la FESP exige « une seule bibliographie regroupant l'ensemble des références du document » (guide ulthese, p. 8). Style bibliographique libre (« en fonction des normes en vigueur dans le domaine d'études »). | Bibliographie unique en fin de thèse : `bib/these.bib` consolidé et dédupliqué. Style **`chicago` + natbib auteur-année** — style dominant dans les chapitres (chap. 1 et 2 : chicago ; chap. 3 : apalike, converti — même famille auteur-année, différences purement typographiques). Les sections « References » internes des articles sont fusionnées dans la bibliographie unique. |
31 +| 11 | **Marges** : haut/bas 25 mm, gauche/droite 30 mm ; folio en bas au centre à 10 mm. **Interligne** 1,5 (simple pour TdM, listes, notes, bibliographie, citations longues). **Pagination** : liminaires en romains (résumé = ii), corps en arabes (introduction = 1), continue jusqu'aux annexes. | Tout est géré nativement par la classe `ulthese` (`\frontmatter`/`\mainmatter` de memoir). Aucun `geometry`, aucun `\pagenumbering`, aucun `\doublespacing` des articles n'est conservé (retirés des copies ; `geometry` est d'ailleurs déclaré incompatible par la classe). |
32 +| 12 | **Police** : LaTeX — Computer Modern/Latin Modern permis, ainsi que Times/Palatino (PSNFSS), Lucida, STIX Two. Corps 10–12 pt. | Police par défaut Latin Modern, 11 pt (défaut ulthese) — conforme. (Les packages `palatino`/`mathpazo`/`sourcesanspro` des articles ne sont pas repris, pour une typographie uniforme.) |
33 +| 13 | **Format de page** : lettre nord-américain. | Défaut d'ulthese (letterpaper). |
34 +| 14 | **Numérotation tableaux/figures par chapitre** (1.1, 2.3…). | Comportement natif de memoir/ulthese avec `\chapter`. Les hacks de numérotation manuelle du chap. 1 (`\thetable → 9(a)`…) sont neutralisés dans les copies (voir INVENTAIRE.md). |
35 +| 15 | **Compilation** : ulthese compatible pdfLaTeX, LuaLaTeX recommandé par la FESP. | `latexmk` (fichier `Makefile` + `.latexmkrc`) avec **pdfLaTeX** (choix : les sources des articles utilisent `inputenc`/`fontenc` et des packages pensés pour pdfLaTeX). Compile sans erreur, références et citations résolues. |
36 +| 16 | **Autorisation des coauteurs** : formulaire FES-100 par article au dépôt initial (sauf article publié avec autorisation de l'éditeur). | Hors document LaTeX — rappelé dans l'avant-propos en commentaire et dans la liste « points à vérifier » du rapport final. ⚠️ **3 formulaires FES-100 à prévoir** (articles coécrits avec M.-H. Gagnon et G. J. Power). |
37 +
38 +## Hypothèses retenues (exigence introuvable ou ambiguë)
39 +
40 +- **A1 — Intitulé du programme** : le libellé exact demandé (« … – finance ») n'existe pas au
41 + répertoire officiel ; retenu « Doctorat en sciences de l'administration -- finance et
42 + assurance ». À valider avec la direction de programme avant dépôt.
43 +- **A2 — Résumés d'articles ≤ 150 mots** : les résumés français des chapitres sont des
44 + traductions-condensés des abstracts originaux, rédigés comme **brouillons** ; l'auteur doit
45 + les réviser et vérifier la limite de 150 mots.
46 +- **A3 — Sections « References » des articles** : la FESP (tableau synthèse 2026) mentionne
47 + encore « Les références » comme élément du chapitre-article, mais le guide ulthese (règle
48 + post-2023) impose une bibliographie unique. **Retenu : bibliographie unique** (règle la plus
49 + récente et gabarit officiel `gabarit-doctorat-articles.tex`).
50 +- **A4 — Annexes d'articles** : les annexes du chap. 3 (preuves, résultats additionnels) et les
51 + annexes désactivées du chap. 1 (COVID/ZLB/full-sample, marquées « available upon request »
52 + dans la source) — retenu : annexes du chap. 3 intégrées comme annexes de thèse (A, B) ;
53 + annexes du chap. 1 **non intégrées** (elles sont désactivées dans la source la plus récente,
54 + décision de l'auteur respectée). Réactivables facilement (voir INVENTAIRE.md).
55 +- **A5 — Exigences FSA spécifiques** : aucune règle de présentation propre à la FSA ou au
56 + programme (au-delà des règles FESP) n'a été trouvée en ligne. Hypothèse : les règles FESP
57 + suffisent ; style de citation choisi selon la norme du domaine (auteur-année, style Chicago,
58 + usuel en finance).
59 +- **A6 — Déclaration IA générative** : la FESP ne fournit pas de texte type public ; un
60 + paragraphe placeholder `[À COMPLÉTER]` est proposé dans l'avant-propos, à adapter par
61 + l'auteur selon sa pratique réelle et les directives de sa direction.
added these-ulaval/annexes/.tex +0 −0
added these-ulaval/annexes/annexe-additionnel.tex +116 −0
@@ -0,0 +1,116 @@
1 +\chapter{Additional Tables and Figures (Chapter 3)}
2 +\label{ch3:sec:app_additional}
3 +
4 +This appendix collects supplementary tables and figures that support the main results but are not essential for following the core argument.
5 +
6 +\section{Descriptive Figures}
7 +
8 +\begin{figure}[H]
9 +\centering
10 +\includegraphics[width=0.85\textwidth]{figD1_return_density.pdf}
11 +\caption{Distribution of 1-minute log returns on FOMC days. Winsorized at 0.5\% tails for display. The heavy tails and leptokurtic shape are consistent with the Jarque--Bera test rejections reported in Table~\ref{ch3:tab:desc_1min_ret}.}
12 +\label{ch3:fig:D1}
13 +\end{figure}
14 +
15 +\begin{figure}[H]
16 +\centering
17 +\includegraphics[width=0.85\textwidth]{figD5_acf_returns.pdf}
18 +\caption{Autocorrelation of 1-minute returns. Grey band shows 95\% confidence interval. Negative first-order autocorrelation reflects bid--ask bounce effects typical of high-frequency data.}
19 +\label{ch3:fig:D5}
20 +\end{figure}
21 +
22 +\begin{figure}[H]
23 +\centering
24 +\includegraphics[width=0.85\textwidth]{figD8b_log_rv_distribution.pdf}
25 +\caption{Distribution of $\log(\mathrm{RV})$. Near-Gaussian shape validates the use of $\log(\mathrm{RV})$ as the dependent variable in the panel regressions.}
26 +\label{ch3:fig:D8b}
27 +\end{figure}
28 +
29 +\section{Ensemble Model Diagnostics}
30 +
31 +\begin{figure}[H]
32 +\centering
33 +\includegraphics[width=0.6\textwidth]{pca_axis_quality.png}
34 +\caption{PCA axis assignment quality. Bar height shows $|\text{Corr}|$ between PC projection and keyword differential. All axes exceed the 0.04 minimum separation threshold.}
35 +\label{ch3:fig:pca_quality}
36 +\end{figure}
37 +
38 +\begin{figure}[H]
39 +\centering
40 +\begin{subfigure}[b]{0.48\textwidth}
41 +\includegraphics[width=\textwidth]{novelty_comparison.png}
42 +\caption{Novelty: MiniLM vs BERT}
43 +\end{subfigure}
44 +\hfill
45 +\begin{subfigure}[b]{0.48\textwidth}
46 +\includegraphics[width=\textwidth]{scatter_comparison.png}
47 +\caption{Score scatter plots}
48 +\end{subfigure}
49 +\caption{Model comparison: MiniLM (384d) vs.\ BERT (768d). Novelty correlation $r = 0.72$; policy stance tone $r = 0.56$. The moderate inter-model correlation supports the use of an ensemble approach.}
50 +\label{ch3:fig:model_comparison}
51 +\end{figure}
52 +
53 +\begin{figure}[H]
54 +\centering
55 +\begin{subfigure}[b]{0.48\textwidth}
56 +\includegraphics[width=\textwidth]{tone_comparison.png}
57 +\caption{Tone comparison across models}
58 +\end{subfigure}
59 +\hfill
60 +\begin{subfigure}[b]{0.48\textwidth}
61 +\includegraphics[width=\textwidth]{confidence_distribution.png}
62 +\caption{Ensemble confidence distribution}
63 +\end{subfigure}
64 +\caption{Tone model comparison and confidence diagnostics. The right panel shows that ensemble confidence is concentrated above 0.7, indicating strong inter-model agreement for most statements.}
65 +\label{ch3:fig:tone_conf}
66 +\end{figure}
67 +
68 +\section{Additional Panel Regression Tables}
69 +
70 +Tables~\ref{ch3:tab:panel_rv_stance}--\ref{ch3:tab:panel_return_novelty} report panel minute-level regressions for the remaining dependent variables (realized volatility in levels, realized beta, and returns), separately for stance and novelty.
71 +
72 +\input{chapitre3/tables/tableP_panel_rv_stance.tex}
73 +\input{chapitre3/tables/tableP_panel_rv_novelty.tex}
74 +\input{chapitre3/tables/tableP_panel_beta_stance.tex}
75 +\input{chapitre3/tables/tableP_panel_beta_novelty.tex}
76 +\input{chapitre3/tables/tableP_panel_return_stance.tex}
77 +\input{chapitre3/tables/tableP_panel_return_novelty.tex}
78 +
79 +\section{Additional Event-Level Regression Tables}
80 +
81 +Tables~\ref{ch3:tab:rolling_rv_ratio_30min} and~\ref{ch3:tab:rolling_log_rv_ratio_30min} report the event-level RV-ratio specifications, which corroborate the pre/post difference results in the main text.
82 +
83 +\input{chapitre3/tables/tableR_rv_ratio_30min.tex}
84 +\input{chapitre3/tables/tableR_log_rv_ratio_30min.tex}
85 +
86 +\section{Additional IRF and Drift Figures}
87 +
88 +\begin{figure}[H]
89 +\centering
90 +\includegraphics[width=0.85\textwidth]{figR9_drift_by_novelty.pdf}
91 +\caption{Cumulative return drift by novelty tercile. Unlike stance-based drift (Figure~\ref{ch3:fig:R8}), novelty terciles show less directional separation, consistent with novelty affecting volatility rather than returns.}
92 +\label{ch3:fig:R9}
93 +\end{figure}
94 +
95 +\begin{figure}[H]
96 +\centering
97 +\includegraphics[width=0.85\textwidth]{figR10_irf_ret_h_novelty.pdf}
98 +\caption{Impulse response: cumulative return to novelty. Effects are generally smaller and less significant than stance effects (Figure~\ref{ch3:fig:R10_stance}), supporting the hypothesis that novelty operates through volatility rather than returns.}
99 +\label{ch3:fig:R10_novelty}
100 +\end{figure}
101 +
102 +\begin{figure}[H]
103 +\centering
104 +\includegraphics[width=0.85\textwidth]{figR10b_irf_placebo_novelty.pdf}
105 +\caption{Pre-announcement placebo: IRF for novelty. All coefficients are near zero, corroborating the placebo results for stance (Figure~\ref{ch3:fig:placebo_stance}).}
106 +\label{ch3:fig:placebo_novelty}
107 +\end{figure}
108 +
109 +\section{Sub-Period Stability}
110 +
111 +\begin{figure}[H]
112 +\centering
113 +\includegraphics[width=0.85\textwidth]{figRob_subperiod_heatmap.pdf}
114 +\caption{Sub-period stability: regression coefficients across six Fed policy regimes. Coefficient signs are generally consistent across sub-periods, though magnitudes vary with the degree of policy uncertainty in each regime.}
115 +\label{ch3:fig:rob_subperiod}
116 +\end{figure}
added these-ulaval/annexes/annexe-preuves.tex +105 −0
@@ -0,0 +1,105 @@
1 +\chapter{Mathematical Proofs (Chapter 3)}
2 +\label{ch3:sec:proofs}
3 +
4 +This appendix provides formal mathematical derivations for the key theoretical relationships presented in Section~\ref{ch3:sec:methodology}.
5 +
6 +\section{Derivation of Pure Stance Definitions}
7 +
8 +\begin{theorem}[Pure Stance Characterization]
9 +The pure dovish and hawkish stance measures satisfy the relationships given in Definition~\ref{ch3:def:pure_stances}.
10 +\end{theorem}
11 +
12 +\begin{proof}
13 +\textbf{Case 1: Pure Dovish Statement.} Assume the Federal Reserve releases a statement that exactly matches the dovish counterfactual: $F_t = F_t^D$.
14 +
15 +The tone measure becomes:
16 +\begin{align}
17 +\text{Tone}_t &= \frac{\text{sim}(F_t, F_t^H) - \text{sim}(F_t, F_t^D)}{1 - \text{sim}(F_t^D, F_t^H)} \notag \\
18 +&= \frac{\text{sim}(F_t^D, F_t^H) - \text{sim}(F_t^D, F_t^D)}{1 - \text{sim}(F_t^D, F_t^H)} \notag \\
19 +&= \frac{\text{sim}(F_t^D, F_t^H) - 1}{1 - \text{sim}(F_t^D, F_t^H)} \notag \\
20 +&= -1
21 +\end{align}
22 +
23 +The novelty measure is:
24 +\begin{equation}
25 +\text{Novelty}_t = 1 - \text{sim}(F_t, F_{t-1}) = 1 - \text{sim}(F_t^D, F_{t-1})
26 +\end{equation}
27 +
28 +Therefore, the stance measure is:
29 +\begin{align}
30 +\text{Stance}_t &= \text{Novelty}_t \times \text{Tone}_t \notag \\
31 +&= \left(1 - \text{sim}(F_t^D, F_{t-1})\right) \times (-1) \notag \\
32 +&= -\left(1 - \text{sim}(F_t^D, F_{t-1})\right) \notag \\
33 +&= \text{Stance}_t^{dove}
34 +\end{align}
35 +
36 +\textbf{Case 2: Pure Hawkish Statement.} Assume $F_t = F_t^H$:
37 +
38 +The tone measure becomes:
39 +\begin{align}
40 +\text{Tone}_t &= \frac{\text{sim}(F_t^H, F_t^H) - \text{sim}(F_t^H, F_t^D)}{1 - \text{sim}(F_t^D, F_t^H)} \notag \\
41 +&= \frac{1 - \text{sim}(F_t^H, F_t^D)}{1 - \text{sim}(F_t^D, F_t^H)} \notag \\
42 +&= 1
43 +\end{align}
44 +
45 +Therefore:
46 +\begin{align}
47 +\text{Stance}_t &= \left(1 - \text{sim}(F_t^H, F_{t-1})\right) \times 1 = \text{Stance}_t^{hawk}
48 +\end{align}
49 +
50 +This completes the proof. \hfill $\square$
51 +\end{proof}
52 +
53 +\section{Derivation of Dovish Weight Parameter}
54 +
55 +\begin{theorem}[Dovish Weight Parameter Formula]
56 +The weight parameter $w_t$ in the weighted stance representation has the form given in Definition~\ref{ch3:def:weighted_stance}.
57 +\end{theorem}
58 +
59 +\begin{proof}
60 +From the weighted stance equation:
61 +\begin{equation}
62 +\text{Stance}_t = w_t \cdot \text{Stance}_t^{dove} + (1 - w_t) \cdot \text{Stance}_t^{hawk}
63 +\end{equation}
64 +
65 +Substituting the expressions for pure stances from Theorem 1:
66 +\begin{align}
67 +\text{Stance}_t &= -w_t\left(1 - \text{sim}(F_t^D, F_{t-1})\right) + (1 - w_t)\left(1 - \text{sim}(F_t^H, F_{t-1})\right) \notag \\
68 +&= 1 - 2w_t + w_t\left(\text{sim}(F_t^D, F_{t-1}) + \text{sim}(F_t^H, F_{t-1})\right) - \text{sim}(F_t^H, F_{t-1})
69 +\end{align}
70 +
71 +Collecting the terms in $w_t$, this reads
72 +\begin{equation}
73 +\text{Stance}_t = 1 - \text{sim}(F_t^H, F_{t-1}) - w_t\left(2 - \text{sim}(F_t^D, F_{t-1}) - \text{sim}(F_t^H, F_{t-1})\right).
74 +\end{equation}
75 +Equating with $\text{Stance}_t = \left(1 - \text{sim}(F_t, F_{t-1})\right) \times \text{Tone}_t$ and solving for $w_t$:
76 +\begin{equation}
77 +w_t = \frac{1 - \text{sim}(F_t^H, F_{t-1}) - \left(1 - \text{sim}(F_t, F_{t-1})\right) \times \text{Tone}_t}{2 - \text{sim}(F_t^D, F_{t-1}) - \text{sim}(F_t^H, F_{t-1})}
78 +\end{equation}
79 +
80 +This establishes the formula. \hfill $\square$
81 +\end{proof}
82 +
83 +\section{Proof of MPS Decomposition}
84 +
85 +\begin{theorem}[Policy Stance Surprise Decomposition]
86 +The policy stance surprise admits the decomposition given in Proposition~\ref{ch3:prop:mps}.
87 +\end{theorem}
88 +
89 +\begin{proof}
90 +From the definitions:
91 +\begin{align}
92 +\text{MPS}_t &= \text{Stance}_t - \mathbb{E}_{t-\Delta}[\text{Stance}_t] \notag \\
93 +&= \text{Novelty}_t \times \text{Tone}_t - (1 - 2p_{t-\Delta}) \cdot \overline{\text{Novelty}}_{t \mid t-\Delta} \notag \\
94 +&= \left(\overline{\text{Novelty}}_{t \mid t-\Delta} + \varepsilon_t\right) \times \text{Tone}_t - (1 - 2p_{t-\Delta}) \cdot \overline{\text{Novelty}}_{t \mid t-\Delta} \notag \\
95 +&= \overline{\text{Novelty}}_{t \mid t-\Delta} \times \text{Tone}_t - \overline{\text{Novelty}}_{t \mid t-\Delta} + 2p_{t-\Delta} \cdot \overline{\text{Novelty}}_{t \mid t-\Delta} + \varepsilon_t \times \text{Tone}_t \notag \\
96 +&= \overline{\text{Novelty}}_{t \mid t-\Delta}(\text{Tone}_t + 2p_{t-\Delta} - 1) + \text{Tone}_t \cdot \varepsilon_t
97 +\end{align}
98 +
99 +This establishes the decomposition. \hfill $\square$
100 +\end{proof}
101 +
102 +\section{Economic Interpretation}
103 +
104 +The mathematical results provide several economic insights. First, the pure stance characterization shows that our measures correctly identify extreme policy communications, with dovish statements receiving negative stance values and hawkish statements receiving positive values. Second, the weight parameter derivation reveals how actual policy communications can be understood as weighted averages of extreme alternatives. Third, the MPS decomposition shows that policy surprises have two distinct sources: unexpected tone conditional on expected information content, and unexpected information content weighted by actual tone.
105 +
added these-ulaval/bib/these.bib +2851 −0
@@ -0,0 +1,2851 @@
1 +% Bibliographie consolidée de la thèse — fusion des master.bib des trois chapitres.
2 +% Dédupliquée par clé (variante la plus complète conservée). Entrée Roll1984 ajoutée
3 +% (la clé citée « Roll1984 » correspond à l'entrée « roll1984 » — BibTeX est
4 +% insensible à la casse des clés). Voir INVENTAIRE.md.
5 +
6 +@article{10.1257/aer.20190964,
7 +Author = {Känzig, Diego R.},
8 +Title = {The Macroeconomic Effects of Oil Supply News: Evidence from {OPEC} Announcements},
9 +Journal = {American Economic Review},
10 +Volume = {111},
11 +Number = {4},
12 +Year = {2021},
13 +Month = {April},
14 +Pages = {1092-1125},
15 +DOI = {10.1257/aer.20190964},
16 +URL = {https://www.aeaweb.org/articles?id=10.1257/aer.20190964}}
17 +
18 +@article{ackert2000arbitrage,
19 + title={Arbitrage and valuation in the market for {S}tandard \& {P}oor's depositary receipts},
20 + author={Ackert, Lucy F and Tian, Yisong S},
21 + journal={Financial Management},
22 + volume={29},
23 + number={3},
24 + pages={71--87},
25 + year={2000},
26 + publisher={Wiley}
27 +}
28 +
29 +@article{agarwal2018etfs,
30 + title={Do {ETFs} increase the commonality in liquidity of underlying stocks?},
31 + author={Agarwal, Vikas and Hanouna, Paul and Moussawi, Rabih and Stahel, Christof W},
32 + journal={Journal of Financial Economics},
33 + volume={127},
34 + number={1},
35 + pages={67--90},
36 + year={2018},
37 + publisher={Elsevier}
38 +}
39 +
40 +@article{almutairi2021value,
41 + author={Almutairi, Hossa and Pierru, Axel and Smith, James A},
42 + doi={10.1111/opec.12199},
43 + title={The Value of OPEC’s Spare Capacity to the Oil Market and Global Economy},
44 + journal={OPEC Energy Review},
45 + year={2021}
46 +}
47 +
48 +@article{alquist2013role,
49 + title = {The Role of Financial Speculation in Driving the Price of Crude Oil},
50 + author = {Alquist, Ron and Gervais, Olivier},
51 + journal = {The Energy Journal},
52 + volume = {34},
53 + number = {3},
54 + pages = {35--54},
55 + year = {2013},
56 + publisher = {International Association for Energy Economics}
57 +}
58 +
59 +@article{andersen1997,
60 + title={Intraday Periodicity and Volatility Persistence in Financial Markets},
61 + author={Andersen, Torben G. and Bollerslev, Tim},
62 + journal={Journal of Empirical Finance},
63 + volume={4},
64 + number={2--3},
65 + pages={115--158},
66 + year={1997}
67 +}
68 +
69 +@article{andersen1998deutsche,
70 + title = {Deutsche Mark--Dollar Volatility: Intraday Activity Patterns, Macroeconomic Announcements, and Longer Run Dependencies},
71 + author = {Andersen, Torben G. and Bollerslev, Tim},
72 + journal = {Journal of Finance},
73 + volume = {53},
74 + number = {1},
75 + pages = {219--265},
76 + year = {1998},
77 + publisher = {Wiley Online Library}
78 +}
79 +
80 +@article{andersen2001distribution,
81 + title = {The Distribution of Realized Stock Return Volatility},
82 + author = {Andersen, Torben G. and Bollerslev, Tim and Diebold, Francis X. and Ebens, Heiko},
83 + journal = {Journal of Financial Economics},
84 + volume = {61},
85 + number = {1},
86 + pages = {43--76},
87 + year = {2001},
88 + publisher = {Elsevier}
89 +}
90 +
91 +@article{andersen2003,
92 + title={Micro Effects of Macro Announcements: Real-Time Price Discovery in Foreign Exchange},
93 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X and Vega, Clara},
94 + journal={American Economic Review},
95 + volume={93},
96 + number={1},
97 + pages={38--62},
98 + year={2003},
99 + publisher={American Economic Association}
100 +}
101 +
102 +@article{andersen2003micro,
103 + title={Micro effects of macro announcements: Real-time price discovery in foreign exchange},
104 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X and Vega, Clara},
105 + journal={American Economic Review},
106 + volume={93},
107 + number={1},
108 + pages={38--62},
109 + year={2003}
110 +}
111 +
112 +@article{andersen2007,
113 + title={Real-Time Price Discovery in Global Stock, Bond and Foreign Exchange Markets},
114 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X and Vega, Clara},
115 + journal={Journal of International Economics},
116 + volume={73},
117 + number={2},
118 + pages={251--277},
119 + year={2007},
120 + publisher={Elsevier}
121 +}
122 +
123 +@article{andersen2007real,
124 + title = {Real-Time Price Discovery in Global Stock, Bond and Foreign Exchange Markets},
125 + author = {Andersen, Torben G. and Bollerslev, Tim and Diebold, Francis X. and Vega, Clara},
126 + journal = {Journal of International Economics},
127 + volume = {73},
128 + number = {2},
129 + pages = {251--277},
130 + year = {2007},
131 + publisher = {Elsevier}
132 +}
133 +
134 +@article{andersen2007roughing,
135 + title={Roughing it up: Including jump components in the measurement, modeling, and forecasting of return volatility},
136 + author={Andersen, Torben G and Bollerslev, Tim and Diebold, Francis X},
137 + journal={The Review of Economics and Statistics},
138 + volume={89},
139 + number={4},
140 + pages={701--720},
141 + year={2007},
142 + publisher={MIT Press}
143 +}
144 +
145 +@article{andersen2008realized,
146 + title={Realized Volatility, Working Paper 2008-14},
147 + author={Andersen, Torben Gustav and Benzoni, Luca},
148 + year={2008}
149 +}
150 +
151 +@techreport{apel2012,
152 + title={The Information Content of Central Bank Minutes},
153 + author={Apel, Mikael and Grimaldi, Marianna Blix},
154 + institution={Sveriges Riksbank},
155 + type={Working Paper Series},
156 + number={261},
157 + year={2012}
158 +}
159 +
160 +@article{araci2019,
161 + title={FinBERT: Financial Sentiment Analysis with Pre-trained Language Models},
162 + author={Araci, Dogu},
163 + journal={arXiv preprint arXiv:1908.10063},
164 + year={2019}
165 +}
166 +
167 +@article{aruoba2009real,
168 + title = {Real-Time Measurement of Business Conditions},
169 + author = {Aruoba, S. Borağan and Diebold, Francis X. and Scotti, Chiara},
170 + journal = {Journal of Business \& Economic Statistics},
171 + volume = {27},
172 + number = {4},
173 + pages = {417-427},
174 + year = {2009},
175 + publisher = {Taylor \& Francis},
176 + doi = {10.1198/jbes.2009.07205},
177 + url = {https://doi.org/10.1198/jbes.2009.07205}
178 +}
179 +
180 +@inproceedings{aulerich2012bubbles,
181 + title = {Bubbles},
182 + author = {Aulerich, Nicole M. and Irwin, Scott H. and Garcia, Philip},
183 + booktitle = {Food Prices, and Speculation: Evidence from the CFTC’s Daily Large Trader Data Files, Paper Prepared for Presentation at the NBER Conference on “Economics of Food Price Volatility” in Seattle, WA},
184 + year = {2012}
185 +}
186 +
187 +@article{bai2003computation,
188 + title={Computation and analysis of multiple structural change models},
189 + author={Bai, Jushan and Perron, Pierre},
190 + journal={Journal of Applied Econometrics},
191 + volume={18},
192 + number={1},
193 + pages={1--22},
194 + year={2003},
195 + publisher={Wiley}
196 +}
197 +
198 +@article{baker2016,
199 + title={Measuring Economic Policy Uncertainty},
200 + author={Baker, Scott R. and Bloom, Nicholas and Davis, Steven J.},
201 + journal={Quarterly Journal of Economics},
202 + volume={131},
203 + number={4},
204 + pages={1593--1636},
205 + year={2016},
206 + publisher={Oxford University Press}
207 +}
208 +
209 +@article{balduzzi2001,
210 + title={Economic News and Bond Prices: Evidence from the US Treasury Market},
211 + author={Balduzzi, Pierluigi and Elton, Edwin J and Green, T Clifton},
212 + journal={Journal of Financial and Quantitative Analysis},
213 + volume={36},
214 + number={4},
215 + pages={523--543},
216 + year={2001},
217 + publisher={Cambridge University Press}
218 +}
219 +
220 +@article{balduzzi2001economic,
221 + title = {Economic News and Bond Prices: Evidence from the US Treasury Market},
222 + author = {Balduzzi, Pierluigi and Elton, Edwin J. and Green, T. Clifton},
223 + journal = {Journal of Financial and Quantitative Analysis},
224 + volume = {36},
225 + number = {4},
226 + pages = {523--543},
227 + year = {2001},
228 + publisher = {JSTOR}
229 +}
230 +
231 +@article{barndorff2002,
232 + title={Econometric Analysis of Realized Volatility and its Use in Estimating Stochastic Volatility Models},
233 + author={Barndorff-Nielsen, Ole E and Shephard, Neil},
234 + journal={Journal of the Royal Statistical Society: Series B (Statistical Methodology)},
235 + volume={64},
236 + number={2},
237 + pages={253--280},
238 + year={2002},
239 + publisher={Wiley Online Library}
240 +}
241 +
242 +@article{barndorff2002econometric,
243 + title={Econometric analysis of realized volatility and its use in estimating stochastic volatility models},
244 + author={Barndorff-Nielsen, Ole E and Shephard, Neil},
245 + journal={Journal of the Royal Statistical Society: Series B},
246 + volume={64},
247 + number={2},
248 + pages={253--280},
249 + year={2002},
250 + publisher={Wiley}
251 +}
252 +
253 +@article{barndorff2002estimating,
254 + title = {Estimating Quadratic Variation Using Realized Variance},
255 + author = {Barndorff-Nielsen, Ole E. and Shephard, Neil},
256 + journal = {Journal of Applied Econometrics},
257 + volume = {17},
258 + number = {5},
259 + pages = {457--477},
260 + year = {2002},
261 + publisher = {Wiley Online Library}
262 +}
263 +
264 +@article{barndorff2004power,
265 + title={Power and bipower variation with stochastic volatility and jumps},
266 + author={Barndorff-Nielsen, Ole E and Shephard, Neil},
267 + journal={Journal of Financial Econometrics},
268 + volume={2},
269 + number={1},
270 + pages={1--37},
271 + year={2004},
272 + publisher={Oxford University Press}
273 +}
274 +
275 +@article{barndorff2009realized,
276 + title={Realized kernels in practice: Trades and quotes},
277 + author={Barndorff-Nielsen, Ole E and Hansen, Peter Reinhard and Lunde, Asger and Shephard, Neil},
278 + journal={The Econometrics Journal},
279 + volume={12},
280 + number={3},
281 + pages={C1--C32},
282 + year={2009},
283 + publisher={Oxford University Press}
284 +}
285 +
286 +@article{barunik2018measuring,
287 + title={Measuring the frequency dynamics of financial connectedness and systemic risk},
288 + author={Baruník, Jozef and Křehlík, Tomáš},
289 + journal={Journal of Financial Econometrics},
290 + volume={16},
291 + number={2},
292 + pages={271--296},
293 + year={2018},
294 + publisher={Oxford University Press}
295 +}
296 +
297 +@article{basak2016model,
298 + title = {A Model of Financialization of Commodities},
299 + author = {Basak, Suleyman and Pavlova, Anna},
300 + journal = {Journal of Finance},
301 + volume = {71},
302 + number = {4},
303 + pages = {1511--1556},
304 + year = {2016},
305 + publisher = {Wiley Online Library}
306 +}
307 +
308 +@article{baumeister2014oil,
309 + title={Do oil price increases cause higher food prices?},
310 + author={Baumeister, Christiane and Kilian, Lutz},
311 + journal={Economic Policy},
312 + volume={29},
313 + number={80},
314 + pages={691--747},
315 + year={2014},
316 + publisher={Oxford University Press}
317 +}
318 +
319 +@article{baur2010gold,
320 + title = {Is Gold a Hedge or a Safe Haven? {An} Analysis of Stocks, Bonds and Gold},
321 + author = {Baur, Dirk G. and Lucey, Brian M.},
322 + journal = {Financial Review},
323 + volume = {45},
324 + number = {2},
325 + pages = {217--229},
326 + year = {2010},
327 + publisher = {Wiley Online Library}
328 +}
329 +
330 +@article{beechey2009high,
331 + title={The high-frequency impact of news on long-term yields and forward rates: Is it real?},
332 + author={Beechey, Meredith J and Wright, Jonathan H},
333 + journal={Journal of Monetary Economics},
334 + volume={56},
335 + number={4},
336 + pages={535--544},
337 + year={2009},
338 + publisher={Elsevier}
339 +}
340 +
341 +@article{ben2017etfs,
342 + title={Do {ETFs} increase volatility?},
343 + author={Ben-David, Itzhak and Franzoni, Francesco and Moussawi, Rabih},
344 + journal={The Journal of Finance},
345 + volume={73},
346 + number={6},
347 + pages={2471--2535},
348 + year={2018},
349 + publisher={Wiley}
350 +}
351 +
352 +@article{ben2018etfs,
353 + title={Do {ETFs} increase volatility?},
354 + author={Ben-David, Itzhak and Franzoni, Francesco and Moussawi, Rabih},
355 + journal={The Journal of Finance},
356 + volume={73},
357 + number={6},
358 + pages={2471--2535},
359 + year={2018},
360 + publisher={Wiley}
361 +}
362 +
363 +@article{benjamini1995controlling,
364 + title={Controlling the false discovery rate: a practical and powerful approach to multiple testing},
365 + author={Benjamini, Yoav and Hochberg, Yosef},
366 + journal={Journal of the Royal Statistical Society: Series B (Methodological)},
367 + volume={57},
368 + number={1},
369 + pages={289--300},
370 + year={1995},
371 + publisher={Wiley Online Library}
372 + }
373 +
374 +@article{bernanke2005,
375 + title={What Explains the Stock Market's Reaction to Federal Reserve Policy?},
376 + author={Bernanke, Ben S and Kuttner, Kenneth N},
377 + journal={Journal of Finance},
378 + volume={60},
379 + number={3},
380 + pages={1221--1257},
381 + year={2005},
382 + publisher={Wiley Online Library}
383 +}
384 +
385 +@article{bernanke2005explains,
386 + title={What explains the stock market's reaction to Federal Reserve policy?},
387 + author={Bernanke, Ben S and Kuttner, Kenneth N},
388 + journal={Journal of Finance},
389 + volume={60},
390 + number={3},
391 + pages={1221--1257},
392 + year={2005},
393 + publisher={Wiley Online Library}
394 +}
395 +
396 +@article{bernile2016can,
397 + title = {Can Information Be Locked Up? Informed Trading Ahead of Macro-News Announcements},
398 + author = {Bernile, Gennaro and Hu, Jianfeng and Tang, Yuehua},
399 + journal = {Journal of Financial Economics},
400 + volume = {121},
401 + number = {3},
402 + pages = {496--520},
403 + year = {2016},
404 + publisher = {Elsevier}
405 +}
406 +
407 +@article{bligh2008,
408 + title={The Decline of Symbolic Politics: Federal Reserve Communications},
409 + author={Bligh, Michelle C and Hess, Gregory D},
410 + journal={Journal of Economic Psychology},
411 + volume={29},
412 + number={4},
413 + pages={535--559},
414 + year={2008},
415 + publisher={Elsevier}
416 +}
417 +
418 +@article{blinder2008,
419 + title={Central Bank Communication and Monetary Policy: A Survey of Theory and Evidence},
420 + author={Blinder, Alan S and Ehrmann, Michael and Fratzscher, Marcel and De Haan, Jakob and Jansen, David-Jan},
421 + journal={Journal of Economic Literature},
422 + volume={46},
423 + number={4},
424 + pages={910--945},
425 + year={2008},
426 + publisher={American Economic Association}
427 +}
428 +
429 +@article{boehm2020us,
430 + title={The {US}, economic news, and the global financial cycle},
431 + author={Boehm, Christoph and Kroner, Niklas},
432 + journal={Economic News, and the Global Financial Cycle (January 10, 2020)},
433 + year={2020}
434 +}
435 +
436 +@article{bohl2012commodity,
437 + title={Do commodity index traders destabilize agricultural futures prices?},
438 + author={Bohl, Martin T and Javed, Farrukh and Stephan, Patrick M and others},
439 + journal={Diskussionspapier, am},
440 + volume={24},
441 + year={2012},
442 + publisher={Citeseer}
443 +}
444 +
445 +@article{bohl2013does,
446 + title={Does futures speculation destabilize spot prices? New evidence for commodity markets},
447 + author={Bohl, Martin T and Stephan, Patrick M},
448 + journal={Journal of Agricultural and Applied Economics},
449 + volume={45},
450 + number={4},
451 + pages={595--616},
452 + year={2013},
453 + publisher={Cambridge University Press}
454 +}
455 +
456 +@article{bollerslev1986generalized,
457 + title={Generalized autoregressive conditional heteroskedasticity},
458 + author={Bollerslev, Tim},
459 + journal={Journal of Econometrics},
460 + volume={31},
461 + number={3},
462 + pages={307--327},
463 + year={1986},
464 + publisher={Elsevier}
465 +}
466 +
467 +@article{boyd2005stock,
468 + title={The stock market's reaction to unemployment news: Why bad news is usually good for stocks},
469 + author={Boyd, John H and Hu, Jian and Jagannathan, Ravi},
470 + journal={Journal of Finance},
471 + volume={60},
472 + number={2},
473 + pages={649--672},
474 + year={2005},
475 + publisher={Wiley Online Library}
476 +}
477 +
478 +@article{boyd2018update,
479 + title={An update on speculation and financialization in commodity markets},
480 + author={Boyd, Naomi E and Harris, Jeffrey H and Li, Bingxin},
481 + journal={Journal of Commodity Markets},
482 + volume={10},
483 + pages={91--104},
484 + year={2018},
485 + publisher={Elsevier}
486 +}
487 +
488 +@article{bredin2015does,
489 + title = {Does Gold Glitter in the Long-Run? Gold as a Hedge and Safe Haven Across Time and Investment Horizon},
490 + author = {Bredin, Don and Conlon, Thomas and Pot{\`\i}, Valerio},
491 + journal = {International Review of Financial Analysis},
492 + volume = {41},
493 + pages = {320--328},
494 + year = {2015},
495 + publisher = {Elsevier}
496 +}
497 +
498 +@incollection{brennan1976supply,
499 + title = {The Supply of Storage},
500 + author = {Brennan, Michael J.},
501 + booktitle = {The Economics of Futures Trading},
502 + pages = {100--107},
503 + year = {1976},
504 + publisher = {Springer}
505 +}
506 +
507 +@article{brennan1991price,
508 + title={The price of convenience and the valuation of commodity contingent claims,[w:] D. Land, B. Oeksendal},
509 + author={Brennan, MJ},
510 + journal={Stochastic Models and Options Values, Elsevier Science Publications},
511 + year={1991}
512 +}
513 +
514 +@article{broman2016liquidity,
515 + title={Liquidity, style investing and excess comovement of exchange-traded fund returns},
516 + author={Broman, Markus S},
517 + journal={Journal of Financial Markets},
518 + volume={30},
519 + pages={27--53},
520 + year={2016},
521 + publisher={Elsevier}
522 +}
523 +
524 +@article{brownlees2014practical,
525 + title={Financial econometric analysis at ultra-high frequency: Data handling concerns},
526 + author={Brownlees, Christian T and Gallo, Giampiero M},
527 + journal={Computational Statistics \& Data Analysis},
528 + volume={51},
529 + number={4},
530 + pages={2232--2245},
531 + year={2006},
532 + publisher={Elsevier}
533 +}
534 +
535 +@article{brunetti2009speculation,
536 + title = {Is Speculation Destabilizing?},
537 + author = {Brunetti, Celso and B{\"u}y{\"u}k{\c{s}}ahin, Bahattin},
538 + journal = {SSRN eLibrary},
539 + year = {2009},
540 + publisher = {Citeseer}
541 +}
542 +
543 +@article{brunetti2011speculators,
544 + author = {Brunetti, Celso and B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Harris, Jeffrey},
545 + title = {Speculators, Prices and Market Volatility},
546 + journal = {SSRN Electronic Journal},
547 + year = {2011},
548 + doi = {10.2139/ssrn.1736737}
549 +}
550 +
551 +@article{brunetti2014commodity,
552 + title = {Commodity Index Trading and Hedging Costs},
553 + author = {Brunetti, Celso and Reiffen, David},
554 + journal = {Journal of Financial Markets},
555 + volume = {21},
556 + pages = {153--180},
557 + year = {2014},
558 + publisher = {Elsevier}
559 +}
560 +
561 +@article{brunetti2016speculators,
562 + title = {Speculators, Prices, and Market Volatility},
563 + author = {Brunetti, Celso and B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Harris, Jeffrey H.},
564 + journal = {Journal of Financial and Quantitative Analysis},
565 + volume = {51},
566 + number = {5},
567 + pages = {1545--1574},
568 + year = {2016},
569 + doi = {10.1017/s0022109016000569},
570 + publisher = {JSTOR}
571 +}
572 +
573 +@article{brusa2015,
574 + title={Asset Allocation and FOMC Announcements},
575 + author={Brusa, Francesca and Savor, Pavel and Wilson, Mungo},
576 + journal={Review of Financial Studies},
577 + volume={28},
578 + number={5},
579 + pages={1398--1446},
580 + year={2015},
581 + publisher={Oxford University Press}
582 +}
583 +
584 +@article{brusa2019,
585 + title={FOMC Announcements and Market Returns: Evidence from the Options Market},
586 + author={Brusa, Francesca and Savor, Pavel and Wilson, Mungo},
587 + journal={Journal of Finance},
588 + volume={75},
589 + number={1},
590 + pages={399--441},
591 + year={2019},
592 + publisher={Wiley Online Library}
593 +}
594 +
595 +@article{bryan1993consumer,
596 + title={The Consumer Price Index as a measure of inflation},
597 + author={Bryan, Michael F and Cecchetti, Stephen G},
598 + journal={Economic Review-Federal Reserve Bank of Cleveland},
599 + volume={29},
600 + number={4},
601 + pages={15},
602 + year={1993},
603 + publisher={Federal Reserve Bank of Cleveland}
604 +}
605 +
606 +@article{bryant2006causality,
607 + title = {Causality in Futures Markets},
608 + author = {Bryant, Henry L. and Bessler, David A. and Haigh, Michael S.},
609 + journal = {Journal of Futures Markets},
610 + volume = {26},
611 + number = {11},
612 + pages = {1039--1057},
613 + year = {2006},
614 + publisher = {Wiley Online Library}
615 +}
616 +
617 +@article{bubak2011volatility,
618 + title={Volatility transmission in emerging European foreign exchange markets},
619 + author={Bub{\'a}k, Vít and Kočenda, Evžen and Žikeš, Filip},
620 + journal={Journal of Banking \& Finance},
621 + volume={35},
622 + number={11},
623 + pages={2829--2841},
624 + year={2011},
625 + publisher={Elsevier}
626 +}
627 +
628 +@article{buckle2018price,
629 + title={Price discovery in the {ETF} market: the role of market makers},
630 + author={Buckle, David J and Curran, Edward and McGroarty, Frank},
631 + journal={Journal of Banking \& Finance},
632 + volume={93},
633 + pages={32--45},
634 + year={2018},
635 + publisher={Elsevier}
636 +}
637 +
638 +@article{buyukcsahin2010matters,
639 + title = {It Matters Who Trades: Hedge Funds, Swap Dealers, and Cross-Market Linkages},
640 + author = {B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Robe, Michel A.},
641 + journal = {SSRN eLibrary},
642 + year = {2010}
643 +}
644 +
645 +@article{buyukcsahin2011speculators,
646 + title={Do speculators drive crude oil futures prices?},
647 + author={B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Harris, Jeffrey H},
648 + journal={The Energy Journal},
649 + volume={32},
650 + number={2},
651 + pages={167--202},
652 + year={2011},
653 + publisher={SAGE Publications Sage CA: Los Angeles, CA}
654 +}
655 +
656 +@article{buyukcsahin2014speculators,
657 + title={Speculators, commodities and cross-market linkages},
658 + author={B{\"u}y{\"u}k{\c{s}}ahin, Bahattin and Robe, Michel A},
659 + journal={Journal of International Money and Finance},
660 + volume={42},
661 + pages={38--70},
662 + year={2014},
663 + publisher={Elsevier}
664 +}
665 +
666 +@article{buyuksahin2010fundamentals,
667 + title={Fundamentals, trader activity and derivative pricing},
668 + author={Buyuksahin, Bahattin and Robe, Michel A},
669 + journal={Journal of Futures Markets},
670 + volume={30},
671 + number={4},
672 + pages={317--341},
673 + year={2010},
674 + publisher={Wiley}
675 +}
676 +
677 +@article{buyuksahin2014speculation,
678 + title={Speculation, commodities and cross-market linkages},
679 + author={Buyuksahin, Bahattin and Robe, Michel A},
680 + journal={Journal of International Money and Finance},
681 + volume={42},
682 + pages={38--70},
683 + year={2014},
684 + publisher={Elsevier}
685 +}
686 +
687 +@article{cai2001moves,
688 + title = {What Moves the Gold Market?},
689 + author = {Cai, Jun and Cheung, Yan-Leung and Wong, Michael C. S.},
690 + journal = {Journal of Futures Markets},
691 + volume = {21},
692 + number = {3},
693 + pages = {257--278},
694 + year = {2001},
695 + publisher = {Wiley Online Library}
696 +}
697 +
698 +@article{campbell2012,
699 + title={Macroeconomic Effects of Federal Reserve Forward Guidance},
700 + author={Campbell, Jeffrey R and Evans, Charles L and Fisher, Jonas DM and Justiniano, Alejandro},
701 + journal={Brookings Papers on Economic Activity},
702 + volume={2012},
703 + number={1},
704 + pages={1--80},
705 + year={2012},
706 + publisher={Brookings Institution Press}
707 +}
708 +
709 +@article{cao2024us,
710 + title={U.S. monetary policy: The pushing hands of crude oil price?},
711 + author={Cao, Fangzhi and Su, Chi-Wei and Sun, Dian and Qin, Meng and Umar, Muhammad},
712 + journal={Energy Economics},
713 + volume={134},
714 + pages={107555},
715 + year={2024},
716 + publisher={Elsevier},
717 + doi={10.1016/j.eneco.2024.107555}
718 +}
719 +
720 +@article{carriero2015forecasting,
721 + title={Forecasting exchange rates with a large {B}ayesian {VAR}},
722 + author={Carriero, Andrea and Kapetanios, George and Marcellino, Massimiliano},
723 + journal={International Journal of Forecasting},
724 + volume={25},
725 + number={2},
726 + pages={400--417},
727 + year={2009},
728 + publisher={Elsevier}
729 +}
730 +
731 +@article{chang1997interday,
732 + title = {Interday Variations in Volume, Variance and Participation of Large Speculators},
733 + author = {Chang, Eric C. and Pinegar, J. Michael and Schachter, Barry},
734 + journal = {Journal of Banking \& Finance},
735 + volume = {21},
736 + number = {6},
737 + pages = {797--810},
738 + year = {1997},
739 + publisher = {Elsevier}
740 +}
741 +
742 +@techreport{cheng2012convective,
743 + title = {Convective Risk Flows in Commodity Futures Markets},
744 + author = {Cheng, Ing-Haw and Kirilenko, Andrei and Xiong, Wei},
745 + year = {2012},
746 + institution = {National Bureau of Economic Research}
747 +}
748 +
749 +@article{cheng2014financialization,
750 + title = {Financialization of Commodity Markets},
751 + author = {Cheng, Ing-Haw and Xiong, Wei},
752 + journal = {Annual Review of Financial Economics},
753 + volume = {6},
754 + number = {1},
755 + pages = {419--441},
756 + year = {2014},
757 + publisher = {Annual Reviews}
758 +}
759 +
760 +@article{cheng2015convective,
761 + title = {Convective Risk Flows in Commodity Futures Markets},
762 + author = {Cheng, Ing-Haw and Kirilenko, Andrei and Xiong, Wei},
763 + journal = {Review of Finance},
764 + volume = {19},
765 + number = {5},
766 + pages = {1733--1781},
767 + year = {2015},
768 + publisher = {Oxford University Press}
769 +}
770 +
771 +@article{chordia2008liquidity,
772 + author = {Chordia, T. and Roll, R. and Subrahmanyam, A.},
773 + title = {Liquidity and market efficiency},
774 + journal = {Journal of Financial Economics},
775 + volume = {87},
776 + number = {2},
777 + pages = {249--268},
778 + year = {2008},
779 + issn = {0304-405X},
780 + doi = {10.1016/j.jfineco.2007.03.005},
781 + url = {https://www.sciencedirect.com/science/article/pii/S0304405X07001833},
782 + keywords = {Liquidity, Market efficiency, Order flow},
783 + abstract = {Short-horizon return predictability from order flows is an inverse indicator of market efficiency. We find that such predictability is diminished when bid-ask spreads are narrower, and has declined over time with the minimum tick size. Variance ratio tests suggest that prices were closer to random walk benchmarks in the more liquid decimal regime than in other ones. These findings indicate that liquidity stimulates arbitrage activity, which, in turn, enhances market efficiency. Further, as the tick size decreased, open-close/close-open return variance ratios increased, while return autocorrelations decreased. This suggests an increased incorporation of private information into prices during more liquid regimes.}
784 +}
785 +
786 +@article{christie2000macroeconomics,
787 + title = {Do Macroeconomics News Releases Affect Gold and Silver Prices?},
788 + author = {Christie--David, Rohan and Chaudhry, Mukesh and Koch, Timothy W.},
789 + journal = {Journal of Economics and Business},
790 + volume = {52},
791 + number = {5},
792 + pages = {405--421},
793 + year = {2000},
794 + publisher = {Elsevier}
795 +}
796 +
797 +@article{clark1997us,
798 + title={{US} inflation developments in 1996},
799 + author={Clark, Todd E and others},
800 + journal={Economic Review-Federal Reserve Bank of Kansas City},
801 + volume={82},
802 + pages={11--30},
803 + year={1997},
804 + publisher={Federal Reserve Bank of Kansas City}
805 +}
806 +
807 +@article{clark2008forecasting,
808 + title={Forecasting {US} inflation using {B}ayesian factor augmented autoregressive models},
809 + author={Clark, Todd E and McCracken, Michael W},
810 + journal={The Review of Economics and Statistics},
811 + volume={90},
812 + number={4},
813 + pages={665--678},
814 + year={2008},
815 + publisher={MIT Press}
816 +}
817 +
818 +@article{connolly2022stock,
819 + title={Stock returns and inflation shocks in weaker economic times},
820 + author={Connolly, Robert A and Stivers, Chris and Sun, Licheng},
821 + journal={Financial Management},
822 + volume={51},
823 + number={3},
824 + pages={827--867},
825 + year={2022},
826 + publisher={Wiley Online Library}
827 +}
828 +
829 +@article{corsi2009simple,
830 + title={A simple approximate long-memory model of realized volatility},
831 + author={Corsi, Fulvio},
832 + journal={Journal of Financial Econometrics},
833 + volume={7},
834 + number={2},
835 + pages={174--196},
836 + year={2009},
837 + publisher={Oxford University Press}
838 +}
839 +
840 +@article{corsi2010threshold,
841 + title={Threshold bipower variation and the impact of jumps on volatility forecasting},
842 + author={Corsi, Fulvio and Pirino, Davide and Reno, Roberto},
843 + journal={Journal of Econometrics},
844 + volume={159},
845 + number={2},
846 + pages={276--288},
847 + year={2010},
848 + publisher={Elsevier}
849 +}
850 +
851 +@article{couleau2020corn,
852 + title={Are corn futures prices getting `jumpy'?},
853 + author={Couleau, Anabelle and Serra, Teresa and Garcia, Philip},
854 + journal={American Journal of Agricultural Economics},
855 + volume={102},
856 + number={2},
857 + pages={569--588},
858 + year={2020},
859 + publisher={Wiley Online Library}
860 +}
861 +
862 +@article{creti2013links,
863 + title={On the links between stock and commodity markets' volatility},
864 + author={Creti, Anna and Jo{\"e}ts, Marc and Mignon, Val{\'e}rie},
865 + journal={Energy Economics},
866 + volume={37},
867 + pages={16--28},
868 + year={2013},
869 + publisher={Elsevier}
870 +}
871 +
872 +@article{da2018exchange,
873 + title={Exchange traded funds and asset return correlations},
874 + author={Da, Zhi and Shive, Sophie},
875 + journal={European Financial Management},
876 + volume={24},
877 + number={1},
878 + pages={136--168},
879 + year={2018},
880 + publisher={Wiley}
881 +}
882 +
883 +@article{da2024financialization,
884 + title={Financialization and commodity markets serial dependence},
885 + author={Da, Zhi and Tang, Ke and Tao, Yubo and Yang, Liyan},
886 + journal={Management Science},
887 + volume={70},
888 + number={4},
889 + pages={2122--2143},
890 + year={2024},
891 + publisher={INFORMS}
892 +}
893 +
894 +@article{daigler1999impact,
895 + title = {The Impact of Trader Type on the Futures Volatility-Volume Relation},
896 + author = {Daigler, Robert T. and Wiley, Marilyn K.},
897 + journal = {Journal of Finance},
898 + volume = {54},
899 + number = {6},
900 + pages = {2297--2316},
901 + year = {1999},
902 + publisher = {Wiley Online Library}
903 +}
904 +
905 +@article{dannhauser2017effect,
906 + title={The impact of innovation: Evidence from corporate bond exchange-traded funds},
907 + author={Dannhauser, Caitlin D},
908 + journal={Journal of Financial Economics},
909 + volume={125},
910 + number={3},
911 + pages={537--560},
912 + year={2017},
913 + publisher={Elsevier}
914 +}
915 +
916 +@article{dannhauser2020etf,
917 + title={{ETF} arbitrage during market turmoil},
918 + author={Dannhauser, Caitlin D},
919 + journal={Journal of Financial Economics},
920 + volume={138},
921 + number={2},
922 + pages={355--374},
923 + year={2020},
924 + publisher={Elsevier}
925 +}
926 +
927 +@article{deaton1992behaviour,
928 + title={On the behaviour of commodity prices},
929 + author={Deaton, Angus and Laroque, Guy},
930 + journal={Review of Economic Studies},
931 + volume={59},
932 + number={1},
933 + pages={1--23},
934 + year={1992},
935 + publisher={Wiley-Blackwell}
936 +}
937 +
938 +@article{diebold2012measuring,
939 + title={Better to give than to receive: Predictive directional measurement of volatility spillovers},
940 + author={Diebold, Francis X and Yilmaz, Kamil},
941 + journal={International Journal of Forecasting},
942 + volume={28},
943 + number={1},
944 + pages={57--66},
945 + year={2012},
946 + publisher={Elsevier}
947 +}
948 +
949 +@article{doi:10.1198/jbes.2009.07205,
950 +author = {S. Borağan Aruoba and Francis X. Diebold and Chiara Scotti},
951 +title = {Real-Time Measurement of Business Conditions},
952 +journal = {Journal of Business \& Economic Statistics},
953 +volume = {27},
954 +number = {4},
955 +pages = {417-427},
956 +year = {2009},
957 +publisher = {Taylor & Francis},
958 +doi = {10.1198/jbes.2009.07205},
959 +
960 +URL = {
961 +
962 + https://doi.org/10.1198/jbes.2009.07205
963 +
964 +
965 +
966 +},
967 +eprint = {
968 +
969 + https://doi.org/10.1198/jbes.2009.07205
970 +
971 +
972 +
973 +}
974 +
975 +}
976 +
977 +@article{domanski2007financial,
978 + title={Financial investors and commodity markets},
979 + author={Domanski, Dietrich and Heath, Alexandra},
980 + journal={BIS Quarterly Review},
981 + volume={3},
982 + number={1},
983 + pages={53--67},
984 + year={2007}
985 +}
986 +
987 +@article{ehrmann2004taking,
988 + title={Taking stock: Monetary policy transmission to equity markets},
989 + author={Ehrmann, Michael and Fratzscher, Marcel},
990 + journal={Journal of Money, Credit and Banking},
991 + pages={719--737},
992 + year={2004},
993 + publisher={JSTOR}
994 +}
995 +
996 +@article{ehrmann2011,
997 + title={Global Crises and Equity Market Contagion},
998 + author={Ehrmann, Michael and Fratzscher, Marcel and Rigobon, Roberto},
999 + journal={Journal of Finance},
1000 + volume={66},
1001 + number={6},
1002 + pages={2597--2649},
1003 + year={2011},
1004 + publisher={Wiley Online Library}
1005 +}
1006 +
1007 +@article{eklund2024,
1008 + title={FOMC Statement Sentiment and Inflation Expectations},
1009 + author={Eklund, Johan and Kim, Soyoung},
1010 + journal={Journal of Monetary Economics},
1011 + volume={141},
1012 + pages={45--62},
1013 + year={2024},
1014 + publisher={Elsevier}
1015 +}
1016 +
1017 +@article{engle1982autoregressive,
1018 + title={Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation},
1019 + author={Engle, Robert F},
1020 + journal={Econometrica},
1021 + pages={987--1007},
1022 + year={1982},
1023 + publisher={JSTOR}
1024 +}
1025 +
1026 +@article{erb2013golden,
1027 + title = {The Golden Dilemma},
1028 + author = {Erb, Claude B and Harvey, Campbell R},
1029 + journal = {Financial Analysts Journal},
1030 + volume = {69},
1031 + number = {4},
1032 + pages = {10--42},
1033 + year = {2013},
1034 + publisher = {Taylor \& Francis}
1035 +}
1036 +
1037 +@article{fama1970,
1038 + title={Efficient Capital Markets: A Review of Theory and Empirical Work},
1039 + author={Fama, Eugene F},
1040 + journal={Journal of Finance},
1041 + volume={25},
1042 + number={2},
1043 + pages={383--417},
1044 + year={1970},
1045 + publisher={Wiley Online Library}
1046 +}
1047 +
1048 +@incollection{fama2016commodity,
1049 + title={Commodity futures prices: Some evidence on forecast power, premiums, and the theory of storage},
1050 + author={Fama, Eugene F and French, Kenneth R},
1051 + booktitle={The World Scientific Handbook of Futures Markets},
1052 + pages={79--102},
1053 + year={2016},
1054 + publisher={World Scientific}
1055 +}
1056 +
1057 +@article{fang2023etf,
1058 + title={{ETF} arbitrage, non-fundamental demand, and return predictability},
1059 + author={Fang, Lily H and Jiang, Joel},
1060 + journal={Review of Financial Studies},
1061 + volume={36},
1062 + number={4},
1063 + pages={1433--1470},
1064 + year={2023},
1065 + publisher={Oxford University Press}
1066 +}
1067 +
1068 +@article{fattouh2013role,
1069 + title={The role of speculation in oil markets: What have we learned so far?},
1070 + author={Fattouh, Bassam and Kilian, Lutz and Mahadeva, Lavan},
1071 + journal={The Energy Journal},
1072 + volume={34},
1073 + number={3},
1074 + pages={7--33},
1075 + year={2013},
1076 + publisher={SAGE Publications Sage CA: Los Angeles, CA}
1077 +}
1078 +
1079 +@article{fishe2012identifying,
1080 + title = {Identifying Informed Traders in Futures Markets},
1081 + author = {Fishe, Raymond P.H. and Smith, Aaron},
1082 + journal = {Journal of Financial Markets},
1083 + volume = {15},
1084 + number = {3},
1085 + pages = {329--359},
1086 + year = {2012},
1087 + url = {https://files.asmith.ucdavis.edu/2012_JFinM_FS_informed.pdf},
1088 + keywords = {finance},
1089 + abstract = {We use daily positions of futures market participants to identify informed traders. These data contain 8,921 unique traders. We identify between 94 and 230 traders as overnight informed and 91 as intraday informed with little overlap. Floor brokers/traders are over-represented in the overnight informed group. The intraday informed group is dominated by managed money traders/hedge funds and swap dealers, with commercial hedgers under-represented. We find that characteristics such as experience, position size, trading activity, and type of positions held offer significant predictive power for who is informed. An analysis of daily trader profits confirms that we select highly profitable traders.},
1090 + publisher = {North-Holland}
1091 +}
1092 +
1093 +@article{flannery2002macroeconomic,
1094 + title={Macroeconomic factors do influence aggregate stock returns},
1095 + author={Flannery, Mark J and Protopapadakis, Aris A},
1096 + journal={Review of Financial Studies},
1097 + volume={15},
1098 + number={3},
1099 + pages={751--782},
1100 + year={2002},
1101 + publisher={Oxford University Press}
1102 +}
1103 +
1104 +@article{fleming1997moves,
1105 + title={What moves the bond market?},
1106 + author={Fleming, Michael J and Remolona, Eli M},
1107 + journal={Federal Reserve Bank of New York Economic Policy Review},
1108 + volume={3},
1109 + number={4},
1110 + year={1997}
1111 +}
1112 +
1113 +@article{fleming1999,
1114 + title={Price Formation and Liquidity in the US Treasury Market: The Response to Public Information},
1115 + author={Fleming, Michael J and Remolona, Eli M},
1116 + journal={Journal of Finance},
1117 + volume={54},
1118 + number={5},
1119 + pages={1901--1915},
1120 + year={1999},
1121 + publisher={Wiley Online Library}
1122 +}
1123 +
1124 +@article{fleming1999price,
1125 + title={Price formation and liquidity in the US Treasury market: The response to public information},
1126 + author={Fleming, Michael J and Remolona, Eli M},
1127 + journal={Journal of Finance},
1128 + volume={54},
1129 + number={5},
1130 + pages={1901--1915},
1131 + year={1999},
1132 + publisher={Wiley Online Library}
1133 +}
1134 +
1135 +@article{flemming1997moves,
1136 + title = {What Moves the Bond Market?},
1137 + author = {Flemming, Michael J. and Remolona, Eli M.},
1138 + journal = {Economic Policy Review},
1139 + volume = {3},
1140 + number = {4},
1141 + year = {1997}
1142 +}
1143 +
1144 +@article{frankel1985commodity,
1145 + title = {Commodity Prices, Money Surprises and Fed Credibility},
1146 + author = {Frankel, Jeffrey A. and Hardouvelis, Gikas A.},
1147 + journal = {Journal of Money, Credit and Banking},
1148 + volume = {17},
1149 + number = {4},
1150 + pages = {425--438},
1151 + year = {1985},
1152 + publisher = {JSTOR}
1153 +}
1154 +
1155 +@article{galkin2018opec,
1156 + author = {Galkin, Philipp and Atalla, Tarek and Ren, Zhongyuan},
1157 + doi = {10.30573/ks--2018-dp38},
1158 + title = {An Estimation of the Drivers Behind OPEC’s Quota Decisions},
1159 + journal = {King Abdullah Petroleum Studies and Research Center Discussion Paper},
1160 + year = {2018}
1161 +}
1162 +
1163 +@article{garbade1983price,
1164 + title = {Price Movements and Price Discovery in Futures and Cash Markets},
1165 + author = {Garbade, Kenneth D. and Silber, William L.},
1166 + journal = {Review of Economics and Statistics},
1167 + volume = {65},
1168 + number = {2},
1169 + pages = {289--297},
1170 + year = {1983},
1171 + publisher = {JSTOR}
1172 +}
1173 +
1174 +@article{gebarowski2015competition,
1175 + author = {Gębarowski, Robert and Drożdż, Stanisław and Górski, Arkadiusz and Oświęcimka, Paweł},
1176 + title = {Competition of Commodities for the Status of Money in an Agent-Based Model},
1177 + journal = {Acta Physica Polonica A},
1178 + volume = {127},
1179 + number = {3a},
1180 + pages = {A-51--A-54},
1181 + year = {2015},
1182 + doi = {10.12693/aphyspola.127.a-51}
1183 +}
1184 +
1185 +@article{gentzkow2019,
1186 + title={Text as Data},
1187 + author={Gentzkow, Matthew and Kelly, Bryan and Taddy, Matt},
1188 + journal={Journal of Economic Literature},
1189 + volume={57},
1190 + number={3},
1191 + pages={535--574},
1192 + year={2019},
1193 + publisher={American Economic Association}
1194 +}
1195 +
1196 +@article{getz1990barometer,
1197 + title={Diffusion indexes: a barometer of the economy},
1198 + author={Getz, Patricia M and Ulmer, Mark G},
1199 + journal={Bureau of Labor Statistics Monthly Labor Review},
1200 + pages={13},
1201 + year={1990}
1202 +}
1203 +
1204 +@article{geweke1992evaluating,
1205 + title={Evaluating the accuracy of sampling-based approaches to the calculation of posterior moments},
1206 + author={Geweke, John},
1207 + journal={Bayesian Statistics},
1208 + volume={4},
1209 + pages={169--193},
1210 + year={1992},
1211 + publisher={Oxford University Press}
1212 +}
1213 +
1214 +@article{ghysels2004midas,
1215 + title={The MIDAS touch: Mixed data sampling regression models},
1216 + author={Ghysels, Eric and Santa-Clara, Pedro and Valkanov, Rossen},
1217 + year={2004}
1218 +}
1219 +
1220 +@article{giannone2015prior,
1221 + title={Prior selection for vector autoregressions},
1222 + author={Giannone, Domenico and Lenza, Michele and Primiceri, Giorgio E},
1223 + journal={The Review of Economics and Statistics},
1224 + volume={97},
1225 + number={2},
1226 + pages={436--451},
1227 + year={2015},
1228 + publisher={MIT Press}
1229 +}
1230 +
1231 +@article{glick2012central,
1232 + title={Central bank announcements of asset purchases and the impact on global financial and commodity markets},
1233 + author={Glick, Reuven and Leduc, Sylvain},
1234 + journal={Journal of International Money and Finance},
1235 + volume={31},
1236 + number={8},
1237 + pages={2078--2101},
1238 + year={2012},
1239 + publisher={Elsevier}
1240 +}
1241 +
1242 +@article{glosten1993relation,
1243 + title={On the relation between the expected value and the volatility of the nominal excess return on stocks},
1244 + author={Glosten, Lawrence R and Jagannathan, Ravi and Runkle, David E},
1245 + journal={Journal of Finance},
1246 + volume={48},
1247 + number={5},
1248 + pages={1779--1801},
1249 + year={1993},
1250 + publisher={Wiley Online Library}
1251 +}
1252 +
1253 +@article{glosten2021etf,
1254 + title={{ETF} activity and informational efficiency of underlying securities},
1255 + author={Glosten, Lawrence and Nallareddy, Suresh and Zou, Yuan},
1256 + journal={Management Science},
1257 + volume={67},
1258 + number={1},
1259 + pages={22--47},
1260 + year={2021},
1261 + publisher={INFORMS}
1262 +}
1263 +
1264 +@techreport{goldberg2013time,
1265 + title = {Time Variation in Asset Price Responses to Macro Announcements},
1266 + author = {Goldberg, Linda S. and Grisse, Christian},
1267 + institution = {National Bureau of Economic Research},
1268 + type = {Working Paper},
1269 + series = {Working Paper Series},
1270 + number = {19523},
1271 + year = {2013},
1272 + month = {October},
1273 + doi = {10.3386/w19523},
1274 + url = {http://www.nber.org/papers/w19523}
1275 +}
1276 +
1277 +@article{goldstein2014speculation,
1278 + title = {Speculation and Hedging in Segmented Markets},
1279 + author = {Goldstein, Itay and Li, Yan and Yang, Liyan},
1280 + journal = {Review of Financial Studies},
1281 + volume = {27},
1282 + number = {3},
1283 + pages = {881--922},
1284 + year = {2014},
1285 + publisher = {Oxford University Press}
1286 +}
1287 +
1288 +@article{goldstein2022commodity,
1289 + title = {Commodity Financialization and Information Transmission},
1290 + author = {Goldstein, Itay and Yang, Liyan},
1291 + journal = {The Journal of Finance},
1292 + volume = {77},
1293 + number = {5},
1294 + pages = {2613--2667},
1295 + year = {2022},
1296 + publisher = {Wiley Online Library}
1297 +}
1298 +
1299 +@article{gorodnichenko2023,
1300 + title={The Voice of Monetary Policy},
1301 + author={Gorodnichenko, Yuriy and Pham, Tho and Talavera, Oleksandr},
1302 + journal={American Economic Review},
1303 + volume={113},
1304 + number={2},
1305 + pages={548--584},
1306 + year={2023},
1307 + publisher={American Economic Association}
1308 +}
1309 +
1310 +@article{gorton2006facts,
1311 + title={Facts and fantasies about commodity futures},
1312 + author={Gorton, Gary and Rouwenhorst, K Geert},
1313 + journal={Financial Analysts Journal},
1314 + volume={62},
1315 + number={2},
1316 + pages={47--68},
1317 + year={2006},
1318 + publisher={CFA Institute}
1319 +}
1320 +
1321 +@article{gospodinov2012effects,
1322 + title = {The Effects of Federal Funds Rate Surprises on S\&P 500 Volatility and Volatility Risk Premium},
1323 + author = {Gospodinov, Nikolay and Jamali, Ibrahim},
1324 + journal = {Journal of Empirical Finance},
1325 + volume = {19},
1326 + number = {4},
1327 + pages = {497--510},
1328 + year = {2012},
1329 + publisher = {Elsevier}
1330 +}
1331 +
1332 +@article{graicer2021opec,
1333 + author = {Graicer, Rafael and Nery, Thiago Affonso and Jucá, Michele Nascimento and Junior, Eli Hadad and Bastos, Douglas Dias},
1334 + doi = {10.5902/1983465963575},
1335 + title = {The Impact of the OPEC Announcement on the Oil Companies' Share Prices},
1336 + journal = {Revista De Administração Da Ufsm},
1337 + year = {2021}
1338 +}
1339 +
1340 +@article{gromb2010limits,
1341 + title={Limits of arbitrage: The state of the theory},
1342 + author={Gromb, Denis and Vayanos, Dimitri},
1343 + journal={Annual Review of Financial Economics},
1344 + volume={2},
1345 + pages={251--275},
1346 + year={2010},
1347 + publisher={Annual Reviews}
1348 +}
1349 +
1350 +@article{grossman1980,
1351 + title={On the Impossibility of Informationally Efficient Markets},
1352 + author={Grossman, Sanford J and Stiglitz, Joseph E},
1353 + journal={American Economic Review},
1354 + volume={70},
1355 + number={3},
1356 + pages={393--408},
1357 + year={1980},
1358 + publisher={American Economic Association}
1359 +}
1360 +
1361 +@article{grossman1980impossibility,
1362 + title = {On the Impossibility of Informationally Efficient Markets},
1363 + author = {Grossman, Sanford J and Stiglitz, Joseph E},
1364 + journal = {American Economic Review},
1365 + volume = {70},
1366 + number = {3},
1367 + pages = {393--408},
1368 + year = {1980},
1369 + publisher = {JSTOR}
1370 +}
1371 +
1372 +@article{gu2018drives,
1373 + title = {What Drives Informed Trading Before Public Releases? Evidence from Natural Gas Inventory Announcements},
1374 + author = {Gu, Chen and Kurov, Alexander},
1375 + journal = {Journal of Futures Markets},
1376 + volume = {38},
1377 + number = {9},
1378 + pages = {1079--1096},
1379 + year = {2018},
1380 + publisher = {Wiley Online Library}
1381 +}
1382 +
1383 +@article{guo2015leveraged,
1384 + title={Leveraged {ETF} performance during periods of market turmoil},
1385 + author={Guo, Hui and Leung, Tim},
1386 + journal={The Journal of Derivatives},
1387 + volume={22},
1388 + number={3},
1389 + pages={61--74},
1390 + year={2015},
1391 + publisher={Institutional Investor Journals}
1392 +}
1393 +
1394 +@article{gurkaynak2005,
1395 + title={Do Actions Speak Louder Than Words? The Response of Asset Prices to Monetary Policy Actions and Statements},
1396 + author={G{\"u}rkaynak, Refet S and Sack, Brian P and Swanson, Eric T},
1397 + journal={International Journal of Central Banking},
1398 + volume={1},
1399 + number={1},
1400 + pages={55--93},
1401 + year={2005}
1402 +}
1403 +
1404 +@article{gurkaynak2005sensitivity,
1405 + title={The sensitivity of long-term interest rates to economic news: Evidence and implications for macroeconomic models},
1406 + author={G{\"u}rkaynak, Refet S and Sack, Brian and Swanson, Eric},
1407 + journal={American Economic Review},
1408 + volume={95},
1409 + number={1},
1410 + pages={425--436},
1411 + year={2005}
1412 +}
1413 +
1414 +@article{gurtler2010,
1415 + title={The Effect of FOMC Statement Language on Financial Markets},
1416 + author={Gürtler, Marc and Gürtler, Oliver},
1417 + journal={Journal of Financial Research},
1418 + volume={33},
1419 + number={4},
1420 + pages={369--391},
1421 + year={2010},
1422 + publisher={Wiley Online Library}
1423 +}
1424 +
1425 +@article{haigh2007hedge,
1426 + title={Hedge funds, volatility, and liquidity provision in energy futures markets},
1427 + author={Haigh, Michael S and Hranaiova, Jana and Overdahl, James A},
1428 + journal={Journal of Alternative Investments},
1429 + volume={9},
1430 + number={4},
1431 + pages={10--38},
1432 + year={2007},
1433 + publisher={Institutional Investor Journals Umbrella}
1434 +}
1435 +
1436 +@article{hakkio1985reaction,
1437 + title={The reaction of exchange rates to economic news},
1438 + author={Hakkio, Craig S and Pearce, Douglas K},
1439 + journal={Economic Inquiry},
1440 + volume={23},
1441 + number={4},
1442 + pages={621--636},
1443 + year={1985},
1444 + publisher={Wiley Online Library}
1445 +}
1446 +
1447 +@article{hamilton2014risk,
1448 + title = {Risk Premia in Crude Oil Futures Prices},
1449 + author = {Hamilton, James D. and Wu, Jing Cynthia},
1450 + journal = {Journal of International Money and Finance},
1451 + volume = {42},
1452 + pages = {9--37},
1453 + year = {2014},
1454 + publisher = {Elsevier}
1455 +}
1456 +
1457 +@article{hamm2011liquidity,
1458 + title={Liquidity and firm value: Evidence from {ETF} holdings},
1459 + author={Hamm, Sophia JW},
1460 + journal={Journal of Corporate Finance},
1461 + volume={20},
1462 + pages={1--15},
1463 + year={2014},
1464 + publisher={Elsevier}
1465 +}
1466 +
1467 +@article{hamm2014liquidity,
1468 + title={Liquidity and firm value: Evidence from {ETF} holdings},
1469 + author={Hamm, Sophia JW},
1470 + journal={Journal of Corporate Finance},
1471 + volume={29},
1472 + pages={1--15},
1473 + year={2014},
1474 + publisher={Elsevier}
1475 +}
1476 +
1477 +@article{hammoudeh2008metal,
1478 + title={Metal volatility in presence of oil and interest rate shocks},
1479 + author={Hammoudeh, Shawkat and Yuan, Yuan},
1480 + journal={Energy Economics},
1481 + volume={30},
1482 + number={2},
1483 + pages={606--620},
1484 + year={2008},
1485 + publisher={Elsevier}
1486 +}
1487 +
1488 +@article{hansen2005realized,
1489 + title={A realized variance for the whole day based on intermittent high-frequency data},
1490 + author={Hansen, Peter Reinhard and Lunde, Asger},
1491 + journal={Journal of Financial Econometrics},
1492 + volume={3},
1493 + number={4},
1494 + pages={525--554},
1495 + year={2005},
1496 + publisher={Oxford University Press}
1497 +}
1498 +
1499 +@article{hansen2017,
1500 + title={Transparency and Deliberation within the FOMC: A Computational Linguistics Approach},
1501 + author={Hansen, Stephen and McMahon, Michael and Prat, Andrea},
1502 + journal={Quarterly Journal of Economics},
1503 + volume={133},
1504 + number={2},
1505 + pages={801--870},
1506 + year={2018},
1507 + publisher={Oxford University Press}
1508 +}
1509 +
1510 +@article{harris2009role,
1511 + title={The role of speculators in the crude oil futures market},
1512 + author={Harris, Jeffrey H and Buyuksahin, Bahattin},
1513 + journal={Available at SSRN 1435042},
1514 + year={2009}
1515 +}
1516 +
1517 +@article{hasbrouck2003intraday,
1518 + title={Intraday price formation in {US} equity index markets},
1519 + author={Hasbrouck, Joel},
1520 + journal={The Journal of Finance},
1521 + volume={58},
1522 + number={6},
1523 + pages={2375--2400},
1524 + year={2003},
1525 + publisher={Wiley}
1526 +}
1527 +
1528 +@article{hautsch2011impact,
1529 + title = {The Impact of Macroeconomic News on Quote Adjustments, Noise, and Informational Volatility},
1530 + author = {Hautsch, Nikolaus and Hess, Dieter and Veredas, David},
1531 + journal = {Journal of Banking \& Finance},
1532 + volume = {35},
1533 + number = {10},
1534 + pages = {2733--2746},
1535 + year = {2011},
1536 + publisher = {Elsevier}
1537 +}
1538 +
1539 +@article{hayo2010,
1540 + title={Do Federal Reserve Communications Help Predict Federal Funds Target Rate Decisions?},
1541 + author={Hayo, Bernd and Neuenkirch, Matthias},
1542 + journal={Journal of Macroeconomics},
1543 + volume={32},
1544 + number={4},
1545 + pages={1014--1024},
1546 + year={2010},
1547 + publisher={Elsevier}
1548 +}
1549 +
1550 +@article{hedegaard2011margins,
1551 + title = {How Margins Are Set and Affect Asset Prices},
1552 + author = {Hedegaard, Esben},
1553 + journal = {Job Market Paper},
1554 + year = {2011}
1555 +}
1556 +
1557 +@article{hedge2007funds,
1558 + title = {Hedge Funds, Volatility, and Liquidity Provision in Energy Futures Markets},
1559 + author = {Haigh, Michael S. and Hranaiova, Jana and Overdahl, James A.},
1560 + journal = {Journal of Alternative Investments},
1561 + volume = {9},
1562 + number = {4},
1563 + pages = {10--38},
1564 + year = {2007},
1565 + publisher = {Institutional Investor Journals Umbrella}
1566 +}
1567 +
1568 +@article{hendershott2013relationship,
1569 + title={Algorithmic trading and the market for liquidity},
1570 + author={Hendershott, Terrence and Riordan, Ryan},
1571 + journal={Journal of Financial and Quantitative Analysis},
1572 + volume={48},
1573 + number={4},
1574 + pages={1001--1024},
1575 + year={2013},
1576 + publisher={Cambridge University Press}
1577 +}
1578 +
1579 +@article{henderson2015new,
1580 + title={New evidence on the financialization of commodity markets},
1581 + author={Henderson, Brian J and Pearson, Neil D and Wang, Li},
1582 + journal={Review of Financial Studies},
1583 + volume={28},
1584 + number={5},
1585 + pages={1285--1311},
1586 + year={2015},
1587 + publisher={Oxford University Press}
1588 +}
1589 +
1590 +@article{hess2008commodity,
1591 + title = {How Do Commodity Futures Respond to Macroeconomic News?},
1592 + author = {Hess, Dieter and Huang, He and Niessen, Alexandra},
1593 + journal = {Financial Markets and Portfolio Management},
1594 + volume = {22},
1595 + number = {2},
1596 + pages = {127--146},
1597 + year = {2008},
1598 + publisher = {Springer}
1599 +}
1600 +
1601 +@article{hicks1975value,
1602 + title = {Value and Capital: An Inquiry into Some Fundamental Principles of Economic Theory},
1603 + author = {Hicks, John Richard and others},
1604 + journal = {OUP Catalogue},
1605 + year = {1975},
1606 + publisher = {Oxford University Press}
1607 +}
1608 +
1609 +@article{hirshleifer1990hedging,
1610 + title = {Hedging Pressure and Futures Price Movements in a General Equilibrium Model},
1611 + author = {Hirshleifer, David},
1612 + journal = {Econometrica},
1613 + pages = {411--428},
1614 + year = {1990},
1615 + publisher = {JSTOR}
1616 +}
1617 +
1618 +@article{hirshleifer1991seasonal,
1619 + title = {Seasonal Patterns of Futures Hedging and the Resolution of Output Uncertainty},
1620 + author = {Hirshleifer, David},
1621 + journal = {Journal of Economic Theory},
1622 + volume = {53},
1623 + number = {2},
1624 + pages = {304--327},
1625 + year = {1991},
1626 + publisher = {Elsevier}
1627 +}
1628 +
1629 +@article{hollstein2020volatility,
1630 + title = {Volatility Term Structures in Commodity Markets},
1631 + author = {Hollstein, Fabian and Prokopczuk, Marcel and W{\"u}rsig, Christoph},
1632 + journal = {Journal of Futures Markets},
1633 + volume = {40},
1634 + number = {4},
1635 + pages = {527--555},
1636 + year = {2020},
1637 + publisher = {Wiley Online Library}
1638 +}
1639 +
1640 +@article{horan2004implied,
1641 + title = {Implied Volatility of Oil Futures Options Surrounding {OPEC} Meetings},
1642 + author = {Horan, Stephen M. and Peterson, Jeffrey H. and Mahar, James},
1643 + journal = {The Energy Journal},
1644 + volume = {25},
1645 + number = {3},
1646 + pages = {103--125},
1647 + year = {2004},
1648 + publisher = {International Association for Energy Economics}
1649 +}
1650 +
1651 +@article{hordahl2020expectations,
1652 + title = {Expectations and Risk Premia at 8:30 am: Deciphering the Responses of Bond Yields to Macroeconomic Announcements},
1653 + author = {H{\"o}rdahl, Peter and Remolona, Eli M. and Valente, Giorgio},
1654 + journal = {Journal of Business \& Economic Statistics},
1655 + volume = {38},
1656 + number = {1},
1657 + pages = {27--42},
1658 + year = {2020},
1659 + publisher = {Taylor \& Francis}
1660 +}
1661 +
1662 +@article{hu2013noise,
1663 + title = {Noise as Information for Illiquidity},
1664 + author = {Hu, Grace Xing and Pan, Jun and Wang, Jiang},
1665 + journal = {Journal of Finance},
1666 + volume = {68},
1667 + number = {6},
1668 + pages = {2341--2382},
1669 + year = {2013},
1670 + publisher = {Wiley Online Library}
1671 +}
1672 +
1673 +@article{hu2017early,
1674 + title={Early peek advantage? Efficient price discovery with tiered information disclosure},
1675 + author={Hu, Grace Xing and Pan, Jun and Wang, Jiang},
1676 + journal={Journal of Financial Economics},
1677 + volume={126},
1678 + number={2},
1679 + pages={399--421},
1680 + year={2017},
1681 + publisher={Elsevier}
1682 +}
1683 +
1684 +@article{hu2019,
1685 + title={Noise as Information for Illiquidity},
1686 + author={Hu, Grace Xing and Pan, Jun and Wang, Jiang},
1687 + journal={Journal of Finance},
1688 + volume={68},
1689 + number={6},
1690 + pages={2341--2382},
1691 + year={2013},
1692 + publisher={Wiley Online Library}
1693 +}
1694 +
1695 +@article{huang2005using,
1696 + title={The relative contribution of jumps to total price variance},
1697 + author={Huang, Xin and Tauchen, George},
1698 + journal={Journal of Financial Econometrics},
1699 + volume={3},
1700 + number={4},
1701 + pages={456--499},
1702 + year={2005},
1703 + publisher={Oxford University Press}
1704 +}
1705 +
1706 +@article{humphreys2010great,
1707 + title = {The Great Metals Boom: A Retrospective},
1708 + author = {Humphreys, David},
1709 + journal = {Resources Policy},
1710 + volume = {35},
1711 + number = {1},
1712 + pages = {1--13},
1713 + year = {2010},
1714 + publisher = {Elsevier}
1715 +}
1716 +
1717 +@article{irwin1987note,
1718 + title = {A Note on the Factors Affecting Technical Trading System Returns},
1719 + author = {Irwin, Scott H. and Brorsen, B. Wade},
1720 + journal = {Journal of Futures Markets},
1721 + volume = {7},
1722 + number = {5},
1723 + pages = {591--595},
1724 + year = {1987},
1725 + publisher = {Wiley Subscription Services, Inc., A Wiley Company New York}
1726 +}
1727 +
1728 +@article{irwin1999managed,
1729 + title = {Managed Futures, Positive Feedback Trading, and Futures Price Volatility},
1730 + author = {Irwin, Scott H. and Yoshimaru, Satoko},
1731 + journal = {Journal of Futures Markets},
1732 + volume = {19},
1733 + number = {7},
1734 + pages = {759--776},
1735 + year = {1999},
1736 + publisher = {Wiley Online Library}
1737 +}
1738 +
1739 +@incollection{irwin2004effect,
1740 + author = {Irwin, Scott H. and Holt, Bryce},
1741 + title = {The Effect of Large Hedge Fund and {CTA} Trading on Futures Market Volatility},
1742 + booktitle = {Commodity Trading Advisors: Risk, Performance Analysis and Selection},
1743 + publisher = {John Wiley and Sons, Inc},
1744 + address = {New York, NY},
1745 + year = {2004}
1746 +}
1747 +
1748 +@article{irwin2009devil,
1749 + title = {Devil or Angel? The Role of Speculation in the Recent Commodity Price Boom (and Bust)},
1750 + author = {Irwin, Scott H. and Sanders, Dwight R. and Merrin, Robert P.},
1751 + journal = {Journal of Agricultural and Applied Economics},
1752 + volume = {41},
1753 + number = {2},
1754 + pages = {377--391},
1755 + year = {2009},
1756 + publisher = {Cambridge University Press}
1757 +}
1758 +
1759 +@article{irwin2011index,
1760 + title = {Index Funds, Financialization, and Commodity Futures Markets},
1761 + author = {Irwin, Scott H. and Sanders, Dwight R.},
1762 + journal = {Applied Economic Perspectives and Policy},
1763 + volume = {33},
1764 + number = {1},
1765 + pages = {1--31},
1766 + year = {2011},
1767 + publisher = {Oxford University Press}
1768 +}
1769 +
1770 +@article{irwin2012financialization,
1771 + title = {Financialization and Structural Change in Commodity Futures Markets},
1772 + author = {Irwin, Scott H. and Sanders, Dwight R.},
1773 + journal = {Journal of Agricultural and Applied Economics},
1774 + volume = {44},
1775 + number = {3},
1776 + pages = {371--396},
1777 + year = {2012},
1778 + publisher = {Cambridge University Press}
1779 +}
1780 +
1781 +@article{irwin2012testing,
1782 + title = {Testing the Masters Hypothesis in Commodity Futures Markets},
1783 + author = {Irwin, Scott H. and Sanders, Dwight R.},
1784 + journal = {Energy Economics},
1785 + volume = {34},
1786 + number = {1},
1787 + pages = {256--269},
1788 + year = {2012},
1789 + publisher = {Elsevier}
1790 +}
1791 +
1792 +@article{israeli2017etf,
1793 + title={Is there a dark side to exchange traded funds? {An} information perspective},
1794 + author={Israeli, Doron and Lee, Charles MC and Sridharan, Suhas A},
1795 + journal={Review of Accounting Studies},
1796 + volume={22},
1797 + number={3},
1798 + pages={1048--1083},
1799 + year={2017},
1800 + publisher={Springer}
1801 +}
1802 +
1803 +@article{jarocinski2020,
1804 + title={Deconstructing Monetary Policy Surprises—The Role of Information Shocks},
1805 + author={Jaro{\'c}inski, Marek and Kar{\'a}di, Peter},
1806 + journal={American Economic Journal: Macroeconomics},
1807 + volume={12},
1808 + number={2},
1809 + pages={1--43},
1810 + year={2020},
1811 + publisher={American Economic Association}
1812 +}
1813 +
1814 +@article{joets2017does,
1815 + title={Does the volatility of commodity prices reflect macroeconomic uncertainty?},
1816 + author={Jo{\"e}ts, Marc and Mignon, Val{\'e}rie and Razafindrabe, Tovonony},
1817 + journal={Energy Economics},
1818 + volume={68},
1819 + pages={313--326},
1820 + year={2017},
1821 + publisher={Elsevier}
1822 +}
1823 +
1824 +@article{jorda2005,
1825 + title={Estimation and Inference of Impulse Responses by Local Projections},
1826 + author={Jord{\`a}, {\`O}scar},
1827 + journal={American Economic Review},
1828 + volume={95},
1829 + number={1},
1830 + pages={161--182},
1831 + year={2005}
1832 +}
1833 +
1834 +@article{kaldor1976speculation,
1835 + title = {Speculation and Economic Stability},
1836 + author = {Kaldor, Nicholas},
1837 + booktitle = {The Economics of Futures Trading},
1838 + pages = {111--123},
1839 + year = {1976},
1840 + publisher = {Springer}
1841 +}
1842 +
1843 +@article{kang2020economic,
1844 + title={Economic determinants of oil futures volatility: A term structure perspective},
1845 + author={Kang, Boda and Nikitopoulos, Christina Sklibosios and Prokopczuk, Marcel},
1846 + journal={Energy Economics},
1847 + volume={88},
1848 + pages={104743},
1849 + year={2020},
1850 + publisher={Elsevier},
1851 + doi={10.1016/j.eneco.2020.104743}
1852 +}
1853 +
1854 +@article{kang2020tale,
1855 + title = {A Tale of Two Premiums: The Role of Hedgers and Speculators in Commodity Futures Markets},
1856 + author = {Kang, Wenjin and Rouwenhorst, K. Geert and Tang, Ke},
1857 + journal = {Journal of Finance},
1858 + volume = {75},
1859 + number = {1},
1860 + pages = {377--417},
1861 + year = {2020},
1862 + publisher = {Wiley Online Library}
1863 +}
1864 +
1865 +@article{kang2023financialization,
1866 + title={Financialization of commodity markets ten years later},
1867 + author={Kang, Wenjin and Tang, Ke and Wang, Ningli},
1868 + journal={Journal of Commodity Markets},
1869 + volume={30},
1870 + pages={100313},
1871 + year={2023},
1872 + publisher={Elsevier}
1873 +}
1874 +
1875 +@article{karali2014macro,
1876 + title={Macro determinants of volatility and volatility spillover in energy markets},
1877 + author={Karali, Berna and Ramirez, Octavio A},
1878 + journal={Energy Economics},
1879 + volume={46},
1880 + pages={413--421},
1881 + year={2014},
1882 + publisher={Elsevier},
1883 + doi={10.1016/j.eneco.2014.06.004}
1884 +}
1885 +
1886 +@article{kellogg2014effect,
1887 + title = {The Effect of Uncertainty on Investment: Evidence from Texas Oil Drilling},
1888 + author = {Kellogg, Ryan},
1889 + journal = {American Economic Review},
1890 + volume = {104},
1891 + number = {6},
1892 + pages = {1698--1734},
1893 + year = {2014}
1894 +}
1895 +
1896 +@inproceedings{kenton2019,
1897 + title={BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding},
1898 + author={Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
1899 + booktitle={Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics},
1900 + pages={4171--4186},
1901 + year={2019}
1902 +}
1903 +
1904 +@article{keynes1923some,
1905 + title = {Some Aspects of Commodity Markets},
1906 + author = {Keynes, John Maynard},
1907 + journal = {Manchester Guardian Commercial: European Reconstruction Series},
1908 + volume = {13},
1909 + pages = {784--786},
1910 + year = {1923}
1911 +}
1912 +
1913 +@article{kilian2011energy,
1914 + title = {Do Energy Prices Respond to US Macroeconomic News? A Test of the Hypothesis of Predetermined Energy Prices},
1915 + author = {Kilian, Lutz and Vega, Clara},
1916 + journal = {Review of Economics and Statistics},
1917 + volume = {93},
1918 + number = {2},
1919 + pages = {660--671},
1920 + year = {2011},
1921 + publisher = {The MIT Press}
1922 +}
1923 +
1924 +@article{kilian2014role,
1925 + title = {The Role of Inventories and Speculative Trading in the Global Market for Crude Oil},
1926 + author = {Kilian, Lutz and Murphy, Daniel P.},
1927 + journal = {Journal of Applied Econometrics},
1928 + volume = {29},
1929 + number = {3},
1930 + pages = {454--478},
1931 + year = {2014},
1932 + publisher = {Wiley Online Library}
1933 +}
1934 +
1935 +@article{kim2015does,
1936 + title = {Does Futures Speculation Destabilize Commodity Markets?},
1937 + author = {Kim, Abby},
1938 + journal = {Journal of Futures Markets},
1939 + volume = {35},
1940 + number = {8},
1941 + pages = {696--714},
1942 + year = {2015},
1943 + publisher = {Wiley Online Library}
1944 +}
1945 +
1946 +@article{knuth2018breakthroughs,
1947 + title = {“Breakthroughs” for a Green Economy? Financialization and Clean Energy Transition},
1948 + author = {Knuth, Sarah},
1949 + journal = {Energy Research \& Social Science},
1950 + volume = {41},
1951 + pages = {220--229},
1952 + year = {2018},
1953 + publisher = {Elsevier}
1954 +}
1955 +
1956 +@article{koop2011forecasting,
1957 + title={Forecasting with medium and large {B}ayesian {VARs}},
1958 + author={Koop, Gary},
1959 + journal={Journal of Applied Econometrics},
1960 + volume={28},
1961 + number={2},
1962 + pages={177--203},
1963 + year={2013},
1964 + publisher={Wiley}
1965 +}
1966 +
1967 +@article{koop2013forecasting,
1968 + title={Forecasting with medium and large {B}ayesian {VARs}},
1969 + author={Koop, Gary},
1970 + journal={Journal of Applied Econometrics},
1971 + volume={28},
1972 + number={2},
1973 + pages={177--203},
1974 + year={2013},
1975 + publisher={Wiley}
1976 +}
1977 +
1978 +@article{kothari2007econometrics,
1979 + title = {Econometrics of Event Studies},
1980 + author = {Kothari, Sagar P. and Warner, Jerold B.},
1981 + booktitle = {Handbook of Empirical Corporate Finance},
1982 + pages = {3--36},
1983 + year = {2007},
1984 + publisher = {Elsevier}
1985 +}
1986 +
1987 +@article{krause2014exchange,
1988 + title={Exchange-traded funds, liquidity and volatility},
1989 + author={Krause, Tanja and Ehsani, Shahram and Lien, Donald},
1990 + journal={Applied Financial Economics},
1991 + volume={24},
1992 + number={24},
1993 + pages={1617--1630},
1994 + year={2014},
1995 + publisher={Taylor \& Francis}
1996 +}
1997 +
1998 +@article{krehbiel2019real,
1999 + title={Real-time price discovery via verbal communication: Method and application},
2000 + author={Krehbiel, Tim and Adkins, Lee C},
2001 + journal={Journal of Futures Markets},
2002 + volume={39},
2003 + number={5},
2004 + pages={596--613},
2005 + year={2019},
2006 + publisher={Wiley}
2007 +}
2008 +
2009 +@article{kurov2018monetary,
2010 + title={Monetary policy uncertainty and the market reaction to macroeconomic news},
2011 + author={Kurov, Alexander and Stan, Raluca},
2012 + journal={Journal of Banking \& Finance},
2013 + volume={86},
2014 + pages={127--142},
2015 + year={2018},
2016 + publisher={Elsevier}
2017 +}
2018 +
2019 +@article{kurov2019price,
2020 + title={Price drift before US macroeconomic news: Private information about public announcements?},
2021 + author={Kurov, Alexander and Sancetta, Alessio and Strasser, Georg and Wolfe, Marketa Halova},
2022 + journal={Journal of Financial and Quantitative Analysis},
2023 + volume={54},
2024 + number={1},
2025 + pages={449--479},
2026 + year={2019},
2027 + publisher={Cambridge University Press}
2028 +}
2029 +
2030 +@article{kuttner2001,
2031 + title={Monetary Policy Surprises and Interest Rates: Evidence from the Fed Funds Futures Market},
2032 + author={Kuttner, Kenneth N},
2033 + journal={Journal of Monetary Economics},
2034 + volume={47},
2035 + number={3},
2036 + pages={523--544},
2037 + year={2001},
2038 + publisher={Elsevier}
2039 +}
2040 +
2041 +@article{kyle1985continuous,
2042 + title = {Continuous Auctions and Insider Trading},
2043 + author = {Kyle, Albert S},
2044 + journal = {Econometrica},
2045 + volume = {53},
2046 + number = {6},
2047 + pages = {1315--1335},
2048 + year = {1985},
2049 + publisher = {JSTOR}
2050 +}
2051 +
2052 +@article{lagi2011food,
2053 + author = {Lagi, Marco and Bertrand, Karla and Bar‐Yam, Yaneer},
2054 + title = {The Food Crises and Political Instability in North Africa and the Middle East},
2055 + journal = {SSRN Electronic Journal},
2056 + year = {2011},
2057 + doi = {10.2139/ssrn.1910031}
2058 +}
2059 +
2060 +@article{lee1991inferring,
2061 + title={Inferring trade direction from intraday data},
2062 + author={Lee, Charles MC and Ready, Mark J},
2063 + journal={The Journal of Finance},
2064 + volume={46},
2065 + number={2},
2066 + pages={733--746},
2067 + year={1991},
2068 + publisher={Wiley}
2069 +}
2070 +
2071 +@article{lee1995oil,
2072 + title={Oil shocks and the macroeconomy: the role of price variability},
2073 + author={Lee, Kiseok and Ni, Shawn and Ratti, Ronald A},
2074 + journal={The Energy Journal},
2075 + volume={16},
2076 + number={4},
2077 + pages = {39--56},
2078 + year={1995},
2079 + publisher={International Association for Energy Economics}
2080 +}
2081 +
2082 +@article{lettau2018exchange,
2083 + title={Exchange-traded funds 101 for economists},
2084 + author={Lettau, Martin and Madhavan, Ananth},
2085 + journal={Journal of Economic Perspectives},
2086 + volume={32},
2087 + number={1},
2088 + pages={135--154},
2089 + year={2018},
2090 + publisher={American Economic Association}
2091 +}
2092 +
2093 +@article{lin2010opec,
2094 + author = {Lin, Sharon Xiaowen and Tamvakis, Michael},
2095 + doi = {10.1016/j.enpol.2009.10.053},
2096 + title = {OPEC Announcements and Their Effects on Crude Oil Prices},
2097 + journal = {Energy Policy},
2098 + volume = {38},
2099 + number = {2},
2100 + pages = {1010--1016},
2101 + year = {2010}
2102 +}
2103 +
2104 +@article{litterman1986forecasting,
2105 + title={Forecasting with {B}ayesian vector autoregressions—five years of experience},
2106 + author={Litterman, Robert B},
2107 + journal={Journal of Business \& Economic Statistics},
2108 + volume={4},
2109 + number={1},
2110 + pages={25--38},
2111 + year={1986},
2112 + publisher={Taylor \& Francis}
2113 +}
2114 +
2115 +@article{liu2015does,
2116 + title={Does anything beat 5-minute {RV}? {A} comparison of realized measures across multiple asset classes},
2117 + author={Liu, Lily Y and Patton, Andrew J and Sheppard, Kevin},
2118 + journal={Journal of Econometrics},
2119 + volume={187},
2120 + number={1},
2121 + pages={293--311},
2122 + year={2015},
2123 + publisher={Elsevier}
2124 +}
2125 +
2126 +@article{loughran2011,
2127 + title={When is a Liability not a Liability? Textual Analysis, Dictionaries, and 10-Ks},
2128 + author={Loughran, Tim and McDonald, Bill},
2129 + journal={Journal of Finance},
2130 + volume={66},
2131 + number={1},
2132 + pages={35--65},
2133 + year={2011},
2134 + publisher={Wiley Online Library}
2135 +}
2136 +
2137 +@article{lucca2012,
2138 + title={The Pre-FOMC Announcement Drift},
2139 + author={Lucca, David O and Moench, Emanuel},
2140 + journal={Journal of Finance},
2141 + volume={70},
2142 + number={1},
2143 + pages={329--371},
2144 + year={2015},
2145 + publisher={Wiley Online Library}
2146 +}
2147 +
2148 +@article{lucey2015precious,
2149 + title = {What Precious Metals Act as Safe Havens, and When? Some {US} Evidence},
2150 + author = {Lucey, Brian M and Li, Sile},
2151 + journal = {Applied Economics Letters},
2152 + volume = {22},
2153 + number = {1},
2154 + pages = {35--45},
2155 + year = {2015},
2156 + publisher = {Taylor \& Francis}
2157 +}
2158 +
2159 +@article{lynch2021market,
2160 + title={Market maker of last resort: The {F}ederal {R}eserve's role in the {COVID}-19 bond market crisis},
2161 + author={Lynch, David P},
2162 + journal={Journal of Financial Economics},
2163 + volume={142},
2164 + number={2},
2165 + pages={588--605},
2166 + year={2021},
2167 + publisher={Elsevier}
2168 +}
2169 +
2170 +@article{madhavan2012exchange,
2171 + title={Exchange-traded funds, market structure, and the flash crash},
2172 + author={Madhavan, Ananth},
2173 + journal={Financial Analysts Journal},
2174 + volume={68},
2175 + number={4},
2176 + pages={20--35},
2177 + year={2012},
2178 + publisher={CFA Institute}
2179 +}
2180 +
2181 +@article{malamud2016portfolio,
2182 + title={Portfolio delegation and market efficiency},
2183 + author={Malamud, Semyon},
2184 + journal={The Journal of Finance},
2185 + volume={71},
2186 + number={6},
2187 + pages={2529--2574},
2188 + year={2016},
2189 + publisher={Wiley}
2190 +}
2191 +
2192 +@article{manela2017,
2193 + title={News Implied Volatility and Disaster Concerns},
2194 + author={Manela, Asaf and Moreira, Alan},
2195 + journal={Journal of Financial Economics},
2196 + volume={123},
2197 + number={1},
2198 + pages={137--162},
2199 + year={2017},
2200 + publisher={Elsevier}
2201 +}
2202 +
2203 +@article{masters2009testimony,
2204 + title = {Testimony Before the Commodity Futures Trading Commission},
2205 + author = {Masters, Michael W.},
2206 + journal = {Testimony to the Commodity Futures Trading Commission},
2207 + year = {2009}
2208 +}
2209 +
2210 +@article{muller1997volatilities,
2211 + title={Volatilities of different time resolutions—analyzing the dynamics of market components},
2212 + author={Müller, Ulrich A and Dacorogna, Michel M and Davé, Rakhal D and Olsen, Richard B and Pictet, Olivier V and von Weizsäcker, Jacob E},
2213 + journal={Journal of Empirical Finance},
2214 + volume={4},
2215 + number={2-3},
2216 + pages={213--239},
2217 + year={1997},
2218 + publisher={Elsevier}
2219 +}
2220 +
2221 +@article{muth1961,
2222 + title={Rational Expectations and the Theory of Price Movements},
2223 + author={Muth, John F},
2224 + journal={Econometrica},
2225 + volume={29},
2226 + number={3},
2227 + pages={315--335},
2228 + year={1961},
2229 + publisher={JSTOR}
2230 +}
2231 +
2232 +@article{nakamura2018,
2233 + title={High-Frequency Identification of Monetary Non-neutrality: The Information Effect},
2234 + author={Nakamura, Emi and Steinsson, J{\'o}n},
2235 + journal={Quarterly Journal of Economics},
2236 + volume={133},
2237 + number={3},
2238 + pages={1283--1330},
2239 + year={2018},
2240 + publisher={Oxford University Press}
2241 +}
2242 +
2243 +@techreport{NBERw19523,
2244 + title = "Time Variation in Asset Price Responses to Macro Announcements",
2245 + author = "Goldberg, Linda S and Grisse, Christian",
2246 + institution = "National Bureau of Economic Research",
2247 + type = "Working Paper",
2248 + series = "Working Paper Series",
2249 + number = "19523",
2250 + year = "2013",
2251 + month = "October",
2252 + doi = {10.3386/w19523},
2253 + URL = "http://www.nber.org/papers/w19523",
2254 + abstract = {Although the effects of economic news announcements on asset prices are well established, these relationships are unlikely to be stable. This paper documents the time variation in the responses of yield curves and exchange rates using high frequency data from January 2000 through August 2011. Significant time variation in news effects is present for those announcements that have the largest effects on asset prices. The time variation in effects is explained by economic conditions, including the level of policy rates at the time of the release, and risk conditions: government bond yields increase in response to "good news", but less so when risk is elevated. Risk conditions matter since they can capture the effects of uncertainty on the information content of news announcements, the interaction of monetary policy and financial stability objectives of central banks, and the effect of news announcements on the risk premium.},
2255 +}
2256 +
2257 +@article{nelson1991conditional,
2258 + title={Conditional heteroskedasticity in asset returns: A new approach},
2259 + author={Nelson, Daniel B},
2260 + journal={Econometrica},
2261 + pages={347--370},
2262 + year={1991},
2263 + publisher={JSTOR}
2264 +}
2265 +
2266 +@misc{newey1986simple,
2267 + title = {A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix},
2268 + author = {Newey, Whitney K. and West, Kenneth D.},
2269 + year = {1986},
2270 + publisher = {National Bureau of Economic Research, Cambridge, MA, USA}
2271 +}
2272 +
2273 +@article{newey1994automatic,
2274 + title={Automatic lag selection in covariance matrix estimation},
2275 + author={Newey, Whitney K and West, Kenneth D},
2276 + journal={Review of Economic Studies},
2277 + volume={61},
2278 + number={4},
2279 + pages={631--653},
2280 + year={1994},
2281 + publisher={Wiley-Blackwell}
2282 +}
2283 +
2284 +@article{ohara2021etf,
2285 + title={Anatomy of a liquidity crisis: Corporate bonds in the {COVID}-19 crisis},
2286 + author={O'Hara, Maureen and Zhou, Xing Alex},
2287 + journal={Journal of Financial Economics},
2288 + volume={142},
2289 + number={1},
2290 + pages={46--68},
2291 + year={2021},
2292 + publisher={Elsevier}
2293 +}
2294 +
2295 +@article{ozdurak2020price,
2296 + title={Price discovery in crude oil markets: Intraday volatility interactions between crude oil futures and energy exchange traded funds},
2297 + author={Ozdurak, Caner and Ulusoy, Veysel},
2298 + journal={International Journal of Energy Economics and Policy},
2299 + volume={10},
2300 + number={3},
2301 + pages={402--413},
2302 + year={2020}
2303 +}
2304 +
2305 +@article{pal2023agricultural,
2306 + author = {Pal, Dipanwita},
2307 + title = {Do Agricultural Prices Respond to Interest on Reserves?},
2308 + journal = {The Economics and Finance Letters},
2309 + volume = {10},
2310 + number = {1},
2311 + pages = {94--101},
2312 + year = {2023},
2313 + doi = {10.18488/29.v10i1.3311}
2314 +}
2315 +
2316 +@article{pan2016etf,
2317 + title={{ETF} arbitrage under liquidity mismatch},
2318 + author={Pan, Kevin and Zeng, Yao},
2319 + journal={Journal of Financial Economics},
2320 + volume={120},
2321 + number={3},
2322 + pages={617--635},
2323 + year={2016},
2324 + publisher={Elsevier}
2325 +}
2326 +
2327 +@article{pan2017market,
2328 + title={Market disruptions and destabilizing speculation},
2329 + author={Pan, Kevin and Zeng, Yao},
2330 + journal={Journal of Financial Economics},
2331 + volume={126},
2332 + number={1},
2333 + pages={450--471},
2334 + year={2017},
2335 + publisher={Elsevier}
2336 +}
2337 +
2338 +@article{park2019speculators,
2339 + author = {Park, Jinwook},
2340 + title = {Effect of Speculators’ Position Changes on the LME Futures Market},
2341 + journal = {International Journal of Financial Studies},
2342 + volume = {7},
2343 + number = {2},
2344 + pages = {32},
2345 + year = {2019},
2346 + doi = {10.3390/ijfs7020032}
2347 +}
2348 +
2349 +@article{patton2013,
2350 + title={Why Do Markets Disagree? Evidence from Variation in Opinion},
2351 + author={Patton, Andrew J and Verardo, Michela},
2352 + journal={Review of Financial Studies},
2353 + volume={25},
2354 + number={12},
2355 + pages={3734--3773},
2356 + year={2012},
2357 + publisher={Oxford University Press}
2358 +}
2359 +
2360 +@techreport{pearce1984stock,
2361 + title={Stock prices and economic news},
2362 + author={Pearce, Douglas K and Roley, V Vance},
2363 + year={1984},
2364 + institution={National Bureau of Economic Research}
2365 +}
2366 +
2367 +@article{pescatori2022opec,
2368 + author = {Pescatori, Andrea and Nazer, Yousef},
2369 + doi = {10.5089/9798400219788.001},
2370 + title = {OPEC and the Oil Market},
2371 + journal = {IMF Working Paper},
2372 + year = {2022}
2373 +}
2374 +
2375 +@article{petajisto2017inefficiencies,
2376 + title={Inefficiencies in the pricing of exchange-traded funds},
2377 + author={Petäjistö, Antti},
2378 + journal={Financial Analysts Journal},
2379 + volume={73},
2380 + number={1},
2381 + pages={24--54},
2382 + year={2017},
2383 + publisher={CFA Institute}
2384 +}
2385 +
2386 +@book{pirrong2011commodity,
2387 + title = {Commodity Price Dynamics: A Structural Approach},
2388 + author = {Pirrong, Craig},
2389 + year = {2011},
2390 + publisher = {Cambridge University Press}
2391 +}
2392 +
2393 +@article{pontiff1996costly,
2394 + title={Costly arbitrage: Evidence from closed-end funds},
2395 + author={Pontiff, Jeffrey},
2396 + journal={The Quarterly Journal of Economics},
2397 + volume={111},
2398 + number={4},
2399 + pages={1135--1151},
2400 + year={1996},
2401 + publisher={Oxford University Press}
2402 +}
2403 +
2404 +@article{primiceri2005time,
2405 + title={Time varying structural vector autoregressions and monetary policy},
2406 + author={Primiceri, Giorgio E},
2407 + journal={The Review of Economic Studies},
2408 + volume={72},
2409 + number={3},
2410 + pages={821--852},
2411 + year={2005},
2412 + publisher={Oxford University Press}
2413 +}
2414 +
2415 +@article{rakowski2021twitter,
2416 + title = {Twitter Activity, Investor Attention, and the Diffusion of Information},
2417 + author = {Rakowski, David and Shirley, Sara E and Stark, Jeffrey R},
2418 + journal = {Financial Management},
2419 + volume = {50},
2420 + number = {1},
2421 + pages = {3--46},
2422 + year = {2021},
2423 + publisher = {Wiley Online Library}
2424 +}
2425 +
2426 +@article{ready2022order,
2427 + title={Order flows and financial investor impacts in commodity futures markets},
2428 + author={Ready, Mark J and Ready, Robert C},
2429 + journal={The Review of Financial Studies},
2430 + volume={35},
2431 + number={10},
2432 + pages={4712--4755},
2433 + year={2022},
2434 + publisher={Oxford University Press}
2435 +}
2436 +
2437 +@article{richie2008examination,
2438 + title={The limits to stock index arbitrage: Examining {S\&P} 500 futures and {SPDRs}},
2439 + author={Richie, Nivine and Daigler, Robert T and Gleason, Kimberly C},
2440 + journal={Journal of Futures Markets},
2441 + volume={28},
2442 + number={12},
2443 + pages={1182--1205},
2444 + year={2008},
2445 + publisher={Wiley}
2446 +}
2447 +
2448 +@article{rlmishra2021financialization,
2449 + author = {Mishra, Ramesh and Mishra, Alok},
2450 + title = {Financialization of Indian Agricultural Commodities: The Case of Index Investments},
2451 + journal = {International Journal of Social Economics},
2452 + volume = {49},
2453 + number = {1},
2454 + pages = {73--96},
2455 + year = {2021},
2456 + doi = {10.1108/ijse-05-2021-0254}
2457 +}
2458 +
2459 +@article{roll1984,
2460 + author = {Roll, R.},
2461 + title = {A Simple Implicit Measure of the Effective Bid-Ask Spread},
2462 + journal = {The Journal of Finance},
2463 + volume = {39},
2464 + number = {4},
2465 + pages = {1127--1139},
2466 + year = {1984},
2467 +}
2468 +
2469 +@article{rosa2013,
2470 + title={The High-Frequency Response of Exchange Rates and Interest Rates to Macroeconomic Announcements},
2471 + author={Rosa, Carlo},
2472 + journal={Journal of Banking \& Finance},
2473 + volume={37},
2474 + number={6},
2475 + pages={2162--2174},
2476 + year={2013},
2477 + publisher={Elsevier}
2478 +}
2479 +
2480 +@article{sanders2010speculative,
2481 + title={A speculative bubble in commodity futures prices? Cross-sectional evidence},
2482 + author={Sanders, Dwight R and Irwin, Scott H},
2483 + journal={Agricultural Economics},
2484 + volume={41},
2485 + number={1},
2486 + pages={25--32},
2487 + year={2010},
2488 + publisher={Wiley Online Library}
2489 +}
2490 +
2491 +@article{sanders2016bubbles,
2492 + title = {Bubbles, Froth and Facts: Another Look at the Masters Hypothesis in Commodity Futures Markets},
2493 + author = {Sanders, Dwight and Irwin, Scott},
2494 + journal = {Journal of Agricultural Economics},
2495 + volume = {68},
2496 + number = {2},
2497 + pages = {345--365},
2498 + year = {2016},
2499 + doi = {10.1111/1477-9552.12191}
2500 +}
2501 +
2502 +@article{savor2013much,
2503 + title={How much do investors care about macroeconomic risk? Evidence from scheduled economic announcements},
2504 + author={Savor, Pavel and Wilson, Mungo},
2505 + journal={Journal of Financial and Quantitative Analysis},
2506 + pages={343--375},
2507 + year={2013},
2508 + publisher={JSTOR}
2509 +}
2510 +
2511 +@article{savor2014,
2512 + title={Asset Pricing: A Tale of Two Days},
2513 + author={Savor, Pavel and Wilson, Mungo},
2514 + journal={Journal of Financial Economics},
2515 + volume={113},
2516 + number={2},
2517 + pages={171--201},
2518 + year={2014},
2519 + publisher={Elsevier}
2520 +}
2521 +
2522 +@article{schmeling2019,
2523 + title={Does Central Bank Tone Move Asset Prices?},
2524 + author={Schmeling, Maik and Wagner, Christian},
2525 + journal={Review of Finance},
2526 + volume={23},
2527 + number={5},
2528 + pages={933--972},
2529 + year={2019},
2530 + publisher={Oxford University Press}
2531 +}
2532 +
2533 +@article{scholtus2014speed,
2534 + title={Speed, algorithmic trading, and market quality around macroeconomic news announcements},
2535 + author={Scholtus, Martin and Van Dijk, Dick and Frijns, Bart},
2536 + journal={Journal of Banking \& Finance},
2537 + volume={38},
2538 + pages={89--105},
2539 + year={2014},
2540 + publisher={Elsevier}
2541 +}
2542 +
2543 +@article{SCOTTI20161,
2544 +title = {Surprise and uncertainty indexes: Real-time aggregation of real-activity macro-surprises},
2545 +journal = {Journal of Monetary Economics},
2546 +volume = {82},
2547 +pages = {1-19},
2548 +year = {2016},
2549 +issn = {0304-3932},
2550 +doi = {https://doi.org/10.1016/j.jmoneco.2016.06.002},
2551 +url = {https://www.sciencedirect.com/science/article/pii/S0304393216300320},
2552 +author = {Chiara Scotti},
2553 +keywords = {Business cycle, Dynamic factor model, State space model, Forecasting weights},
2554 +abstract = {Two daily, real-time, real-activity indexes are constructed for the United States, euro area, United Kingdom, Canada, and Japan: (i) a surprise index summarizing recent economic data surprises and measuring optimism/pessimism about the state of the economy, and (ii) an uncertainty index measuring uncertainty related to the state of the economy. The surprise index parsimoniously preserves the properties of the underlying series when affecting asset prices. For the United States, the real-activity uncertainty index is compared to other uncertainty proxies to show that, when uncertainty is strictly related to real activity only, it has a potentially milder effect on economic activity.}
2555 +}
2556 +
2557 +@article{scotti2016surprise,
2558 + title = {Surprise and Uncertainty Indexes: Real-Time Aggregation of Real-Activity Macro-Surprises},
2559 + journal = {Journal of Monetary Economics},
2560 + volume = {82},
2561 + pages = {1--19},
2562 + year = {2016},
2563 + issn = {0304-3932},
2564 + doi = {https://doi.org/10.1016/j.jmoneco.2016.06.002},
2565 + url = {https://www.sciencedirect.com/science/article/pii/S0304393216300320},
2566 + author = {Scotti, Chiara},
2567 + keywords = {Business cycle, Dynamic factor model, State space model, Forecasting weights},
2568 + abstract = {Two daily, real-time, real-activity indexes are constructed for the United States, euro area, United Kingdom, Canada, and Japan: (i) a surprise index summarizing recent economic data surprises and measuring optimism/pessimism about the state of the economy, and (ii) an uncertainty index measuring uncertainty related to the state of the economy. The surprise index parsimoniously preserves the properties of the underlying series when affecting asset prices. For the United States, the real-activity uncertainty index is compared to other uncertainty proxies to show that, when uncertainty is strictly related to real activity only, it has a potentially milder effect on economic activity.}
2569 +}
2570 +
2571 +@article{shanker2017new,
2572 + title = {New Indices of Adequate and Excess Speculation and Their Relationship with Volatility in the Crude Oil Futures Market},
2573 + author = {Shanker, Latha},
2574 + journal = {Journal of Commodity Markets},
2575 + volume = {5},
2576 + pages = {18--35},
2577 + year = {2017},
2578 + publisher = {Elsevier}
2579 +}
2580 +
2581 +@article{shapiro2019,
2582 + title={Taking the Fed at its Word: A New Approach to Estimating Central Bank Objectives using Text Analysis},
2583 + author={Shapiro, Adam Hale and Sudhof, Moritz and Wilson, Daniel J},
2584 + journal={Review of Economics and Statistics},
2585 + volume={104},
2586 + number={4},
2587 + pages={768--784},
2588 + year={2022},
2589 + publisher={MIT Press}
2590 +}
2591 +
2592 +@article{shin2013basis,
2593 + title={Basis and trading volume in international {ETF} markets},
2594 + author={Shin, Sujeong and Soydemir, Gokce},
2595 + journal={Journal of Futures Markets},
2596 + volume={33},
2597 + number={10},
2598 + pages={967--992},
2599 + year={2013},
2600 + publisher={Wiley}
2601 +}
2602 +
2603 +@article{shleifer1990noise,
2604 + title={The noise trader approach to finance},
2605 + author={Shleifer, Andrei and Summers, Lawrence H},
2606 + journal={Journal of Economic perspectives},
2607 + volume={4},
2608 + number={2},
2609 + pages={19--33},
2610 + year={1990},
2611 + publisher={American Economic Association}
2612 +}
2613 +
2614 +@article{sims1999error,
2615 + title={Error bands for impulse responses},
2616 + author={Sims, Christopher A and Zha, Tao},
2617 + journal={Econometrica},
2618 + volume={67},
2619 + number={5},
2620 + pages={1113--1155},
2621 + year={1999},
2622 + publisher={Wiley}
2623 +}
2624 +
2625 +@article{singleton2014investor,
2626 + author = {Singleton, Kenneth},
2627 + title = {Investor Flows and the 2008 Boom/Bust in Oil Prices},
2628 + journal = {Management Science},
2629 + volume = {60},
2630 + number = {2},
2631 + pages = {300--318},
2632 + year = {2014},
2633 + doi = {10.1287/mnsc.2013.1756}
2634 +}
2635 +
2636 +@article{soana2020index,
2637 + author = {Soana, Maria and Verga, Giovanna and Volpi, Maurizio},
2638 + title = {Did Index Trader and Swap Dealer Activity Produce a Bubble in the Agricultural Commodity Market?},
2639 + journal = {African Journal of Business Management},
2640 + volume = {14},
2641 + number = {1},
2642 + pages = {9--24},
2643 + year = {2020},
2644 + doi = {10.5897/ajbm2019.8877}
2645 +}
2646 +
2647 +@article{staer2017asset,
2648 + title={Asset management via {ETFs}},
2649 + author={Staer, Alexander},
2650 + journal={The Review of Financial Studies},
2651 + volume={30},
2652 + number={9},
2653 + pages={3225--3264},
2654 + year={2017},
2655 + publisher={Oxford University Press}
2656 +}
2657 +
2658 +@article{stoll2010commodity,
2659 + title={Commodity index investing and commodity futures prices},
2660 + author={Stoll, Hans R and Whaley, Robert E},
2661 + journal={Journal of Applied Finance (Formerly Financial Practice and Education)},
2662 + volume={20},
2663 + number={1},
2664 + year={2010}
2665 +}
2666 +
2667 +@article{swanson2014,
2668 + title={Measuring the Effect of the Zero Lower Bound on Medium- and Longer-Term Interest Rates},
2669 + author={Swanson, Eric T and Williams, John C},
2670 + journal={American Economic Review},
2671 + volume={104},
2672 + number={10},
2673 + pages={3154--3185},
2674 + year={2014},
2675 + publisher={American Economic Association}
2676 +}
2677 +
2678 +@article{tang2012index,
2679 + title = {Index Investment and the Financialization of Commodities},
2680 + author = {Tang, Ke and Xiong, Wei},
2681 + journal = {Financial Analysts Journal},
2682 + volume = {68},
2683 + number = {6},
2684 + pages = {54--74},
2685 + year = {2012},
2686 + publisher = {Taylor \& Francis}
2687 +}
2688 +
2689 +@article{todorov2021etf,
2690 + title={When passive funds affect prices: Evidence from volatility and commodity {ETFs}},
2691 + author={Todorov, Karamfil},
2692 + journal={Review of Finance},
2693 + year={2024},
2694 + volume={28},
2695 + number={3},
2696 + pages={831--863},
2697 + publisher={Oxford University Press}
2698 +}
2699 +
2700 +@book{veldkamp2011,
2701 + title={Information Choice in Macroeconomics and Finance},
2702 + author={Veldkamp, Laura L},
2703 + year={2011},
2704 + publisher={Princeton University Press}
2705 +}
2706 +
2707 +@article{vivian2012commodity,
2708 + title = {Commodity Volatility Breaks},
2709 + author = {Vivian, Andrew and Wohar, Mark E.},
2710 + journal = {Journal of International Financial Markets, Institutions and Money},
2711 + volume = {22},
2712 + number = {2},
2713 + pages = {395--422},
2714 + year = {2012},
2715 + publisher = {Elsevier}
2716 +}
2717 +
2718 +@inproceedings{wang2021tsdae,
2719 + title={TSDAE: Using Transformer-based Sequential Denoising Auto-Encoder for Unsupervised Sentence Embedding Learning},
2720 + author={Wang, Kexin and Reimers, Nils and Gurevych, Iryna},
2721 + booktitle={Findings of the Association for Computational Linguistics: EMNLP 2021},
2722 + pages={671--688},
2723 + year={2021}
2724 +}
2725 +
2726 +@article{wirl2004impact,
2727 + title={The impact of {OPEC} Conference outcomes on world oil prices 1984-2001},
2728 + author={Wirl, Franz and Kujundzic, Azra},
2729 + journal={The Energy Journal},
2730 + volume={25},
2731 + number={1},
2732 + pages = {45--62},
2733 + year={2004},
2734 + publisher={International Association for Energy Economics}
2735 +}
2736 +
2737 +@article{wongswan2009,
2738 + title={The Response of Global Equity Indexes to U.S. Monetary Policy Announcements},
2739 + author={Wongswan, Jon},
2740 + journal={Journal of International Money and Finance},
2741 + volume={28},
2742 + number={2},
2743 + pages={344--365},
2744 + year={2009},
2745 + publisher={Elsevier}
2746 +}
2747 +
2748 +@article{working1949theory,
2749 + title = {The Theory of Price of Storage},
2750 + author = {Working, Holbrook},
2751 + journal = {American Economic Review},
2752 + volume = {39},
2753 + number = {6},
2754 + pages = {1254--1262},
2755 + year = {1949},
2756 + publisher = {JSTOR}
2757 +}
2758 +
2759 +@article{working1960speculation,
2760 + title = {Speculation on Hedging Markets},
2761 + author = {Working, Holbrook},
2762 + journal = {Food Research Institute Studies},
2763 + volume = {1},
2764 + number = {2},
2765 + pages = {185--220},
2766 + year = {1960}
2767 +}
2768 +
2769 +@article{wu2019does,
2770 + title = {Does Gold or Bitcoin Hedge Economic Policy Uncertainty?},
2771 + author = {Wu, Shan and Tong, Mu and Yang, Zhongyi and Derbali, Abdelkader},
2772 + journal = {Finance Research Letters},
2773 + volume = {31},
2774 + pages = {171--178},
2775 + year = {2019},
2776 + publisher = {Elsevier}
2777 +}
2778 +
2779 +@article{yang2005futures,
2780 + title = {Futures Trading Activity and Commodity Cash Price Volatility},
2781 + author = {Yang, Jian and Balyeat, R. Brian and Leatham, David J.},
2782 + journal = {Journal of Business Finance \& Accounting},
2783 + volume = {32},
2784 + number = {1-2},
2785 + pages = {297--323},
2786 + year = {2005},
2787 + publisher = {Wiley Online Library}
2788 +}
2789 +
2790 +@article{ye2021macroeconomic,
2791 + title = {Macroeconomic Forecasts and Commodity Futures Volatility},
2792 + author = {Ye, Wuyi and Guo, Ranran and Deschamps, Bruno and Jiang, Ying and Liu, Xiaoquan},
2793 + journal = {Economic Modelling},
2794 + volume = {94},
2795 + pages = {981--994},
2796 + year = {2021},
2797 + publisher = {Elsevier}
2798 +}
2799 +
2800 +@article{zhang2018informed,
2801 + title = {Informed Options Trading Prior to Dividend Change Announcements},
2802 + author = {Zhang, Jun},
2803 + journal = {Financial Management},
2804 + volume = {47},
2805 + number = {1},
2806 + pages = {81--103},
2807 + year = {2018},
2808 + publisher = {Wiley Online Library}
2809 +}
2810 +
2811 +@article{zhang2022hedging,
2812 + author = {Zhang, Tao},
2813 + title = {Hedging Pressure and Liquidity Provision in Commodity Options Markets},
2814 + journal = {Journal of Futures Markets},
2815 + volume = {42},
2816 + number = {7},
2817 + pages = {1212--1233},
2818 + year = {2022},
2819 + doi = {10.1002/fut.22327}
2820 +}
2821 +
2822 +@article{zhu2022oil,
2823 + title={Oil price shocks and stock market anomalies},
2824 + author={Zhu, Zhaobo and Sun, Licheng and Tu, Jun and Ji, Qiang},
2825 + journal={Financial Management},
2826 + volume={51},
2827 + number={2},
2828 + pages={573--612},
2829 + year={2022},
2830 + publisher={Wiley Online Library}
2831 +}
2832 +
2833 +@article{zorn1983sovereignty,
2834 + author = {Zorn, Stephen A},
2835 + doi = {10.1111/j.1477-8947.1983.tb00276.x},
2836 + title = {Permanent Sovereignty Over Natural Resources},
2837 + journal = {Natural Resources Forum},
2838 + volume = {7},
2839 + number = {4},
2840 + pages = {321--328},
2841 + year = {1983}
2842 +}
2843 +
2844 +@inproceedings{boons2014price,
2845 + author = {Martijn Boons and Frans A. de Roon and Marta Szymanowska},
2846 + title = {The Price of Commodity Risk in Stock and Futures Markets},
2847 + booktitle = {AFA 2012 Chicago Meetings Paper},
2848 + year = {2014},
2849 + note = {Available at SSRN: \url{https://ssrn.com/abstract=1785728} or \url{http://dx.doi.org/10.2139/ssrn.1785728}}
2850 +}
2851 +
added these-ulaval/chapitre1/chapitre1.tex +70 −0
@@ -0,0 +1,70 @@
1 +%% ---------------------------------------------------------------------------
2 +%% Chapitre 1 — adapté de phd_chap1_20260731/ (version révisée pour
3 +%% The Energy Journal). Corps scientifique inchangé ; voir INVENTAIRE.md pour
4 +%% la liste des adaptations mécaniques.
5 +%% Le sous-titre d'article « Revised version for The Energy Journal » et la
6 +%% note de remerciements de la page de titre de l'article ne sont pas repris
7 +%% ici : les remerciements liés à l'article figurent dans l'avant-propos.
8 +%% ---------------------------------------------------------------------------
9 +
10 +\chapter{Speculative Trading in Energy Markets: Evidence from Macroeconomic Surprises}
11 +\label{chap-chapitre1}
12 +
13 +\section{Résumé}
14 +
15 +% BROUILLON À RÉVISER PAR L'AUTEUR (traduction condensée de l'abstract ; la
16 +% FESP limite le résumé de chaque article inséré à 150 mots).
17 +\begin{otherlanguage*}{french}
18 + La spéculation sur les marchés de l'énergie et des matières premières est
19 + souvent accusée d'accroître la volatilité, de distordre les prix et de nuire
20 + à l'efficience des marchés. Nous adoptons une approche nouvelle, fondée sur
21 + les annonces macroéconomiques et des données à haute fréquence, pour mesurer
22 + l'effet de l'intensité de la spéculation. Nous étudions l'effet de
23 + 26~annonces macroéconomiques sur les contrats à terme énergétiques (pétrole
24 + brut, gaz naturel), comparés aux métaux (or, argent, cuivre, palladium). Une
25 + intensité spéculative plus élevée atténue l'effet des surprises
26 + macroéconomiques sur la dérive des prix, la volatilité et les écarts
27 + acheteur-vendeur : la spéculation améliore la liquidité et la découverte des
28 + prix tout en réduisant la volatilité. Cet effet d'atténuation est plus
29 + marqué pour les matières premières procycliques que pour l'or, valeur
30 + refuge. En désagrégeant par type de trader, ces effets bénéfiques
31 + proviennent des \emph{money managers} plutôt que des \emph{swap dealers}.
32 +\end{otherlanguage*}
33 +
34 +\section{Abstract}
35 +
36 +Speculative trading in energy and commodity markets has been blamed for
37 +increased volatility, price distortions and market inefficiency, with negative
38 +effects on the real economy. We take a new approach to investigate the impact
39 +of speculative trading using macroeconomic announcements and high-frequency
40 +data. We study the impact of 26 macroeconomic announcement releases on energy
41 +commodities (crude oil, natural gas) as our baseline case, which we contrast
42 +with metals (gold, silver, copper, and palladium). We find that increased
43 +speculative trading lessens the impact of macroeconomic surprises on futures
44 +markets, as measured by price drift, volatility, and bid-ask spreads. Our
45 +full-sample results show that increased trading by speculators improves
46 +liquidity and price discovery, while reducing volatility. We document a
47 +damping effect on volatility that is stronger for procyclical commodities such
48 +as crude oil and natural gas than for precious metals such as gold, which is a
49 +safe haven. In sub-sample analysis where we separate the effects of money
50 +managers and swap dealers, we find that the positive effects that we document
51 +are driven by money managers. Since traditional market participants prefer
52 +stability, our results suggest a beneficial impact of increased trading and
53 +speculation.
54 +
55 +\medskip
56 +\noindent\textbf{Keywords:} energy markets, crude oil, futures,
57 +high-frequency, speculation, trading, commercial, volatility, macroeconomic,
58 +announcements.
59 +
60 +%% --- Corps de l'article (sections Introduction à Conclusion) ---
61 +\input{chapitre1/corps}
62 +
63 +%% --- Tableaux et figures de l'article ---
64 +%% (les annexes A, B et C de l'article — full sample NLS, COVID, ZLB — sont
65 +%% désactivées dans la source d'origine : « available upon request »)
66 +\input{chapitre1/tables}
67 +
68 +\section{Figures}
69 +
70 +\input{chapitre1/figures}
added these-ulaval/chapitre1/corps.tex +442 −0
@@ -0,0 +1,442 @@
1 +\section{Introduction}
2 +\label{ch1:sec:introduction}
3 +
4 +%SETUP: EASIER TO TRADE COMMO FUTURES, IT’S BIG BUSINESS PROFITABLE, IT MATTERS
5 +The appeal of energy and other commodities as an asset class has grown since the Commodity Futures Modernization Act of 2000 (CFMA). By partially deregulating derivatives, the CFMA has made it easier to trade commodity futures contracts for investment purposes. Rather than invest in physicals or in shares of commodity-linked firms, investors can use futures to gain exposure to "commodity beta" \citep{boons2014price}. This evolution is particularly relevant for energy markets, where futures trading volume has grown tremendously. For instance, the total trading volume of commodity derivatives was 137.3 bn contracts in 2023, which is 64\% more than in 2022 (Futures Industry Association, 2024).
6 +An important reason for this trend is that commodities have periodically benefited from bull cycles (most notably in 2004-2008), attracting a growing number of speculators and institutional investors. As a result, the commodities asset class has become an important but volatile revenue source for trading firms and investment banks. %\footnote{Gross margins from commodities trading reached about \$57 bn during the 2007-09 period. While profits were more modest in the following years, they increased once more from \$36 bn in 2018 to \$100 bn in 2021 as a result of increased worldwide demand and supply chain bottlenecks, and up to \$150 bn in 2022. A substantial portion of these profits came from energy trading, particularly during periods of high oil price volatility.}
7 +Three firms, Goldman Sachs, Citi, and Macquarie earned together \$20 bn from commodities trading in 2022, much of it from energy-related contracts.\footnote{Sources: The Financial Times, Bloomberg, S\&P Global and Euronews.}
8 +
9 +
10 +%ARE FINANCIAL INVESTORS BAD? PRESSURE ON POLICYMAKERS, DESPITE WEAK EVIDENCE
11 +Is the presence of more financial investors harmful to traditional market participants, such as hedgers? While many think so, the evidence is unclear. The idea that poorly informed investors can disrupt markets has a long history \citep{shleifer1990noise} and has been revived in a recent theoretical literature on financialization \citep{basak2016model,goldstein2022commodity}. These papers are motivated by the commodity price run-up of 2004-2008, which occurred shortly after the CFMA was passed \citep{domanski2007financial}. Critics argue that the activities of financial investors, who are not directly involved in producing or processing commodities, can distort prices and increase volatility.\footnote{A high-profile example is \citet{masters2009testimony}, who testified before the U.S. Congress in 2008 and before the CFTC in 2009 about ``Ending excessive speculation in commodity markets.'' While his argument is not supported by empirical evidence, as shown by \citet{irwin2012testing}, the Masters hypothesis reflects beliefs held at the time by many market participants.} These critics claim that energy and commodity markets have become more sensitive to financial market fluctuations, and less to supply and demand fundamentals.
12 +Energy futures markets have attracted attention due to the popular perception that large price swings affect the real economy (e.g., through higher gasoline and heating costs) \citep{cheng2014financialization}. Research, however, generally does not support this claim \citep{baumeister2014oil}. Whether or not these fears are justified, policymakers have taken notice and the CFTC has progressively implemented rule changes such as new position limits.%\footnote{In 2009 and 2010, the CFTC held public hearings to review the impact of speculative trading on market integrity. In 2010, the CFTC proposed new regulations to impose position limits on commodity derivatives (e.g., number of contracts held by any single trader). The ``Final rule on position limits'' was only adopted on March 15, 2021.}
13 +
14 +
15 +%COMMODITY BULL CYCLE WHO IS TO BLAME, NOT CLEAR
16 +
17 +A large empirical literature debates the causes of periodic price and volatility run-ups in commodity and energy markets. Researchers emphasize the importance of differentiating between speculative traders and passive investors (e.g., index traders). The latter category of traders is more recent and trades energy and commodity contracts for diversification purposes rather than for speculative profit. While \citet{singleton2014investor} suggests that financial investors may be to blame for higher energy prices, \citet{kilian2014role} use a structural model to show that speculation can be ruled out as a cause of the oil price surge during 2003-2008 -- even though speculative demand played a role in previous oil price spikes. Further evidence against the hypothesis that index traders are responsible for the sharp increase in commodity prices is provided by \citet{irwin2011index} and \citet{irwin2012testing}. In a different strand of the literature, \citet{buyukcsahin2011speculators} use Granger causality tests and daily data to investigate whether speculators increase crude oil futures prices. They find little evidence to support that claim. Also using daily position-level data, \citet{brunetti2016speculators} show that speculators reduce price volatility in commodity and energy markets. Reviewing this early literature, \citet{fattouh2013role} conclude that speculation is unlikely to explain the commodity and energy bull cycle of 2004-2008.
18 +
19 +
20 +%RECENT FINANCIALIZATION PAPERS THEORY AND EMPIRICAL
21 +Recent research provides new theoretical grounds to establish how trading activity could affect energy and commodity prices \citep{basak2016model,goldstein2022commodity}. The subsequent empirical literature, however, does not reach a consensus. \citet{henderson2015new} use data on commodity-linked notes to show that uninformed trading flows affect commodity prices, but \citet{ready2022order} argue that the economic magnitude of this effect is too small to matter. Other recent papers find instances of futures price overshooting, reversals, and greater noise in markets \citep{da2024financialization}. They also find that commodities seem to display higher correlations with equities and with each other \citep{kang2023financialization}.
22 +
23 +
24 +%POSITIONING OUR PAPER AS ABOUT SPECULATIVE TRADING, NOT FINANCIALIZATION (REFEREE 2) %MH J'ai collé un petit paragraphe du haut avec celui-ci. Je pense que ça va et que c'est ensemble.
25 +
26 +
27 +
28 +%Although financialization has been linked to the trading activities of any non-traditional, financial investors in commodity markets (e.g., hedge funds), a more commonly accepted definition focuses on the activities of passive (index) traders \citep{tang2012index}. In contrast with speculation, which is likely to reflect informed trades, index traders are considered uninformed. Since this paper is motivated by the economic role played by all non-commercial traders, we will refer to speculative trading as a general description of this broad class of market participants. It is important to distinguish, however, speculators from index or institutional investors. It is the latter who are the more recent financial actors in energy and commodity markets \citep{irwin2011index}. Our paper therefore investigates the impact of speculative trading as well as sub-categories of non-commercial traders.%, rather than the narrower class of index traders.
29 +
30 +
31 +%WHAT WE DO AND WHY IT IS INNOVATIV
32 +Thus, our main contribution is to provide sharply identified evidence on the impact of speculative trading on energy (crude oil and natural gas) and metal markets (gold, silver, copper, and palladium), with additional evidence on sub-categories of traders. We focus on speculation rather than financialization, which has been linked to the trading activities of passive (index) traders \citep{tang2012index}. Specifically, our paper investigates the impact of speculative trading of sub-categories of non-commercial traders. Our rationale is that speculative trading is likely to reflect informed trades, which is our focus, in contrast to index traders who are considered uninformed. Energy commodities serve as our baseline case due to their economic importance and high trading volumes, while metals offer a useful comparison, particularly as gold is perceived as a safe-haven asset. We use high-frequency (5-minute) data to measure the instantaneous reaction of commodity futures returns, volatility, and bid-ask spreads to the surprise component in macroeconomic announcement releases \citep{andersen2007real, kurov2019price}. The data runs from April 4th, 2007, to February 11th, 2024. Our framework also accounts for the time-varying intensity of speculative trading activity. This study builds on \citet{kilian2011energy}, who find no evidence, at a daily frequency, that energy prices react to macroeconomic announcements. By using intraday data, we can better identify the impact of specific macro surprises. We also avoid a common criticism of event study methods, namely that using daily frequency data may reduce the power of statistical tests and could lead the researcher to misattribute the effect of a specific announcement, as other market events occur the same day \citep*{kothari2007econometrics}.
33 +
34 +
35 +%OUR FINDINGS, ONE BY ONE, BRIEFLY
36 +We find evidence of beneficial effects (price stability and market efficiency) from increased trading activity in energy and commodity markets. Our first finding is a damping effect on price reactions: while macro surprises generate a positive abnormal return for good news (and negative for bad news), the magnitude of this reaction is significantly weaker when speculative trading is higher. Second, we find a similar damping effect on volatility reactions. While all surprises (good or bad) generate a volatility increase, this reaction is lessened when the futures market shows more speculative trading. Third, we document lower bid-ask spreads when speculative trading is higher, controlling for the surprise environment. Fourth and last, these beneficial effects are linked to the trading activities of money managers. In contrast, increased trading by swap dealers appears to have an amplifying effect on reactions to macro surprises. These new insights are made possible by investigating this issue using a new angle, namely their sensitivity to macroeconomic surprises, and with high-frequency data. Our findings have important implications for energy market investment and regulation, and to the broader debate about speculation in energy and commodity markets.
37 +
38 +
39 +By investigating a broad range of traders in energy and commodity markets, our findings also extend the work of \citet{brunetti2016speculators}. They find that financial investors, especially money managers and hedge funds, help commodity markets by supplying liquidity, reducing volatility, and generally improving market efficiency. Moreover, our results relate to \citet{cheng2015convective} who show that financial investors, being better informed about markets, contribute to price discovery and liquidity. This is particularly relevant for energy markets, where accurate price discovery is crucial for physical market participants and investors. Thus, speculative traders help markets by distributing and assimilating new information into prices. These insights are valuable given the ongoing energy transition and the importance of efficient price discovery in energy markets. Two papers are probably closest to ours. First, \citet{brunetti2016speculators} who find that hedge funds add liquidity to commodity markets, resulting in more efficient prices and lower volatility. They argue that it is merchant positions (i.e., hedgers) that are linked to greater volatility, and that the presence of hedge funds allows for faster and more efficient price discovery. Second, using daily data, \citet{kilian2011energy} study how energy prices react to macroeconomic announcements. Our paper extends this work to high frequency data.
40 +
41 +
42 +
43 +
44 +
45 +
46 +
47 +
48 +
49 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
50 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
51 +\section{Background}
52 +% personnellement, je pense qu'on peut enlever le paragraphe en entier. le reste de la section s'applique bien je pense. et on en a parlé dans l'intro de ça.
53 +%\alert{\subsection{Speculative trading activity in energy and commodity futures markets}
54 +%In the literature, financialization usually refers to the process by which an asset class, such as commodities, attracts significantly more attention from financial investors. As a result, prices and volatility may no longer be determined only by commodity supply and demand \citep{cheng2014financialization}. \citet{brunetti2016speculators} refers to financialization as ``the changing mix of participant positions''. These financial investors include hedge funds, commodity trading firms, swap dealers and index traders, but specifically exclude hedgers. Traders may take commodity futures positions for purposes of speculation, diversification, or factor exposure to commodity beta. REVISE TO AVOID FINANCIALIZATION.}
55 +
56 +
57 +Price discovery in commodity markets occurs mainly in futures markets and is affected by informational frictions around supply and demand. Thus, risk sharing and information discovery represent a potential channel for financial investors to generate distortions in energy and commodity markets \citep{cheng2014financialization}. In their model, \citet{basak2016model} predict that the increased presence of financial actors can increase commodity futures volatility, as well as correlations between commodity and equity returns. \citet{goldstein2022commodity} also argue that under some conditions, a greater presence of financial investors can be harmful to commodity markets.
58 + Theory shows how a change in the participant mix in commodity markets could generate undesirable distortions, but the empirical literature is far from settled. \citet{singleton2014investor} argues that trading activity by financial investors creates informational frictions, leading commodity prices to become more volatile and to diverge from their fundamental values. Using a no-arbitrage argument, however, \citet*{hamilton2014risk} show that the positions of commodity traders included in index funds cannot be used to achieve excess returns in futures markets. \citet{ready2022order} show that while index traders do have a positive price impact, it is much too small to explain the apparent price distortions or bull cycles observed since 2004. In addition, financial investors do not have a uniform impact on market liquidity. Investors affect liquidity risk by either providing liquidity to meet the hedging needs of other traders or consuming liquidity when they trade for their own needs \citep{kang2020tale}. Indeed, \citet{brunetti2014commodity} show using data on commodity trader positions that index traders provide insurance against price risk.
59 +
60 +\subsection{Macroeconomic announcements}
61 +Surprises in macroeconomic announcements affect financial markets, whether in stocks \citep*{scholtus2014speed} or in bonds \citep{fleming1997moves}. In a key study, \citet*{balduzzi2001economic} find that 17 public news releases affect bond prices, trading volume, and bid-ask spreads. \citet{karali2014macro} show that energy futures markets exhibit asymmetric responses to macroeconomic news, with significant volatility spillovers between natural gas and crude oil markets. \citet{cao2024us} document a time-varying relationship between U.S. monetary policy and crude oil prices, finding that unexpected oil price increases can push monetary policy from expansionary to restrictive stance. \citet{kang2020economic} further show that after 2004, short-term oil price volatility is driven by industrial production, term spreads, and credit spreads, along with traditional market factors.
62 +
63 +
64 +The literature on commodity-specific announcements is smaller and less conclusive. \citet*{hollstein2020volatility} look at how different economic variables affect the term structure of commodity futures volatility. They show that speculation and jobs-related macro variables have the largest impact on volatility. \citet{zhu2022oil} further show that stock market anomalies can be explained to some extent by oil price shocks, separately from the effect of other macroeconomic variables and investor sentiment.
65 +While the literature finds a clear impact of macroeconomic announcements on stock and bond prices, there is no clear answer as to whether they affect commodity futures prices, or whether increased trading by financial participants accentuates these reactions. This issue is especially relevant for energy markets, given their macroeconomic importance. Our research provides new insights by using high-frequency data, expanding the set of announcements, and considering a time-varying measure of speculative trading intensity to capture trading activities for each of the commodities in the sample.
66 +
67 +
68 +
69 +
70 +
71 +
72 +
73 +
74 +
75 +
76 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
77 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
78 +\section{Data}
79 +
80 +We now present a detailed description of our data. Since this paper relies on several types of data, we describe: i) how to obtain the macroeconomic announcement surprises, ii) the commodity futures data, and iii) how to capture speculative trading activity.%, and iv) a PCA analysis for financial investor activity.
81 +
82 +\subsection{Data on macroeconomic announcements}
83 +
84 +The macroeconomic announcement release data are obtained from Bloomberg and Refinitiv Eikon. We collect information on 22 announcements that are standard to the literature \citep[see e.g.,][]{andersen2003micro}. Our sample for macroeconomic announcements is matched to our high-frequency data and therefore runs from April 2nd, 2007 to February 11th, 2024. The announcements belong to ten categories: Income, Employment, Industrial Activity, Investment, Consumption, Housing Sector, Government, Net Exports, Inflation, and Forward-looking. Most of the announcements are released on a monthly basis. Table \ref{ch1:tab:stat1} summarizes the announcements and provides more detail such as the number of observations, release frequency, source, unit of measure, and time of release. Bloomberg provides analyst forecasts for all announcements, as well as the actual value of the announcement release. For all announcements except the Consumer Price Index and Initial Jobless Claims releases, a positive surprise will be interpreted by investors as signaling a strong economy \citep*{fleming1997moves}.
85 +%Based on a survey of U.S. announcements and their impacts on financial markets, the only ones that signal a weaker economy are \citep*{fleming1997moves}. For the other announcements, a positive surprise will be interpreted by investors as signaling a strong economy.
86 +In addition, we include energy sector-specific announcements published by the U.S. Energy Information Administration. The first is the weekly crude oil storage report, which provides an update on the quantity of crude oil held in storage in the U.S. The second is the weekly natural gas storage report. We do not include OPEC announcements, as they cannot be reliably used in a high-frequency econometric design \citep{10.1257/aer.20190964}.\footnote{There are a few issues with the OPEC announcements: First, they are not released at a specific time. Second, it is impossible to know precisely when a given OPEC announcement was made available to investors. Third, OPEC's influence has weakened since the 1980s.}
87 +
88 +It is common practice in this literature to use the standardized surprise of an announcement rather than its realized value to quantify the unexpected component of the release. To calculate surprises, we follow \citet*{balduzzi2001economic}. Let $A_{kt}$ be the realized value (i.e., release) of macroeconomic announcement $k$ at time $t$, and let $E_{kt}$ be the median value of all Bloomberg analyst forecasts for announcement $k$ at time $t$. To standardize the surprise, we divide the raw surprise $(A_{kt} - E_{kt})$ by $\sigma_k$, the sample standard deviation of the surprise for announcement $k$. Thus, equation (\ref{ch1:eqn:SURPRISE}) describes the standardized surprise for announcement $k$ at time $t$:
89 +
90 +\begin{equation}\label{ch1:eqn:SURPRISE}
91 +S_{kt} = \frac{A_{kt} - E_{kt}}{\sigma_k}
92 +\end{equation}
93 +
94 +The sample period is used to compute $\sigma_k$, as in \citet{balduzzi2001economic} and \citet{kurov2019price}.\footnote{The literature argues that measuring $\sigma_k$ in this way is reasonable because the standardized surprise is not used for forecasting purposes. Using raw surprises is not recommended due to scaling issues, nor is using analyst dispersion for $\sigma_k$ because announcement coverage sometimes involves only a few analysts. For robustness, we also estimate our models using surprises where $\sigma_k$ is computed using only past observations. The main findings are unchanged. In this case, we exclude the first $M$ observations (e.g., $M=10$) to get a reasonable sample size for $\sigma_k$.} Table \ref{ch1:tab:stat2} presents the minimum, 1st quartile, median, mean, 3rd quartile, and maximum of the surprise for each announcement.
95 +
96 +
97 +
98 +
99 +
100 +
101 +
102 +
103 +
104 +
105 +\subsection{Commodity futures price data}
106 +
107 +For intraday data on commodity futures prices, we use Barchart's API.\footnote{See the \url{https://www.barchart.com/futures} website.} Our dataset for prices contains some of the most economically significant commodity futures contracts traded in the U.S. We use a high-frequency price series that runs from April 2nd, 2007 to February 11, 2024. Among these contracts, crude oil and natural gas are pro-cyclical, while gold and silver behave as safe havens. High-grade copper and palladium are industrial metals used in the manufacturing of consumer products.
108 +
109 +For each of the commodities in our sample, price returns $R_t$ are calculated as the log return over a 5-minute period $(\tau=5)$ beginning at time $t$. The database provides the futures contract close price ($p_{t}^{close}$) of each 5-minute period. Thus, $R_t$ is obtained as in equation (\ref{ch1:eqn:RETURN}):
110 +
111 +\begin{equation}\label{ch1:eqn:RETURN}
112 +R_t^{t+\tau} = \ln \left( \frac{p_{t+\tau}^{close}}{p_{t}^{close}} \right) = \ln (p_{t+\tau}^{close}) - \ln(p_{t}^{close})
113 +\end{equation}
114 +
115 +Descriptive statistics for the 5-minute log returns are presented in Table \ref{ch1:tab:stat4}. The most extreme outlier observations belong to crude oil, while gold has the fewest outliers.\footnote{The main findings are robust to using different window lengths by estimating equation \ref{ch1:eqn:RETURN} using 30-minute returns.}
116 +
117 +\subsection{Measures of speculative trading activity and trader categories}
118 +%To measure the impact of financialization, we need a measure that captures the intensity of speculation in commodity markets.
119 +
120 +
121 +%We investigate the impact of speculative trading activity increasing relative to productive activity. % We consider proxies designed to capture the intensity of speculation in commodity markets and we use them as indicators of financial investor activity.
122 +%For instance, swap dealers are not speculators, but they belong to our broad definition of financialization.
123 +The index of speculative trading is constructed using data in the \emph{Commitment of Traders (CoT) Report} published weekly by the Commodity Futures Trading Commission (CFTC). The data provided by the CFTC includes the number of positions held by different types of participants in commodity markets. The CFTC separates trader types as follows: \textit{Commercials} refer to trader-reported futures positions which the trader claims are used for hedging purposes, while \textit{Non-Commercials} is obtained by subtracting the total long and short commercial positions from the total open interest.\footnote{The CFTC defines commercial traders as participants in commodity markets who primarily use futures contracts to hedge their business activities (e.g., buying or selling commodities). All traders who are not classified as Commercial are automatically classified as Non-Commercial traders. To obtain the number of long positions held by Non-Commercial traders, we subtract the total long Commercial positions from the total open interest. For the number of short positions held by Non-Commercial traders, we subtract the total short Commercial Positions from the total open interest.}
124 +%Given that $CoT$ reports are not available in real time, they are not considered to be sources of information that traders could act upon, but rather a way to capture the state of the market.
125 +We use the following information presented in the $CoT$ report: for a futures contract $i$, the number of long and short positions held by Non-Commercial traders are $SL_i$ and $SS_i$, respectively, while for Commercial traders they are $HL_i$ and $HS_i$.\footnote{In an earlier draft, we also reported results based on two alternative proxies as well as a proxy constructed using principal component analysis. The alternative proxies are Working's $T$ \citep{working1960speculation} and the market share of non-commercial traders \citep*{buyukcsahin2014speculators}. These results, which are available upon request, are consistent with our main findings and do not change the paper's implications.}
126 +
127 +%The first proxy we consider to assess levels of speculative and hedging activity is Working’s $T$ \citep{working1960speculation}. This index compares the activity levels of Non-Commercial commodity futures traders (e.g., speculators) to those of Commercial traders (e.g., hedgers). Working’s $T$ measures the extent to which speculation exceeds the level required to offset any unbalanced hedging at the market clearing price. This index, denoted $WT_i$, is computed as follows:
128 +%Historically this is true but today more nuanced Typically, Commercial traders take short positions in futures contracts while Non-Commercial traders take long positions.
129 +
130 +%\begin{equation} \label{ch1:eqn:Working}
131 +%WT_i = \left\{
132 +%\begin{matrix}
133 +%1 + \frac{SS_i}{HL_i + HS_i} & \mbox{if} & HS_i \ge HL_i \\
134 +%1 + \frac{SL_i}{HL_i + HS_i} & \mbox{if} & HS_i < HL_i
135 +%\end{matrix}
136 +%\right.
137 +%\end{equation}
138 +
139 + %\citet*{buyukcsahin2014speculators} suggest a measure that emphasizes the \emph{market share of Non-Commercial traders (MSCT)}. This ratio is expressed as the sum of the short and long positions of Non-Commercial traders over twice the total open interest in a market:
140 +%%Instead of Working’s $T$,
141 +%
142 +%
143 +%\begin{equation} \label{ch1:eqn:MSCT}
144 +%MSCT_i = \frac{SL_i + SS_i}{2 \times OI_i}
145 +%\end{equation}
146 +
147 +The specific measure we use follows \citet{hedegaard2011margins}, who suggests an index of speculative activity computed as the ratio of net long speculative positions over total open interest ($NLS_i$):
148 +
149 +\begin{equation} \label{ch1:eqn:NLS}
150 +NLS_i = \frac{SL_i - SS_i}{OI_i}
151 +\end{equation}
152 +
153 +In addition to computing $NLS$ using the full sample data, we use disaggregated data from the CFTC to compute the NLS index separately for money manager (MM) and swap dealer (SD) positions, which allows for additional empirical analysis. The data on money manager and swap dealer positions come from Quandl's API.\footnote{See the \url{https://data.nasdaq.com/data/CFTC-commodity-futures-trading-commission-reports} website.}
154 +\emph{Money managers} typically refer to Non-Commercial market participants who are involved in managing funds and investing in commodity futures and options markets \citep{fishe2012identifying}.\footnote{This category is also called ``Managed money.'' The CFTC writes that they are ``registered commodity trading advisor (CTA); a registered commodity pool operator (CPO); or an unregistered fund identified by CFTC.'' There is some overlap between Money managers and hedge funds, but they are distinct.} Their activities are influenced by financial and economic factors related to commodities. Money managers are often considered to be more informed investors because they actively manage portfolios and adjust their positions based on market information and analysis.
155 +% factors such as speculative activity, imperfect information about real economic activity as well as supply, demand, and inventory accumulation in commodity markets \citep{singleton2014}.
156 +\emph{Swap dealers} are considered as Non-Commercial traders by the CFTC. They typically use futures contracts to hedge risk generated by their swap positions. Swap dealers have been studied in relation to index investors, as their positions are distinct from those of other market participants. Their activities are influenced by the need to manage large exposures and to facilitate trading for clients. While money managers regularly take long or short futures positions, swap dealers mainly take long positions \citep{fishe2012identifying}. Hedgers, who we exclude from the analysis, tend to take short positions.
157 +%often leading them to take significant speculative positions in the market.
158 +%and contribute to price discovery and liquidity \citep{brunetti2016speculators}
159 +%\citep{sandrs2016bubbles}
160 +
161 +
162 +%\subsection{A financialization proxy using principal component analysis}
163 +%
164 +%Descriptive statistics for the three financialization variables are shown in Table \ref{ch1:tab:stat5} and computed separately for each of the six commodities in our sample, based on the number of open positions for a given futures contract. The variables themselves are scale-free, unlike the number of open positions. The MSCT variable fluctuates between 0 and 0.5, while NLS varies between -0.4 and 0.8, and Working's $T$ between 1 and 2. Since the literature does not show that one is superior to the others, and for brevity’s sake, we use Principal Component Analysis to construct a new proxy using information from the three variables. This step allows us to present results based on a single variable.\footnote{In an earlier draft, we showed results using the three different variables and found that our main findings were similar.}
165 +%Our approach is as follows: First, since the variables have different scales, we standardize them to have a mean of zero and a variance of one. Second, we compute the covariance matrix. Third, we calculate the eigenvalues and eigenvectors. Fourth, we obtain the principal components as the eigenvectors of the covariance matrix, and we use the first principal component as our new proxy for financialization. Table \ref{ch1:tab:PCA} shows the outcome of the PCA analysis and in particular, that the first principal component explains a large part of the individual variances.
166 +
167 +
168 +
169 +
170 +
171 +
172 +
173 +
174 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
175 +\section{Econometric framework and methods}
176 +
177 +\subsection{Modeling the impact of surprises on returns}\label{ch1:return}
178 +
179 +Our high-frequency regression model is based on \citet{kurov2019price}.\footnote{In unreported results, we run the regressions using the approach shown in \citet{andersen2003micro}. The results are similar.} We run the following regression using the specification in equation (\ref{ch1:eq:Model 1}):
180 +
181 +\begin{equation}\label{ch1:eq:Model 1}
182 +R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{j} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_j)+\beta R_{t-\tau}^{t}+\epsilon_{t}
183 +\end{equation}
184 +%%% MH :donc on enlève les notes en rouge???
185 +where $R_{t}^{t+\tau}$ is the continuously compounded futures return from time $t$ to $t+\tau$, $S_{mt}$ is the surprise for macroeconomic announcement $m$ published at time $t$, and $X_{j}$ is the NLS speculative trading intensity variable, which is updated at a weekly frequency, with $j$ the index for the week.
186 +%$X_{t}$ is the value of the speculative intensity proxy using the $NLS$ variable.
187 +The impact of macro announcements on commodity futures returns can be assessed by looking at the $\gamma_m$ coefficient in the mean equation, while the $\delta$ coefficient controls for the level of speculative trading intensity as it relates to futures returns. The key coefficient to help answer our main research question is $\theta_m$, which relates the effect of time-varying speculative trading intensity on the impact of the news release.% This last coefficient is the most important one to help answer our main research question.
188 +The regression is estimated using a two-step weighted least squares (WLS) procedure. %For robustness, appendix A presents results using the NLS variable.
189 +To account for heteroskedasticity, we construct a volatility estimate by means of an exponential moving average, using the regression residuals obtained in the first step. This auxiliary regression is presented in equation~(\ref{ch1:eqn:auxiliary 2}), with a smoothing parameter $\alpha=0.9$ and a starting parameter value set to $\sigma_1=\epsilon_t$:
190 +
191 +\begin{equation}\label{ch1:eqn:auxiliary 2}
192 +\sigma_t=\alpha \sigma_{t-1}+(1-\alpha) \mid \epsilon_t \mid
193 +\end{equation}
194 +
195 +After obtaining $\sigma_t$ for each observation, we apply the transformation $w_t = \hat{\sigma_t}^{-2}$ to obtain the WLS regression weight. Then, we multiply each variable by $w_t$ and run an OLS regression to estimate the model.
196 +
197 +
198 +
199 +
200 +
201 +\subsection{Modeling the impact of surprises on volatility}\label{ch1:variance}
202 +
203 +To estimate the volatility equation, we use a GARCH specification, as it is well known that the variance of commodity futures returns displays time variation and clustering \citep*[see e.g.,][]{brunetti2014commodity}. We specify a GARCH (1,1) model and extend the equation by including our NLS speculative intensity proxy as well as the macroeconomic news surprise variables.
204 +First, we estimate the mean equation (\ref{ch1:eqn:MeanEqn}):
205 +
206 +\begin{equation}\label{ch1:eqn:MeanEqn}
207 +%R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+\beta R_{t-\tau}^{t}+\epsilon_{t} ICI À VOUS AVEC sp PARCE QUE JE NE VOIS PAS CE QUI A CHANGÉ
208 +R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+\beta R_{t-\tau}^{t}+\epsilon_{t}
209 +\end{equation}
210 +
211 +Then, we estimate the following equation for conditional variance:
212 +
213 +\begin{equation}\label{ch1:eqn:VarianceEqn}
214 +%\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h
215 +\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{j}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h
216 +\end{equation}
217 +%j,ai change le j en bas
218 +
219 +
220 +
221 +where $I_{k,t}=D_{m,t} \cdot X_{j}$ and $D_{m,t}$ is a dummy variable for macro announcement $m$. The latter equals 1 if an announcement takes place at time $t$ (5-minute frequency) and equals 0 otherwise. $X_{j}$ is the NLS speculative intensity variable as defined earlier. The $\rho_h$ coefficient captures intraday periodicity, while the $D_h$ dummy equals 1 at hour $h$ and 0 otherwise. The impact of macro announcement $m$ on conditional variance is captured by the $\Phi_m$ coefficient in equation~(\ref{ch1:eqn:VarianceEqn}), while the $\beta$ coefficient shows the impact of the speculative intensity variable $X_{j}$. Finally, the $\phi_k$ coefficient shows the interaction effect from speculative trading and the macro surprise $m$.
222 +The standard errors are computed using the Newey-West heteroskedasticity and autocorrelation consistent (HAC) estimator with automatic lag selection, following the procedure outlined in \citet{newey1994automatic}.%\footnote{This approach accounts for both heteroskedasticity and serial correlation in the residuals, which is standard practice in high-frequency financial data analysis.}
223 + This methodology mirrors the approach used in \citet{andersen2003micro, andersen2007real} and \citet{kurov2019price} to study announcement effects in other asset classes. In unpublished results, we consider the mixed-data sampling (MIDAS) approach proposed by \citet{ghysels2004midas}. We find that the results are similar.%\footnote{We thank a reviewer for suggesting a mixed-frequency approach. Results are available upon request.}
224 +
225 +% Among the announcements in our sample, all but one are ``good news.'' Only a positive surprise in Initial Jobless Claims indicates a deterioration in economic conditions. Therefore, the surprise coefficient is expected to be positive for all pro-cyclical commodities (i.e., all but gold and silver) for all announcements except Initial Jobless Claims, for which it should be negative (since a positive surprise is ``bad news’’). In the case of gold and silver, which are safe-haven commodities, the reverse is expected for coefficient signs.
226 +
227 +
228 +
229 +
230 +
231 +
232 +
233 +
234 +
235 +\subsection{Modeling the impact on bid-ask spreads}
236 + Speculative trading could make markets more efficient by improving information. We test this hypothesis by measuring the effect of macro surprises on the futures price bid-ask spread in high-frequency regressions.
237 +%We aim to investigate whether increased financialization affects the responsiveness of commodity futures returns to macroeconomic surprises by analyzing the impact on the bid-ask spread.
238 +The bid-ask spread is widely recognized as a measure of market efficiency. A narrower spread suggests less uncertainty about the asset's true value and reflects lower transaction costs, improved liquidity, and lower information asymmetry \citep{Roll1984}.
239 +%A narrower bid-ask spread typically indicates higher liquidity and better informational efficiency, suggesting less uncertainty about the asset's value.
240 +Furthermore, \citet{chordia2008liquidity} show that the bid-ask spread is an indicator of market quality and market efficiency. Since a smaller spread is associated with a more efficient price discovery process, an increase in the quality of market information should decrease the spread. Therefore, we estimate the following equation, where the relative bid-ask spread is defined as $(Ask_t-Bid_t)/Mid_t$:
241 +
242 +\begin{equation}\label{ch1:eq:Model 2}
243 +%\text{Spread}_{t}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta \text{Spread}_{t-\tau}+\epsilon_{t}
244 +\text{Spread}_{t}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{j} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_j)+\beta \: \text{Spread}_{t-\tau}+\epsilon_{t}.
245 +\end{equation}
246 +
247 +In equation ~(\ref{ch1:eq:Model 2}) , $\text{Spread}_{t}$ is the relative bid-ask spread measured at 5-minute frequency $t$ using the high-frequency data and $\text{Spread}_{t-\tau}$ is the lagged spread. In addition, $\sum_{m=1}^{22} \gamma_m D_{m,t}$ accounts for the macroeconomic announcements, where each dummy variable $D_{m,t}$ is multiplied by its respective coefficient $\gamma_m$. We also include the speculative trading intensity variable $X_{j}$ with its coefficient $\delta$. The interaction terms $\sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_j)$ capture the effect of speculative trading intensity on the bid-ask spread at the time of a release, with each term multiplied by its respective coefficient $\theta_m$. To test whether increased speculative trading activity improves informational efficiency at the time of a macroeconomic announcement, we check whether the sign on $\theta_m$ is negative and significant, thus reducing the spread. %Indeed, the interaction terms $\theta_m (D_{m,t} \cdot X_t)$ capture the effect of financialization on the spread at the time of a release.% macroeconomic announcements.
248 +%(measured at daily frequency, where $j$ represents the trading day)
249 +
250 +
251 +
252 +
253 +
254 +
255 +
256 +
257 +
258 +
259 +%\subsection{Impact according to the type of non-commercial trader}
260 +%\alert{TOO SHORT AS A STAND ALONE SECTION. FIGURE OUT HOW TO INTEGRATE IT IN THE TEXT.}
261 +
262 +
263 +
264 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
265 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
266 +\section{Results} \label{ch1:sec:result}
267 +
268 +\subsection{The impact of surprises on cumulative abnormal returns}
269 +
270 +We begin by documenting the impact of macroeconomic announcement surprises on commodity futures returns. The impact of surprises is shown in the following graphs of high-frequency cumulative abnormal returns (CARs). Figures \ref{ch1:fig:cl} to \ref{ch1:fig:ng} show CARs for each commodity as measured over a window of 60 minutes before to 60 minutes after a macroeconomic announcement release. The figures are constructed similarly to those shown in \citet{kurov2019price}. The magnitude of the CARs after announcement releases is comparable to those shown in their paper for stock index and Treasury futures. Unlike them, however, we do not see evidence of a pre-announcement drift.
271 +
272 +
273 +A red line denotes the average CAR for announcement releases that are seen as negative surprises (i.e., worse than anticipated news), while a green line denotes the average CAR for positive surprises.
274 +For brevity, we discuss only the CARs for crude oil and gold, as they are representative of pro-cyclical energy markets and safe haven assets. Figure \ref{ch1:fig:cl} shows the average CAR for crude oil futures. The CAR increases following a positive surprise and decreases following a negative one, confirming that crude oil is a pro-cyclical commodity. In contrast, figure \ref{ch1:fig:gc} shows that gold futures react in the opposite manner. The red line indicates that CAR is positive after bad news, while the green line shows that CAR is negative after good news. %These results support the hypothesis that gold is a safe haven asset.%finanthe one presented for stocks in their paper.\footnote{A difference is that the commodity markets do not seem to display pre-announcement drift whereas \citet{kurov2019price} documented one for the S\&P 500 futures.}
275 +%The increase (decrease) in CAR in reaction to a positive (negative) surprise confirms that crude oil is a pro-cyclical commodity.
276 +
277 +
278 +
279 +
280 +\subsection{Macroeconomic surprises, speculative trading activity and futures returns}
281 +
282 +Table~\ref{ch1:tab:macro_fin_nls_fut_returns_full} presents the results of high-frequency regressions that explain commodity futures returns immediately after a macroeconomic announcement release. Our discussion focuses on coefficients that are statistically significant at the 5\% level.\footnote{In an earlier draft, we also reported results for different sub-periods, such as the Zero Lower Bound period, the 2008-2010 financial crisis and Great Recession, and the COVID-19 period. These results do not materially affect our findings or conclusions, and they are available upon request. } %For brevity, the table only reports results for regressions in which the PCA proxy is used.\footnote{That being said, there is no material difference in the results if we use one of the other measures.}
283 +We begin with energy commodities, which serve as our baseline case. The $\gamma_m$ coefficient shows the immediate impact of a macro surprise on commodity futures returns. For crude oil futures, we find that several macroeconomic announcements exhibit significant effects. Consider for instance Initial Jobless Claims, for which a greater than expected value indicates bad economic news. The table shows that for this announcement, $\gamma_m$ is negative, indicating that crude oil prices tend to drop in response to unexpected increases in jobless claims. This result is consistent with the expectation that higher jobless claims signal weaker economic conditions, which reduce the demand for crude oil. The corresponding $\theta_m$ coefficient for Initial Jobless Claims is positive, however, suggesting that increased speculative trading mitigates the negative impact of bad macroeconomic news on crude oil prices. This damping effect suggests that markets are better informed as a result of the increased participation of speculators. Indeed, the announcement release creates a smaller surprise and a smaller shock.
284 +
285 +%TOUT CA EST A CHANGER AVEC NLS car les coefficients changent de signes parfois et ne sont plus significatifs au mêmes endroits. For natural gas, the ADP Employment announcements show a pattern similar to crude oil, with a positive $\gamma_m$ coefficient implying that better than expected employment figures boost energy prices, reflecting increased economic activity and demand. The negative $\theta_m$ coefficient indicates that speculative trading dampens this positive reaction. The CB Consumer Confidence and Advance Retail Sales announcements also show positive $\gamma_m$ coefficients and negative $\theta_m$ coefficients for both energy commodities, supporting the claim that increased speculative trading intensity smooths out the market's response to macroeconomic surprises.
286 +
287 +In the case of natural gas futures, we find that the results across announcements are less frequently significant. Increased speculative trading tends to increase the magnitude of the surprise's effects, when the results are significant. An exception is the natural gas market-specific announcement release (inventories), for which the coefficient is negative but not significant. Overall for natural gas futures, we do not find as much support that speculative trading dampens the effect of macroeconomic announcements on returns. The reason why the results for natural gas futures are less conclusive is most likely that this futures contract displays greater volatility, that it has a lower trading volume \citep{irwin2012testing}, and it has less speculative trading activity (thus, fewer informed traders) \citep{buyukcsahin2014speculators}.
288 +%results confirm the specific nature of natural gas, in terms of lower trading volume , and less speculative trading activity such as international traders \citep{buyukcsahin2014speculators}.
289 +
290 +Copper is a pro-cyclical, industrial commodity, so it is expected that the results for copper futures should resemble those for crude oil. The results are highly significant for many announcements. If we look at announcements such as ADP Employment and CB Consumer Confidence, which are ``good news'', the $\gamma_m$ coefficients are positive, confirming copper's pro-cyclical nature and its ties to industrial production. The corresponding $\theta_m$ coefficients are negative, consistent with the damping effect that we document for energy commodities. Thus, the main finding for copper futures is that, as with crude oil, increased speculative trading intensity has the effect of weakening the impact of a macro surprise.
291 +%Comparing these results with other commodities, the behavior of copper futures closely resembles that of crude oil -- indeed, both are pro-cyclical. For
292 +
293 +
294 +Gold is considered to be a safe haven asset \citep{baur2010gold}. Therefore, it is expected that the reaction of gold futures returns to macro surprises will be the opposite to what we have found for crude oil, natural gas, and copper. This is indeed what we find: gold futures returns are lower after "good news" and higher after "bad news". These results support the idea that energy commodities are pro-cyclical, while gold is a safe haven asset. In particular, the positive $\gamma_m$ coefficient for Initial Jobless Claims indicates that gold prices rise in response to unexpected increases in jobless claims, as investors seek safety in gold positions during economic uncertainty. The $\theta_m$ coefficient is negative, indicating that speculative trading activity tempers this flight to safety, leading to less pronounced price increases.%Turning to gold futures, the reaction to macroeconomic surprises is opposite to what we find for our energy baseline case:
295 +In the case of the ADP Employment release, the $\gamma_m$ coefficient is negative for gold futures, suggesting that strong employment figures reduce gold prices as investors move away from safe-haven assets towards pro-cyclical assets, such as energy commodities. The positive $\theta_m$ coefficient suggests that speculative trading reduces the extent of this price drop. The CB Consumer Confidence and Advance Retail Sales announcements also generate negative $\gamma_m$ coefficients for gold futures and positive $\theta_m$ coefficients at a 10 percent level, providing additional evidence of a moderating influence of speculative trading on the reaction of gold to economic news. The main finding is therefore that whether a commodity is pro-cyclical or a safe haven asset, increased speculative trading has a damping effect on the reactions to macro surprises.
296 +
297 +The last two commodities in our sample, silver and palladium futures, behave more like gold futures. We find that for Initial Jobless Claims for palladium and ADP Employment in silver, significant $\gamma_m$ coefficients are found in directions consistent with their status as safe haven assets. The $\theta_m$ coefficients for silver and palladium also indicate a damping effect on price reactions. Taken together, our results show that the damping effect of speculative trading is stronger in energy markets, suggesting that the beneficial effects of financial participants are particularly important for energy commodities.
298 +
299 +
300 +
301 +
302 +
303 +
304 +\subsection{Macroeconomic surprises, speculative trading and volatility}
305 +
306 +Table~\ref{ch1:tab:macro_fin_nls_var_full} presents regression results to explain the conditional variance of high-frequency commodity futures returns after macroeconomic announcements. We first examine our baseline assets, energy commodities, as volatility in energy markets is of particular concern given their economic importance and direct impact on consumer prices.
307 +For crude oil, macroeconomic surprises generally lead to an increase in conditional variance, as the $\Phi_m$ coefficients are consistently positive. For instance, a surprise in Initial Jobless Claims significantly increases crude oil volatility. Natural gas exhibits similar patterns, with some variations in magnitude and a greater impact on inventory-related announcements. This result suggests that unexpected economic news generates greater uncertainty and price fluctuations in energy markets.
308 +In contrast, the interaction coefficients $\phi_m$, which inform us about the impact of speculative trading, tend to be negative for the two energy commodities. The implication is that increased trading activity by speculative traders lowers volatility following a macro surprise. Taking Initial Jobless Claims as an example, we find that higher levels of speculation reduce the impact of news on volatility in crude oil futures markets. Thus, speculative trading can act as a stabilizing force by damping the heightened volatility that occurs after a surprise in macroeconomic news. %\textcolor{red}{This stabilizing effect is especially valuable in energy markets, where excessive volatility can have significant economic consequences}.
309 +
310 +Comparing these results with those for other commodities, we find that gold, copper, silver, and palladium also show positive $\Phi_m$ coefficients across various announcements, indicating increased volatility following macro surprises. However, the magnitude of these effects is generally smaller than what we see in energy markets, especially in the case of precious metals. The corresponding $\phi_m$ coefficients, measuring the damping effect of financial investor activity, are negative across commodities and announcements, and the strongest effects are found in crude oil futures markets. This pattern holds for other macroeconomic announcements as well. The coefficients for ADP Employment, CB Consumer Confidence, and Advance Retail Sales, for instance, generally indicate increased volatility after surprises, as shown by the positive $\Phi_m$ coefficients, while the corresponding $\phi_m$ coefficients are negative, supporting the finding of a stabilizing effect of increased speculative trading. %\textcolor{red}{The consistency of these results across different types of announcements is particularly noteworthy in energy markets, where price stability is crucial for both consumers and industrial users}.
311 +
312 +Therefore, our findings highlight the valuable role of speculative traders in reducing volatility, particularly in crude oil futures markets, where price stability has important implications for the broader economy. While macroeconomic news tends to increase volatility across commodity markets, the damping effect of speculative trading appears strongest in energy markets. We show that this relationship is consistent across different types of announcements and remains robust when controlling for various market conditions.
313 +
314 +
315 +
316 +
317 +\subsection{Macroeconomic surprises, speculation and bid-ask spreads}
318 +
319 +Table~\ref{ch1:tab:return-fin-full} shows our results for the impact of macroeconomic surprises and speculative trading intensity on futures prices bid-ask spreads. This empirical analysis provides a test of informational efficiency. We first focus on energy markets. In the bid-ask spread regressions, the $\gamma_m$ coefficient denotes the effect of surprises on the spread, while the $\theta_m$ coefficient shows the interaction effect between surprises and speculative trading intensity. The main hypothesis is whether $\theta_m < 0$, which would indicate that greater speculative activity improves informational efficiency through narrower spreads. Such a finding would be consistent with what we have documented above for futures returns and volatility. In the case of crude oil futures, we find that $\gamma_m$ is negative for the initial jobless claims announcement, which means that the market becomes more efficient immediately after a news release. This is consistent with the resolution of uncertainty. The $\theta_m$ coefficient is always negative when it is significant, indicating that the bid-ask spread narrows even more (implying greater informational efficiency) after a macro announcement if crude oil futures markets benefit from greater speculative activity relative to hedging activity, as measured by the NLS proxy. The results for natural gas futures are similar to those for crude oil. The relationship between speculative trading and market efficiency is clearest during periods of higher trading volume, such as the release of storage reports and weather-related announcements.
320 +%\textcolor{red}{CHECK TO MAKE SURE THAT THIS IS STILL ACCURATE.}
321 +
322 +Comparing these results to those obtained for the other commodities in our sample, we find that the results for $\theta_m$ (speculative intensity) are less often significant, but that they are negative when they are significant. The $\theta_m$ coefficient being negative suggests that greater speculative trading intensity improves market efficiency, as bid-ask spreads tend to be lower when the coefficient is significant. This improvement in market efficiency appears to be most pronounced in energy markets, where accurate price discovery is particularly important given their relevance for the real economy.
323 +%The stronger effects we observe in energy markets may reflect their greater integration with the broader financial system and higher trading volumes, which allow for more efficient price discovery. This is particularly relevant given the increasing importance of energy price stability for economic policy and planning}.
324 +Overall, the results for the bid-ask spread provide additional support for our claim that greater trading activity has beneficial effects on commodity derivatives markets, with these benefits being especially notable in energy markets where efficient price discovery has important implications for both market participants and the broader economy.
325 +
326 +
327 +
328 +
329 +\subsection{Differences in results according to trader type}
330 +
331 +To investigate whether differences in trader type are relevant in explaining our findings, we provide disaggregated results in this section. To this end, we estimate equations (\ref{ch1:eqn:MeanEqn}), (\ref{ch1:eqn:VarianceEqn}) and (\ref{ch1:eq:Model 2}) for two categories of Non-Commercial traders, namely, swap dealers (SD) and money managers (MM). For each of the two, the CFTC reports the number of long and short positions in their disaggregated Commitment of Traders (COT) reports. We compute the NLS index for each trader category over time, allowing us to separately quantify the intensity of trading activity by money managers and swap dealers.%\footnote{We use the NLS variable for this disaggregated analysis because it is the only one we can construct using the available data on MM and SD positions.}
332 +
333 +
334 +%This section shows disaggregated results for the two types of financial participants that are reported separately, namely swap dealers and money managers. %We continue to use the NLS variable, but it is computed separately for MM and SD futures positions. %\footnote{The main reason is that the MSCT and Working's $T$ variables cannot be computed only for MM or SD because the calculations involve the positions of commercial traders.}
335 + First, we examine the returns equation for money manager positions, as shown in table \ref{ch1:tab:macro_fin_mm_fut_returns_full}. Increased trading activity by money managers has the same effect as in our baseline results. If we consider crude oil futures, for example, the $\gamma_m$ macro surprise coefficient is positive while the $\theta_m$ coefficient for speculative trading is negative. Since $\theta$ has the opposite sign to $\gamma$, the implication is that increased futures trading activity by money managers lowers the impact of surprises on futures returns. This is similar to our baseline, aggregate findings.
336 + Next, table \ref{ch1:tab:macro_fin_sd_fut_returns_full} presents results using only swap dealer positions. Here we find a notable difference relative to money managers. This table shows that the speculative intensity coefficient $\theta$ for swap dealers has the same sign as the macro surprise coefficient $\gamma$. This result means that increased swap dealer trading activity seems to amplify the reaction of futures returns to macro surprises. The exception to these results is in the case of natural gas futures, where increased trading by swap dealers for some announcements appears to have the opposite effect on returns. To contrast this finding with prior research, \citet{brunetti2016speculators} find, using daily data, that the positions of swap dealers are not correlated with contemporaneous returns and volatility in commodity futures markets. They further show that hedge funds decrease, and hedgers increase, volatility. Our empirical analysis extends their findings using high-frequency data and the setting of macro surprises as a source of new information affecting energy and commodity futures markets.
337 +
338 +Tables \ref{ch1:tab:macro_fin_mm_var_full} and \ref{ch1:tab:macro_fin_sd_var_full} show our disaggregated results for the variance equation using sample data for money managers and swap dealers, respectively. The money manager results are similar to what we find in the aggregate sample, namely that both good and bad surprises increase volatility (as shown by $\Phi_{mm}>0$), while greater money manager trading activity lowers the impact of news on volatility (as shown by $\phi_{mm}<0$). This result strengthens prior evidence in the literature about beneficial effects of speculators, which were based on daily data \citep{brunetti2016speculators}. In contrast, our results suggest that increased activity by swap dealers appears to increase volatility after macro news (since $\phi_{sd}>0$). Together, the results line up with our findings for the returns equation and with our main message, which is that informed traders stabilize markets by contributing new information. Swap dealers, as intermediaries, typically do not trade based on information but rather following the needs of their clients. This economic motivation can explain why our results suggest that their trading activities amplify price and volatility reactions to news.
339 +%. that the signs for $\gamma_m$ and $\theta_m$ are equal. Therefore, unlike money managers, it seems that more trading activity by swap dealers causes even more strong market reactions to macroeconomic news announcements. Swap dealers' trading operations are motivated by supporting their clients' needs rather than information.
340 +%Findings for the variance equation using only money managers and swap dealers can be found in tables respectively. We find that whereas for swap dealers the financialization interaction coefficient $\phi_k$ is positive, for money managers it is negative. These findings support the economic interpretation of previous findings for returns. Although rising trade by money managers lessens the impact of macro surprises on volatility, a higher presence of swap dealers seems to enhance the influence of surprises on volatility. Therefore, the trading actions of money managers help to minimize market volatility after market news; yet, the activities of swap dealers exacerbate this effect.
341 +Although our discussion centers on crude oil as a benchmark commodity, the results for the other pro-cyclical commodities support our interpretation of the findings. In the case of gold futures, the disaggregated results continue to support a safe haven interpretation \citep{erb2013golden}. Indeed, gold has features of a commodity and a currency, but earlier studies have also found that its value increases with investor risk aversion, since it is perceived as a safe haven during times of economic uncertainty and market volatility.
342 +
343 +
344 +%\subsubsection{Bid-ask spread analysis for money managers and swap dealers}
345 +
346 +%The examination of bid-ask spreads following macroeconomic announcements is expanded to include a NLS financialization proxy computed using just money manager (MM) or swap dealer (SD) positions, to account for potential differences in the impact of their trading activities.
347 +Lastly, we examine whether the effects on bid-ask spreads differ between trader types.
348 +Table \ref{ch1:tab:macro_fin_mm_fut_spread_full} presents the results for money managers. As the $\theta_m$ coefficients tend to be negative, it seems that money managers enhance market efficiency by lowering bid-ask spreads. The results for swap dealers are shown in Table \ref{ch1:tab:macro_fin_sd_fut_spread_full}. Unlike for money managers, we find that $\theta_m$ coefficients for swap dealers tend to be positive, indicating that greater swap dealer activity increases the bid-ask spread after a macroeconomic release. The results for swap dealers therefore suggest a decline in market efficiency due to their increased trading activities. %meaning that a higher degree of activity by swap dealers reduces market efficiency by increasing bid-ask spreads following a macroeconomic announcement, indicating a decline in market efficiency.
349 +These findings are in line with our main results and further support the claim that traders in energy and commodity markets do not all have the same effect on market efficiency. The results point to the importance of information acquisition and investor attention as an economic channel.
350 +%These results also relate to \citet{rakowski2021twitter} who show that Twitter activity, by serving as a distinct form of investor attention, has significant effects on trading volume and asset prices.
351 +
352 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
353 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
354 +
355 +%Important: keep this in LaTex format; use % to comment out; keep references as is
356 +\section{Discussion and implications}
357 +
358 +
359 +
360 +Our findings contribute new insights to an unsettled literature on the impact of different types of financial participants in energy and commodity markets \citep{ready2022order}. In addition to being an important alternative asset class, energy commodities are central to economic activity, while energy futures trading is important for price stability. Being closely related to macroeconomic risk \citep{cheng2015convective}, energy commodities therefore play a key role in macro-finance. By taking a novel angle of high-frequency market reactions to macroeconomic surprises, our results contribute a more nuanced picture of the impact of speculative trading and help to reconcile previous findings in the literature.
361 +%Our analysis builds on the models proposed by \citet{goldstein2022commodity} and \citet{basak2016model}, which explain how institutional investors in particular can influence commodity prices, volatility and market efficiency.
362 +
363 +%Our paper is motivated by recent theoretical advances.
364 +\citet{goldstein2022commodity} develop a theoretical model to examine how financial traders improve information transmission between futures and spot markets. They show how market efficiency should benefit from the presence of informed financial traders, whose actions lead to increases in pricing transparency and to lower information asymmetry. Our empirical results provide support for this model. We show that increased speculative trading activity lowers bid-ask spreads after a macro news release, in addition to reducing the magnitude of price and volatility reactions to macro surprises. This improvement in informational efficiency is especially valuable in energy markets, where price discovery has direct implications for industrial users and consumers. We find that this effect is driven by money managers. In addition, our results point to smaller, negative effects linked to swap dealers. The difference between the two sets of disaggregated empirical results can be explained by the fact that they trade for different purposes. Money managers aim to make a profit based on information, while swap dealers are intermediaries who, while perhaps informed, primarily provide services to customers such as institutional investors, in addition to hedging their swap positions. In line with this interpretation, our results also build on \citet{fishe2012identifying}, who show that some financial traders (e.g., money managers) are better informed than others in commodity markets.
365 +%\textcolor{red}{This information advantage appears particularly valuable in energy markets, where complex supply-demand dynamics and geopolitical factors make price discovery especially challenging}.
366 +
367 +%DO NOT INCLUDE THIS PART SINCE WE WANT TO DOWNPLAY INDEX TRADERS
368 +%Our findings also provide evidence to support the predictions in the model proposed by \citet{basak2016model}. They argue that financialization should increase commodity futures prices and volatility, especially for indexed futures. They argue that the activities of institutional investors enhance the link between commodities and equities. This relationship is particularly relevant in energy markets, where crude oil and natural gas prices have become increasingly correlated with financial markets. We find some evidence to this effect, as increased trading by swap dealers, who tend to be long and deal with index funds, increases price reactions and volatility. However, our results suggest that overall, these effects are offset by the stabilizing presence of informed traders such as money managers.
369 +
370 +Our findings also extend and build on the results of an earlier empirical literature that uses daily-level data. This literature includes \citet{brunetti2016speculators} who find that speculative traders, especially money managers and hedge funds, are helpful to energy and commodity markets, as well as \citet{buyukcsahin2011speculators} and \citet{alquist2013role} who show that the futures positions of financial firms such as hedge funds do not predict next-day changes in crude oil prices.
371 +Moreover, our results relate to \citet{cheng2015convective} who show that financial investors, being better informed about markets, contribute to price discovery and liquidity. The key message is therefore that speculative trading is helpful to markets by distributing and assimilating new information into prices.
372 +%This is particularly relevant for energy markets, where accurate price discovery is crucial not only for market participants but also for energy policy and planning}.
373 +The findings shown in this paper have important implications for energy markets regulation and policy. First, they suggest that attempts to limit speculative trading in energy markets could, in fact, increase price volatility and reduce informational efficiency and price discovery. Second, they indicate that different types of financial participants have distinct effects on market quality, suggesting that regulatory frameworks should pay careful attention to market composition. Third, they highlight the importance of maintaining a robust price discovery mechanism in energy markets, given their crucial role in the economy.
374 +%REDONDANT
375 +%To summarize, the evidence we present suggests that financial investors play an important role in improving the functioning of energy markets, which is especially important given current challenges including the energy transition, geopolitical tensions, and the need for stable energy prices to support economic growth.
376 +
377 +
378 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
379 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
380 +
381 +%\section{Robustness in sub-periods}
382 +%
383 +%To rule out the possibility that our main results are driven by a specific period, we present in the following robustness checks our results for different sub-sample periods, namely the Zero Lower Bound period from December 22nd, 2008 to December 21st, 2015, and the COVID-19 pandemic period from January 31st, 2020 to June 10th, 2022. The start and end dates are selected following the literature and sources such as the NBER and the CDC. Our main finding is that the sub-sample results are consistent with our full sample results, despite some minor discrepancies. Therefore, our main findings are not explained by any particular period, but rather hold for the entire sample period from 2007 to 2024.
384 +%
385 +%% The full tables are presented in appendices B and C, respectively. In order to investigate whether our results are period-specific, we now present the results for different sub-periods. The results presented in the appendix are for specific sub-periods, namely the Zero Lower Bound (ZLB) (2008-12-22 to 2015-12-21) and COVID-19 sub-periods (2020-01-31 to 2022-06-10). The main finding is that the sub-period results are consistent with the full-sample results presented earlier.
386 +
387 +%\subsection{Price reactions during the Zero Lower Bound and COVID-19 sub-periods}
388 +%
389 +%Tables~\ref{ch1:tab:macro_fin_fut_returns_covid} and \ref{ch1:tab:macro_fin_fut_returns_zlb} present the results of high-frequency regressions to explain returns after macroeconomic announcements, estimated separately for the Zero Lower Bound (ZLB) and COVID-19 sub-periods. We discuss a few major announcements here. Looking at Initial Jobless Claims, the surprise coefficient $\gamma_m$ is negative for crude oil and positive for gold in both sub-periods, consistent with the full sample results.
390 +%%indicating that more jobless claims than expected decrease crude oil prices and increase gold prices.
391 +%
392 +%The financialization $\theta_m$ coefficient is negative for crude oil during the COVID-19 sub-period and positive during the ZLB sub-period, suggesting that the interaction effect of financialization varies to some degree over the years. In the case of ADP Employment, $\gamma_m$ is positive for crude oil and copper while negative for gold in both sub-periods. The $\theta_m$ coefficient has the opposite sign to $\gamma_m$, similar to our baseline results. In the case of CB Consumer Confidence, $\gamma_m$ is negative for gold during the ZLB sub-period but is not significant during COVID-19, while the $\theta_m$ coefficients have the opposite sign, consistent with the damping effect we discuss. Lastly, for Advance Retail Sales, $\gamma_m$ is significant for crude oil, gold, copper, and silver in both sub-periods while $\theta_m$ generally has the opposite sign to $\gamma_m$. Thus, our sub-period results are for the most part similar to the full sample results, and we do not see any important differences in the economic interpretation of the coefficient signs.
393 +%
394 +%\subsection{Volatility reactions during the Zero Lower Bound and COVID-19 sub-periods}
395 +%
396 +%Table \ref{ch1:tab:macro_fin_fut_var_covid} and Table \ref{ch1:tab:macro_fin_fut_var_zlb} present the results of regressions to estimate the reaction of conditional variance to macro surprises in different sub-periods. This analysis confirms our full sample results, namely that surprises increase volatility while financialization acts to lower this impact. This can be seen from the $\Phi_m$ surprise coefficients, which remain positive, while the financialization coefficients $\phi_m$ are negative. Thus, our main findings are not driven by a particular sub-period.
397 +%
398 +%\subsection{Bid-ask spreads during the Zero Lower Bound and COVID-19 sub-periods}
399 +%
400 +%%This section examines the impact of macroeconomic surprises and financialization on the bid-ask spread during the Zero Lower Bound (ZLB) and COVID-19 sub-periods, aiming to determine whether the relationship between financialization and market efficiency, as measured by the bid-ask spread, remains consistent across different economic environments.
401 +%
402 +%Tables \ref{ch1:tab:macro_fin_fut_spread_zlb} and \ref{ch1:tab:macro_fin_fut_spread_covid} present our regression results for the bid-ask spread during the ZLB and COVID-19 sub-periods. We find that in both sub-periods the $\gamma_m$ coefficients tend to be negative, indicating that macroeconomic surprises generally reduce the bid-ask spread across our sample of commodities. This is the case, for instance, for Initial Jobless Claims for crude oil and copper futures in the COVID-19 sub-period, which implies increased efficiency. Furthermore, the $\theta_m$ coefficients tend to be negative in the two sub-periods, suggesting that financial investor activity improves efficiency by narrowing spreads after the release of macro news.
403 +%
404 +%%during the COVID-19 period are generally negative, indicating that increased financialization amplifies the efficiency improvements induced by macroeconomic surprises. Specifically, for crude oil, the negative $\theta_m$ coefficient suggests that the participation of financial investors, such as money managers, enhances the market's ability to process new information, further reducing the bid-ask spread.
405 +%%% POURQUOI ON PARLE DE MONEY MANAGERS ICI?
406 +%%shows the results for the ZLB period. The $\gamma_m$ coefficients are mostly negative, consistent with the full-sample findings, indicating that macroeconomic surprises during the ZLB period also tend to reduce the bid-ask spread, enhancing market efficiency. The $\theta_m$ coefficients are negative for most commodities, similar to the COVID-19 period, suggesting that financialization continues to improve market efficiency by further reducing the bid-ask spread following macroeconomic announcements.
407 +%
408 +%While the overall patterns in the $\gamma_m$ and $\theta_m$ coefficients are similar to the full sample results, the coefficients tend to be less statistically significant in these sub-periods. This may be due to smaller sample sizes or to unique market conditions during these periods.
409 +%%That being said, the results are consistent with our main findings for bid-ask spreads and suggest there are no meaningful differences in sub-periods.%, the general trend indicates that macroeconomic surprises reduce the bid-ask spread, and financialization enhances this effect, thereby improving market efficiency.
410 +%
411 +%That being said, we find that the beneficial impact of financialization on market efficiency, as measured by a narrower bid-ask spread, is robust across different economic conditions.
412 +%
413 +%%% JE NE CROIS PAS QUE LES PHRASES SUIVANTES AJOUTENT BEAUCOUP; ON SE REPETE.
414 +%%The consistent negative $\theta_m$ coefficients imply that the presence of financial investors, particularly money managers, helps markets process information more efficiently, regardless of the broader economic environment. This supports the argument that financialization contributes to improved market functioning by enhancing the dissemination and incorporation of new information into asset prices. Overall, the sub-period analysis reinforces our full-sample results, highlighting the beneficial role of financialization in commodity markets, particularly in terms of market efficiency as reflected in narrower bid-ask spreads following macroeconomic surprises.
415 +%
416 +%\subsection{Sub-period analysis for money managers and swap dealers}
417 +%
418 +%We discuss in this section a robustness check for the disaggregated analysis using only positions of money managers and swap dealers. In short, these results confirm the baseline finding that money managers bring more stability to markets through a damping effect after a macro surprise, while swap dealers have the opposite impact. The results for money managers are shown in table \ref{ch1:tab:macro_fin_mm_fut_returns_covid} (returns equation, COVID-19 period), table \ref{ch1:tab:macro_fin_mm_var_covid} (variance equation, COVID-19 period), \ref{ch1:tab:macro_fin_mm_fut_returns_zlb} (returns equation, ZLB period), and table \ref{ch1:tab:macro_fin_mm_var_zlb} (variance equation, ZLB period). The regression results for swap dealers, in the same order, are presented in tables \ref{ch1:tab:macro_fin_sd_fut_returns_covid}, \ref{ch1:tab:macro_fin_sd_var_covid}, \ref{ch1:tab:macro_fin_sd_fut_returns_zlb}, and \ref{ch1:tab:macro_fin_sd_var_zlb}.
419 +%
420 +%%The results during the ZLB sub-period reveal that the $\gamma_m$ coefficients for money managers are generally positive for crude oil and copper and negative for gold, similar to the full-sample findings. This suggests that the presence of money managers mitigates the impact of macroeconomic surprises on commodity futures returns. The $\theta_m$ coefficients are negative, further indicating that financialization by money managers dampens the effects of macro surprises. For swap dealers, the $\gamma_m$ coefficients are positive for most commodities, suggesting that their activities amplify market reactions to macroeconomic announcements. The $\theta_m$ coefficients for swap dealers are positive, supporting the notion that increased trading by swap dealers leads to greater market reactions to surprises.
421 +%
422 +%%During the COVID-19 sub-period, the results are consistent with the ZLB findings. The $\gamma_m$ coefficients for money managers show a similar pattern, with positive values for crude oil and copper and negative values for gold. The $\theta_m$ coefficients remain negative, confirming that financialization by money managers continues to reduce the impact of macroeconomic surprises on returns. For swap dealers, the $\gamma_m$ coefficients are again positive for most commodities, and the $\theta_m$ coefficients are positive, indicating that swap dealers' activities exacerbate market reactions to economic news. The variance equation results during both sub-periods show that the $\phi_m$ coefficients for money managers are negative, indicating that their presence reduces volatility following macroeconomic announcements. In contrast, the $\phi_m$ coefficients for swap dealers are positive, suggesting that their trading activities increase volatility in response to macro surprises.
423 +%
424 +%Overall, the sub-period analysis reinforces our full-sample findings. It highlights how money managers and swap dealers have different impacts on the market's reactions to macroeconomic announcements. While money managers tend to enhance market stability by damping the effects of surprises, swap dealers appear to increase market volatility and amplify the reactions to economic news.
425 +
426 +
427 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
428 +\section{Conclusion} \label{ch1:sec:conclusion}
429 +
430 +This paper investigates the impact of speculative trading on the real economy and energy markets through a new angle, namely high-frequency surprises in macroeconomic announcement releases, which allows for better-identified effects. %We study energy commodities as our baseline case, given their economic importance.
431 +We empirically test whether increased speculative trading activity amplifies or dampens the impact of macro surprises on prices and volatility in commodity futures markets.
432 +%ce paragraphe est bon, mais la conslusion est très longue et selon les commentaires on devrait focuser sur les trader types.
433 +%In fact, it is well known that the equity and bond markets react to these surprises. Suppose that a consequence of increased speculative trading intensity is to make energy and commodity futures behave more like financial assets. Then, we should see futures prices display greater reactions to macro surprises. This hypothesis is the basis of our investigation. Moreover, since the increased involvement of financial investors cannot be easily categorized into before and after sub-sample, we measure the intensity of speculative trading by means of a time-varying and commodity-specific proxy.
434 +Our results suggest that increased speculative trading activity has beneficial effects for energy and commodity markets. This is accomplished by reducing volatility and improving price discovery, as indeed price stability and efficient price discovery are crucial for market participants and the broader economy. We find that a greater intensity of speculative trading does not amplify the effects of macro announcement surprises on prices or volatility. On the contrary, an increase in speculative trading in a given commodity has a damping effect: prices and volatility react \emph{less} to macro surprises when speculative trading is higher. This stabilizing effect of speculation is especially valuable to energy markets, where price volatility can have significant economic consequences.
435 +
436 +What is more, our findings are consistent with information diffusion economic arguments. Our analysis of bid-ask spreads in futures contracts further confirms that speculative trading tends to improve market efficiency.
437 +% j'ai ajouté le début du paragraphe pour mettre l'emphase sur le main results
438 +Our results show that these effects are mostly linked to the trading activities of money managers rather than swap dealers. Thus, we contribute to a literature that emphasizes how non-commercial market participants such as money managers are beneficial to commodity markets by supplying liquidity, reducing volatility, and generally improving market efficiency. This finding is particularly relevant for energy markets, which have seen substantial increases in trading volume and complexity. %The results we present are robust to the use of a non-parametric variance estimator, different proxies for speculative trading intensity, and to alternative empirical specifications (e.g., regression specification, high-frequency window, etc.).
439 +
440 +Our findings have important implications for energy market regulation and policy. The damping effect on volatility shocks documented in this paper implies that speculative traders contribute to market stability, which is essential for energy security and economic planning. This stability could also facilitate investment in energy infrastructure and support the ongoing energy transition. By lowering the magnitude of volatility shocks, and thus reducing the real option value of delaying investments, our findings suggest that a greater involvement by speculative traders may also help with sustainability efforts to finance a green energy transition, alongside other instruments such as green bonds and portfolio screens for sustainable investments.
441 +Looking forward, our results suggest several promising avenues for future research in energy markets. First, the role of financial investors (speculators as well as passive investors) in facilitating the energy transition warrants further investigation. Second, the connection between speculative trading and energy market regulation remains an important area for study, given that we find different impacts for money managers and swap dealers. Finally, the impact on energy price discovery of new trading technologies and market participants is an emerging research frontier. These questions are particularly relevant given the increasing importance of energy markets in addressing climate uncertainty and in ensuring economic stability.
442 +
added these-ulaval/chapitre1/figures.tex +49 −0
@@ -0,0 +1,49 @@
1 +
2 + \begin{figure}[h]
3 + \centering
4 + \includegraphics[width=\textwidth]{pi_cl.png}
5 + \caption{Cumulative Abnormal Return of Crude Oil
6 +60 Minutes Before and After Macroeconomic Announcements}
7 + \label{ch1:fig:cl}
8 + \end{figure}
9 +
10 + \begin{figure}[h]
11 + \centering
12 + \includegraphics[width=\textwidth]{pi_gc.png}
13 + \caption{Cumulative Abnormal Return of Gold
14 +60 Minutes Before and After Macroeconomic Announcements}
15 + \label{ch1:fig:gc}
16 + \end{figure}
17 +
18 + \begin{figure}[h]
19 + \centering
20 + \includegraphics[width=\textwidth]{pi_hg.png}
21 + \caption{Cumulative Abnormal Return of Copper
22 +60 Minutes Before and After Macroeconomic Announcements}
23 + \label{ch1:fig:hg}
24 + \end{figure}
25 +
26 + \begin{figure}[h]
27 + \centering
28 + \includegraphics[width=\textwidth]{pi_si.png}
29 + \caption{Cumulative Abnormal Return of Silver
30 +60 Minutes Before and After Macroeconomic Announcements}
31 + \label{ch1:fig:pi}
32 + \end{figure}
33 +
34 + \begin{figure}[h]
35 + \centering
36 + \includegraphics[width=\textwidth]{pi_pa.png}
37 + \caption{Cumulative Abnormal Return of Palladium
38 +60 Minutes Before and After Macroeconomic Announcements}
39 + \label{ch1:fig:pa}
40 + \end{figure}
41 +
42 + \begin{figure}[h]
43 + \centering
44 + \includegraphics[width=\textwidth]{pi_ng.png}
45 + \caption{Cumulative Abnormal Return of Natural Gas
46 +60 Minutes Before and After Macroeconomic Announcements}
47 + \label{ch1:fig:ng}
48 + \end{figure}
49 +
\ No newline at end of file
added these-ulaval/chapitre1/figures/pi_cl.png +0 −0

Binary file not shown.

added these-ulaval/chapitre1/figures/pi_gc.png +0 −0

Binary file not shown.

added these-ulaval/chapitre1/figures/pi_hg.png +0 −0

Binary file not shown.

added these-ulaval/chapitre1/figures/pi_ng.png +0 −0

Binary file not shown.

added these-ulaval/chapitre1/figures/pi_pa.png +0 −0

Binary file not shown.

added these-ulaval/chapitre1/figures/pi_si.png +0 −0

Binary file not shown.

added these-ulaval/chapitre1/tables.tex +656 −0
@@ -0,0 +1,656 @@
1 +\section{Tables}
2 +
3 +
4 +
5 +
6 +\begin{landscape}
7 +\begin{table}[]
8 +\caption{Summary of the literature: Effect of non-commercial trading activity on commodity futures volatility}
9 +\label{ch1:tab:fin}
10 +\centering
11 +\begin{tabular}{@{}lll@{}}
12 +\toprule
13 +\textbf{References} & \textbf{Proxy used for financialization or speculation} & \textbf{Impact on volatility} \\ \midrule
14 +\citet{chang1997interday} & CFTC’s definition of speculators & \multirow{4}{*}{\textbf{Increase}} \\
15 +\citet{daigler1999impact} & CFTC’s definition of speculators & \\
16 +\citet{irwin2004effect} & Set speculators & \\
17 +\citet{tang2012index} & Commodity index trader (CIT) positions & \\ \midrule
18 +\citet{irwin1987note} & Amount of money invested in traded futures funds & \multirow{4}{*}{\textbf{No change}} \\
19 +\citet{irwin1999managed} & Trading volume of large-commodity pool operators & \\
20 +\citet{bryant2006causality} & CFTC’s definition of speculators & \\
21 +\citet{haigh2007hedge} & Number and positions of commodity pool operators and hedge funds & \\ \midrule
22 +\citet{brunetti2016speculators} & Net positions of hedge funds and floor brokers & \multirow{2}{*}{\textbf{Decrease}} \\
23 +\citet{aulerich2012bubbles} & Commodity index trader (CIT) positions & \\ \bottomrule
24 +\end{tabular}
25 +\begin{tablenotes}\item[]
26 +\singlespacing
27 +\footnotesize
28 +This table summarizes the findings of a range of studies on the effect of financialization and speculation on commodity futures volatility. The impact on volatility is categorized as Positive, Neutral, or Negative based on the results reported by each study. The proxies used for financialization or speculation include definitions and positions from the CFTC, set speculators, commodity index trader (CIT) positions, trading volume, and net positions of hedge funds and floor brokers.
29 +\end{tablenotes}
30 +\end{table}
31 +\end{landscape}
32 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
33 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
34 +\begin{table}[]
35 +\begin{center}
36 +\caption{List of macroeconomic announcements in our sample}
37 +\label{ch1:tab:stat1}
38 +\begin{tabular}{@{}lcccc@{}}
39 +\toprule
40 +\multicolumn{1}{c}{\textbf{Announcement}} & \textbf{Frequency} & \textbf{Source*} & \textbf{Unit} & \textbf{Time} \\ \midrule
41 +\textbf{GDP advance} & Quarterly & BEA & \% & 8:30 \\
42 +\textbf{GDP preliminary} & Quarterly & BEA & \% & 8:30 \\
43 +\textbf{GDP final} & Quarterly & BEA & \% & 8:30 \\
44 +\textbf{Personal income} & Monthly & BEA & \% & 8:30 \\
45 +\textbf{ADP employment} & Monthly & ADP & Number of jobs & 8:15 \\
46 +\textbf{Initial jobless claims} & Weekly & ETA & Number of claims & 8:30 \\
47 +\textbf{Non-farm employment} & Monthly & BLS & Number of jobs & 8:30 \\
48 +\textbf{Factory orders} & Monthly & BC & \% & 10:00 \\
49 +\textbf{Industrial production} & Monthly & FRB & \% & 9:15 \\
50 +\textbf{Construction spending} & Monthly & BC & \% & 10:00 \\
51 +\textbf{Durable goods orders} & Monthly & BC & \% & 8:30 \\
52 +\textbf{Advance retail sales} & Monthly & BC & \% & 8:30 \\
53 +\textbf{Consumer credit} & Monthly & FRB & USD & 15:00 \\
54 +\textbf{Personal consumption} & Monthly & BEA & \% & 8:30 \\
55 +\textbf{Building permits} & Monthly & BC & Number of permits & 8:30 \\
56 +\textbf{Existing home sales} & Monthly & NAR & Number of homes & 10:00 \\
57 +\textbf{Housing starts} & Monthly & BC & Number of homes & 8:30 \\
58 +\textbf{New home sales} & Monthly & BC & Number of homes & 10:00 \\
59 +\textbf{Pending home sales} & Monthly & NAR & \% & 10:00 \\
60 +\textbf{Trade balance} & Monthly & BEA & USD & 8:30 \\
61 +\textbf{Consumer price index} & Monthly & BLS & \% & 8:30 \\
62 +\textbf{Producer price index} & Monthly & BLS & \% & 8:30 \\
63 +\textbf{CB Consumer confidence index} & Monthly & CB & Index & 10:00 \\
64 +\textbf{UM Consumer sentiment} & Monthly & TR/UM & Index & 9:55 \\
65 +\textbf{Weekly Crude Oil Stock} & Weekly & EIA & number of barrels & 10:30 \\
66 +\textbf{Weekly Natural Gas Stock} & Weekly & EIA & number of cubic feet & 11:00 \\\bottomrule
67 +\end{tabular}
68 +\end{center}
69 +\begin{tablenotes}\item[]
70 + \singlespacing
71 + \footnotesize
72 +This table shows the category, frequency, source, unit of measure, and release time for each macroeconomic announcements. *(Automatic Data Processing, Inc. (ADP), Bureau of the Census (BC), Bureau of Economic Analysis (BEA), Bureau of Labor Statistics (BLS), Conference Board (CB), Employment and Training Administration (ETA), Federal Reserve Board (FRB), Institute for Supply Management (ISM), National Association of Realtors (NAR), Thomson Reuters/University of Michigan (TR/UM), and U.S. Department of the Treasury (USDT).)
73 +\end{tablenotes}
74 +\end{table}
75 +
76 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
77 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
78 +\begin{landscape}
79 +\begin{table}[]
80 +\begin{center}
81 +\caption{Descriptive statistics for the standardized surprise calculated for each of the macroeconomic announcements}
82 +\label{ch1:tab:stat2}
83 +\begin{tabular}{@{}lccccccc@{}}
84 +\toprule
85 +\multicolumn{1}{c}{\textbf{Announcements}} & \textbf{Nb. obs.} & \textbf{Min.} & \textbf{1st Qu.} & \textbf{Med.} & \textbf{Mean} & \textbf{3rd Qu.} & \textbf{Max.} \\ \midrule
86 +\textbf{Initial jobless claims} & 825 & -3.407 & -0.0720 & -0.007 & 0.068 & 0.065 & 22.672 \\
87 +\textbf{ADP Employment} & 202 & -2.751 & -0.0640 & 0.008 & 0.046 & 0.078 & 12.880 \\
88 +\textbf{CB Consumer} & 201 & -2.635 & -0.4638 & 0.093 & 0.102 & 0.872 & 2.412 \\
89 +\textbf{Advance retail sales} & 202 & -4.028 & -0.3661 & -0.092 & 0.023 & 0.183 & 8.879 \\
90 +\textbf{Building permit} & 198 & -2.375 & -0.5356 & 0.025 & 0.082 & 0.627 & 3.205 \\
91 +\textbf{Construction spending} & 202 & -3.054 & -0.5912 & -0.099 & -0.130 & 0.493 & 4.335 \\
92 +\textbf{Consumer\_credit} & 202 & -2.055 & -0.5217 & 0.104 & 0.061 & 0.619 & 3.131 \\
93 +\textbf{Consumer price index} & 201 & -3.483 & -0.6966 & 0.000 & -0.035 & 0.697 & 4.180 \\
94 +\textbf{Durable goods orders} & 193 & -2.702 & -0.5757 & 0.022 & 0.037 & 0.531 & 6.688 \\
95 +\textbf{Existing home sales} & 202 & -4.729 & -0.4627 & 0.000 & -0.067 & 0.488 & 2.467 \\
96 +\textbf{Factory orders} & 202 & -3.040 & -0.3378 & 0.000 & 0.081 & 0.507 & 2.534 \\
97 +\textbf{GDP} & 185 & -2.589 & -0.3698 & 0.000 & -0.026 & 0.370 & 2.958 \\
98 +\textbf{Housing starts} & 199 & -2.285 & -0.6178 & 0.000 & 0.042 & 0.624 & 3.401 \\
99 +\textbf{Industrial production} & 385 & -4.773 & -0.5727 & 0.000 & -0.080 & 0.573 & 2.291 \\
100 +\textbf{Michigan Sentiment Index} & 202 & -3.922 & -0.4100 & 0.036 & -0.049 & 0.463 & 3.244 \\
101 +\textbf{New home sales} & 202 & -3.062 & -0.3466 & 0.116 & 0.118 & 0.631 & 3.562 \\
102 +\textbf{Non-farm employment} & 201 & -0.892 & -0.0564 & 0.005 & 0.078 & 0.071 & 13.169 \\
103 +\textbf{Pending home sales} & 202 & -2.949 & -0.4244 & 0.022 & 0.075 & 0.581 & 5.674 \\
104 +\textbf{Personal consumption} & 201 & -3.666 & -0.3666 & 0.000 & -0.026 & 0.367 & 2.566 \\
105 +\textbf{Personal income} & 201 & -1.079 & -0.0771 & 0.000 & 0.095 & 0.077 & 13.108 \\
106 +\textbf{Producer price index} & 188 & -3.168 & -0.5760 & 0.000 & 0.083 & 0.864 & 2.880 \\
107 +\textbf{Trade balance} & 202 & -1.831 & -0.1801 & -0.018 & 0.013 & 0.207 & 2.359 \\ \bottomrule
108 +\end{tabular}
109 +\end{center}
110 +\begin{tablenotes}\item[]
111 + \singlespacing
112 + \footnotesize
113 +This table presents descriptive statistics for the standardized surprise $(A_{kt}-E_{kt})/\sigma_{kt}$ for each of the macroeconomic announcements. The column (Nb. Observations) shows the number of individual surprises that can be calculated over the whole analysis period. The columns (Min.), (1st Qu.), (Median), (Mean), (3rd Qu.) and (Max) present respectively the minimum value, the first quartile, the median, the mean, the third quartile and the maximum value for the standardized surprise of each macroeconomic announcement
114 +\end{tablenotes}
115 +\end{table}
116 +\end{landscape}
117 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
118 +
119 +%%%%%%%%%%%%%%%%%%%%%%%%%%%
120 +
121 +
122 +
123 +
124 +
125 +\begin{landscape}
126 +\begin{table}[]
127 +\begin{center}
128 +\caption{Descriptive statistics: 5-minute intraday futures price returns}
129 +\label{ch1:tab:stat4}
130 +\begin{tabular}{@{}lllllll@{}}
131 +\toprule
132 +\textbf{Commodity Futures} & \textbf{Min (\%)} & \textbf{1st Qu. (\%)} & \textbf{Med. (\%)} & \textbf{Mean (\%)} & \textbf{3rd Qu. (\%)} & \textbf{Max (\%)} \\ \midrule
133 +\textbf{Crude Oil (CL=F)} & -33.91 & -0.0441 & 0.000 & 0.000 & 0.0447 & 41.64 \\
134 +\textbf{Gold (GC=F)} & -2.782 & -0.0241 & 0.000 & 0.0001 & 0.0244 & 3.064 \\
135 +\textbf{Copper (HG=F)} & -4.534 & -0.0363 & 0.000 & -0.0001 & 0.0365 & 8.877 \\
136 +\textbf{Natural Gas (NG=F)} & -6.735 & -0.0528 & 0.000 & -0.0003 & 0.0532 & 15.62 \\
137 +\textbf{Palladium (PA=F)} & -13.350 & -0.034 & 0.000 & 0.0001 & 0.0348 & 9.467 \\
138 +\textbf{Silver (SI=F)} & -7.504 & -0.0394 & 0.000 & 0.0001 & 0.0415 & 4.242 \\ \bottomrule
139 +\end{tabular}
140 +\end{center}
141 +\begin{tablenotes}\item[]
142 + \singlespacing
143 + \footnotesize
144 +Shows descriptive statistics of the 5-minute intraday returns, for each commodity futures. The columns (Min.), (1st Qu.), (Median), (Mean), (3rd Qu.) and (Max) present respectively the minimum value, the first quartile, the median, the mean, the third quartile and the maximum value for the 5 minute intraday returns.
145 +\end{tablenotes}
146 +\end{table}
147 +\end{landscape}
148 +
149 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
150 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
151 +
152 +\begin{table}[ht]
153 +\centering
154 +\caption{Descriptive statistics: Computed value of the NLS proxy (respectively: full sample, money managers-only sample, swap dealers-only sample)}
155 +\label{ch1:tab:stat5}
156 +\small
157 +\begin{tabular}{lcccccc}
158 +\toprule
159 + & \textbf{CL} & \textbf{GC} & \textbf{HG} & \textbf{SI} & \textbf{PA} & \textbf{NG} \\
160 +\midrule
161 + & & & \textbf{$NLS$} & & & \\
162 +\midrule
163 +Min. & -0.1667 & -0.4505 & -0.3238 & -0.1364 & -0.6246 & -0.2745 \\
164 +1st Qu. & 0.0358 & 0.2285 & -0.1067 & 0.1547 & 0.3181 & -0.1731 \\
165 +Median & 0.0642 & 0.2397 & 0.0235 & 0.2163 & 0.3056 & -0.0563 \\
166 +Mean & 0.0774 & 0.1780 & 0.0278 & 0.2123 & 0.2685 & -0.0579 \\
167 +3rd Qu. & 0.1834 & 0.4012 & 0.1417 & 0.3334 & 0.5652 & -0.0321 \\
168 +Max. & 0.2941 & 0.5269 & 0.4413 & 0.5748 & 0.7343 & 0.0794 \\
169 +\midrule
170 + & & & \textbf{$NLS_{MM}$} & & & \\
171 +\midrule
172 +Min. & -0.0370 & -0.2353 & -0.2727 & -0.2303 & -0.6005 & -0.2471 \\
173 +1st Qu. & 0.0358 & 0.2177 & 0.0238 & 0.1550 & 0.3919 & -0.0345 \\
174 +Median & 0.1059 & 0.2177 & 0.0238 & 0.1550 & 0.3919 & -0.0345 \\
175 +Mean & 0.1011 & 0.2125 & 0.0350 & 0.1425 & 0.3294 & -0.0371 \\
176 +3rd Qu. & 0.1834 & 0.4012 & 0.1417 & 0.3334 & 0.5652 & -0.0321 \\
177 +Max. & 0.2051 & 0.4563 & 0.3923 & 0.4477 & 0.7330 & 0.1867 \\
178 +\midrule
179 + & & & \textbf{$NLS_{SD}$} & & & \\
180 +\midrule
181 +Min. & -0.2645 & -0.4092 & -0.2263 & -0.2263 & -0.3640 & -0.0689 \\
182 +1st Qu. & -0.1119 & -0.1457 & 0.0238 & -0.0255 & -0.0251 & 0.1091 \\
183 +Median & -0.1119 & -0.1457 & 0.2398 & -0.0255 & -0.0251 & 0.1091 \\
184 +Mean & -0.0897 & -0.1612 & 0.2395 & -0.0276 & 0.0086 & 0.1087 \\
185 +3rd Qu. & 0.1847 & 0.1290 & 0.3923 & 0.3334 & 0.5652 & 0.2743 \\
186 +Max. & 0.1847 & 0.1290 & 0.3923 & 0.3334 & 0.5652 & 0.2743 \\
187 +\bottomrule
188 +\end{tabular}
189 +\begin{tablenotes}\item[]
190 +\singlespacing
191 +\footnotesize
192 +This table provides descriptive statistics of the NLS proxy for speculative trading intensity for each commodity futures contract in our sample. The lines (Min.), (1st Qu.), (Median), (Mean), (3rd Qu.) and (Max) present respectively the minimum value, the first quartile, the median, the mean, the third quartile, and the maximum value. CL: crude oil, GC: gold, HG: high-grade copper, SI: silver, PA: palladium, NG: natural gas.
193 +\end{tablenotes}
194 +\end{table}
195 +
196 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
197 +
198 +\begin{sidewaystable}
199 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (NLS) on Futures Returns - Full Sample}
200 +\label{ch1:tab:macro_fin_nls_fut_returns_full}
201 +\centering
202 +\resizebox{\linewidth}{!}{%
203 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
204 +%\begin{tabular}{@{}lllllllllllll@{}}
205 +\toprule
206 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
207 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
208 +
209 +\textbf{Initial jobless claims} & -0.19763*** & 0.87888*** & 0.70022*** & -1.33955*** & -0.44930*** & 1.98801*** & -0.00740 & -0.01540 & 0.02913** & -0.22057*** & -0.00266 & -0.10938 \\
210 +\textbf{ADP Employment} & 0.34291*** & -1.06756** & -1.18741*** & 2.52498*** & 0.58107*** & -2.36989*** & -0.02591*** & 0.16900 & 0.00801 & -0.14170 & -0.11464*** & -2.79764*** \\
211 +\textbf{CB Consumer} & 0.11672*** & -0.46895*** & -0.05938*** & 0.07030* & 0.07957*** & -0.15159** & 0.00049 & 0.00567 & -0.06841*** & 0.11631** & 0.02933 & -0.01440 \\
212 +\textbf{Advance retail sales} & 0.19991*** & -0.74542*** & -0.22861*** & 0.44747*** & 0.10409*** & -0.42536** & 0.01333 & -0.08638 & -0.01767 & 0.00664 & -0.02725 & -0.47149 \\
213 +\textbf{Building permit} & -0.00037 & 0.03995 & -0.02651** & 0.03697 & 0.06270*** & -0.15653** & 0.00698 & -0.00358 & -0.00233 & -0.03748 & 0.04272* & 0.36164** \\
214 +\textbf{Construction spending} & 0.02190 & -0.14609 & -0.04079** & 0.09211** & 0.04044* & -0.14376* & -0.00070 & -0.18868*** & 0.00951 & -0.00257 & -0.02886 & -0.17899 \\
215 +\textbf{Consumer credit} & -0.01646 & 0.08344 & -0.01359* & 0.02955 & -0.01554* & 0.06217 & 0.00217 & 0.02437 & 0.01087 & -0.01043 & -0.01175 & -0.11173 \\
216 +\textbf{Consumer price index} & 0.08391*** & -0.29308** & -0.17224*** & 0.31022*** & 0.26865*** & -0.75945*** & -0.03002*** & -0.02265 & -0.14233*** & 0.23064*** & 0.01302 & 0.15012 \\
217 +\textbf{Durable goods orders} & 0.15300*** & -0.73816*** & -0.07265*** & 0.14703*** & 0.06435*** & -0.14930* & -0.01052* & 0.00024 & -0.05354*** & 0.14525*** & -0.00328 & -0.09972 \\
218 +\textbf{Existing home sales} & 0.08722*** & -0.59238*** & -0.02000 & 0.05026 & -0.02244 & 0.10822 & -0.02238*** & -0.06905 & 0.01177 & -0.04488 & 0.04083 & 0.21897 \\
219 +\textbf{Factory orders} & -0.00112 & 0.01036 & -0.02949 & 0.00285 & -0.05976** & 0.14591 & 0.00357 & 0.02488 & -0.01887 & 0.00026 & 0.10729*** & 0.68503*** \\
220 +\textbf{Gross domestic product} & 0.04571* & -0.18390 & -0.16085*** & 0.24165*** & 0.16953*** & -0.33056*** & 0.00920 & -0.00478 & -0.05405*** & 0.00789 & -0.03601 & -0.33730* \\
221 +\textbf{Housing starts} & 0.03224 & -0.12690 & -0.06263*** & 0.11383*** & 0.07118*** & -0.14876** & 0.00503 & 0.00563 & 0.00159 & -0.03648* & 0.02340 & 0.38343** \\
222 +\textbf{Industrial production} & 0.02075 & -0.15376 & -0.00085 & -0.07097 & -0.01332 & -0.11571 & -0.00947 & -0.00797 & -0.02227 & 0.02109 & -0.02279 & -0.02294 \\
223 +\textbf{New home sales} & 0.09853*** & -0.43638** & -0.06940*** & 0.13595*** & -0.03309* & -0.04095 & -0.02368*** & 0.08652* & -0.00796 & 0.02235 & -0.00278 & -0.08427 \\
224 +\textbf{Non-farm employment} & 1.40422*** & -5.11125*** & -3.19414*** & 6.81185*** & 1.38897*** & -6.04367*** & -0.02702*** & 0.62417** & 0.00013 & 0.15340 & -0.21810*** & -6.29434*** \\
225 +\textbf{Pending home sales} & 0.07333*** & -0.34943** & -0.01840 & 0.00310 & -0.01292 & -0.02301 & -0.01712** & -0.11126 & -0.00877 & 0.03099 & -0.02417 & -0.39913 \\
226 +\textbf{Personal consumption} & -0.01478 & 0.06799 & -0.03594** & 0.06306* & 0.00926 & -0.05828 & 0.00274 & 0.08690 & 0.01300 & -0.05027 & 0.00370 & 0.36897 \\
227 +\textbf{Personal income} & 0.00873 & -0.13628 & -0.07190 & 0.15000 & 0.25364*** & -1.05220** & -0.01709* & 0.27874** & -0.01271 & -0.02240 & -0.02828 & -0.82024 \\
228 +\textbf{Producer price index} & 0.03513* & -0.23561** & -0.07698*** & 0.15026*** & -0.01476 & -0.04153 & -0.00480 & -0.01293 & -0.02035 & 0.04451 & -0.04862* & -0.22346 \\
229 +\textbf{Trade balance} & 0.00320 & -0.05635 & -0.05207** & 0.11911** & -0.01111 & 0.04144 & -0.00354 & -0.13232 & -0.03871 & 0.07731 & -0.04667 & -0.45064 \\
230 +\textbf{Crude Oil Weekly inventory} & -0.07659*** & -0.39633*** & & & & & & & & & & \\
231 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.36472*** & -0.07672 \\ \midrule
232 +
233 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
234 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{0.198} & \multicolumn{2}{c}{0.204} & \multicolumn{2}{c}{0.101} & \multicolumn{2}{c}{0.0290} & \multicolumn{2}{c}{0.0246} & \multicolumn{2}{c}{0.118} \\ \bottomrule
235 +\end{tabular}
236 +}
237 +\begin{tablenotes}\item[]
238 +\singlespacing
239 +\footnotesize
240 +This table presents estimates of eq. \ref{ch1:eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and speculative trading intensity variable $X_{t}=NLS_t$. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in the futures return when an announcement has just occurred. The $\theta_m$ coefficients capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of speculative trading intensity.
241 +\end{tablenotes}
242 +\end{sidewaystable}
243 +
244 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
245 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
246 +
247 +\begin{sidewaystable}
248 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (NLS) on Futures Conditional Variance - Full Sample}
249 +\label{ch1:tab:macro_fin_nls_var_full}
250 +\centering
251 +\resizebox{\linewidth}{!}{%
252 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
253 +%\begin{tabular}{@{}lllllllllllll@{}}
254 +\toprule
255 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
256 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
257 +
258 +\textbf{Initial jobless claims} & 0.10116*** & -0.43618*** & 0.04743*** & 0.01989 & 0.11651*** & -0.18846*** & 0.02474*** & -0.05555*** & 0.11786*** & -0.11355*** & -0.01167 & -0.02107 \\
259 +\textbf{ADP Employment} & 0.02792 & -0.20043 & 0.06377*** & -0.08427*** & 0.05327*** & -0.03006 & 0.02831*** & 0.03382 & -0.02858* & 0.01538 & 0.01600 & 0.05689 \\
260 +\textbf{CB Consumer} & 0.10356*** & -0.41440*** & 0.03139*** & 0.01209 & 0.04540*** & 0.02788 & 0.02838*** & -0.10290** & 0.06091*** & -0.09895*** & -0.00770 & -0.01276 \\
261 +\textbf{Advance retail sales} & 0.16868*** & -0.86165*** & 0.09604*** & -0.05612* & 0.12379*** & -0.05358 & 0.02819*** & -0.15572*** & 0.11351*** & -0.16926*** & 0.05769*** & 0.25105* \\
262 +\textbf{Building permit} & 0.10470*** & -0.54035*** & 0.04592*** & -0.06034* & 0.06133*** & -0.19301*** & 0.00484 & -0.07408* & 0.07672*** & -0.06975*** & 0.02190 & -0.02627 \\
263 +\textbf{Construction spending} & 0.15338*** & -0.62724*** & 0.09518*** & -0.11889*** & 0.08606*** & -0.09793 & 0.04108*** & -0.18959*** & 0.04626*** & 0.01330 & -0.01904 & -0.32089** \\
264 +\textbf{Consumer credit} & 0.04798* & -0.26103* & 0.00778 & -0.00318 & 0.01661 & 0.00378 & 0.00280 & 0.03858 & -0.00633 & 0.03412 & 0.00152 & 0.01392 \\
265 +\textbf{Consumer price index} & 0.07237** & -0.06680 & 0.13915*** & -0.07513* & 0.26627*** & -0.52898*** & 0.09389*** & -0.37452*** & 0.18224*** & -0.27404*** & 0.02706 & 0.03922 \\
266 +\textbf{Durable goods orders} & 0.09826*** & -0.41079*** & 0.02310** & 0.01561 & 0.04156*** & 0.02205 & 0.00773 & -0.02411 & 0.06781*** & -0.03730 & -0.02225 & -0.25576* \\
267 +\textbf{Existing home sales} & 0.04757* & -0.05948 & 0.01879* & 0.04102 & 0.03838*** & -0.02961 & 0.02231*** & -0.13249*** & 0.04815*** & -0.06749** & 0.02201 & -0.11251 \\
268 +\textbf{Factory orders} & 0.08762*** & -0.31063** & -0.00678 & 0.16162*** & 0.01842 & 0.15052** & 0.02641*** & -0.13159*** & 0.03966*** & -0.04030 & -0.03041 & -0.69494*** \\
269 +\textbf{Gross domestic product} & 0.11632*** & -0.61837*** & 0.04659*** & 0.05303 & 0.09959*** & -0.06323 & 0.02464*** & 0.01653 & 0.11025*** & -0.12093*** & -0.05116** & -0.30947** \\
270 +\textbf{Housing starts} & 0.11313*** & -0.60745*** & 0.03660*** & -0.03112 & 0.06213*** & -0.19012*** & 0.00848 & -0.09479** & 0.05963*** & -0.04924 & 0.01829 & 0.04047 \\
271 +\textbf{Industrial production} & 0.09578*** & -0.63833*** & 0.01928* & 0.00308 & 0.02059 & -0.00706 & 0.00182 & -0.04794 & 0.01964 & -0.07145** & -0.03281 & -0.18705 \\
272 +\textbf{New home sales} & 0.10149*** & -0.52688*** & 0.04068*** & 0.01515 & 0.05778*** & 0.02800 & 0.02485*** & -0.00096 & 0.06768*** & -0.11522*** & -0.02516 & -0.22929* \\
273 +\textbf{Non-farm employment} & 0.36475*** & -1.20583*** & 0.23592*** & 0.04682 & 0.38126*** & -0.14107** & 0.12881*** & -0.15602*** & 0.17336*** & -0.05048 & 0.02807 & -0.35383** \\
274 +\textbf{Pending home sales} & 0.10920*** & -0.49393*** & 0.01999* & -0.01624 & 0.02548* & 0.02096 & 0.02833*** & -0.00831 & 0.05870*** & -0.08511*** & -0.01374 & -0.30813** \\
275 +\textbf{Personal consumption} & 0.00638 & 0.02122 & 0.06203*** & -0.08145** & 0.10177*** & -0.26781*** & 0.00662 & -0.00781 & 0.11081*** & -0.14798*** & 0.02410 & -0.37604** \\
276 +\textbf{Personal income} & 0.00650 & 0.04657 & 0.09149*** & -0.15016*** & 0.12450*** & -0.34101*** & 0.01322** & -0.02338 & 0.09417*** & -0.10866*** & 0.00283 & -0.43167*** \\
277 +\textbf{Producer price index} & 0.09735*** & -0.34959** & 0.01720 & 0.08462** & 0.09045*** & -0.09926 & 0.01106* & -0.05096 & 0.11987*** & -0.06696** & 0.00810 & 0.00477 \\
278 +\textbf{Trade balance} & 0.07844*** & -0.12829 & -0.00650 & 0.20643*** & 0.00273 & 0.39295*** & 0.00919 & 0.02977 & 0.10323*** & -0.08686*** & 0.01821 & 0.11158 \\
279 +\textbf{Crude Oil Weekly inventory} & 0.00456 & 0.62373*** & & & & & & & & & & \\
280 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.53064*** & -0.68665*** \\ \midrule
281 +
282 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
283 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{7.065} & \multicolumn{2}{c}{8.064} & \multicolumn{2}{c}{7.396} & \multicolumn{2}{c}{7.057} & \multicolumn{2}{c}{7.629} & \multicolumn{2}{c}{13.60} \\ \bottomrule
284 +\end{tabular}
285 +}
286 +\begin{tablenotes}\item[]
287 +\singlespacing
288 +\footnotesize
289 +This table presents estimates of eq. \ref{ch1:eqn:VarianceEqn} using the speculative trading intensity variable $NLS_t$. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2007-04-01 to 2024-02-11. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of speculative trading intensity.
290 +\end{tablenotes}
291 +\end{sidewaystable}
292 +
293 +
294 +
295 +
296 +
297 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
298 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
299 +\begin{sidewaystable}
300 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (NLS) on Futures Price Bid-Ask Spreads - Full Sample}
301 +\label{ch1:tab:return-fin-full}
302 +\centering
303 +\resizebox{\linewidth}{!}{%
304 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
305 +%\begin{tabular}{@{}lllllllllllll@{}}
306 +\toprule
307 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Palladium}} \\ \midrule
308 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule
309 +\multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
310 +\textbf{Initial jobless claims} & -2.7096*** & 13.6978 & 5.4253 & -10.3796*** & 5.8502 & -26.3969*** & -0.2493 & -4.5426*** & 0.0148 & 0.8567 & -0.1912 & 0.5368 \\
311 +\textbf{ADP Employment} & 3.6287 & -10.9417*** & -10.5416*** & 22.3535 & -7.3992*** & 31.3360 & -0.6515 & -13.8892*** & 0.2498 & 2.0429 & 0.0364 & -0.4670\\
312 +\textbf{CB Consumer} & 1.4188 & -6.2213*** & -0.8502 & 1.4474 & -1.1319 & 3.0482 & 0.5121 & 0.3016 & 0.0264 & 0.2419 & -0.7750 & 1.9280 \\
313 +\textbf{Advance retail sales} & 2.0886 & -7.0546*** & -2.2792 & 4.4308 & -0.8497 & 2.8479 & -0.2272 & -2.0227 & 0.3505 & -3.2196*** & -0.4188 & 1.5128 \\
314 +\textbf{Building permit} & -0.1970 & 1.7823 & -0.4372 & 0.6514 & -0.7742 & 2.5610 & 0.3011 & 3.7647 & 0.0299 & 0.3677 & -0.1989 & -0.2028 \\
315 +\textbf{Construction spending} & 0.1768 & -1.0346 & -0.0637 & 0.0761 & -0.2650 & 0.8472 & -0.1867 & -2.6962*** & 0.0609 & -2.4199*** & -0.1259 & 0.2982 \\
316 +\textbf{Consumer credit} & 0.3706 & -1.8987 & -0.0269 & 0.0457 & -0.0288 & 0.1356 & -0.0409 & 0.0473 & 0.0549 & 0.3589 & -0.0071 & 0.0208 \\
317 +\textbf{Consumer price index} & -0.8292 & 1.3319 & -2.2389 & 4.8692 & -3.0533 & 9.4014 & -0.0417 & 1.9823 & -0.5620 & 1.0961 & -1.8277 & 3.2679 \\
318 +\textbf{Durable goods orders} & 1.3175 & -7.3678*** & -0.9559 & 2.3132 & -1.3099 & 5.5532 & -0.1211 & -1.4967 & 0.2027 & -0.1838 & -0.3876 & 1.2514 \\
319 +\textbf{Existing home sales} & 1.0542 & -5.3658*** & -0.3977 & 1.0670 & -0.1607 & 1.4199 & 1.0635 & 5.2952 & 0.4218 & -1.8117 & 0.0162 & -0.0411 \\
320 +\textbf{Factory orders} & -0.3378 & 1.5881 & -0.8512 & 1.3429 & -0.7687 & 1.8326 & 1.5490 & 11.3206 & 0.0137 & 1.0877 & -0.0331 & -0.3245 \\
321 +\textbf{GDP} & 0.3581 & -1.5995 & -1.8905 & 3.7376 & -2.2481 & 6.3831 & -0.3939 & -3.6342*** & 0.1213 & -0.7319 & -0.5648 & 0.3738 \\
322 +\textbf{Housing starts} & 0.1898 & -1.2327 & -0.9901 & 1.8960 & -0.9695 & 2.5715 & 0.5413 & 6.5382 & 0.0832 & 0.5107 & 0.0362 & -0.5628 \\
323 +\textbf{Industrial production} & 0.1195 & -2.0990 & -0.1954 & -0.4140 & -0.0606 & -1.9941 & -0.2034 & -1.1923 & -0.0584 & -0.4020 & -0.1814 & 0.2817 \\
324 +\textbf{Manufacturing capacity} & -0.1028 & 2.2785 & -0.6122 & 0.9215 & -0.8221 & 1.9754 & -0.3658 & -2.8980*** & -0.0373 & 0.6966 & -0.1408 & 0.5487 \\
325 +\textbf{New home sales} & 1.6308 & -8.3434*** & -0.9063 & 1.6516 & -0.3849 & -0.4070 & -0.2523 & -0.8854 & 0.3688 & -0.4561 & -0.0098 & 0.0570 \\
326 +\textbf{Non-farm payroll} & 10.7879 & -38.8878*** & -29.7105*** & 63.3421 & -10.2858*** & 43.9569 & -2.4909 & -71.4713*** & 0.2717 & 8.5221 & -0.0000 & 1.4673 \\
327 +\textbf{Pending home sales} & 0.8251 & -2.5452 & -0.4223 & 0.5242 & -0.1941 & -0.1573 & -0.5748 & -9.0134*** & 0.1567 & -2.3599*** & -0.3537 & 0.2819 \\
328 +\textbf{Personal consumption} & -0.2150 & -0.0394 & -0.8262 & 1.6288 & -0.0344 & 0.1194 & -0.1174 & 3.9321 & 0.0919 & 2.0435 & -0.2523 & 0.5773 \\
329 +\textbf{Personal income} & -0.0876 & -0.8200 & -1.5351 & 3.2898 & -4.4002*** & 17.6153 & -0.4605 & -12.8607*** & -0.2459 & -3.8381*** & -0.1619 & 0.2354 \\
330 +\textbf{Producer price index} & 0.7142 & -4.4244*** & -0.7227 & 1.4344 & 0.2196 & -1.8938 & -0.5144 & -2.8027*** & 0.0608 & -0.1209 & -0.7231 & 1.4795 \\
331 +\textbf{Treasury balance} & 0.0459 & -1.6195 & -0.8212 & 1.9771 & -0.8034 & 2.9669 & -0.1261 & -0.7379 & -0.0332 & -1.5029 & -0.3137 & 0.8416 \\
332 +\midrule
333 +\multicolumn{13}{c}{\textbf{Announcements specific to commodity markets}} \\ \midrule
334 +\textbf{Weekly crude oil stock} & -0.9773 & -3.4730*** & & & & & & & & & & \\
335 +\textbf{Natural Gas Weekly inventory} & & & & & & & -1.7473 & -7.0273*** & & & & \\
336 +\midrule
337 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{1,101,836} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{749,168} \\
338 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{1.98} & \multicolumn{2}{c}{1.77} & \multicolumn{2}{c}{0.56} & \multicolumn{2}{c}{1.85} & \multicolumn{2}{c}{1.28} & \multicolumn{2}{c}{1.51} \\
339 +\bottomrule
340 +\end{tabular}
341 +}
342 +\begin{tablenotes}\item[]
343 + \singlespacing
344 + \footnotesize
345 + This table presents estimates of the equation $R_{SPREAD_{t}}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+\delta X_{t,i}+\sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{SPREAD_{t-\tau}}^{t}+\epsilon_t$, analyzing the effects of speculative trading intensity and macroeconomic announcements on the bid-ask spread using the speculative trading intensity variable $NLS_t$. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the speculative trading intensity variable $NLS_t$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and speculative trading intensity.
346 +\end{tablenotes}
347 +\end{sidewaystable}
348 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
349 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
350 +
351 +
352 +\begin{sidewaystable}
353 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Money Managers only) on Futures Returns}
354 +\label{ch1:tab:macro_fin_mm_fut_returns_full}
355 +\centering
356 +\resizebox{\linewidth}{!}{%
357 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
358 +%\begin{tabular}{@{}lllllllllllll@{}}
359 +\toprule
360 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
361 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
362 +
363 +\textbf{Initial jobless claims} & -0.02956 & 0.01952 & 0.30064*** & -0.96225*** & -0.02630 & 0.33366 & -0.00912 & -0.02954 & 0.04621** & -0.27526*** & 0.00955 & 0.06074 \\
364 +\textbf{ADP Employment} & 0.29189** & -1.23728* & -0.06824 & 0.15437 & 0.22281***& -0.97429** & -0.02291***& 0.14952 & 0.01012 & -0.11814 & -0.02204 & -0.49202 \\
365 +\textbf{CB Consumer} & 0.11505*** & -0.65428** & -0.04763***& 0.05599 & 0.06964***& -0.13979** & 0.00036 & 0.00188 & -0.06685***& 0.11545** & 0.03748** & 0.12201 \\
366 +\textbf{Advance retail sales} & 0.18018*** & -0.95056***& 0.00625 & -0.22754*** & 0.09507***& -0.52284***& 0.01327 & -0.06932 & -0.02139 & 0.02073 & -0.00518 & -0.24418 \\
367 +\textbf{Building permit} & -0.03075 & 0.30773* & -0.01409** & -0.00246 & 0.04883***& -0.12171** & 0.00701 & 0.00070 & -0.00304 & -0.03782 & 0.00348 & 0.08695 \\
368 +\textbf{Construction spending} & 0.04684 & -0.41529 & -0.04794***& 0.17091*** & 0.03370** & -0.15213* & 0.00487 & -0.12621** & 0.00961 & -0.00307 & -0.01120 & -0.09293 \\
369 +\textbf{Consumer credit} & -0.00932 & 0.05736 & -0.01044** & 0.04101* & -0.01023 & 0.05293 & 0.00149 & 0.02019 & 0.01095 & -0.01252 & -0.00177 & -0.04766 \\
370 +\textbf{Consumer price index} & 0.08568*** & -0.42002** & -0.15785***& 0.47194*** & 0.21663***& -0.72688***& -0.02944***& -0.01458 & -0.13577***& 0.22337*** & -0.00076 & 0.07776 \\
371 +\textbf{Durable goods orders} & 0.09316*** & -0.69602***& -0.05849***& 0.15716*** & 0.05268***& -0.11874 & -0.01489** & -0.05119 & -0.06161***& 0.17723*** & 0.00976 & -0.02018 \\
372 +\textbf{Existing home sales} & 0.09536*** & -0.80184***& -0.00676 & 0.01184 & -0.01163 & 0.07853 & -0.02082** & -0.01303 & 0.01267 & -0.05028 & 0.01320 & 0.05330 \\
373 +\textbf{Factory orders} & -0.02429 & 0.21872 & -0.04389***& 0.06866 & 0.05024** & -0.14478 & 0.00322 & 0.01249 & -0.01854 & -0.00031 & 0.04152** & 0.38741** \\
374 +\textbf{Gross domestic product} & 0.04408* & -0.25171 & -0.08308***& 0.02306 & 0.12802***& -0.18740** & 0.00637 & 0.02704 & -0.05987***& 0.02416 & -0.00388 & -0.28244* \\
375 +\textbf{Housing starts} & -0.01672 & 0.19837 & -0.02757***& 0.00801 & 0.04577***& -0.03650 & 0.00348 & 0.02420 & 0.00214 & -0.04053* & -0.01298 & 0.14226 \\
376 +\textbf{Industrial production} & -0.00324 & -0.03528 & -0.02254** & -0.02049 & -0.02215 & -0.10057 & -0.00675 & -0.03402 & -0.03174 & 0.04511 & -0.01321 & 0.23336 \\
377 +\textbf{New home sales} & 0.03959 & -0.20140 & -0.03508***& 0.05297 & 0.03238** & 0.06133 & -0.02410***& -0.04529 & -0.01226 & 0.03240 & 0.00382 & -0.18259 \\
378 +\textbf{Non-farm employment} & 0.79766*** & -4.07313***& -0.51670***& 2.14516*** & 0.77114***& -4.43004***& -0.01496* & 0.61309*** & 0.02467 & -0.16206 & -0.00450 & -0.95575 \\
379 +\textbf{Pending home sales} & 0.05641 & -0.31219 & -0.01885* & 0.00739 & -0.02053 & 0.01626 & -0.01554* & -0.03397 & -0.01511 & 0.05025 & 0.01033 & -0.26951 \\
380 +\textbf{Personal consumption} & 0.09082** & -0.64077***& -0.01948** & 0.04335 & 0.00547 & -0.06372 & -0.00024 & 0.08100 & 0.01162 & -0.04886 & -0.02503 & 0.23224 \\
381 +\textbf{Personal income} & 0.05847 & -0.50101 & 0.01713 & -0.09655 & 0.14463** & -0.83394** & -0.00826 & -0.20026 & -0.01654 & 0.00889 & 0.00405 & -1.05049* \\
382 +\textbf{Producer price index} & 0.03892* & -0.32583* & -0.05523***& 0.13297*** & -0.00600 & -0.12185 & -0.00347 & -0.03082 & -0.02220 & 0.05468 & -0.02632 & -0.11214 \\
383 +\textbf{Trade balance} & 0.00732 & -0.10801 & -0.03076** & 0.08094* & -0.00911 & 0.03998 & 0.00382 & -0.11800 & -0.02494 & 0.06033 & -0.00010 & -0.26851 \\
384 +\textbf{Crude Oil Weekly inventory} & -0.07233*** & -0.61043***& & & & & & & & & & \\
385 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.35459*** & -0.12230 \\ \midrule
386 +
387 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
388 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{0.181} & \multicolumn{2}{c}{0.151} & \multicolumn{2}{c}{0.0923} & \multicolumn{2}{c}{0.0284} & \multicolumn{2}{c}{0.0235} & \multicolumn{2}{c}{0.113} \\ \bottomrule
389 +\end{tabular}
390 +}
391 +\begin{tablenotes}\item[]
392 +\singlespacing
393 +\footnotesize
394 +This table presents estimates of eq. \ref{ch1:eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and the money manager-only speculative trading intensity variable $NLS_{t,MM}$, calculated with the money manager positions. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of speculative trading intensity.
395 +\end{tablenotes}
396 +\end{sidewaystable}
397 +
398 +
399 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
400 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
401 +
402 +
403 +\begin{sidewaystable}
404 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Swap Dealers only) on Futures Returns}
405 +\label{ch1:tab:macro_fin_sd_fut_returns_full}
406 +\centering
407 +\resizebox{\linewidth}{!}{%
408 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
409 +%\begin{tabular}{@{}lllllllllllll@{}}
410 +\toprule
411 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
412 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
413 +
414 +\textbf{Initial jobless claims} & -0.18601*** & -0.76285*** & 0.36069*** & 1.02251*** & 0.01407 & -0.51053 & 0.08238** & 0.44801** & 0.00571 & -0.03460 & -0.00709 & 0.23566 \\
415 +\textbf{ADP Employment} & 0.42554*** & 1.77009*** & -0.73172***& -1.92332*** & 0.14323*** & 1.90790*** & 0.10514 & -0.41655 & -0.08432* & -0.37900* & -0.07015***& 1.98134*** \\
416 +\textbf{CB Consumer} & 0.08606*** & 0.38665*** & -0.02682***& 0.06149 & 0.04702*** & 0.04493 & -0.05678***& -0.25240***& -0.02421** & -0.13691** & 0.02902 & 0.01534 \\
417 +\textbf{Advance retail sales} & 0.18764*** & 0.83656*** & -0.22074***& -0.62235*** & 0.02334* & 0.32997 & -0.05224** & -0.34386** & -0.02373 & 0.03729 & -0.01204 & 0.18256 \\
418 +\textbf{Building permit} & 0.01182 & 0.03350 & -0.02794***& -0.06415** & 0.03852*** & 0.19589* & 0.00878 & -0.00836 & -0.01490** & 0.06228 & 0.06712** & -0.57136***\\
419 +\textbf{Construction spending} & 0.00921 & 0.10371 & -0.01531* & -0.03974 & 0.01559 & 0.15444 & -0.04579** & -0.19927** & 0.01295 & 0.06733 & -0.00588 & -0.03512 \\
420 +\textbf{Consumer credit} & -0.00057 & 0.00723 & -0.00297 & 0.00152 & -0.00353 & 0.00472 & 0.00671 & -0.01905 & 0.00686 & 0.02748 & -0.02010 & 0.18485 \\
421 +\textbf{Consumer price index} & -0.05142*** & -0.12282 & -0.04196***& -0.11608** & 0.10927*** & 0.50120*** & -0.07969***& -0.20611** & -0.05426***& -0.28370*** & 0.04152 & -0.40373 \\
422 +\textbf{Durable goods orders} & 0.08229*** & 0.40797*** & -0.02824***& -0.02785 & 0.02337** & 0.27759*** & -0.00067 & 0.04659 & 0.00927 & -0.09008** & 0.01138 & -0.01219 \\
423 +\textbf{Existing home sales} & 0.02521** & 0.28450*** & -0.00669 & -0.02050 & -0.00093 & 0.07534 & -0.06009***& -0.33603***& -0.00636 & 0.05054 & 0.05433 & -0.33599 \\
424 +\textbf{Factory orders} & -0.00312 & -0.03814 & -0.00410 & 0.14500** & 0.03244*** & -0.19383 & 0.01462 & -0.04673 & -0.01972 & -0.03200 & 0.09608*** & -0.62225** \\
425 +\textbf{Gross domestic product} & 0.03848** & 0.17661** & -0.09123***& -0.06216 & 0.09968*** & -0.15355 & -0.02494 & 0.14602** & -0.05188***& -0.00867 & -0.03176 & 0.30154 \\
426 +\textbf{Housing starts} & 0.03038** & 0.13425* & -0.05461***& -0.14857*** & 0.05147*** & 0.21799** & -0.01782 & 0.11965* & -0.01043 & 0.02270 & 0.01927 & -0.32116 \\
427 +\textbf{Industrial production} & 0.02354 & 0.19850* & -0.00350 & 0.10713** & 0.03408*** & 0.14657 & -0.02953 & 0.09807 & -0.01335 & -0.01264 & -0.02947 & 0.09028 \\
428 +\textbf{New home sales} & 0.07156*** & 0.34303*** & -0.04011***& -0.08154* & 0.03797*** & 0.09004 & -0.03774** & -0.28795***& 0.00009 & -0.02615 & 0.01621 & -0.11025 \\
429 +\textbf{Non-farm employment} & 1.81964*** & 8.77192*** & -3.07930***& -8.25792*** & 0.22593*** & 3.88087*** & -0.48535***& -2.66148***& -0.14565* & -0.68396* & -0.06665** & 2.42302** \\
430 +\textbf{Pending home sales} & 0.04227*** & 0.27467*** & -0.01662 & 0.00210 & -0.01694 & 0.07454 & -0.03279 & 0.19949** & 0.00391 & -0.02496 & -0.00429 & 0.25203 \\
431 +\textbf{Personal consumption} & -0.00158 & -0.00543 & -0.01737* & -0.02729 & 0.00538 & 0.32148** & -0.00779 & 0.03095 & -0.00577 & 0.07145 & 0.00273 & -0.38325 \\
432 +\textbf{Personal income} & -0.02221 & 0.00621 & -0.05774* & -0.15405 & 0.08167*** & 1.82948*** & -0.09051 & 0.43742 & -0.02297 & 0.03880 & -0.00815 & 0.53016 \\
433 +\textbf{Producer price index} & 0.00818 & 0.07471 & -0.02420***& -0.00196 & 0.02453*** & 0.08862 & -0.03150 & 0.11098 & -0.00586 & -0.03594 & -0.03630 & 0.15142 \\
434 +\textbf{Trade balance} & -0.00089 & 0.03751 & -0.01618 & -0.06246 & -0.01214 & -0.23857 & -0.04471 & 0.14389 & -0.01350 & -0.22026* & 0.00043 & 0.12505 \\
435 +\textbf{Crude Oil Weekly inventory} & -0.14367*** & 0.03576 & & & & & & & & & & \\
436 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -0.47323***& 0.90803*** \\ \midrule
437 +
438 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
439 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{0.196} & \multicolumn{2}{c}{0.206} & \multicolumn{2}{c}{0.0809} & \multicolumn{2}{c}{0.0337} & \multicolumn{2}{c}{0.0195} & \multicolumn{2}{c}{0.123} \\ \bottomrule
440 +\end{tabular}
441 +}
442 +\begin{tablenotes}\item[]
443 +\singlespacing
444 +\footnotesize
445 +This table presents estimates of eq. \ref{ch1:eq:Model 1}, $R_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m S_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (S_{m,t} \cdot X_t)+\beta R_{t-\tau}^{t}+\epsilon_{t}$ using the method proposed by \citet{kurov2019price} and the swap dealer-specific speculative trading intensity variable $NLS_{t,SD}$, calculated with the swap dealer positions. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in return when an announcement has just occurred. The coefficients $\theta_m$ capture the instantaneous change in return when an announcement has just occurred in conjunction with the level of speculative trading intensity.
446 +\end{tablenotes}
447 +\end{sidewaystable}
448 +
449 +
450 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
451 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
452 +
453 +
454 +\begin{sidewaystable}
455 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Money Managers only) on Futures Conditional Variance}
456 +\label{ch1:tab:macro_fin_mm_var_full}
457 +\centering
458 +\resizebox{\linewidth}{!}{%
459 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
460 +%\begin{tabular}{@{}lllllllllllll@{}}
461 +\toprule
462 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
463 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
464 +
465 +\textbf{Initial jobless claims} & 0.08555*** & -0.47011*** & 0.05181*** & 0.01050 & 0.10409*** & -0.17922*** & 0.02685*** & -0.04784*** & 0.12581*** & -0.13984*** & -0.01102** & -0.04846 \\
466 +\textbf{ADP Employment} & 0.00118 & -0.03201 & 0.04021*** & -0.02019 & 0.05068*** & -0.02457 & 0.02543*** & -0.06310* & -0.02832* & 0.01564 & 0.01156 & 0.05686 \\
467 +\textbf{CB Consumer} & 0.11197*** & -0.66649*** & 0.03980*** & -0.01979 & 0.04828*** & 0.01872 & 0.03063*** & -0.04699 & 0.06245*** & -0.10441*** & -0.00788 & -0.04287 \\
468 +\textbf{Advance retail sales} & 0.14174*** & -0.96296*** & 0.08087*** & -0.01399 & 0.12479*** & -0.08493 & 0.03330*** & -0.11206*** & 0.12530*** & -0.20972*** & 0.03867*** & 0.27501** \\
469 +\textbf{Building permit} & 0.10251*** & -0.75666*** & 0.02103*** & 0.02773 & 0.04613*** & -0.16591*** & 0.00635 & -0.03797 & 0.08523*** & -0.10006*** & 0.02777** & 0.07714 \\
470 +\textbf{Construction spending} & 0.14202*** & -0.78417*** & 0.06648*** & -0.04959* & 0.07206*** & -0.04082 & 0.04891*** & -0.17441*** & 0.05608*** & -0.01314 & 0.00738 & -0.30331*** \\
471 +\textbf{Consumer credit} & 0.03931 & -0.28727 & 0.00369 & 0.01455 & 0.00970 & 0.05397 & 0.00119 & 0.03617 & -0.00823 & 0.03933 & 0.00029 & 0.00914 \\
472 +\textbf{Consumer price index} & 0.05875* & 0.03855 & 0.15937*** & -0.21038***& 0.23368*** & -0.51978*** & 0.10425*** & -0.21797*** & 0.18249*** & -0.28192*** & 0.02276 & 0.01481 \\
473 +\textbf{Durable goods orders} & 0.05652** & -0.21010 & 0.02385*** & 0.02141 & 0.04912*** & -0.02468 & 0.00903 & -0.03221 & 0.07695*** & -0.06578* & 0.00235 & -0.15833 \\
474 +\textbf{Existing home sales} & 0.11456*** & -0.69145*** & 0.02386*** & 0.03978 & 0.04052*** & -0.05757 & 0.02726*** & -0.11502*** & 0.05097*** & -0.07660** & 0.03224*** & -0.08380 \\
475 +\textbf{Factory orders} & 0.16239*** & -1.11410*** & 0.03236*** & -0.07028** & 0.02909*** & -0.13809** & 0.03208*** & -0.13161*** & 0.04159*** & -0.04631 & 0.03273*** & -0.46006*** \\
476 +\textbf{Gross domestic product} & 0.08125*** & -0.53798** & 0.05453*** & 0.04432 & 0.09089*** & -0.02946 & 0.02457*** & 0.00785 & 0.12198*** & -0.15871*** & -0.02217* & -0.24014** \\
477 +\textbf{Housing starts} & 0.11207*** & -0.85733*** & 0.02082*** & 0.02774 & 0.04880*** & -0.17498*** & 0.01059* & -0.05686 & 0.06737*** & -0.07388** & 0.01719 & 0.09484 \\
478 +\textbf{Industrial production} & 0.12021*** & -1.14614*** & 0.01389** & 0.03136 & 0.03199*** & -0.09631 & 0.00429 & -0.06702* & 0.02154 & -0.07965** & -0.01891 & -0.21217* \\
479 +\textbf{New home sales} & 0.07830*** & -0.53409** & 0.04935*** & -0.01730 & 0.06779*** & -0.03253 & 0.02495*** & -0.00382 & 0.06925*** & -0.12089*** & -0.00394 & -0.15829 \\
480 +\textbf{Non-farm employment} & 0.30168*** & -1.11309*** & 0.19942*** & -0.24293***& 0.34791*** & 0.03828 & 0.13642*** & -0.17279*** & 0.17368*** & -0.05267 & 0.05711*** & -0.32357*** \\
481 +\textbf{Pending home sales} & 0.12382*** & -0.82687*** & 0.02174*** & -0.03272 & 0.02609** & 0.02470 & 0.02837*** & 0.00193 & 0.06077*** & -0.09181*** & 0.01005 & -0.33617*** \\
482 +\textbf{Personal consumption} & 0.03205 & -0.20100 & 0.06603*** & -0.14244***& 0.08686*** & -0.27403*** & 0.00835 & -0.03042 & 0.12020*** & -0.17807*** & 0.06250*** & -0.19183 \\
483 +\textbf{Personal income} & 0.03464 & -0.18771 & 0.06633*** & -0.11505***& 0.09950*** & -0.30582*** & 0.01641** & -0.05830 & 0.10262*** & -0.13314*** & 0.04515*** & -0.21897* \\
484 +\textbf{Producer price index} & 0.12303*** & -0.72893*** & 0.03575*** & 0.04448 & 0.08899*** & -0.12987** & 0.01285** & -0.03913 & 0.12713*** & -0.09331** & 0.00855 & 0.02351 \\
485 +\textbf{Trade balance} & 0.10961*** & -0.46684* & 0.01732** & -0.20946***& 0.03656*** & -0.31451*** & 0.00883 & 0.00687 & 0.10595*** & -0.09640*** & 0.00668 & 0.03687 \\
486 +\textbf{Crude Oil Weekly inventory} & 0.05271* & 0.55262** & & & & & & & & & & \\
487 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.59765*** & -0.38025*** \\ \midrule
488 +
489 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
490 +\textbf{$R^2$} & \multicolumn{2}{c}{7.869} & \multicolumn{2}{c}{8.237} & \multicolumn{2}{c}{7.398} & \multicolumn{2}{c}{7.119} & \multicolumn{2}{c}{7.50} & \multicolumn{2}{c}{13.643} \\ \bottomrule
491 +\end{tabular}
492 +}
493 +\begin{tablenotes}\item[]
494 +\singlespacing
495 +\footnotesize
496 +This table presents estimates of eq. \ref{ch1:eqn:VarianceEqn} using the speculative trading intensity variable $NLS_{t,MM}$, calculated with the money manager positions. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2007-04-01 to 2024-02-11. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of speculative trading intensity.
497 +\end{tablenotes}
498 +\end{sidewaystable}
499 +
500 +
501 +
502 +
503 +
504 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
505 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
506 +
507 +
508 +
509 +\begin{sidewaystable}
510 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Swap Dealers only) on Futures Conditional Variance }
511 +\label{ch1:tab:macro_fin_sd_var_full}
512 +\centering
513 +\resizebox{\linewidth}{!}{%
514 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
515 +%\begin{tabular}{@{}lllllllllllll@{}}
516 +\toprule
517 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
518 +\textbf{Announcements} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} & \multicolumn{1}{c}{\textbf{$\Phi_m$}} & \multicolumn{1}{c}{\textbf{$\phi_m$}} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
519 +
520 +\textbf{Initial jobless claims} & 0.05822*** & 0.23343*** & 0.03845*** & 0.09404*** & 0.08327*** & 0.26382*** & 0.01604** & 0.17118*** & 0.07776*** & 0.21244*** & -0.02694** & 0.14805* \\
521 +\textbf{ADP Employment} & 0.01013 & 0.12754 & 0.03466*** & -0.00761 & 0.04490*** & -0.11581 & -0.00362 & 0.13183** & -0.02324** & -0.06171 & 0.01926 & -0.08546 \\
522 +\textbf{CB Consumer} & 0.06461*** & 0.24214*** & 0.02474*** & 0.06479** & 0.05249*** & 0.09368 & 0.06041***& 0.37330*** & 0.02534** & 0.16265*** & 0.00279 & -0.07654 \\
523 +\textbf{Advance retail sales} & 0.09553*** & 0.54512*** & 0.07545*** & -0.01412 & 0.11206*** & -0.05012 & 0.06957***& 0.41281*** & 0.05474*** & 0.28103*** & 0.04847** & -0.17485 \\
524 +\textbf{Building permit} & 0.04924*** & 0.24970*** & 0.03390*** & 0.04372 & 0.02824*** & 0.23562*** & 0.06755***& 0.32009*** & 0.05311*** & 0.13539*** & -0.01950 & 0.37466** \\
525 +\textbf{Construction spending} & 0.08554*** & 0.28910*** & 0.05709*** & 0.00763 & 0.06924*** & 0.19964** & 0.07736***& 0.49540*** & 0.05126*** & 0.03006 & 0.00924 & 0.08375 \\
526 +\textbf{Consumer credit} & 0.02409* & 0.15378* & 0.00977 & 0.01706 & 0.01612** & -0.06572 & 0.01001 & -0.03267 & 0.00416 & -0.08387 & -0.00125 & 0.00774 \\
527 +\textbf{Consumer price index} & 0.03913** & 0.23500** & 0.08146*** & 0.20522*** & 0.16575*** & 0.29275*** & 0.05399** & 0.17749** & 0.08766*** & 0.44880*** & -0.02157 & 0.35991* \\
528 +\textbf{Durable goods orders} & 0.06383*** & 0.28226*** & 0.02366*** & -0.02745 & 0.04815*** & 0.19932** & 0.03407** & 0.17902*** & 0.05428*** & 0.10969** & -0.04285** & 0.43043***\\
529 +\textbf{Existing home sales} & 0.03173** & -0.05715 & 0.03054*** & -0.00958 & 0.03451*** & 0.13272 & -0.02703 & 0.21180*** & 0.02430** & 0.16807*** & 0.03280 & 0.02110 \\
530 +\textbf{Factory orders} & 0.06008*** & 0.19180** & 0.01488** & 0.19247*** & 0.04458*** & 0.23291*** & -0.02611 & 0.21673*** & 0.02510** & 0.08760 & -0.02845 & 0.67174***\\
531 +\textbf{Gross domestic product} & 0.05612*** & 0.34455*** & 0.02232*** & 0.24770*** & 0.08973*** & 0.16653* & -0.00972 & 0.13666* & 0.06717*** & 0.24490*** & -0.05860** & 0.36187** \\
532 +\textbf{Housing starts} & 0.04932*** & 0.28860*** & 0.02698*** & 0.00304 & 0.02806*** & 0.19249** & 0.06481***& 0.31655*** & 0.04260*** & 0.09632* & -0.01498 & 0.24401 \\
533 +\textbf{Industrial production} & 0.03118** & 0.31942*** & 0.03257*** & 0.07688** & 0.01915*** & 0.02450 & -0.01125 & 0.05783 & -0.00415 & 0.13347** & -0.02739 & 0.14104 \\
534 +\textbf{New home sales} & 0.04779*** & 0.27850*** & 0.03898*** & -0.04063 & 0.06594*** & 0.18551** & -0.02060 & 0.19254*** & 0.02760*** & 0.20740*** & -0.02547 & 0.23051 \\
535 +\textbf{Non-farm employment} & 0.26824*** & 0.84150*** & 0.24170*** & -0.05819* & 0.36074*** & 0.48117*** & 0.01816 & 0.47516*** & 0.15490*** & 0.18734*** & 0.06170*** & 0.06634 \\
536 +\textbf{Pending home sales} & 0.06020*** & 0.26392*** & 0.01544** & 0.00401 & 0.03015*** & 0.03614 & 0.03653** & 0.26870*** & 0.02869*** & 0.15020*** & 0.01378 & 0.07586 \\
537 +\textbf{Personal consumption} & 0.00517 & -0.04382 & 0.03137*** & -0.01852 & 0.05089*** & 0.26220*** & 0.01457 & -0.04014 & 0.05675*** & 0.31626*** & -0.04473* & 0.91871***\\
538 +\textbf{Personal income} & 0.00784 & -0.05332 & 0.06246*** & 0.12366*** & 0.06217*** & 0.45491*** & 0.01766 & -0.02283 & 0.05419*** & 0.23258*** & -0.06483***& 0.99017***\\
539 +\textbf{Producer price index} & 0.06156*** & 0.16722* & 0.03764*** & -0.04623 & 0.07112*** & 0.05176 & -0.02949 & 0.16895** & 0.09461*** & 0.18704*** & 0.02017 & -0.10796 \\
540 +\textbf{Trade balance} & 0.07471*** & 0.15035* & 0.03543*** & 0.15511*** & 0.06670*** & 0.75961*** & 0.01410 & -0.02210 & 0.07275*** & 0.12279** & 0.01722 & -0.10359 \\
541 +\textbf{Crude Oil Weekly inventory} & 0.11644*** & 0.00487 & & & & & & & & & & \\
542 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & 0.48444*** & 1.02519***\\ \midrule
543 +
544 +\textbf{Observations} & \multicolumn{2}{c}{1,193,455} & \multicolumn{2}{c}{1,190,001} & \multicolumn{2}{c}{1,180,816} & \multicolumn{2}{c}{1,138,696} & \multicolumn{2}{c}{749,168} & \multicolumn{2}{c}{1,101,836} \\
545 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{7.01} & \multicolumn{2}{c}{7.716} & \multicolumn{2}{c}{8.023} & \multicolumn{2}{c}{7.975} & \multicolumn{2}{c}{7.732} & \multicolumn{2}{c}{13.66} \\ \bottomrule
546 +\end{tabular}
547 +}
548 +
549 +\begin{tablenotes}\item[]
550 +\singlespacing
551 +\footnotesize
552 +This table presents estimates of eq. \ref{ch1:eqn:VarianceEqn} using the speculative trading intensity variable $NLS_{t,SD}$, calculated with the swap dealer positions. The equation is $\sigma_{t}^2=\alpha_0+\alpha_1 \sigma_{t-1}^2+\alpha_2 \epsilon_t^2 +\sum_{m=1}^{22} \Phi_m D_{m,t}+\beta X_{i,t}+\sum_{k=1}^n \phi_k I_{kt} + \sum_{h=1}^{23} \rho_h D_h$ where $I_{kt}=D_{m,t} \cdot X_{i,t}$ and $D_{m,t}$ is a macro announcement dummy variable for release $m$. The period covered is from 2007-04-01 to 2024-02-11. The $\Phi_m$ coefficients capture the instantaneous change in the conditional variance when an announcement has just occurred. The $\phi_m$ coefficients capture the conditional variance when an announcement has just occurred in conjunction with the level of speculative trading intensity.
553 +\end{tablenotes}
554 +\end{sidewaystable}
555 +
556 +
557 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
558 +
559 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
560 +
561 +\begin{sidewaystable}
562 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Money Managers only) on Futures Price Bid-Ask Spreads}
563 +\label{ch1:tab:macro_fin_mm_fut_spread_full}
564 +\centering
565 +\resizebox{\linewidth}{!}{%
566 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
567 +%\begin{tabular}{@{}lllllllllllll@{}}
568 +\toprule
569 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
570 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
571 +
572 +\textbf{Initial jobless claims} & -2.46565** & 9.75050 & 0.22202 & -6.10718 & -0.41391 & -2.99267 & -1.52013** & -5.80854 & -7.73754* & 10.17312 & -2.06909** & 2.89216 \\
573 +\textbf{ADP Employment} & -1.10495 & 4.87377 & -0.03119 & 0.31326 & -0.44597 & -0.60609 & -0.46357 & 0.04112 & 11.32547 & -20.08663 & -1.56664 & 9.51315 \\
574 +\textbf{CB Consumer} & -0.88291 & -7.59047 & -1.07264** & -21.88274***& -0.60931 & -6.32195 & -2.26961* & -5.00564 & -4.89368 & 6.32172 & -6.10967***& 29.78721 \\
575 +\textbf{Advance retail sales} & -1.75295 & 6.53863 & -1.84813 & 4.41188 & -1.50715* & -4.13973 & -1.11161 & -5.34223 & -8.47285 & 16.81488 & -2.52082 & -23.72607 \\
576 +\textbf{Building permit} & 0.04804 & -5.08377 & 0.12315 & -2.62534 & -0.43797 & -3.66026 & -0.85890 & -0.39189 & -16.80677** & -9.47013 & -2.48078 & -3.85074 \\
577 +\textbf{Construction spending} & 0.71484 & -21.50516 & 0.41072 & -19.63649** & -2.03618** & -6.82194 & -0.78189 & 1.91535 & -3.65053 & 14.68763 & -6.18198***& 14.39348 \\
578 +\textbf{Consumer credit} & -0.39090 & -2.14483 & -0.23236 & 0.92328 & -0.63429 & -0.72773 & 0.72304 & 4.26292 & 2.89371 & 0.82550 & -2.08503 & -10.93259 \\
579 +\textbf{Consumer price index} & -1.79444 & 13.11087 & -0.32839 & -2.43101 & -0.59389 & -0.57386 & -1.52632 & -7.46228 & -28.00983** & 33.08946 & -2.16619 & 17.81488 \\
580 +\textbf{Durable goods orders} & -2.20101 & 5.98876 & 0.23721 & -7.01173 & -1.02867 & -5.37622 & -2.46602* & -19.42954** & -19.84697** & -52.26667***& -3.50474 & 22.77345 \\
581 +\textbf{Existing home sales} & -0.49431 & -6.37862 & 0.15230 & -2.61367 & -0.83207 & -5.40951 & -1.45784 & -2.82054 & -5.47968 & 6.15830 & -5.71583***& 23.65943 \\
582 +\textbf{Factory orders} & -2.69030 & 15.96650 & -0.51700 & -6.28637 & -1.62052* & -9.04768 & -1.96193 & -1.64059 & -0.55217 & 28.54841 & -4.09505** & 17.20082 \\
583 +\textbf{Gross domestic product} & -3.81281 & 16.72542 & -1.27400 & -5.90060 & -1.18407 & -6.39056 & -1.55842 & -17.41209* & -10.42842 & 13.46385 & -2.79300 & 6.61661 \\
584 +\textbf{Housing starts} & -0.26839 & -3.67286 & 0.02239 & -2.47882 & -0.32593 & -3.62060 & -0.12310 & -1.59499 & -14.93911* & -11.91340 & -2.48037 & -3.10561 \\
585 +\textbf{Industrial production} & 0.99908 & -10.14181 & -0.03976 & 8.87670 & -0.26830 & -0.37440 & 0.52140 & 0.47759 & 7.05215 & -13.19391 & -3.89105* & 10.56246 \\
586 +\textbf{New home sales} & 0.51742 & -13.67619 & -0.31951 & -4.68388 & -0.54238 & -2.95472 & -0.87358 & -12.35468 & -8.91154 & 6.44362 & -4.41648** & 11.79880 \\
587 +\textbf{Non-farm employment} & -1.43941 & 1.49780 & -0.83880 & -7.00960 & -3.09250***& -24.19358***& -0.26216 & -5.05771 & -19.80034** & 22.46053 & -2.06886 & 19.68747 \\
588 +\textbf{Pending home sales} & -1.22436 & 4.32860 & 0.23288 & -3.30352 & -0.65997 & -2.14049 & -2.38112* & -5.64287 & 4.60257 & -14.70125 & -5.08845***& 10.99935 \\
589 +\textbf{Personal consumption} & -1.58014 & 8.27394 & -0.38176 & 2.58232 & 0.28583 & -1.30713 & -0.71712 & -10.63446 & -14.71023* & -0.85336 & -2.60387 & 16.06990 \\
590 +\textbf{Personal income} & -1.83332 & 4.18659 & 0.03450 & -0.93297 & -0.54847 & -3.16160 & -1.78660 & -3.12819 & -18.44907** & -7.56018 & -3.68559 & 15.01741 \\
591 +\textbf{Producer price index} & 1.23242 & -25.45917 & 0.17337 & -5.42396 & -1.58493 & -8.72610 & 0.46325 & -4.84467 & -3.85292 & -6.53376 & -0.31231 & 7.49785 \\
592 +\textbf{Trade balance} & 0.77994 & -13.30115 & -0.00374 & 0.05686 & 0.13732 & -10.60780 & -0.44476 & -5.85479 & -11.44187 & 39.18528 & -0.22393 & 5.23703 \\
593 +\textbf{Crude Oil Weekly inventory} & 0.49358 & -12.74294 & & & & & & & & & & \\
594 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -7.59576***& -19.17343* \\ \midrule
595 +
596 +\textbf{Observations} & \multicolumn{2}{c}{1,041,497} & \multicolumn{2}{c}{1,022,592} & \multicolumn{2}{c}{1,023,897} & \multicolumn{2}{c}{1,022,006} & \multicolumn{2}{c}{683,875} & \multicolumn{2}{c}{995,919} \\
597 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{0.0572} & \multicolumn{2}{c}{0.115} & \multicolumn{2}{c}{0.12} & \multicolumn{2}{c}{0.118} & \multicolumn{2}{c}{0.034} & \multicolumn{2}{c}{0.0942} \\ \bottomrule
598 +\end{tabular}
599 +}
600 +\begin{tablenotes}\item[]
601 + \singlespacing
602 + \footnotesize
603 + This table presents estimates of the equation ${R_{\text{SPREAD}}}_{t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta {R_{\text{SPREAD}}}_{t-\tau}^{t}+\epsilon_{t}$, analyzing the effects of speculative trading intensity and macroeconomic announcements on the bid-ask spread using the speculative trading intensity variable $NLS_{t,MM}$, calculated with the money manager positions. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the speculative trading intensity variable $NLS_{t,MM}$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and speculative trading intensity. The $\beta$ coefficient represents the effect of the lagged bid-ask spread.
604 +\end{tablenotes}
605 +
606 +\end{sidewaystable}
607 +
608 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
609 +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
610 +
611 +
612 +\begin{sidewaystable}
613 +\caption{Effects of Macro Announcements and Speculative Trading Intensity (Swap Dealers only) on Futures Price Bid-Ask Spreads}
614 +\label{ch1:tab:macro_fin_sd_fut_spread_full}
615 +\centering
616 +\resizebox{\linewidth}{!}{%
617 +\begin{tabular}{@{}lSSSSSSSSSSSS@{}}
618 +%\begin{tabular}{@{}lllllllllllll@{}}
619 +\toprule
620 +\textbf{Commodities} & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Copper}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Palladium}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\ \midrule
621 +\textbf{Announcements} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} & \multicolumn{1}{c}{$\gamma_m$} & \multicolumn{1}{c}{$\theta_m$} \\ \midrule \multicolumn{13}{c}{\textbf{Macroeconomic News Announcements}} \\ \midrule
622 +
623 +\textbf{Initial jobless claims} & -2.76158*** & 8.99275* & -2.06041 & -5.31071 & -0.63502 & -2.29125 & 1.03040 & -14.26906 & -4.55195 & -14.41688 & -0.23562 & -20.00611 \\
624 +\textbf{ADP Employment} & -1.53692 & -6.19939 & 1.56579 & 4.65731 & -1.00409 & -6.69557 & -0.02006 & -3.16419 & 4.42780 & 8.98661 & 1.48355 & -32.33990 \\
625 +\textbf{CB Consumer} & -2.10022 & -2.32545 & -8.62678** & -22.90121 & -1.03518 & -2.63864 & 1.08755 & -19.13223 & -2.84047 & -14.63955 & 1.02725 & 76.98272*** \\
626 +\textbf{Advance retail sales} & -0.39770 & 4.05260 & -4.03721 & -14.26039 & -2.88363** & 16.59467* & -1.58737 & 4.11864 & -3.70598 & -15.43327 & -3.91203 & 24.52836 \\
627 +\textbf{Building permit} & -0.47588 & 0.18138 & -1.72855 & -6.97984 & -0.98877 & -5.01949 & -2.65095 & 12.39262 & -20.04496***& 17.71476 & -1.26962 & -10.40198 \\
628 +\textbf{Construction spending} & -2.17096 & -3.44095 & -3.30488 & 5.82834 & -3.51547** & -15.86227 & 1.24637 & -12.79038 & 1.03029 & -16.52376 & -2.31365 & -39.77206 \\
629 +\textbf{Consumer credit} & -1.20889 & -3.80213 & -4.20991 & -19.23425 & -1.53730 & -10.76448 & -2.64266 & 18.44068 & 3.18871 & 5.41810 & -0.78792 & -12.38910 \\
630 +\textbf{Consumer price index} & -0.73150 & -2.25239 & -3.39799 & -10.56428 & -1.30151 & -8.59060 & 0.41782 & -9.96604 & -16.99259** & -50.19107 & 1.32430 & -41.28821 \\
631 +\textbf{Durable goods orders} & -1.77394 & -1.82834 & -2.58254 & -8.90295 & -1.86417 & -8.85290 & -1.15772 & -7.40161 & -3.72137 & 56.42319* & -1.13899 & -26.70623 \\
632 +\textbf{Existing home sales} & -3.86759** & 17.19577* & -1.30608 & -5.41876 & -1.44836 & -5.31788 & 1.86846 & -18.96790 & -3.19545 & -16.48154 & -0.81606 & -51.32055* \\
633 +\textbf{Factory orders} & -1.79369 & -5.86609 & -7.11295* & -20.33911 & -3.07893** & -13.68582 & 6.84761** & 52.38087*** & 9.27291 & -49.32580 & 1.51334 & 62.45159** \\
634 +\textbf{Gross domestic product} & -3.16022* & -8.26499 & -2.81458 & -1.40512 & -2.08926 & -8.87196 & -1.76528 & -1.47472 & -5.93161 & -23.15106 & -0.13820 & -30.61457 \\
635 +\textbf{Housing starts} & -0.58545 & 0.77527 & -2.08499 & -8.11537 & -0.88969 & -4.66326 & -3.39511 & 20.13366 & -18.99183***& 23.19105 & -0.84224 & -15.89329 \\
636 +\textbf{Industrial production} & -0.29436 & -1.05059 & 5.50709 & 13.45477 & -0.04290 & 1.71610 & -2.71137 & 19.60256 & 2.81663 & 10.15067 & -0.86143 & -33.18089 \\
637 +\textbf{New home sales} & -2.88999 & -11.62913 & -0.67553 & 2.29246 & -1.33152 & -7.42828 & 0.00223 & -5.27181 & -6.63642 & -19.19409 & 1.59693 & 66.71395** \\
638 +\textbf{Non-farm employment} & -2.46100 & -7.53641 & -3.23439 & -3.80813 & -5.00635*** & 12.32649 & 0.73351 & -7.12626 & -13.54711* & -21.50040 & 2.51278 & 55.39346* \\
639 +\textbf{Pending home sales} & -2.61983 & -11.86787 & 1.00528 & 1.47548 & -1.26851 & -4.63955 & 0.42822 & -15.44296 & -0.19188 & 5.13867 & -2.25446 & -28.30922 \\
640 +\textbf{Personal consumption} & -1.67718 & -6.90320 & -9.79488** & 50.08330** & 0.15770 & 0.20581 & 0.92260 & -10.78664 & -14.99575* & 0.79995 & -0.64875 & -19.87802 \\
641 +\textbf{Personal income} & -4.57034** & -20.24571 & -4.80584 & -25.18481 & -1.16029 & -5.97089 & 1.28405 & -17.01679 & -21.24693***& 17.20807 & -0.65659 & -31.76262 \\
642 +\textbf{Producer price index} & -1.37042 & 1.60079 & -4.05858 & -13.59722 & -3.14320** & -14.84227 & -2.95390 & 17.63070 & -5.66177 & 0.27909 & 0.31519 & -8.87114 \\
643 +\textbf{Trade balance} & -1.05000 & -2.68201 & 0.70212 & 3.23724 & -0.23813 & 6.58851 & 2.77586 & -18.99333 & -0.38237 & -34.84505 & -0.46430 & 2.48295 \\
644 +\textbf{Crude Oil Weekly inventory} & 0.05943 & 7.79488 & & & & & & & & & & \\
645 +\textbf{Natural Gas Weekly inventory} & & & & & & & & & & & -1.31214 & 67.43257*** \\ \midrule
646 +
647 +\textbf{Observations} & \multicolumn{2}{c}{1,041,497} & \multicolumn{2}{c}{1,022,592} & \multicolumn{2}{c}{1,023,897} & \multicolumn{2}{c}{1,022,006} & \multicolumn{2}{c}{683,875} & \multicolumn{2}{c}{995,919} \\
648 +\textbf{$R^2 (\%)$} & \multicolumn{2}{c}{0.071} & \multicolumn{2}{c}{0.063} & \multicolumn{2}{c}{0.20} & \multicolumn{2}{c}{0.29} & \multicolumn{2}{c}{0.031} & \multicolumn{2}{c}{0.48} \\ \bottomrule
649 +\end{tabular}
650 +}
651 +\begin{tablenotes}\item[]
652 + \singlespacing
653 + \footnotesize
654 + This table presents estimates of the equation $R_{\text{SPREAD}, t}^{t+\tau}=\alpha+\sum_{m=1}^{22} \gamma_m D_{m,t}+ \delta X_{t,i} + \sum_{m=1}^{22} \theta_m (D_{m,t} \cdot X_t)+\beta R_{\text{SPREAD}, t-\tau}^{t}+\epsilon_{t}$, analyzing the effects of financialization and macroeconomic announcements on the bid-ask spread using the speculative trading intensity variable $NLS_{t,SD}$, calculated with the swap dealer positions. The period covered is from 2007-04-01 to 2024-02-11. The $\gamma_m$ coefficients capture the instantaneous change in the bid-ask spread when a macroeconomic announcement occurs. The $\delta$ coefficient represents the effect of the speculative trading intensity variable $NLS_{t,SD}$. The $\theta_m$ coefficients capture the interaction effect between macroeconomic announcements and speculative trading intensity. The $\beta$ coefficient represents the effect of the lagged bid-ask spread.
655 +\end{tablenotes}
656 +\end{sidewaystable}
added these-ulaval/chapitre2/01_introduction.tex +22 −0
@@ -0,0 +1,22 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Introduction}
4 +
5 +The exchange-traded fund (ETF) market has grown rapidly over the past two decades, with global assets under management (AUM) exceeding \$23 trillion as of May 2026. The growth in the ETF market has also altered the structure of underlying asset markets \citep{petajisto2017inefficiencies}. ETF ownership has been found to increase non-fundamental volatility in underlying equities \citep{ben2018etfs}, weaken the link between prices and fundamentals \citep{israeli2017etf}, and strengthen correlations during periods of market stress \citep{da2018exchange}. The number and economic size of commodity ETFs, such as the SPDR Gold Trust (GLD), has rapidly grown in recent years. These products offer exposure to commodity markets without the need for direct futures trading or physical storage \citep{gorton2006facts}, thus increasing the participation of financial investors, creating new economic links between commodity and equity prices, and establishing the ETF as an alternative venue for price discovery \citep{basak2016model, buyuksahin2014speculation}.
6 +
7 +Commodities offer a particularly interesting setting to study ETF pricing and volatility transmission. Exchange-traded funds are usually studied as near-transparent wrappers around their holdings. For example, VOO and IVV replicate the U.S. S\&P~500 market index and hold the stocks that comprise this index according to their weights. For a typical equity ETF this framing is close to exact. The underlying securities trade on the same exchanges during the same hours, and creation-redemption arbitrage pins the fund's price to its net asset value quickly and in essentially one direction. Thus, for an equity ETF, asking whether the ETF leads the basket would be similar to asking whether a shadow leads the object that casts it. In contrast, commodity ETFs break this identity in a way that is not feasible in other major ETF classes. This is because their underlying may not be the physical good, but could be instead a derivative claim. For instance, if it is a futures position in crude oil or natural gas, there are economic implications for cost-of-carry, roll, and the shape of the term structure. In the case of a stored physical claim in gold and silver there is lease-and-storage economics to consider.
8 +
9 +Thus, for commodity ETFs, the arbitrage relationship is not as simple as it is for equities or bonds, and volatility transmission becomes a genuine economic question rather than a purely mechanical one. To the point of this paper, it is important to investigate the fund's indicative net asset value (iNAV), which is the real-time fair value of that basket. Commodity ETFs also play a role in the financialization debate \citep{basak2016model}, as they are popular retail-accessible instruments. It is important for investors, traders, hedgers and policymakers to understand how volatility travels between these vehicles and their underlyings. The direction of this volatility flow is not clear a priori. Indeed, commodity ETFs differ along another significant dimension, namely that unlike equities, in commodities the ETF may well be more liquid and more continuously accessible than its underlying. This feature provides a clear reason why the ETF might contribute to price discovery rather than simply inherit it. Whether it does, however, remains an empirical question.
10 +
11 +
12 +The theoretical foundation of ETF pricing rests on arbitrage. Authorized participants (APs) maintain price alignment through creation and redemption, buying undervalued ETFs while selling their underlying constituents, or vice versa \citep{ackert2000arbitrage}. Classic arbitrage theory predicts that volatility transmission would be unidirectional, flowing from the underlying constituents (NAV) to the ETF: when underlying volatility rises, arbitrageurs trade more actively to maintain alignment, which transmits volatility to the ETF. Recent evidence challenges this prediction, documenting bidirectional transmission in which ETF trading influences underlying asset volatility \citep{ben2018etfs, da2018exchange}. This outcome arises because ETFs often trade more frequently and with smaller spreads than their constituents, making the ETF a primary venue for price formation \citep{glosten2021etf, pan2016etf}. The relative magnitude and direction of transmission therefore reveal which market dominates information discovery and whether arbitrage functions efficiently. These questions have direct implications for hedging (directional dependencies) and for regulation, e.g., as feedback effects may amplify volatility during crises \citep{madhavan2012exchange, petajisto2017inefficiencies}.
13 +
14 +This paper studies how volatility is transmitted between commodity ETFs and their underlying assets using high-frequency realized variance over 2010--2023. We focus on four major single-commodity ETFs spanning two market structures: physically-backed precious metals (SPDR Gold Trust, GLD; iShares Silver Trust, SLV) and futures-based energy funds (United States Oil Fund, USO; United States Natural Gas Fund, UNG). Our central question is whether transmission is unidirectional or bidirectional, and how its direction, strength, and time horizon vary with commodity type, sampling frequency, and the continuous versus jump nature of volatility. We study volatility transmission rather than price discovery because it reveals how risk, not just information, propagates across linked markets and whether ETF arbitrage stabilizes or amplifies it---a question central to risk management and systemic-risk regulation.
15 +
16 +The empirical literature on ETF volatility transmission is incomplete in three respects. First, it focuses on equity ETFs using daily data \citep{ben2018etfs, israeli2017etf}, leaving the intraday dynamics that govern arbitrage largely unexplored. Indeed, as arbitrage operates continuously through the day, daily aggregation may obscure rapid transmission. Second, the literature has not examined how spillovers differ across commodity types, even though precious metals trade in liquid global markets with physical arbitrage while energy commodities rely on futures with rollover costs and storage constraints---features that should generate systematically different transmission. Third, it has not distinguished the roles of underlying assets versus ETFs in driving volatility, leaving open the questionwhether ETFs passively follow their constituents or actively feed back into them.
17 +
18 +We address these gaps by constructing minute-by-minute indicative Net Asset Value (iNAV) series for the four ETFs over thirteen years spanning the European sovereign debt crisis, the 2014--2016 commodity collapse, the COVID-19 pandemic, and the subsequent inflation surge. We combine Heterogeneous Autoregressive (HAR) models that capture the long memory of realized volatility across daily, weekly, and monthly horizons \citep{corsi2009simple} with Bayesian Vector Autoregression (BVAR) models to accommodate time-varying dependence while avoiding overfitting \citep{koop2011forecasting}. We measure realized variance from high-frequency returns \citep{andersen2001distribution} and decompose it into continuous and jump components \citep{barndorff2004power}. We separate jumps because commodity prices respond to discrete shocks---geopolitical events, supply disruptions---that may transmit through different channels than smooth price movements, with distinct consequences for tail-risk hedging. On this basis, we test four hypotheses, stated formally in Section~\ref{ch2:sec:methodolog}: (i) whether transmission is uni- or bidirectional and varies by commodity type; (ii) whether high-frequency sampling reveals dynamics hidden in daily data; (iii) whether jumps or the continuous component dominate transmission; and (iv) whether transmission is stable over time.
19 +
20 +Our results reveal substantial heterogeneity across commodity types and sampling frequencies. For precious metals (GLD, SLV), transmission is strongly unidirectional from iNAV to ETF, with spillover coefficients ranging from 0.42 (silver) to 0.63 (gold) at one-minute frequency and negligible reverse effects. Energy ETFs (USO, UNG) show bidirectional transmission, with the iNAV-to-ETF direction dominant for crude oil (about four to one) and the two directions comparable for natural gas. Sampling frequency matters: the daily iNAV-to-ETF spillover is up to roughly twice as large in one-minute as in thirty-minute data. Jump components dominate continuous transmission, especially for precious metals. The BVAR analysis confirms these patterns through impulse responses and variance decompositions in which underlying volatility rivals or exceeds ETF self-persistence in explaining ETF volatility.
21 +
22 +The paper makes two contributions. Methodologically, we construct the first comprehensive high-frequency iNAV series for commodity ETFs over an extended period, enabling precise measurement of intraday arbitrage relationships that are unobservable with daily NAV data. Empirically, we document systematic differences in transmission across commodity categories and sampling frequencies, showing that market structure shapes information transmission and that empirical inference is affected by temporal aggregation. Thus, we extend the literature on ETF volatility transmission from equity to commodity markets \citep{ben2018etfs, israeli2017etf}. We further connect it to the microstructure literature \citep{hasbrouck2003intraday, richie2008examination}, and we contribute to research on commodity-ETF financialization \citep{todorov2021etf, buyuksahin2014speculation} by showing that the effects vary in systematic ways between physically-backed and futures-based ETFs. The remainder of the paper is as follows. Section~\ref{ch2:sec:data} describes the data and the construction of high-frequency iNAV and realized-volatility series. Section~\ref{ch2:sec:methodolog} presents the econometric framework and states the hypotheses. Section~\ref{ch2:sec:results} reports the empirical results. Section~\ref{ch2:sec:conclusion} concludes.
added these-ulaval/chapitre2/03_data.tex +76 −0
@@ -0,0 +1,76 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Data and Sample Construction} \label{ch2:sec:data}
4 +
5 +Our data processing approach improves on what has been previously used in the ETF volatility literature, which typically relies on daily data and may not detect high-frequency transmission channels \citep{ben2018etfs, israeli2017etf}. By constructing minute-by-minute price and indicative NAV series for four commodity ETFs over thirteen years, we capture volatility dynamics at horizons not previously examined and compare volatility transmission across heterogeneous commodity market structures.
6 +
7 +\subsection{Sample Selection and Data Sources}
8 +
9 +We study four large single-commodity ETFs, each among the top funds in its category ranked by assets under management:
10 +
11 +\begin{itemize}
12 +\item \textbf{SPDR Gold Trust (GLD)} and \textbf{iShares Silver Trust (SLV)} are physically-backed precious-metals ETFs holding bullion in trust (over \$50 billion and \$10 billion in assets, respectively). Physical backing minimizes tracking error but creates arbitrage frictions tied to delivery, storage, and insurance.
13 +\item \textbf{United States Oil Fund (USO)} and \textbf{United States Natural Gas Fund (UNG)} are futures-based energy ETFs tracking WTI crude oil and Henry Hub natural gas through NYMEX futures. Continuous rolling of expiring contracts generates tracking errors related to contango and backwardation \citep{todorov2021etf}, and natural gas adds extreme seasonality and storage constraints.
14 +\end{itemize}
15 +
16 +We select these ETFs for three reasons: high liquidity (top-five funds by assets in each category), continuous tradability over the full sample (January 2010 to January 2023) and-, most importantly, contrasting tracking mechanisms. Physically-backed ETFs (GLD, SLV) settle through physical delivery against London Bullion Market Association (LBMA) spot prices, with no rollover costs but slower arbitrage. Futures-based ETFs (USO, UNG) are arbitraged electronically against NYMEX futures, with faster settlement but involving roll costs, contango/backwardation patterns, and basis risk. Since these market structures create different frictions, transaction costs and lags, we expect them to generate different patterns of volatility transmission. We use single-commodity rather than index ETFs to avoid cross-commodity correlation effects.
17 +
18 +Our primary data source is the Bloomberg terminal, which provides tick-by-tick trade prices, bid--ask quotes, and volume with millisecond timestamps for the ETFs and their underlying assets. We supplement this with futures contract price data from the Chicago Mercantile Exchange (CME) and spot price data from the LBMA. We begin the sample in 2010 for three reasons: By this date, the ETFs (launched 2004--2007) had plausibly reached the trading volume and market-making infrastructure needed for reliable high-frequency data and meaningful arbitrage. Moreover, starting in 2010 avoids the atypical microstructure effects during certain periods of the 2008--2009 crisis. Lastly, consistent tick-level data became available across all instruments in our sample. The resulting thirteen-year period spans the 2010--2012 European sovereign debt crisis, the 2014--2016 commodity collapse, the 2020 COVID-19 pandemic, and the 2021--2022 inflation surge, covering diverse volatility regimes.
19 +
20 +\subsection{High-Frequency Data Construction and Cleaning}
21 +
22 +High-frequency data require careful filtering to remove microstructure noise. Starting from raw tick data over regular U.S. market hours (9:30 AM--4:00 PM EST), we apply standard procedures adapted to the ETF market structure \citep{barndorff2009realized}. First, we remove outliers using the \citet{brownlees2014practical} method, deleting observations more than 10 standard deviations from a rolling 20-minute median (about 0.03\% of observations). Second, we apply the duration filter suggested by \citet{hansen2005realized}, excluding trades separated by more than 30 minutes, which typically indicates closures or technical failures (less than 0.1\% of observations). We then build synchronized price series at 1-, 5-, and 30-minute frequencies using previous-tick interpolation \citep{andersen2001distribution}. The final cleaned dataset contains roughly 45 million price observations. Data loss is minimal and concentrated in thinly traded periods.
23 +
24 +\subsection{Indicative Net Asset Value (iNAV) Construction}
25 +
26 +Official NAV is published only at the end of each trading day \citep{petajisto2017inefficiencies}. This variable cannot capture the intraday arbitrage activities that actually drive ETF pricing. We therefore construct an indicative NAV (iNAV) at a high frequency, which provides a real-time estimate of fundamental value from current underlying prices. This approach lets us measure arbitrage and transmission at the frequency where authorized participants make creation/redemption decisions, and let us separate fundamental ETF volatility (changes in underlying value) from non-fundamental volatility (liquidity shocks, inventory effects, or temporary arbitrage breakdowns).
27 +
28 +For physically-backed ETFs (GLD, SLV), iNAV follows a composition-based approach:
29 +
30 +\begin{equation}
31 +\text{iNAV}_{t} = \frac{1}{N_t} \left[ \text{Cash}_t + \sum_{i} \left( P_{it} \cdot f_{it} \cdot q_{it} \cdot c_{it} \right) \right]
32 +\end{equation}
33 +
34 +where $N_t$ is the number of outstanding ETF shares, $\text{Cash}_t$ is the fund's cash holdings, $P_{it}$ is the price of underlying asset $i$ in local currency, $f_{it}$ the currency conversion factor, $q_{it}$ the quantity held, and $c_{it}$ an adjustment for accrued interest, dividends, or other cash flows. We use LBMA gold and silver prices converted to U.S. dollars at real-time exchange rates. Physical holdings are updated daily with creation/redemption activity, while intraday changes reflect only price movements.
35 +
36 +For futures-based ETFs (USO, UNG), iNAV follows a futures-position model:
37 +
38 +\begin{equation}
39 +\text{iNAV}_{t} = \frac{1}{N_t} \left[ \text{Cash}_t + \sum_{j} \left( F_{jt} \cdot cc_{jt} \cdot q_{jt} \cdot m_{jt} \right) \right] \cdot FX_t
40 +\end{equation}
41 +
42 +where $F_{jt}$ is the price of futures contract $j$, $cc_{jt}$ the contract conversion factor, $q_{jt}$ the number of contracts held, $m_{jt}$ the contract multiplier, and $FX_t$ any currency conversion. We use real-time NYMEX WTI crude oil and Henry Hub natural gas futures, accounting for the funds' actual contract positions and monthly roll schedules.
43 +
44 +A technical consideration is the mismatch between ETF trading hours and underlying market hours: gold and silver trade nearly around the clock in London and Asia, while energy futures have defined sessions. To handle this issue, we weight each market session by its share of price discovery and carry forward the most recent adjusted prices when an underlying market is closed. We validate the construction against published end-of-day NAV. We find that our iNAV has a correlation greater than 0.999 with the official NAV, and mean absolute deviations of less than 5 basis points.
45 +
46 +\subsection{Realized Variance Construction and Jump Detection}
47 +
48 +We measure volatility using realized variance, i.e., the sum of squared intraday returns \citep{andersen2001distribution}. For asset $i$ on day $t$:
49 +
50 +\begin{equation}
51 +RV_{i,t} = \sum_{j=1}^{M} r_{i,t,j}^2
52 +\end{equation}
53 +
54 +where $r_{i,t,j} = \log(P_{i,t,j}) - \log(P_{i,t,j-1})$ and $M$ is the number of intraday returns, giving $M=390$ (1-minute), $M=78$ (5-minute), and $M=13$ (30-minute). Under standard conditions, realized variance converges to integrated variance as sampling intensifies \citep{barndorff2002econometric}. The choice of frequency trades statistical efficiency against microstructure bias from bid--ask bounce \citep{hansen2005realized, liu2015does}, so we use three frequencies to assess robustness.
55 +
56 +To separate continuous movements from discrete jumps, we use the bipower variation \citep{barndorff2004power}:
57 +
58 +\begin{equation}
59 +BV_{i,t} = \mu_1^{-2} \sum_{j=2}^{M} |r_{i,t,j}| \cdot |r_{i,t,j-1}|
60 +\end{equation}
61 +
62 +where $\mu_1 = \sqrt{2/\pi} \approx 0.798$. Bipower variation consistently estimates integrated variance even in the presence of jumps. The jump component is
63 +
64 +\begin{equation}
65 +J_{i,t} = \max(RV_{i,t} - BV_{i,t}, 0)
66 +\end{equation}
67 +
68 +and the continuous component is $C_{i,t} = RV_{i,t} - J_{i,t}$ \citep{huang2005using}. This continuous component is reported as quadratic power variation (QPV) in the estimation tables. Separating the two matters because jumps---from supply disruptions or geopolitical events---may transmit across markets differently than smooth volatility.
69 +
70 +\subsection{Descriptive Statistics and Stylized Facts}
71 +
72 +Tables \ref{ch2:tab:desc_stats_5min}, \ref{ch2:tab:desc_stats_1min}, and \ref{ch2:tab:desc_stats_30min} report descriptive statistics for realized variance, bipower variation, and jumps across the three frequencies. Several stylized facts emerge. Realized variance is heterogeneous across commodities and between ETFs and their iNAVs. Natural gas and crude oil have the highest average volatility: the 5-minute mean realized variance is 0.104\%, and 0.081\% for the iNAV. For crude oil, the iNAV (0.081\%) is more volatile than the ETF value (0.066\%), as expected from the damping effect of arbitrage. The heaviest distributional tails occur in crude oil and gold: For crude oil, the iNAV realized variance reaches a maximum above 40\%. For gold, the ETF series shows occasional extreme outliers (maximum near 79\%), reflecting occasional disruptions in physical gold arbitrage adjustments. Precious metals are otherwise less volatile on average than energy, though for gold the ETF mean volatility (0.034\%) exceeds its iNAV mean (0.012\%). Jump activity is most pronounced for crude oil, consistent with this commodity's sensitivity to geopolitical and supply shocks, and it is weakest for precious metals, for which prices move more smoothly.
73 +
74 +Figures \ref{ch2:fig:rv_uso}, \ref{ch2:fig:rv_gld}, \ref{ch2:fig:rv_slv}, and \ref{ch2:fig:rv_ung} show the evolution of realized variance over time. All series show strong volatility clustering, which is most pronounced during the 2014--2016 oil collapse, the 2016 Brexit referendum (precious metals), and the 2020 pandemic (for all commodities). ETF--iNAV synchronization is high for precious metals but more variable for energy commodities, where periods of close co-movement alternate with divergence. As these relationships are time-varying, a simple correlation analysis would fail to detect economically significant dynamics, which motivates the modeling framework in Section~\ref{ch2:sec:methodolog}.
75 +
76 +Finally, all realized-variance series are stationary in both levels and logs (augmented Dickey--Fuller tests), and we validate the GLD iNAV against the NYSE indicative optimized portfolio value (IOPV), obtaining correlations above 0.995. Tests for structural breaks and the full set of robustness checks are reported with the econometric framework in Section~\ref{ch2:sec:methodolog}.
added these-ulaval/chapitre2/04_methods.tex +103 −0
@@ -0,0 +1,103 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Econometric Methodology} \label{ch2:sec:methodolog}
4 +
5 +We analyze volatility transmission between commodity ETFs and their iNAVs with two complementary frameworks. Heterogeneous Autoregressive (HAR) models \citep{corsi2009simple} capture the long memory of realized volatility across daily, weekly, and monthly horizons and, in cross-market form, yield interpretable spillover coefficients. Bayesian Vector Autoregression (BVAR) models \citep{koop2011forecasting, carriero2015forecasting} treat ETF and iNAV volatility as jointly endogenous, using Minnesota-prior shrinkage to control parameter proliferation and providing impulse responses and variance decompositions with proper uncertainty bands. We favor realized-variance models over GARCH because realized variance is a nearly model-free volatility estimate that forecasts better \citep{andersen2001distribution}, and HAR over fractionally integrated (ARFIMA) specifications because it captures long memory parsimoniously through a horizon cascade \citep{corsi2009simple}.
6 +
7 +\subsection{Theoretical Framework for Volatility Transmission}
8 +
9 +Under frictionless arbitrage, ETF prices equal their NAV and volatility transmission is instantaneous and bidirectional \citep{ackert2000arbitrage, petajisto2017inefficiencies}. Real-world frictions---transaction costs, inventory and funding constraints---can instead create asymmetric transmission that varies across horizons and market conditions. We study volatility transmission rather than price discovery because it reveals how risk propagates and whether arbitrage stabilizes or amplifies fluctuations, with direct implications for hedging effectiveness and systemic risk. Examining multiple horizons follows the heterogeneous market hypothesis \citep{muller1997volatilities}, under which day traders, institutions, and longer-horizon participants generate volatility that persists over their characteristic time scales. Finally, the contrast between physically-backed ETFs (GLD, SLV), whose arbitrage requires physical delivery, and futures-based ETFs (USO, UNG), which face roll costs and basis risk \citep{todorov2021etf}, motivates our cross-sectional comparison.
10 +
11 +\subsection{Heterogeneous Autoregressive (HAR) Models}
12 +
13 +The HAR model \citep{corsi2009simple} is the dominant framework for realized-volatility dynamics \citep{andersen2007roughing}. The baseline specification is
14 +
15 +\begin{equation} \label{ch2:eq:har_baseline}
16 +\log(RV_{i,t}) = \beta_0 + \beta_1 \log(RV_{i,t-1}) + \beta_2 \log(\overline{RV}_{i,t-5:t-1}) + \beta_3 \log(\overline{RV}_{i,t-22:t-1}) + \varepsilon_{i,t}
17 +\end{equation}
18 +
19 +where $\overline{RV}_{i,t-h:t-1} = \frac{1}{h}\sum_{j=1}^{h} RV_{i,t-j}$ averages realized variance over the previous $h$ days, capturing daily, weekly ($h=5$), and monthly ($h=22$) persistence. The log transform keeps fitted volatility positive and stabilizes the residual variance.
20 +
21 +To measure transmission, we add cross-market terms. For iNAV volatility,
22 +
23 +\begin{align} \label{ch2:eq:har_x_nav}
24 +\log(RV_{t,\text{NAV}}) &= \beta_0 + \beta_1 \log(RV_{t-1,\text{NAV}}) + \beta_2 \log(\overline{RV}_{t-5:t-1,\text{NAV}}) + \beta_3 \log(\overline{RV}_{t-22:t-1,\text{NAV}}) \\
25 +&\quad + \alpha_1 \log(RV_{t-1,\text{ETF}}) + \varepsilon_{t,\text{NAV}} \nonumber
26 +\end{align}
27 +
28 +and symmetrically for ETF volatility,
29 +
30 +\begin{align} \label{ch2:eq:har_x_etf}
31 +\log(RV_{t,\text{ETF}}) &= \gamma_0 + \gamma_1 \log(RV_{t-1,\text{ETF}}) + \gamma_2 \log(\overline{RV}_{t-5:t-1,\text{ETF}}) + \gamma_3 \log(\overline{RV}_{t-22:t-1,\text{ETF}}) \\
32 +&\quad + \delta_1 \log(RV_{t-1,\text{NAV}}) + \varepsilon_{t,\text{ETF}} \nonumber
33 +\end{align}
34 +
35 +Each equation includes the own daily, weekly, and monthly terms and a single cross-market term at the daily lag, following the parsimony of the HAR cascade. The coefficient $\alpha_1$ measures the daily ETF-to-iNAV spillover and $\delta_1$ the daily iNAV-to-ETF spillover; the own weekly ($\beta_2,\gamma_2$) and monthly ($\beta_3,\gamma_3$) terms capture longer-horizon volatility persistence.
36 +
37 +To separate the transmission of continuous and jump volatility, we estimate a HAR-CJ-X specification using the components $C$ and $J$ defined in Section~\ref{ch2:sec:data}:
38 +
39 +\begin{align} \label{ch2:eq:har_cj_x_nav}
40 +\log(RV_{t,\text{NAV}}) &= \beta_0 + \beta_1 \log(C_{t-1,\text{NAV}}) + \beta_2 \log(\overline{C}_{t-5:t-1,\text{NAV}}) + \beta_3 \log(\overline{C}_{t-22:t-1,\text{NAV}}) \\
41 +&\quad + \beta_4 \log(1+J_{t-1,\text{NAV}}) + \beta_5 \log(1+\overline{J}_{t-5:t-1,\text{NAV}}) + \beta_6 \log(1+\overline{J}_{t-22:t-1,\text{NAV}}) \nonumber \\
42 +&\quad + \alpha_1 \log(C_{t-1,\text{ETF}}) + \alpha_2 \log(1+J_{t-1,\text{ETF}}) + \varepsilon_{t,\text{NAV}} \nonumber
43 +\end{align}
44 +
45 +with an analogous ETF equation. The own continuous and jump components enter at all three horizons through $\beta_1$--$\beta_3$ and $\beta_4$--$\beta_6$; the daily cross-market continuous and jump spillovers are $\alpha_1$ and $\alpha_2$. The transform $\log(1+J_{t})$ keeps jump terms defined when $J_{t}=0$.
46 +
47 +\subsection{Bayesian Vector Autoregression (BVAR)}
48 +
49 +The HAR cascade imposes a fixed horizon structure. As a flexible complement, we estimate a VAR treating ETF and iNAV volatility as jointly endogenous:
50 +
51 +\begin{equation} \label{ch2:eq:var_baseline}
52 +\mathbf{y}_t = \mathbf{c} + \sum_{k=1}^{p} \mathbf{A}_k \mathbf{y}_{t-k} + \mathbf{u}_t
53 +\end{equation}
54 +
55 +where $\mathbf{y}_t = [\log(RV_{t,\text{ETF}}), \log(RV_{t,\text{NAV}})]'$ and $\mathbf{u}_t \sim \mathcal{N}(\mathbf{0}, \mathbf{\Sigma})$. Information criteria (BIC) select $p=2$. Because unrestricted VARs over-parameterize, we apply the Minnesota prior \citep{litterman1986forecasting}:
56 +
57 +\begin{align} \label{ch2:eq:minnesota_prior}
58 +\beta_{ij}^{(k)} &\sim \mathcal{N}(0, \lambda_1^2 \cdot k^{-\lambda_3}) \quad \text{for } i \neq j \\
59 +\beta_{ii}^{(1)} &\sim \mathcal{N}(1, \lambda_1^2) \nonumber \\
60 +\beta_{ii}^{(k)} &\sim \mathcal{N}(0, \lambda_1^2 \cdot k^{-\lambda_3}) \quad \text{for } k > 1 \nonumber
61 +\end{align}
62 +
63 +where $\beta_{ij}^{(k)}$ is the coefficient on variable $j$ at lag $k$ in equation $i$; $\lambda_1$ controls overall tightness, $\lambda_2$ cross-variable shrinkage, and $\lambda_3$ lag decay. We set $\lambda_1=0.2$, $\lambda_2=0.5$, $\lambda_3=2$ and vary them in sensitivity analysis. The prior encodes that own lags matter more than others, recent lags more than distant ones, and coefficients are not extreme---assumptions well suited to persistent, mean-reverting volatility.
64 +
65 +We estimate by Gibbs sampling, alternating between the coefficients (multivariate normal given $\mathbf{\Sigma}$) and the covariance (inverse-Wishart given the coefficients):
66 +
67 +\begin{equation} \label{ch2:eq:coefficient_posterior}
68 +\boldsymbol{\beta} | \mathbf{\Sigma}, \mathbf{Y} \sim \mathcal{N}(\hat{\boldsymbol{\beta}}, \mathbf{\Sigma} \otimes (\mathbf{X}'\mathbf{X} + \mathbf{V}_0^{-1})^{-1}), \qquad
69 +\hat{\boldsymbol{\beta}} = (\mathbf{X}'\mathbf{X} + \mathbf{V}_0^{-1})^{-1}(\mathbf{X}'\text{vec}(\mathbf{Y}) + \mathbf{V}_0^{-1}\boldsymbol{\beta}_0)
70 +\end{equation}
71 +
72 +\begin{equation} \label{ch2:eq:covariance_posterior}
73 +\mathbf{\Sigma} | \boldsymbol{\beta}, \mathbf{Y} \sim \text{IW}(\mathbf{S} + \mathbf{S}_0, T + \nu_0)
74 +\end{equation}
75 +
76 +where $\mathbf{Y}$ and $\mathbf{X}$ are the stacked data, $\mathbf{V}_0$, $\mathbf{S}_0$, and $\nu_0$ are prior parameters, $\mathbf{S}$ the residual sum of squares, and $T$ the sample size. We run 50{,}000 iterations, discard 10{,}000 as burn-in, and thin every tenth draw; convergence is assessed with trace plots and the Geweke test.
77 +
78 +From the posterior we compute orthogonalized impulse responses and forecast error variance decompositions:
79 +
80 +\begin{equation} \label{ch2:eq:impulse_response}
81 +\text{IRF}(h) = \mathbf{C}_h \mathbf{P}, \qquad
82 +\text{FEVD}_{i,j}(h) = \frac{\sum_{k=0}^{h-1} [\mathbf{C}_k \mathbf{P}]_{i,j}^2}{\sum_{k=0}^{h-1} [\mathbf{C}_k \mathbf{\Sigma} \mathbf{C}_k']_{i,i}}
83 +\end{equation}
84 +
85 +where $\mathbf{C}_h$ is the $h$-step moving-average matrix and $\mathbf{P}$ the Cholesky factor of $\mathbf{\Sigma}$. We order iNAV before ETF, so iNAV innovations may affect ETF volatility contemporaneously but not the reverse, reflecting that underlying price movements lead ETF adjustments through arbitrage. The confidence bands follow \citet{sims1999error}. A high $\text{FEVD}_{\text{ETF},\text{NAV}}(h)$ indicates that iNAV innovations explain ETF volatility, and conversely for $\text{FEVD}_{\text{NAV},\text{ETF}}(h)$.
86 +
87 +\subsection{Hypotheses and Testing}
88 +
89 +We test four hypotheses, mapping each to coefficients in the equations above.
90 +
91 +\textbf{H1 (Direction and cross-commodity heterogeneity).} Theory predicts unidirectional iNAV-to-ETF transmission, but a more liquid ETF can reverse the flow \citep{glosten2021etf}. We expect physically-backed precious metals to show unidirectional iNAV-to-ETF transmission and futures-based energy ETFs more balanced bidirectional transmission. In equations~\eqref{ch2:eq:har_x_nav}--\eqref{ch2:eq:har_x_etf}, ETF-to-iNAV transmission tests $H_0:\alpha_1=0$ against $H_a:\alpha_1\neq 0$, and iNAV-to-ETF transmission tests $H_0:\delta_1=0$ against $H_a:\delta_1\neq 0$. Transmission is unidirectional when only $\delta_1$ is significant and bidirectional when both $\alpha_1$ and $\delta_1$ are significant. Heterogeneity means these outcomes differ across commodities.
92 +
93 +\textbf{H2 (Frequency dependence).} If arbitrage occurs within minutes, daily aggregation would understate transmission. Using the daily spillover coefficients $\delta_1$ and $\alpha_1$ estimated at the 1-, 5-, and 30-minute frequencies, we test $H_0: \delta_1^{(1\text{m})}=\delta_1^{(5\text{m})}=\delta_1^{(30\text{m})}$ against $H_a:$ the coefficients differ across frequencies.
94 +
95 +\textbf{H3 (Jumps versus continuous component).} Tail risk from discrete shocks is harder to hedge than smooth volatility, which is why we decompose volatility into continuous and jump parts. In the HAR-CJ-X model~\eqref{ch2:eq:har_cj_x_nav}, we compare the daily cross-market jump and continuous spillovers, testing $H_0:\alpha_2=\alpha_1$ (jump and continuous transmission equal) against $H_a:\alpha_2>\alpha_1$ (jump transmission dominates), and analogously for the ETF equation.
96 +
97 +\textbf{H4 (Stability over time).} Over a data sample spanning several crises, patterns of transmission may evolve. We re-estimate the BVAR over sub-periods (2010--2014, 2015--2019, 2020--2023) and apply the structural-break test of \citet{bai2003computation}, testing $H_0:$ transmission coefficients are constant across regimes against $H_a:$ they change.
98 +
99 +\subsection{Estimation, Model Selection, and Robustness}
100 +
101 +We select lag lengths and compare nested models with the AIC and BIC (the DIC for Bayesian models, which accounts for shrinkage), and evaluate out-of-sample forecasts with rolling windows (a 1{,}000-day estimation window re-estimated every 250 days) at the 1-, 5-, and 22-day horizons using RMSE and MAE. Residual diagnostics include the Ljung--Box (serial correlation), Breusch--Pagan (heteroskedasticity), and Jarque--Bera (normality) tests, plus posterior predictive checks for the BVAR. We also gauge economic magnitude through the effect of one-standard-deviation shocks on forecast volatility.
102 +
103 +We assess robustness along three dimensions, consolidating the checks noted in Section~\ref{ch2:sec:data}. First, \emph{specifications}: alternative HAR lag structures, BVAR lags from one to four, level and square-root (rather than log) transforms, and time-varying-parameter versions. Second, \emph{sample stability}: the sub-period and regime estimation underlying H4, structural-break tests \citep{bai2003computation}---which flag the 2014 oil collapse, the 2016 Brexit referendum, and the 2020 pandemic---and bootstrap inference. Our transmission results hold across regimes, indicating they capture general mechanisms rather than period-specific effects. Third, \emph{data construction}: additional 15- and 60-minute frequencies, alternative estimators (truncated realized variance, realized kernels, range-based measures), and alternative iNAV constructions (currency conversion, cash treatment), including validation against the GLD IOPV. Our main findings are robust throughout.
added these-ulaval/chapitre2/05_results.tex +59 −0
@@ -0,0 +1,59 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Empirical Results} \label{ch2:sec:results}
4 +
5 +This section reports the results of our empirical analysis of volatility transmission between commodity ETFs and their underlying assets. We organize the presentation of results around the four hypotheses stated in Section~\ref{ch2:sec:methodolog}, analyzing them in order: namely, the direction of transmission and its variation across commodity types (H1), the role of sampling frequency (H2), the relative importance of jump and continuous components (H3), and the stability of transmission over time (H4).
6 +
7 +\subsection{Direction of Transmission and Commodity-Specific Asymmetries}
8 +
9 +Our first hypothesis concerns whether transmission between an ETF and its iNAV is unidirectional or bidirectional, and whether this varies across commodities (H1). As discussed in Section~\ref{ch2:sec:methodolog}, theory predicts a flow only from iNAV to ETF, but a more liquid ETF can reverse the direction of the flow. In the case of commodities, the storability and settlement mechanism is also expected to matter. We therefore expect physically-backed precious metals to show unidirectional iNAV-to-ETF transmission, and futures-based energy ETFs more balanced bidirectional transmission.
10 +
11 +The HAR-X estimates in Tables~\ref{ch2:tab:HAR_5min} through~\ref{ch2:tab:HAR_30min} reject the null of uniform bidirectional transmission across all commodities, extending the heterogeneity documented by \citet{gorton2006facts} and \citet{buyuksahin2014speculation} to the ETF setting and consistent with \citet{basak2016model} on how financialization effects vary across commodity types.
12 +
13 +The results for precious metals show a strongly unidirectional transmission from iNAV to ETF, with little evidence of reverse transmission. Consider the iNAV-to-ETF coefficient, which is the coefficient for lagged NAV volatility in the ETF equation. For gold at a 1-minute frequency, this coefficient is 0.632 (Table~\ref{ch2:tab:HAR_1min}), which is among the largest across commodities, while the ETF-to-iNAV effect (i.e., lagged ETF volatility in the NAV equation) is not statistically different from zero ($-0.001$, $p=0.96$). The results for silver show a similar but weaker pattern: the NAV-to-ETF coefficient is 0.418 while the reverse effect is 0.004 and not significant. On the other hand, our results for energy commodities show bidirectional transmission. For crude oil at a 1-minute frequency, the iNAV-to-ETF coefficient is 0.378 and the ETF-to-iNAV coefficient is 0.089. Both are significant at the 1\% level, and the coefficient for iNAV-to-ETF is roughly four times larger. The results for natural gas are the most directionally balanced: the effects in both directions are significant and of similar magnitude (iNAV-to-ETF 0.079; ETF-to-iNAV 0.127). This contrast between the strongly unidirectional results for metals and the bidirectional results for energy supports our hypothesis H1. Indeed, the evidence confirms that physical versus futures-based arbitrage, settlement mechanism, and liquidity shape the transmission of volatility between the commodity underlying and the ETF.
14 +
15 +These patterns are confirmed by the results for the formal tests described in Section~\ref{ch2:sec:methodolog}. The ETF-to-iNAV restriction $H_0:\alpha_1=0$ in equation~\eqref{ch2:eq:har_x_nav} is not rejected for gold or silver---their reverse coefficients are insignificant at every frequency---but is rejected at the 1\% level for crude oil and natural gas. The iNAV-to-ETF restriction $H_0:\delta_1=0$ in equation~\eqref{ch2:eq:har_x_etf} is rejected at the 1\% level for all four commodities. Transmission is therefore unidirectional (iNAV-to-ETF) for precious metals and bidirectional for energy. Moreover, the iNAV-to-ETF channel dominates for crude oil (with a magnitude of about four to one), while for natural gas the effects for the two directions are comparable.
16 +
17 +The economic interpretation is consistent across models. In precious metals, arbitrage requires physical delivery against LBMA bullion, which is costly and slow. Authorized participants readily create or redeem ETF shares in response to underlying price moves, but cannot easily push ETF-specific shocks back into the tightly arbitraged spot market, so volatility flows essentially one way. The reverse coefficients for both metals hover near zero at all frequencies, confirming that ETF activity does not transmit volatility back to the spot market. In energy markets, by contrast, both the ETF and the underlying futures settle electronically and trade with comparable liquidity, so shocks propagate in both directions, even though fundamental supply-and-demand information still enters first through the futures-based iNAV.
18 +
19 +Within precious metals, gold shows stronger unidirectional effects than silver at every frequency. The iNAV-to-ETF coefficients are 0.632 at a 1-minute and 0.387 at a 30-minute frequency for gold, versus 0.418 to 0.277 for silver. These results are consistent with the hypothesis that gold is a financial store of value, while silver carries additional industrial demand-related volatility. The reverse (ETF-to-iNAV) coefficients for both metals stay close to zero across frequencies: gold between $-0.015$ and 0.021, silver between $-0.034$ and 0.004, and neither is economically meaningful. In the category of energy commodities, crude oil shows a clear iNAV-to-ETF dominance across frequencies. However, the effects are more balanced for natural gas. For iNAV-to-ETF and ETF-to-iNAV, respectively, the coefficients are 0.105 and 0.126 at the 5-minute, and 0.081 and 0.074 at the 30-minute frequency. These results reflect the illiquidity, storage limits, and contango that impede arbitrage in the ETF and futures markets.
20 +
21 +As the HAR-X specification includes cross-market terms only at the daily lag, directional transmission is identified at the daily horizon. The weekly and monthly coefficients measure each series' own persistence rather than spillovers. This own-persistence is high and, for precious metals, stable across horizons and frequencies: for instance, gold's weekly own-volatility coefficient stays near 0.33, confirming the presence of long memory in realized volatility.
22 +
23 +\subsection{The Relevance of Using High-Frequency Data}
24 +
25 +Our second hypothesis (H2) is that higher-frequency sampling reveals transmission that is obscured in daily data. Tables~\ref{ch2:tab:HAR_5min}, \ref{ch2:tab:HAR_1min}, and~\ref{ch2:tab:HAR_30min} report the HAR-X estimates at 5-, 1-, and 30-minute frequencies. The daily iNAV-to-ETF transmission is markedly larger at a finer sampling frequency. For crude oil, it is measured at 0.378 at a 1-minute, 0.311 at a 5-minute, and 0.20 at a 30-minute frequency. The 1-minute estimate represents a 22\% increase over the 5-minute estimate and nearly double the 30-minute estimate. Thus, much of the same-day arbitrage transmission occurs within minutes, and measurements of this mechanism would be understated at a coarser sampling, as is traditionally used in the literature.
26 +
27 +The sensitivity of estimates according to frequency also varies by commodity. The iNAV-to-ETF transmission decreases substantially from 1-minute to 30-minute sampling in the case of crude oil (0.378 to 0.20, i.e., a 47\% drop), gold (0.632 to 0.387, or 39\% smaller), and silver (0.418 to 0.277, or 34\% smaller). It is, however, essentially flat for natural gas (0.079 versus 0.081). The economic interpretation is that transmission in the actively and continuously arbitraged crude oil and bullion markets clears within minutes, while for natural gas this adjusts over longer horizons because storage and pipeline constraints slow down the arbitrage activities that would otherwise help equalize ETF and underlying volatility.
28 +
29 +The reverse channel, ETF-to-iNAV transmission, is empirically weaker. Moreover, unlike the iNAV-to-ETF channel, the estimated effects do not strengthen monotonically with the sampling frequency. For crude oil, the daily ETF-to-iNAV coefficient is 0.089 at a 1-minute, 0.111 at a 5-minute, and 0.103 at a 30-minute frequency, peaking at the intermediate horizon rather than rising with the frequency. The contrast between the strongly frequency-dependent iNAV-to-ETF channel and this flatter reverse channel reinforces the directional dominance finding which we describe under hypothesis H1.
30 +
31 +Overall, we reject the null hypothesis of frequency-invariant transmission. A Wald test of $H_0:\delta_1^{(1\text{m})}=\delta_1^{(5\text{m})}=\delta_1^{(30\text{m})}$ on the daily iNAV-to-ETF coefficient rejects equality at the 1\% level for crude oil, gold, and silver; for natural gas, whose spillover is small and flat across frequencies, the difference is not significant. Daily data thus understate short-horizon transmission by up to roughly a factor of two for the most affected commodities, which would lead to incorrect conclusions about arbitrage effectiveness, addressing the sampling-frequency question raised by \citet{hansen2005realized}. For practitioners, this finding supports the relevance of high-frequency market monitoring to detect transmission patterns that are understated in daily analysis.
32 +
33 +\subsection{Jump Components and Discontinuous Volatility Transmission}
34 +
35 +Our third hypothesis (H3) concerns whether transmission is driven by the continuous (diffusion) component or by jumps. Tables~\ref{ch2:tab:HAR_CJ_5min}, \ref{ch2:tab:HAR_CJ_1min}, and~\ref{ch2:tab:HAR_CJ_30min} report the HAR-CJ-X model estimates. Across the four commodities, the daily cross-market jump coefficient dwarfs the continuous one: discrete price moves transmit volatility, while smooth price moves essentially do not. In the dominant iNAV-to-ETF direction at a 1-minute frequency, the jump transmission (lagged NAV jump in the ETF equation) is large and positive---0.892 for gold, 0.578 for silver, 0.432 for crude oil, 0.09 for natural gas---while the corresponding continuous transmission is near zero or negative for every commodity (gold $-0.123$, silver $-0.065$, crude oil $-0.019$, natural gas $-0.002$). The effect is strongest for precious metals, where the jump component is the largest in the sample.
36 +
37 +The same pattern holds in the reverse (ETF-to-iNAV) direction but with smaller economic magnitudes: the daily jump coefficient is 0.096 for crude oil and 0.128 for natural gas, while the continuous coefficients are near $-0.01$. Natural gas shows the weakest iNAV-to-ETF jump coefficient (0.090), which is consistent with natural gas volatility spikes arising from idiosyncratic, localized events---hurricanes, pipeline failures, extreme weather---that do not propagate systematically. In contrast, there is a larger value for the \emph{own} continuous effect: i.e., the weekly own-continuous coefficient reaches 0.116 at 1-minute frequency. This result suggests a separate, within-market phenomenon rather than a cross-market transmission pattern. Testing $H_0:\alpha_2=\alpha_1$ (daily jump versus continuous effect) against $H_a:\alpha_2>\alpha_1$, we find that the dominance of the jump component is confirmed at the 1\% level for all four commodities. This result supports H3: transmission mainly occurs through discrete jumps rather than continuous diffusion, which has implications for tail-risk hedging. Indeed, jump-driven volatility is harder to hedge with strategies that are usually built for the assumption of continuous price processes and volatility diffusions.
38 +
39 +\subsection{Bayesian VAR Analysis and Stability Over Time}
40 +
41 +Our fourth hypothesis (H4) concerns whether transmission is stable over time. The BVAR results shown in Tables~\ref{ch2:tab:VAR_USO}--\ref{ch2:tab:VAR_UNG} characterize the joint dynamics of ETF and iNAV volatility and, through sub-period estimation, the stability of the coefficient estimates. Since the BVAR treats the two volatilities as endogenous, it further provides a check on the HAR-X spillover coefficients without imposing a cascade structure. Across the four commodities, the results for the BVAR model confirm the asymmetries documented using the HAR-X model. Moreover, the cross-market coefficients remain stable across the 2010--2014, 2015--2019, and 2020--2023 sub-periods. Thus, we do not find evidence that the transmission mechanisms meaningfully change over time. The structural break test due to \citet{bai2003computation} indicates breaks in the volatility time series during the 2014 crude oil price collapse, the 2016 Brexit referendum, and the 2020 Covid-19 pandemic. However, reestimating the BVAR model for each regime period does not change the sign and relative magnitude of the cross-market coefficients. Therefore, we fail to reject the null hypothesis (H4) of a volatility transmission model that is stable over the full sample period.
42 +
43 +For crude oil (Table~\ref{ch2:tab:VAR_USO}), both volatilities are strongly persistent (iNAV first-lag coefficient 0.5704, 95\% credible interval [0.5237, 0.6171]; second lag 0.2606 [0.2206, 0.3005]). The ETF has only a weak effect on the iNAV (first lag 0.0735 [0.0343, 0.1135]), whereas the iNAV strongly drives the ETF: its first-lag effect (0.2861 [0.2315, 0.3433]) is comparable to ETF self-persistence (0.2941 [0.2460, 0.3418]) and its second-lag effect remains economically large (0.1934 [0.1443, 0.2410]). That the iNAV's first-lag effect rivals the ETF's own self-persistence is striking: for crude oil, underlying volatility is about as important as the ETF's recent volatility in explaining today's ETF volatility---the BVAR counterpart of the strong iNAV-to-ETF spillover found in the HAR-X estimates.
44 +
45 +Gold (Table~\ref{ch2:tab:VAR_GLD}) shows the most asymmetric configuration. The ETF has essentially no effect on the iNAV (first lag $-0.0101$ [$-0.0547$, 0.0336], with only a marginal second-lag effect of 0.0383 [0.0001, 0.0759]), while the iNAV dominates the ETF: its first-lag effect (0.3487 [0.2848, 0.4126]) exceeds ETF self-persistence (0.1549 [0.1004, 0.2083]) by more than twofold, with a large second lag (0.2758 [0.2221, 0.3307]). Silver (Table~\ref{ch2:tab:VAR_SLV}) is intermediate: iNAV self-persistence is the highest in the sample (first lag 0.6192 [0.5631, 0.6738]), the ETF-to-iNAV effect remains negligible ($-0.0446$ [$-0.0932$, 0.0051], turning weakly positive but still economically trivial at the second lag, 0.0584 [0.0176, 0.0996]), and the iNAV-to-ETF effect is strong (first lag 0.3760 [0.3102, 0.4404] versus ETF self-persistence 0.1790 [0.1226, 0.2362]) and remains economically large at the second lag (0.2016 [0.1471, 0.2563]). Natural gas (Table~\ref{ch2:tab:VAR_UNG}) shows the clearest bidirectionality. Its iNAV is moderately persistent (first lag 0.3919 [0.3488, 0.4366], second lag 0.2313 [0.1915, 0.2715]) and, unlike the other commodities, the ETF affects the iNAV through both lags, with a larger second-lag effect (0.0799 [0.0381, 0.1214] then 0.1567 [0.1189, 0.1938]), indicating that ETF activity feeds back to the underlying market through delayed channels tied to natural-gas storage operations. The forecast error variance decompositions reinforce this interpretation. Indeed, iNAV innovations explain a large and increasing share of ETF volatility forecast errors at longer horizons, while the share of iNAV forecast-error variance attributable to ETF innovations stays small for every commodity and is near zero for gold and silver. The impulse responses examined below confirm the same asymmetry.
46 +
47 +\subsection{Graphical Evidence on Volatility Patterns and Dynamic Responses}
48 +
49 +The realized volatility series shown in Figures~\ref{ch2:fig:rv_uso}--\ref{ch2:fig:rv_ung} corroborate the findings reported above. For crude oil (Figure~\ref{ch2:fig:rv_uso}), ETF and iNAV volatility co-move closely with synchronized peaks during stress periods, consistent with bidirectional transmission. For gold (Figure~\ref{ch2:fig:rv_gld}), iNAV volatility consistently precedes ETF volatility in periods of significant spikes, which is the visual counterpart of unidirectional iNAV-to-ETF transmission. Silver (Figure~\ref{ch2:fig:rv_slv}) shows the same pattern, but with occasional divergence, while natural gas (Figure~\ref{ch2:fig:rv_ung}) shows the most frequent ETF--iNAV divergences.
50 +
51 +The impulse responses shown in Figures~\ref{ch2:fig:irf1}--\ref{ch2:fig:irf4} tell the same story but add evidence on how the shocks evolve over time. For crude oil (Figure~\ref{ch2:fig:irf1}), iNAV shocks produce large, persistent responses in ETF volatility while ETF shocks produce small, transitory responses in the iNAV. Gold (Figure~\ref{ch2:fig:irf2}) shows the most pronounced asymmetry, with negligible responses of the iNAV to ETF shocks. Silver (Figure~\ref{ch2:fig:irf3}) is similar but noisier, while natural gas (Figure~\ref{ch2:fig:irf4}) shows sizable responses in both directions with delayed peaks, confirming bidirectional transmission.
52 +
53 +\subsection{Summary of the Evidence for the Hypotheses}
54 +
55 +Taken together, the estimates from the HAR-X, HAR-CJ-X and BVAR models support our four research hypotheses, and the formal tests defined in Section~\ref{ch2:sec:methodolog} reject each null hypothesis. We can summarize as follows.
56 +
57 +\textbf{H1.} We find that transmission is heterogeneous: the reverse (ETF-to-iNAV) coefficient is not significant for gold and silver but it is significant for crude oil and natural gas, while the forward (iNAV-to-ETF) coefficient is significant everywhere. Transmission is therefore unidirectional from iNAV to ETF for precious metals (a magnitude about four to one greater) and bidirectional for energy commodities (the coefficients are similar in magnitude). Thus, we reject the null hypothesis of uniform bidirectional transmission and we extend the evidence from \citet{gorton2006facts} to setting of commodity volatility. \textbf{H2.} Transmission is frequency-dependent: the daily iNAV-to-ETF spillover is up to roughly twice as large at the 1-minute as the 30-minute frequency. It is flat only for natural gas. These results confirm that daily data understate these volatility dynamics \citep{hansen2005realized, liu2015does}. \textbf{H3.} The daily cross-market jump component of transmission dwarfs the continuous part for all four commodities---most sharply for precious metals (i.e., the jump coefficient for gold iNAV-to-ETF jump is 0.892 versus $-0.123$ for the continuous component)---so transmission occurs mainly through discrete jumps. \textbf{H4.} The cross-market coefficients are stable across sub-periods and are robust to astructural-break test, so the documented mechanisms hold in general and are not specific to sub-periods.
58 +
59 +These findings build around a single mechanism, namely that the market structure and the arbitrage mechanism that links an ETF to its underlying asset---physical delivery for precious metals versus electronic futures settlement for energy---strongly shapes the characteristics of volatility transmission (e.g., direction, magnitude, horizon, and importance of jump vs diffusion parts), consistent with the arguments of \citet{petajisto2017inefficiencies} and \citet{basak2016model}. Finally, we explainthe implications for investors, market makers, and regulators in the conclusion.
added these-ulaval/chapitre2/06_conclusion.tex +14 −0
@@ -0,0 +1,14 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Conclusion} \label{ch2:sec:conclusion}
4 +
5 +Using high-frequency realized variance and a combination of HAR and Bayesian VAR models, this paper examines volatility transmission between four commodity ETFs and their underlying assets over the period 2010--2023. We find that volatility transmission is unidirectional from iNAV to ETF for physically-backed precious metals (GLD, SLV), while it is bidirectional (though still dominated by iNAV-to-ETF effects) for futures-based energy commodity funds (USO, UNG). Short-horizon transmission is far stronger in high-frequency data than in daily data, while longer-horizon effects are frequency-invariant; jump components transmit more strongly than continuous ones across all commodities; and these patterns are stable across sub-periods.
6 +
7 +These results show how arbitrage operates in practice and how its effectiveness depends on market structure, extending the limits-to-arbitrage frameworks of \citet{ackert2000arbitrage}, \citet{pontiff1996costly}, and \citet{gromb2010limits}. The unidirectional transmission which we document for precious metals is consistent with physical-delivery frictions that limit ETF activity from influencing underlying prices. In contrast, the more balanced volatility transmission in energy commodity markets fits the concept of futures-based arbitrage with electronic settlement, as argued by \citet{basak2016model}. The frequency-specific results suggest that arbitrage operates primarily through high-frequency channels. The evidence of longer-horizon persistence reflects separate fundamental forces, building on the intraday microstructure analysis of \citet{hasbrouck2003intraday} and \citet{richie2008examination}. The dominance of jumps in the transmission of volatility, which we identify using the bipower variation decomposition due to \citet{barndorff2004power} and \citet{huang2005using}, indicates that models used for ETF pricing should pay separate attention to the continuous and discontinuous components of volatility.
8 +
9 +In terms of methodology, our high-frequency iNAV series is the first that are built specifically for commodity ETFs over an extended sample period. These new series allow us to measure intraday arbitrage relationships that are unobservable when one uses end-of-day NAV \citep{petajisto2017inefficiencies}. Moreover, our comparison of HAR and BVAR models, together with a thorough analysis by sampling frequency, shows that temporal aggregation strongly affects the reliability of the empirical conclusions that can be drawn \citep{corsi2009simple, koop2011forecasting, andersen2007roughing}.
10 +
11 +The findings in this paper carry practical implications for different market participants and extend the risk management discussion of \citet{madhavan2012exchange} and \citet{staer2017asset}. For investors and risk managers, we show that precious metals ETF volatility can be forecast from underlying volatility alone, while energy ETF models must account for bidirectional feedback. Both benefit from high-frequency information for short horizons. For market makers and authorized participants, arbitrage in precious metals flows mainly from underlying markets to ETFs, while energy markets contain more balanced bidirectional opportunities \citep{hendershott2013relationship}. For regulators, the asymmetries that we document suggest monitoring that is tailored to the commodity type: precious metals markets show little ETF feedback and thus have a lower destabilization risk, while energy markets exhibit stronger bidirectional links warranting closer monitoring during stress periods \citep{ohara2021etf, dannhauser2017effect}. The strength of the jump transmission further suggests that stress testing should account for discontinuous shock scenarios.
12 +
13 +
14 +Finally, for future work, we note that the high-frequency iNAV methodology shown in this paper could be applied to equity, international, and fixed-income ETFs where similar arbitrage mechanisms operate. Future research could also investigate the relationship between jumps, which we find drive volatility transmission, to specific news events, order-flow imbalances, and market-maker inventory constraints.
added these-ulaval/chapitre2/chapitre2.tex +64 −0
@@ -0,0 +1,64 @@
1 +%% ---------------------------------------------------------------------------
2 +%% Chapitre 2 — adapté de phd_chap2_20260731/ (version préparée pour le
3 +%% Journal of Futures Markets). Corps scientifique inchangé ; voir
4 +%% INVENTAIRE.md pour la liste des adaptations mécaniques.
5 +%% ---------------------------------------------------------------------------
6 +
7 +\chapter{Seeing Through the ETF: Indicative NAV and Commodity Volatility Transmission}
8 +\label{chap-chapitre2}
9 +
10 +\section{Résumé}
11 +
12 +% BROUILLON À RÉVISER PAR L'AUTEUR (traduction condensée de l'abstract ; la
13 +% FESP limite le résumé de chaque article inséré à 150 mots).
14 +\begin{otherlanguage*}{french}
15 + Cet article construit une base de données inédite de valeurs liquidatives
16 + indicatives (iNAV) pour les fonds négociés en bourse (FNB) de matières
17 + premières et mesure la transmission de volatilité entre les FNB et leurs
18 + paniers sous-jacents à l'aide de la variance réalisée à haute fréquence.
19 + Appliquée au pétrole brut, à l'or, à l'argent et au gaz naturel, cette
20 + approche montre que l'iNAV offre une image plus nette de la relation de
21 + volatilité. En décomposant la variance réalisée en composantes continue et
22 + de sauts, nous établissons que la transmission passe principalement par les
23 + sauts plutôt que par la diffusion, un canal que les mesures usuelles de
24 + connectivité peuvent masquer. La fréquence d'échantillonnage importe : les
25 + données à une minute révèlent une transmission quotidienne jusqu'à deux fois
26 + plus grande que les estimations à 30~minutes. Enfin, la transmission est
27 + unidirectionnelle — de l'iNAV vers le FNB — pour l'or et l'argent, mais
28 + bidirectionnelle et asymétrique pour les FNB énergétiques.
29 +\end{otherlanguage*}
30 +
31 +\section{Abstract}
32 +
33 +This paper builds a novel dataset of indicative Net Asset Value (iNAV)
34 +observations for commodity ETFs and measures volatility transmission between
35 +ETFs and their underlying baskets using high-frequency realized variance.
36 +Applying this design to crude oil, gold, silver, and natural gas, we show that
37 +the iNAV provides a sharper image of the volatility relationship. Decomposing
38 +realized variance into continuous and jump components, we establish that
39 +transmission runs primarily through jumps rather than diffusion, a channel
40 +that standard volatility connectedness measures may obscure. Sampling
41 +frequency also matters: 1-minute data reveal daily volatility transmission up
42 +to twice as large as 30-minute estimates. Finally, transmission differs across
43 +commodity types. For gold and silver, it is unidirectional from iNAV to ETF,
44 +consistent with passive arbitrage. For energy ETFs, which are more liquid and
45 +hold futures as underlying assets, transmission is bidirectional and
46 +asymmetric.
47 +
48 +\medskip
49 +\noindent\textbf{Keywords:} ETF, volatility, transmission, commodity markets,
50 +futures, intraday, realized variance, jump, HAR, arbitrage, microstructure.\\
51 +\noindent\textbf{JEL Classification:} G12, G13, G14, C32
52 +
53 +%% --- Corps de l'article ---
54 +%% (pas de fichier 02_* dans la source : la revue de littérature est intégrée
55 +%% à l'introduction)
56 +\input{chapitre2/01_introduction}
57 +\input{chapitre2/03_data}
58 +\input{chapitre2/04_methods}
59 +\input{chapitre2/05_results}
60 +\input{chapitre2/06_conclusion}
61 +
62 +%% --- Tableaux et figures de l'article ---
63 +\input{chapitre2/tables}
64 +\input{chapitre2/figures}
added these-ulaval/chapitre2/figures.tex +77 −0
@@ -0,0 +1,77 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Figures}
4 +
5 +
6 +\begin{landscape}
7 +\begin{figure}[t]
8 +\includegraphics[height=0.78\textheight,keepaspectratio]{oil_rv.pdf}
9 +\centering
10 +\caption{Realized volatility of the Crude Oil ETF and its iNAV. The figure plots daily realized volatility (sum of squared intraday log returns, $\times 10^{-3}$) for the USO ETF (blue) and its iNAV, proxied by the CL futures contract (red), from 2008-03-27 to 2025-08-01. Panels A, B, and C use returns sampled at the 1-minute, 5-minute, and 30-minute frequencies, respectively. For readability, each series is winsorized at its 99th percentile within each panel.}
11 +\label{ch2:fig:rv_uso}
12 +\end{figure}
13 +\end{landscape}
14 +
15 +\begin{landscape}
16 +\begin{figure}[t]
17 +\includegraphics[height=0.78\textheight,keepaspectratio]{gold_rv.pdf}
18 +\centering
19 +\caption{Realized volatility of the Gold ETF and its iNAV. The figure plots daily realized volatility (sum of squared intraday log returns, $\times 10^{-3}$) for the GLD ETF (blue) and its iNAV, proxied by the GC futures contract (red), from 2008-03-27 to 2025-08-01. Panels A, B, and C use returns sampled at the 1-minute, 5-minute, and 30-minute frequencies, respectively. For readability, each series is winsorized at its 99th percentile within each panel.}
20 +\label{ch2:fig:rv_gld}
21 +\end{figure}
22 +\end{landscape}
23 +
24 +\begin{landscape}
25 +\begin{figure}[t]
26 +\includegraphics[height=0.78\textheight,keepaspectratio]{silver_rv.pdf}
27 +\centering
28 +\caption{Realized volatility of the Silver ETF and its iNAV. The figure plots daily realized volatility (sum of squared intraday log returns, $\times 10^{-3}$) for the SLV ETF (blue) and its iNAV, proxied by the SI futures contract (red), from 2008-03-27 to 2025-08-01. Panels A, B, and C use returns sampled at the 1-minute, 5-minute, and 30-minute frequencies, respectively. For readability, each series is winsorized at its 99th percentile within each panel.}
29 +\label{ch2:fig:rv_slv}
30 +\end{figure}
31 +\end{landscape}
32 +
33 +\begin{landscape}
34 +\begin{figure}[t]
35 +\includegraphics[height=0.78\textheight,keepaspectratio]{gas_rv.pdf}
36 +\centering
37 +\caption{Realized volatility of the Natural Gas ETF and its iNAV. The figure plots daily realized volatility (sum of squared intraday log returns, $\times 10^{-3}$) for the UNG ETF (blue) and its iNAV, proxied by the NG futures contract (red), from 2008-03-27 to 2025-08-01. Panels A, B, and C use returns sampled at the 1-minute, 5-minute, and 30-minute frequencies, respectively. For readability, each series is winsorized at its 99th percentile within each panel.}
38 +\label{ch2:fig:rv_ung}
39 +\end{figure}
40 +\end{landscape}
41 +
42 +
43 +\begin{landscape}
44 +\begin{figure}[t]
45 +\includegraphics[width=16cm,trim=0 0 0 22,clip]{oil__uso_cl__irf.pdf}
46 +\centering
47 +\caption{Impulse Response Functions (IRF) for USO ETF and Crude Oil (CL) iNAV Volatility. Bayesian VAR impulse responses showing fast shock absorption with systematic asymmetry: iNAV shocks cause large and persistent responses in ETF volatility while ETF shocks generate smaller and more transitory effects on iNAV, validating bidirectional but asymmetric transmission in crude oil markets.}
48 +\label{ch2:fig:irf1}
49 +\end{figure}
50 +\end{landscape}
51 +
52 +\begin{landscape}
53 +\begin{figure}[t]
54 +\includegraphics[width=16cm,trim=0 0 0 22,clip]{gold__gld_gc__irf.pdf}
55 +\centering
56 +\caption{Impulse Response Functions (IRF) for GLD ETF and Gold (GC) iNAV Volatility. Bayesian VAR impulse responses exhibiting the most pronounced asymmetries: iNAV shocks create large and persistent responses in ETF volatility that decay smoothly, while ETF shocks generate negligible responses in iNAV, providing strong dynamic evidence for unidirectional transmission in gold markets.}
57 +\label{ch2:fig:irf2}
58 +\end{figure}
59 +\end{landscape}
60 +
61 +\begin{landscape}
62 +\begin{figure}[t]
63 +\includegraphics[width=16cm,trim=0 0 0 22,clip]{silver__slv_si__irf.pdf}
64 +\centering
65 +\caption{Impulse Response Functions (IRF) for SLV ETF and Silver (SI) iNAV Volatility. Bayesian VAR impulse responses preserving the gold market's asymmetric structure but with qualitatively richer adjustment dynamics and greater variability, reflecting silver's dual function as both an industrial and a precious metal.}
66 +\label{ch2:fig:irf3}
67 +\end{figure}
68 +\end{landscape}
69 +
70 +\begin{landscape}
71 +\begin{figure}[t]
72 +\includegraphics[width=16cm,trim=0 0 0 22,clip]{gas__ung_ng__irf.pdf}
73 +\centering
74 +\caption{Impulse Response Functions (IRF) for UNG ETF and Natural Gas (NG) iNAV Volatility. Bayesian VAR impulse responses exhibiting the most elaborate structures with large responses in both directions, delayed peaks, and periodic behavior, substantiating bidirectional transmission involving numerous channels and horizons in natural gas markets.}
75 +\label{ch2:fig:irf4}
76 +\end{figure}
77 +\end{landscape}
added these-ulaval/chapitre2/figures/gas__ung_ng__irf.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre2/figures/gas_rv.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre2/figures/gold__gld_gc__irf.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre2/figures/gold_rv.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre2/figures/oil__uso_cl__irf.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre2/figures/oil_rv.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre2/figures/silver__slv_si__irf.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre2/figures/silver_rv.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre2/tables.tex +678 −0
@@ -0,0 +1,678 @@
1 +% For submission to the Journal of Futures Markets (JFM)
2 +
3 +\section{Tables}
4 +
5 +\begin{table}[htbp]
6 +\centering
7 +\caption{Descriptive Statistics for Realized Volatility, Quadratic Power Variation, and Jump Variables}
8 +\label{ch2:tab:desc_stats_5min}
9 +\begin{threeparttable}
10 +\begin{tabular}{@{}lcccccc@{}}
11 +\toprule
12 +\multicolumn{1}{l}{Commodity} & Variable & Obs & Mean & Std. Dev. & Min & Max \\
13 +\midrule
14 +\multicolumn{7}{l}{\textit{Panel A: Realized Volatility}} \\
15 +\addlinespace[0.1cm]
16 +Crude Oil & $RV_{t,NAV}$ & 3,935 & 0.081 & 0.717 & 0.002 & 42.082 \\
17 + & $RV_{t,ETF}$ & 3,935 & 0.066 & 0.204 & 0.001 & 10.335 \\
18 +\addlinespace[0.05cm]
19 +Gold & $RV_{t,NAV}$ & 3,935 & 0.012 & 0.017 & 0.001 & 0.327 \\
20 + & $RV_{t,ETF}$ & 3,935 & 0.034 & 1.259 & 0.001 & 78.988 \\
21 +\addlinespace[0.05cm]
22 +Silver & $RV_{t,NAV}$ & 3,935 & 0.043 & 0.063 & 0.002 & 1.120 \\
23 + & $RV_{t,ETF}$ & 3,935 & 0.042 & 0.067 & 0.004 & 1.286 \\
24 +\addlinespace[0.05cm]
25 +Natural Gas & $RV_{t,NAV}$ & 3,935 & 0.104 & 0.173 & 0.010 & 6.635 \\
26 + & $RV_{t,ETF}$ & 3,935 & 0.103 & 0.116 & 0.004 & 1.651 \\
27 +\addlinespace[0.2cm]
28 +
29 +\multicolumn{7}{l}{\textit{Panel B: Quadratic Power Variation}} \\
30 +\addlinespace[0.1cm]
31 +Crude Oil & $QPV_{t,NAV}$ & 3,935 & 0.110 & 6.797 & 0.000 & 426.363 \\
32 + & $QPV_{t,ETF}$ & 3,935 & 0.003 & 0.121 & 0.000 & 7.295 \\
33 +\addlinespace[0.05cm]
34 +Gold & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.007 \\
35 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.027 \\
36 +\addlinespace[0.05cm]
37 +Silver & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.003 & 0.000 & 0.139 \\
38 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.043 \\
39 +\addlinespace[0.05cm]
40 +Natural Gas & $QPV_{t,NAV}$ & 3,935 & 0.003 & 0.152 & 0.000 & 9.555 \\
41 + & $QPV_{t,ETF}$ & 3,935 & 0.001 & 0.030 & 0.000 & 1.881 \\
42 +\addlinespace[0.2cm]
43 +
44 +\multicolumn{7}{l}{\textit{Panel C: Jump Component}} \\
45 +\addlinespace[0.1cm]
46 +Crude Oil & $J_{t,NAV}$ & 3,935 & $-$0.028 & 6.132 & 0.000 & 8.941 \\
47 + & $J_{t,ETF}$ & 3,935 & 0.063 & 0.124 & 0.000 & 3.040 \\
48 +\addlinespace[0.05cm]
49 +Gold & $J_{t,NAV}$ & 3,935 & 0.012 & 0.017 & 0.001 & 0.324 \\
50 + & $J_{t,ETF}$ & 3,935 & 0.034 & 1.259 & 0.001 & 78.981 \\
51 +\addlinespace[0.05cm]
52 +Silver & $J_{t,NAV}$ & 3,935 & 0.042 & 0.061 & 0.002 & 1.094 \\
53 + & $J_{t,ETF}$ & 3,935 & 0.042 & 0.066 & 0.004 & 1.243 \\
54 +\addlinespace[0.05cm]
55 +Natural Gas & $J_{t,NAV}$ & 3,935 & 0.101 & 0.209 & 0.000 & 6.634 \\
56 + & $J_{t,ETF}$ & 3,935 & 0.102 & 0.113 & 0.000 & 1.651 \\
57 +\bottomrule
58 +\end{tabular}
59 +\begin{tablenotes}
60 +\small
61 +\item This table presents descriptive statistics for realized volatility (RV), quadratic power variation (QPV), and jump component (J) variables constructed using 5-minute price data. All values are expressed in percentages. NAV refers to net asset value prices, and ETF refers to exchange-traded fund prices. The sample period includes 3,935 daily observations for each commodity.
62 +\end{tablenotes}
63 +\end{threeparttable}
64 +\end{table}
65 +
66 +
67 +\begin{table}[htbp]
68 +\centering
69 +\caption{Descriptive Statistics for Realized Volatility, Quadratic Power Variation, and Jump Variables (1-minute data)}
70 +\label{ch2:tab:desc_stats_1min}
71 +\begin{threeparttable}
72 +\begin{tabular}{@{}lcccccc@{}}
73 +\toprule
74 +\multicolumn{1}{l}{Commodity} & Variable & Obs & Mean & Std. Dev. & Min & Max \\
75 +\midrule
76 +\multicolumn{7}{l}{\textit{Panel A: Realized Volatility}} \\
77 +\addlinespace[0.1cm]
78 +Crude Oil & $RV_{t,NAV}$ & 3,935 & 0.083 & 0.959 & 0.001 & 58.167 \\
79 + & $RV_{t,ETF}$ & 3,935 & 0.058 & 0.250 & 0.001 & 14.051 \\
80 +\addlinespace[0.05cm]
81 +Gold & $RV_{t,NAV}$ & 3,935 & 0.012 & 0.018 & 0.001 & 0.302 \\
82 + & $RV_{t,ETF}$ & 3,935 & 0.031 & 1.259 & 0.001 & 78.986 \\
83 +\addlinespace[0.05cm]
84 +Silver & $RV_{t,NAV}$ & 3,935 & 0.040 & 0.064 & 0.001 & 1.347 \\
85 + & $RV_{t,ETF}$ & 3,935 & 0.036 & 0.061 & 0.002 & 1.367 \\
86 +\addlinespace[0.05cm]
87 +Natural Gas & $RV_{t,NAV}$ & 3,935 & 0.099 & 0.174 & 0.007 & 6.699 \\
88 + & $RV_{t,ETF}$ & 3,935 & 0.086 & 0.096 & 0.004 & 1.634 \\
89 +\addlinespace[0.2cm]
90 +
91 +\multicolumn{7}{l}{\textit{Panel B: Quadratic Power Variation}} \\
92 +\addlinespace[0.1cm]
93 +Crude Oil & $QPV_{t,NAV}$ & 3,935 & 0.019 & 1.152 & 0.000 & 72.288 \\
94 + & $QPV_{t,ETF}$ & 3,935 & 0.004 & 0.191 & 0.000 & 11.936 \\
95 +\addlinespace[0.05cm]
96 +Gold & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.004 \\
97 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.016 \\
98 +\addlinespace[0.05cm]
99 +Silver & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.060 \\
100 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.025 \\
101 +\addlinespace[0.05cm]
102 +Natural Gas & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.004 & 0.000 & 0.232 \\
103 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.039 \\
104 +\addlinespace[0.2cm]
105 +
106 +\multicolumn{7}{l}{\textit{Panel C: Jump Component}} \\
107 +\addlinespace[0.1cm]
108 +Crude Oil & $J_{t,NAV}$ & 3,935 & 0.064 & 0.330 & 0.000 & 8.933 \\
109 + & $J_{t,ETF}$ & 3,935 & 0.055 & 0.112 & 0.001 & 3.172 \\
110 +\addlinespace[0.05cm]
111 +Gold & $J_{t,NAV}$ & 3,935 & 0.012 & 0.018 & 0.001 & 0.300 \\
112 + & $J_{t,ETF}$ & 3,935 & 0.031 & 1.259 & 0.001 & 78.969 \\
113 +\addlinespace[0.05cm]
114 +Silver & $J_{t,NAV}$ & 3,935 & 0.040 & 0.063 & 0.001 & 1.328 \\
115 + & $J_{t,ETF}$ & 3,935 & 0.036 & 0.060 & 0.002 & 1.344 \\
116 +\addlinespace[0.05cm]
117 +Natural Gas & $J_{t,NAV}$ & 3,935 & 0.098 & 0.173 & 0.007 & 6.697 \\
118 + & $J_{t,ETF}$ & 3,935 & 0.086 & 0.095 & 0.004 & 1.634 \\
119 +\bottomrule
120 +\end{tabular}
121 +\begin{tablenotes}
122 +\small
123 +\item This table presents descriptive statistics for realized volatility (RV), quadratic power variation (QPV), and jump component (J) variables constructed using 1-minute price data. All values are expressed in percentages. NAV refers to net asset value prices, and ETF refers to exchange-traded fund prices. The sample period includes 3,935 daily observations for each commodity.
124 +\end{tablenotes}
125 +\end{threeparttable}
126 +\end{table}
127 +
128 +\begin{table}[htbp]
129 +\centering
130 +\caption{Descriptive Statistics for Realized Volatility, Quadratic Power Variation, and Jump Variables (30-minute data)}
131 +\label{ch2:tab:desc_stats_30min}
132 +\begin{threeparttable}
133 +\begin{tabular}{@{}lcccccc@{}}
134 +\toprule
135 +\multicolumn{1}{l}{Commodity} & Variable & Obs & Mean & Std. Dev. & Min & Max \\
136 +\midrule
137 +\multicolumn{7}{l}{\textit{Panel A: Realized Volatility}} \\
138 +\addlinespace[0.1cm]
139 +Crude Oil & $RV_{t,NAV}$ & 3,935 & 0.073 & 0.459 & 0.001 & 23.817 \\
140 + & $RV_{t,ETF}$ & 3,935 & 0.052 & 0.176 & 0.000 & 8.995 \\
141 +\addlinespace[0.05cm]
142 +Gold & $RV_{t,NAV}$ & 3,935 & 0.011 & 0.019 & 0.001 & 0.407 \\
143 + & $RV_{t,ETF}$ & 3,935 & 0.030 & 1.257 & 0.000 & 78.854 \\
144 +\addlinespace[0.05cm]
145 +Silver & $RV_{t,NAV}$ & 3,935 & 0.039 & 0.069 & 0.001 & 1.701 \\
146 + & $RV_{t,ETF}$ & 3,935 & 0.033 & 0.060 & 0.001 & 1.618 \\
147 +\addlinespace[0.05cm]
148 +Natural Gas & $RV_{t,NAV}$ & 3,935 & 0.093 & 0.176 & 0.004 & 6.829 \\
149 + & $RV_{t,ETF}$ & 3,935 & 0.077 & 0.093 & 0.002 & 1.668 \\
150 +\addlinespace[0.2cm]
151 +
152 +\multicolumn{7}{l}{\textit{Panel B: Quadratic Power Variation}} \\
153 +\addlinespace[0.1cm]
154 +Crude Oil & $QPV_{t,NAV}$ & 3,935 & 0.002 & 0.067 & 0.000 & 4.036 \\
155 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.013 & 0.000 & 0.831 \\
156 +\addlinespace[0.05cm]
157 +Gold & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.002 \\
158 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.000 & 0.000 & 0.002 \\
159 +\addlinespace[0.05cm]
160 +Silver & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.050 \\
161 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.037 \\
162 +\addlinespace[0.05cm]
163 +Natural Gas & $QPV_{t,NAV}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.044 \\
164 + & $QPV_{t,ETF}$ & 3,935 & 0.000 & 0.001 & 0.000 & 0.035 \\
165 +\addlinespace[0.2cm]
166 +
167 +\multicolumn{7}{l}{\textit{Panel C: Jump Component}} \\
168 +\addlinespace[0.1cm]
169 +Crude Oil & $J_{t,NAV}$ & 3,935 & 0.072 & 0.401 & 0.001 & 19.781 \\
170 + & $J_{t,ETF}$ & 3,935 & 0.051 & 0.165 & 0.000 & 8.164 \\
171 +\addlinespace[0.05cm]
172 +Gold & $J_{t,NAV}$ & 3,935 & 0.011 & 0.018 & 0.001 & 0.404 \\
173 + & $J_{t,ETF}$ & 3,935 & 0.030 & 1.257 & 0.000 & 78.852 \\
174 +\addlinespace[0.05cm]
175 +Silver & $J_{t,NAV}$ & 3,935 & 0.038 & 0.068 & 0.001 & 1.665 \\
176 + & $J_{t,ETF}$ & 3,935 & 0.033 & 0.059 & 0.001 & 1.583 \\
177 +\addlinespace[0.05cm]
178 +Natural Gas & $J_{t,NAV}$ & 3,935 & 0.093 & 0.175 & 0.004 & 6.823 \\
179 + & $J_{t,ETF}$ & 3,935 & 0.077 & 0.093 & 0.002 & 1.633 \\
180 +\bottomrule
181 +\end{tabular}
182 +\begin{tablenotes}
183 +\small
184 +\item This table presents descriptive statistics for realized volatility (RV), quadratic power variation (QPV), and jump component (J) variables constructed using 30-minute price data. All values are expressed in percentages. NAV refers to net asset value prices, and ETF refers to exchange-traded fund prices. The sample period includes 3,935 daily observations for each commodity.
185 +\end{tablenotes}
186 +\end{threeparttable}
187 +\end{table}
188 +
189 +
190 +\begin{landscape}
191 +\begin{table}[htbp]
192 +\centering
193 +\caption{HAR-X Model Estimates with 5-minute Realized Variance}
194 +\label{ch2:tab:HAR_5min}
195 +\begin{threeparttable}
196 +\footnotesize
197 +\begin{tabular}{@{}lcccccccc@{}}
198 +\toprule
199 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
200 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
201 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
202 +\midrule
203 +$RV_{t-1,NAV}$ & 0.288*** & 0.311*** & 0.253*** & 0.515*** & 0.345*** & 0.383*** & 0.092** & 0.105*** \\
204 + & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) \\
205 +\addlinespace[0.1cm]
206 +$\overline{RV}_{t-5,NAV}$ & 0.408*** & & 0.338*** & & 0.318*** & & 0.393*** & \\
207 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
208 +\addlinespace[0.1cm]
209 +$\overline{RV}_{t-22,NAV}$ & 0.144*** & & 0.321*** & & 0.271*** & & 0.313*** & \\
210 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
211 +\addlinespace[0.1cm]
212 +$RV_{t-1,ETF}$ & 0.111*** & 0.098*** & 0.021 & $-$0.084* & $-$0.009 & $-$0.015 & 0.126*** & 0.084*** \\
213 + & (0.000) & (0.000) & (0.158) & (0.062) & (0.612) & (0.464) & (0.000) & (0.000) \\
214 +\addlinespace[0.1cm]
215 +$\overline{RV}_{t-5,ETF}$ & & 0.297*** & & 0.270*** & & 0.209*** & & 0.431*** \\
216 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
217 +\addlinespace[0.1cm]
218 +$\overline{RV}_{t-22,ETF}$ & & 0.227*** & & 0.164*** & & 0.342*** & & 0.332*** \\
219 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
220 +\midrule
221 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
222 +\bottomrule
223 +\end{tabular}
224 +\begin{tablenotes}
225 +\small
226 +\item This table presents estimation results for the HAR-X model using 5-minute realized variance data. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $RV_{t-1}$ represents the lagged daily realized variance, $\overline{RV}_{t-5}$ is the average of the past 5 days' realized variances, and $\overline{RV}_{t-22}$ is the average of the past 22 days' realized variances.
227 +\end{tablenotes}
228 +\end{threeparttable}
229 +\end{table}
230 +\end{landscape}
231 +
232 +
233 +\begin{landscape}
234 +\begin{table}[htbp]
235 +\centering
236 +\caption{HAR-X Model Estimates with 1-minute Realized Variance}
237 +\label{ch2:tab:HAR_1min}
238 +\begin{threeparttable}
239 +\footnotesize
240 +\begin{tabular}{@{}lcccccccc@{}}
241 +\toprule
242 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
243 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
244 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
245 +\midrule
246 +$RV_{t-1,NAV}$ & 0.371*** & 0.378*** & 0.331*** & 0.632*** & 0.404*** & 0.418*** & 0.121*** & 0.079*** \\
247 + & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) \\
248 +\addlinespace[0.1cm]
249 +$\overline{RV}_{t-5,NAV}$ & 0.387*** & & 0.329*** & & 0.297*** & & 0.386*** & \\
250 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
251 +\addlinespace[0.1cm]
252 +$\overline{RV}_{t-22,NAV}$ & 0.109*** & & 0.280*** & & 0.226*** & & 0.285*** & \\
253 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
254 +\addlinespace[0.1cm]
255 +$RV_{t-1,ETF}$ & 0.089*** & 0.097*** & $-$0.001 & $-$0.059* & 0.004 & 0.015 & 0.127*** & 0.124*** \\
256 + & (0.000) & (0.000) & (0.961) & (0.065) & (0.804) & (0.410) & (0.000) & (0.000) \\
257 +\addlinespace[0.1cm]
258 +$\overline{RV}_{t-5,ETF}$ & & 0.272*** & & 0.196*** & & 0.206*** & & 0.438*** \\
259 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
260 +\addlinespace[0.1cm]
261 +$\overline{RV}_{t-22,ETF}$ & & 0.200*** & & 0.113*** & & 0.276*** & & 0.318*** \\
262 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
263 +\midrule
264 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
265 +\bottomrule
266 +\end{tabular}
267 +\begin{tablenotes}
268 +\small
269 +\item This table presents estimation results for the HAR-X model using 1-minute realized variance data. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $RV_{t-1}$ represents the lagged daily realized variance, $\overline{RV}_{t-5}$ is the average of the past 5 days' realized variances, and $\overline{RV}_{t-22}$ is the average of the past 22 days' realized variances.
270 +\end{tablenotes}
271 +\end{threeparttable}
272 +\end{table}
273 +\end{landscape}
274 +
275 +
276 +\begin{landscape}
277 +\begin{table}[htbp]
278 +\centering
279 +\caption{HAR-X Model Estimates with 30-minute Realized Variance}
280 +\label{ch2:tab:HAR_30min}
281 +\begin{threeparttable}
282 +\footnotesize
283 +\begin{tabular}{@{}lcccccccc@{}}
284 +\toprule
285 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
286 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
287 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
288 +\midrule
289 +$RV_{t-1,NAV}$ & 0.133*** & 0.200*** & 0.150*** & 0.387*** & 0.226*** & 0.277*** & 0.043 & 0.081*** \\
290 + & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.121) & (0.000) \\
291 +\addlinespace[0.1cm]
292 +$\overline{RV}_{t-5,NAV}$ & 0.461*** & & 0.335*** & & 0.319*** & & 0.399*** & \\
293 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
294 +\addlinespace[0.1cm]
295 +$\overline{RV}_{t-22,NAV}$ & 0.243*** & & 0.445*** & & 0.394*** & & 0.393*** & \\
296 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
297 +\addlinespace[0.1cm]
298 +$RV_{t-1,ETF}$ & 0.103*** & 0.049 & $-$0.015 & $-$0.128** & $-$0.034 & $-$0.075** & 0.074** & 0.012 \\
299 + & (0.000) & (0.149) & (0.476) & (0.019) & (0.140) & (0.013) & (0.030) & (0.662) \\
300 +\addlinespace[0.1cm]
301 +$\overline{RV}_{t-5,ETF}$ & & 0.363*** & & 0.332*** & & 0.238*** & & 0.405*** \\
302 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
303 +\addlinespace[0.1cm]
304 +$\overline{RV}_{t-22,ETF}$ & & 0.319*** & & 0.212*** & & 0.460*** & & 0.447*** \\
305 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
306 +\midrule
307 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
308 +\bottomrule
309 +\end{tabular}
310 +\begin{tablenotes}
311 +\small
312 +\item This table presents estimation results for the HAR-X model using 30-minute realized variance data. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $RV_{t-1}$ represents the lagged daily realized variance, $\overline{RV}_{t-5}$ is the average of the past 5 days' realized variances, and $\overline{RV}_{t-22}$ is the average of the past 22 days' realized variances.
313 +\end{tablenotes}
314 +\end{threeparttable}
315 +\end{table}
316 +\end{landscape}
317 +
318 +
319 +\begin{landscape}
320 +\begin{table}[htbp]
321 +\centering
322 +\caption{HAR-CJ-X Model Estimates with 5-minute Realized Variance}
323 +\label{ch2:tab:HAR_CJ_5min}
324 +\begin{threeparttable}
325 +\footnotesize
326 +\begin{tabular}{@{}lcccccccc@{}}
327 +\toprule
328 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
329 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
330 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
331 +\midrule
332 +\multicolumn{9}{l}{\textit{Panel A: Quadratic Power Variation}} \\
333 +\addlinespace[0.1cm]
334 +$QPV_{t-1,NAV}$ & 0.049** & 0.008 & 0.026 & $-$0.001 & $-$0.015 & $-$0.026 & 0.065*** & 0.016 \\
335 + & (0.025) & (0.661) & (0.239) & (0.957) & (0.467) & (0.273) & (0.001) & (0.400) \\
336 +\addlinespace[0.05cm]
337 +$\overline{QPV}_{t-5,NAV}$ & 0.032 & & $-$0.027 & & $-$0.030 & & 0.110*** & \\
338 + & (0.167) & & (0.317) & & (0.236) & & (0.000) & \\
339 +\addlinespace[0.05cm]
340 +$\overline{QPV}_{t-22,NAV}$ & $-$0.046*** & & $-$0.021 & & $-$0.005 & & 0.044** & \\
341 + & (0.007) & & (0.341) & & (0.793) & & (0.018) & \\
342 +\addlinespace[0.05cm]
343 +$QPV_{t-1,ETF}$ & 0.001 & 0.043** & $-$0.015 & $-$0.030 & 0.006 & 0.010 & 0.011 & 0.028* \\
344 + & (0.969) & (0.038) & (0.545) & (0.248) & (0.779) & (0.686) & (0.620) & (0.088) \\
345 +\addlinespace[0.05cm]
346 +$\overline{QPV}_{t-5,ETF}$ & & 0.017 & & $-$0.008 & & $-$0.003 & & 0.062*** \\
347 + & & (0.454) & & (0.737) & & (0.876) & & (0.001) \\
348 +\addlinespace[0.05cm]
349 +$\overline{QPV}_{t-22,ETF}$ & & $-$0.040** & & 0.133*** & & 0.006 & & 0.038 \\
350 + & & (0.043) & & (0.000) & & (0.763) & & (0.137) \\
351 +\addlinespace[0.2cm]
352 +
353 +\multicolumn{9}{l}{\textit{Panel B: Jump Component}} \\
354 +\addlinespace[0.1cm]
355 +$J_{t-1,NAV}$ & 0.207*** & 0.280*** & 0.212*** & 0.491*** & 0.375*** & 0.431*** & $-$0.003 & 0.057* \\
356 + & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.923) & (0.082) \\
357 +\addlinespace[0.05cm]
358 +$\overline{J}_{t-5,NAV}$ & 0.338*** & & 0.407*** & & 0.389*** & & 0.220*** & \\
359 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
360 +\addlinespace[0.05cm]
361 +$\overline{J}_{t-22,NAV}$ & 0.267*** & & 0.354*** & & 0.272*** & & 0.213*** & \\
362 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
363 +\addlinespace[0.05cm]
364 +$J_{t-1,ETF}$ & 0.094*** & 0.019 & 0.041 & $-$0.023 & $-$0.016 & $-$0.030 & 0.044 & 0.039 \\
365 + & (0.000) & (0.473) & (0.156) & (0.497) & (0.527) & (0.291) & (0.190) & (0.195) \\
366 +\addlinespace[0.05cm]
367 +$\overline{J}_{t-5,ETF}$ & & 0.264*** & & 0.246*** & & 0.214*** & & 0.307*** \\
368 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
369 +\addlinespace[0.05cm]
370 +$\overline{J}_{t-22,ETF}$ & & 0.329*** & & $-$0.014 & & 0.327*** & & 0.250*** \\
371 + & & (0.000) & & (0.631) & & (0.000) & & (0.000) \\
372 +\midrule
373 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
374 +\bottomrule
375 +\end{tabular}
376 +\begin{tablenotes}
377 +\small
378 +\item This table presents estimation results for the HAR-CJ-X model using 5-minute realized variance data. The model incorporates both continuous (quadratic power variation, QPV) and jump (J) components. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $QPV_{t-1}$ and $J_{t-1}$ represent the lagged daily components, while $\overline{QPV}_{t-5}$, $\overline{J}_{t-5}$, $\overline{QPV}_{t-22}$, and $\overline{J}_{t-22}$ are the corresponding weekly and monthly averages.
379 +\end{tablenotes}
380 +\end{threeparttable}
381 +\end{table}
382 +\end{landscape}
383 +
384 +\begin{landscape}
385 +\begin{table}[htbp]
386 +\centering
387 +\caption{HAR-CJ-X Model Estimates with 1-minute Realized Variance}
388 +\label{ch2:tab:HAR_CJ_1min}
389 +\begin{threeparttable}
390 +\footnotesize
391 +\begin{tabular}{@{}lcccccccc@{}}
392 +\toprule
393 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
394 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
395 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
396 +\midrule
397 +\multicolumn{9}{l}{\textit{Panel A: Quadratic Power Variation}} \\
398 +\addlinespace[0.1cm]
399 +$QPV_{t-1,NAV}$ & 0.036 & $-$0.019 & $-$0.033* & $-$0.123*** & $-$0.041** & $-$0.065*** & 0.008 & $-$0.002 \\
400 + & (0.246) & (0.525) & (0.096) & (0.000) & (0.039) & (0.002) & (0.773) & (0.920) \\
401 +\addlinespace[0.05cm]
402 +$\overline{QPV}_{t-5,NAV}$ & $-$0.021 & & $-$0.076*** & & $-$0.068*** & & 0.116*** & \\
403 + & (0.444) & & (0.004) & & (0.009) & & (0.000) & \\
404 +\addlinespace[0.05cm]
405 +$\overline{QPV}_{t-22,NAV}$ & $-$0.031 & & $-$0.001 & & 0.009 & & $-$0.040** & \\
406 + & (0.309) & & (0.976) & & (0.716) & & (0.046) & \\
407 +\addlinespace[0.05cm]
408 +$QPV_{t-1,ETF}$ & $-$0.011 & 0.009 & $-$0.020* & $-$0.032* & $-$0.011 & $-$0.014 & $-$0.011 & $-$0.021* \\
409 + & (0.665) & (0.698) & (0.062) & (0.089) & (0.617) & (0.514) & (0.671) & (0.089) \\
410 +\addlinespace[0.05cm]
411 +$\overline{QPV}_{t-5,ETF}$ & & $-$0.014 & & 0.016 & & 0.003 & & 0.016 \\
412 + & & (0.591) & & (0.566) & & (0.895) & & (0.554) \\
413 +\addlinespace[0.05cm]
414 +$\overline{QPV}_{t-22,ETF}$ & & $-$0.024** & & 0.060*** & & $-$0.017 & & $-$0.014 \\
415 + & & (0.036) & & (0.003) & & (0.463) & & (0.617) \\
416 +\addlinespace[0.2cm]
417 +
418 +\multicolumn{9}{l}{\textit{Panel B: Jump Component}} \\
419 +\addlinespace[0.1cm]
420 +$J_{t-1,NAV}$ & 0.306*** & 0.432*** & 0.426*** & 0.892*** & 0.510*** & 0.578*** & 0.092* & 0.090*** \\
421 + & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.000) & (0.075) & (0.000) \\
422 +\addlinespace[0.05cm]
423 +$\overline{J}_{t-5,NAV}$ & 0.435*** & & 0.509*** & & 0.468*** & & 0.219*** & \\
424 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
425 +\addlinespace[0.05cm]
426 +$\overline{J}_{t-22,NAV}$ & 0.190*** & & 0.234*** & & 0.161*** & & 0.346*** & \\
427 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
428 +\addlinespace[0.05cm]
429 +$J_{t-1,ETF}$ & 0.096*** & 0.077** & 0.031 & 0.011 & 0.028 & 0.051 & 0.128*** & 0.166*** \\
430 + & (0.000) & (0.019) & (0.290) & (0.689) & (0.313) & (0.106) & (0.000) & (0.000) \\
431 +\addlinespace[0.05cm]
432 +$\overline{J}_{t-5,ETF}$ & & 0.300*** & & 0.132** & & 0.184*** & & 0.401*** \\
433 + & & (0.000) & & (0.023) & & (0.000) & & (0.000) \\
434 +\addlinespace[0.05cm]
435 +$\overline{J}_{t-22,ETF}$ & & 0.250*** & & 0.028 & & 0.295*** & & 0.350*** \\
436 + & & (0.000) & & (0.455) & & (0.000) & & (0.000) \\
437 +\midrule
438 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
439 +\bottomrule
440 +\end{tabular}
441 +\begin{tablenotes}
442 +\small
443 +\item This table presents estimation results for the HAR-CJ-X model using 1-minute realized variance data. The model incorporates both continuous (quadratic power variation, QPV) and jump (J) components. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $QPV_{t-1}$ and $J_{t-1}$ represent the lagged daily components, while $\overline{QPV}_{t-5}$, $\overline{J}_{t-5}$, $\overline{QPV}_{t-22}$, and $\overline{J}_{t-22}$ are the corresponding weekly and monthly averages.
444 +\end{tablenotes}
445 +\end{threeparttable}
446 +\end{table}
447 +\end{landscape}
448 +
449 +
450 +\begin{landscape}
451 +\begin{table}[htbp]
452 +\centering
453 +\caption{HAR-CJ-X Model Estimates with 30-minute Realized Variance}
454 +\label{ch2:tab:HAR_CJ_30min}
455 +\begin{threeparttable}
456 +\footnotesize
457 +\begin{tabular}{@{}lcccccccc@{}}
458 +\toprule
459 + & \multicolumn{2}{c}{\textbf{Crude Oil}} & \multicolumn{2}{c}{\textbf{Gold}} & \multicolumn{2}{c}{\textbf{Silver}} & \multicolumn{2}{c}{\textbf{Natural Gas}} \\
460 +\cmidrule(lr){2-3} \cmidrule(lr){4-5} \cmidrule(lr){6-7} \cmidrule(lr){8-9}
461 + & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ & $RV_{t,NAV}$ & $RV_{t,ETF}$ \\
462 +\midrule
463 +\multicolumn{9}{l}{\textit{Panel A: Quadratic Power Variation}} \\
464 +\addlinespace[0.1cm]
465 +$QPV_{t-1,NAV}$ & 0.019 & $-$0.026 & 0.058*** & 0.067*** & 0.013 & 0.033 & 0.057*** & 0.027 \\
466 + & (0.535) & (0.421) & (0.005) & (0.003) & (0.596) & (0.235) & (0.008) & (0.278) \\
467 +\addlinespace[0.05cm]
468 +$\overline{QPV}_{t-5,NAV}$ & 0.029 & & $-$0.051** & & $-$0.008 & & 0.114*** & \\
469 + & (0.375) & & (0.044) & & (0.779) & & (0.000) & \\
470 +\addlinespace[0.05cm]
471 +$\overline{QPV}_{t-22,NAV}$ & $-$0.001 & & 0.001 & & $-$0.045* & & 0.156*** & \\
472 + & (0.975) & & (0.977) & & (0.098) & & (0.000) & \\
473 +\addlinespace[0.05cm]
474 +$QPV_{t-1,ETF}$ & $-$0.010 & 0.015 & $-$0.018 & $-$0.027 & $-$0.006 & 0.001 & $-$0.014 & 0.001 \\
475 + & (0.683) & (0.562) & (0.489) & (0.313) & (0.799) & (0.984) & (0.575) & (0.972) \\
476 +\addlinespace[0.05cm]
477 +$\overline{QPV}_{t-5,ETF}$ & & 0.055** & & $-$0.017 & & $-$0.001 & & 0.079*** \\
478 + & & (0.033) & & (0.517) & & (0.966) & & (0.001) \\
479 +\addlinespace[0.05cm]
480 +$\overline{QPV}_{t-22,ETF}$ & & 0.016 & & 0.234*** & & $-$0.035 & & 0.168*** \\
481 + & & (0.517) & & (0.000) & & (0.191) & & (0.000) \\
482 +\addlinespace[0.2cm]
483 +
484 +\multicolumn{9}{l}{\textit{Panel B: Jump Component}} \\
485 +\addlinespace[0.1cm]
486 +$J_{t-1,NAV}$ & 0.110** & 0.221*** & 0.054 & 0.206*** & 0.201*** & 0.216*** & $-$0.006 & 0.027 \\
487 + & (0.019) & (0.000) & (0.190) & (0.000) & (0.000) & (0.000) & (0.870) & (0.381) \\
488 +\addlinespace[0.05cm]
489 +$\overline{J}_{t-5,NAV}$ & 0.403*** & & 0.436*** & & 0.339*** & & 0.204*** & \\
490 + & (0.000) & & (0.000) & & (0.000) & & (0.000) & \\
491 +\addlinespace[0.05cm]
492 +$\overline{J}_{t-22,NAV}$ & 0.248*** & & 0.440*** & & 0.489*** & & 0.089 & \\
493 + & (0.000) & & (0.000) & & (0.000) & & (0.123) & \\
494 +\addlinespace[0.05cm]
495 +$J_{t-1,ETF}$ & 0.110*** & 0.043 & 0.009 & $-$0.059 & $-$0.023 & $-$0.080* & 0.016 & 0.007 \\
496 + & (0.000) & (0.181) & (0.772) & (0.183) & (0.428) & (0.077) & (0.608) & (0.823) \\
497 +\addlinespace[0.05cm]
498 +$\overline{J}_{t-5,ETF}$ & & 0.259*** & & 0.274*** & & 0.241*** & & 0.243*** \\
499 + & & (0.000) & & (0.000) & & (0.000) & & (0.000) \\
500 +\addlinespace[0.05cm]
501 +$\overline{J}_{t-22,ETF}$ & & 0.283*** & & $-$0.082* & & 0.532*** & & 0.088 \\
502 + & & (0.000) & & (0.058) & & (0.000) & & (0.224) \\
503 +\midrule
504 +Observations & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 & 3,935 \\
505 +\bottomrule
506 +\end{tabular}
507 +\begin{tablenotes}
508 +\small
509 +\item This table presents estimation results for the HAR-CJ-X model using 30-minute realized variance data. The model incorporates both continuous (quadratic power variation, QPV) and jump (J) components. The dependent variables are the realized variances for NAV and ETF prices of each commodity. Standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1\%, 5\%, and 10\% levels, respectively. $QPV_{t-1}$ and $J_{t-1}$ represent the lagged daily components, while $\overline{QPV}_{t-5}$, $\overline{J}_{t-5}$, $\overline{QPV}_{t-22}$, and $\overline{J}_{t-22}$ are the corresponding weekly and monthly averages.
510 +\end{tablenotes}
511 +\end{threeparttable}
512 +\end{table}
513 +\end{landscape}
514 +
515 +
516 +\begin{landscape}
517 +\begin{table}[htbp]
518 +\centering
519 +\caption{Bayesian Vector Autoregression Results: USO ETF and Net Asset Value}
520 +\label{ch2:tab:VAR_USO}
521 +\begin{threeparttable}
522 +\footnotesize
523 +\begin{tabular}{@{}lccccccc@{}}
524 +\toprule
525 + & \multicolumn{6}{c}{\textbf{Panel A: $\log(RV_{t,NAV})$}} \\
526 +\cmidrule(lr){2-7}
527 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
528 +\cmidrule(lr){6-7}
529 + & & & & & Lower & Upper \\
530 +\midrule
531 +$\log(RV_{t-1,NAV})$ & 0.570 & 0.024 & 0.000 & 0.570 & 0.524 & 0.617 \\
532 +$\log(RV_{t-2,NAV})$ & 0.261 & 0.021 & 0.000 & 0.261 & 0.221 & 0.300 \\
533 +$\log(RV_{t-1,ETF})$ & 0.074 & 0.020 & 0.000 & 0.074 & 0.034 & 0.113 \\
534 +$\log(RV_{t-2,ETF})$ & 0.016 & 0.018 & 0.000 & 0.016 & $-$0.019 & 0.051 \\
535 +\addlinespace[0.3cm]
536 +
537 + & \multicolumn{6}{c}{\textbf{Panel B: $\log(RV_{t,ETF})$}} \\
538 +\cmidrule(lr){2-7}
539 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
540 +\cmidrule(lr){6-7}
541 + & & & & & Lower & Upper \\
542 +\midrule
543 +$\log(RV_{t-1,NAV})$ & 0.286 & 0.028 & 0.000 & 0.286 & 0.231 & 0.343 \\
544 +$\log(RV_{t-2,NAV})$ & 0.193 & 0.025 & 0.000 & 0.193 & 0.144 & 0.241 \\
545 +$\log(RV_{t-1,ETF})$ & 0.294 & 0.024 & 0.000 & 0.294 & 0.246 & 0.342 \\
546 +$\log(RV_{t-2,ETF})$ & 0.106 & 0.022 & 0.000 & 0.106 & 0.064 & 0.148 \\
547 +\bottomrule
548 +\end{tabular}
549 +\begin{tablenotes}
550 +\small
551 +\item This table presents posterior statistics from a Bayesian Vector Autoregression (VAR) model analyzing the relationship between the USO ETF and its Net Asset Value (NAV). Panel A shows results for the NAV equation, while Panel B shows results for the ETF equation. The model includes two lags of both log realized variances. Mean represents the posterior mean, Std. Dev. is the posterior standard deviation, MCSE is the Monte Carlo standard error, Median is the posterior median, and the 95\% credible interval provides the range of plausible parameter values. All realized variance measures are constructed using 5-minute price data.
552 +\end{tablenotes}
553 +\end{threeparttable}
554 +\end{table}
555 +\end{landscape}
556 +
557 +\begin{landscape}
558 +\begin{table}[htbp]
559 +\centering
560 +\caption{Bayesian Vector Autoregression Results: GLD ETF and Net Asset Value}
561 +\label{ch2:tab:VAR_GLD}
562 +\begin{threeparttable}
563 +\footnotesize
564 +\begin{tabular}{@{}lccccccc@{}}
565 +\toprule
566 + & \multicolumn{6}{c}{\textbf{Panel A: $\log(RV_{t,NAV})$}} \\
567 +\cmidrule(lr){2-7}
568 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
569 +\cmidrule(lr){6-7}
570 + & & & & & Lower & Upper \\
571 +\midrule
572 +$\log(RV_{t-1,NAV})$ & 0.551 & 0.027 & 0.000 & 0.551 & 0.498 & 0.603 \\
573 +$\log(RV_{t-2,NAV})$ & 0.248 & 0.023 & 0.000 & 0.248 & 0.203 & 0.293 \\
574 +$\log(RV_{t-1,ETF})$ & $-$0.010 & 0.023 & 0.000 & $-$0.010 & $-$0.055 & 0.034 \\
575 +$\log(RV_{t-2,ETF})$ & 0.038 & 0.019 & 0.000 & 0.038 & 0.000 & 0.076 \\
576 +\addlinespace[0.3cm]
577 +
578 + & \multicolumn{6}{c}{\textbf{Panel B: $\log(RV_{t,ETF})$}} \\
579 +\cmidrule(lr){2-7}
580 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
581 +\cmidrule(lr){6-7}
582 + & & & & & Lower & Upper \\
583 +\midrule
584 +$\log(RV_{t-1,NAV})$ & 0.349 & 0.033 & 0.000 & 0.348 & 0.285 & 0.413 \\
585 +$\log(RV_{t-2,NAV})$ & 0.276 & 0.028 & 0.000 & 0.276 & 0.222 & 0.331 \\
586 +$\log(RV_{t-1,ETF})$ & 0.155 & 0.028 & 0.000 & 0.155 & 0.100 & 0.208 \\
587 +$\log(RV_{t-2,ETF})$ & 0.048 & 0.023 & 0.000 & 0.048 & 0.003 & 0.095 \\
588 +\bottomrule
589 +\end{tabular}
590 +\begin{tablenotes}
591 +\small
592 +\item This table presents posterior statistics from a Bayesian Vector Autoregression (VAR) model analyzing the relationship between the GLD ETF and its Net Asset Value (NAV). Panel A shows results for the NAV equation, while Panel B shows results for the ETF equation. The model includes two lags of both log realized variances. Mean represents the posterior mean, Std. Dev. is the posterior standard deviation, MCSE is the Monte Carlo standard error, Median is the posterior median, and the 95\% credible interval provides the range of plausible parameter values. All realized variance measures are constructed using 5-minute price data.
593 +\end{tablenotes}
594 +\end{threeparttable}
595 +\end{table}
596 +\end{landscape}
597 +
598 +\begin{landscape}
599 +\begin{table}[htbp]
600 +\centering
601 +\caption{Bayesian Vector Autoregression Results: SLV ETF and Net Asset Value}
602 +\label{ch2:tab:VAR_SLV}
603 +\begin{threeparttable}
604 +\footnotesize
605 +\begin{tabular}{@{}lccccccc@{}}
606 +\toprule
607 + & \multicolumn{6}{c}{\textbf{Panel A: $\log(RV_{t,NAV})$}} \\
608 +\cmidrule(lr){2-7}
609 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
610 +\cmidrule(lr){6-7}
611 + & & & & & Lower & Upper \\
612 +\midrule
613 +$\log(RV_{t-1,NAV})$ & 0.619 & 0.028 & 0.000 & 0.619 & 0.563 & 0.674 \\
614 +$\log(RV_{t-2,NAV})$ & 0.214 & 0.024 & 0.000 & 0.214 & 0.168 & 0.261 \\
615 +$\log(RV_{t-1,ETF})$ & $-$0.045 & 0.025 & 0.000 & $-$0.045 & $-$0.093 & 0.005 \\
616 +$\log(RV_{t-2,ETF})$ & 0.058 & 0.021 & 0.000 & 0.058 & 0.018 & 0.100 \\
617 +\addlinespace[0.3cm]
618 +
619 + & \multicolumn{6}{c}{\textbf{Panel B: $\log(RV_{t,ETF})$}} \\
620 +\cmidrule(lr){2-7}
621 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
622 +\cmidrule(lr){6-7}
623 + & & & & & Lower & Upper \\
624 +\midrule
625 +$\log(RV_{t-1,NAV})$ & 0.376 & 0.033 & 0.000 & 0.377 & 0.310 & 0.440 \\
626 +$\log(RV_{t-2,NAV})$ & 0.202 & 0.028 & 0.000 & 0.202 & 0.147 & 0.256 \\
627 +$\log(RV_{t-1,ETF})$ & 0.179 & 0.029 & 0.000 & 0.179 & 0.123 & 0.236 \\
628 +$\log(RV_{t-2,ETF})$ & 0.078 & 0.024 & 0.000 & 0.078 & 0.031 & 0.126 \\
629 +\bottomrule
630 +\end{tabular}
631 +\begin{tablenotes}
632 +\small
633 +\item This table presents posterior statistics from a Bayesian Vector Autoregression (VAR) model analyzing the relationship between the SLV ETF and its Net Asset Value (NAV). Panel A shows results for the NAV equation, while Panel B shows results for the ETF equation. The model includes two lags of both log realized variances. Mean represents the posterior mean, Std. Dev. is the posterior standard deviation, MCSE is the Monte Carlo standard error, Median is the posterior median, and the 95\% credible interval provides the range of plausible parameter values. All realized variance measures are constructed using 5-minute price data.
634 +\end{tablenotes}
635 +\end{threeparttable}
636 +\end{table}
637 +\end{landscape}
638 +
639 +\begin{landscape}
640 +\begin{table}[htbp]
641 +\centering
642 +\caption{Bayesian Vector Autoregression Results: UNG ETF and Net Asset Value}
643 +\label{ch2:tab:VAR_UNG}
644 +\begin{threeparttable}
645 +\footnotesize
646 +\begin{tabular}{@{}lccccccc@{}}
647 +\toprule
648 + & \multicolumn{6}{c}{\textbf{Panel A: $\log(RV_{t,NAV})$}} \\
649 +\cmidrule(lr){2-7}
650 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
651 +\cmidrule(lr){6-7}
652 + & & & & & Lower & Upper \\
653 +\midrule
654 +$\log(RV_{t-1,NAV})$ & 0.392 & 0.022 & 0.000 & 0.392 & 0.349 & 0.437 \\
655 +$\log(RV_{t-2,NAV})$ & 0.231 & 0.020 & 0.000 & 0.231 & 0.192 & 0.272 \\
656 +$\log(RV_{t-1,ETF})$ & 0.080 & 0.021 & 0.000 & 0.080 & 0.038 & 0.121 \\
657 +$\log(RV_{t-2,ETF})$ & 0.157 & 0.019 & 0.000 & 0.157 & 0.119 & 0.194 \\
658 +\addlinespace[0.3cm]
659 +
660 + & \multicolumn{6}{c}{\textbf{Panel B: $\log(RV_{t,ETF})$}} \\
661 +\cmidrule(lr){2-7}
662 +Variable & Mean & Std. Dev. & MCSE & Median & \multicolumn{2}{c}{95\% Credible Interval} \\
663 +\cmidrule(lr){6-7}
664 + & & & & & Lower & Upper \\
665 +\midrule
666 +$\log(RV_{t-1,NAV})$ & 0.130 & 0.024 & 0.000 & 0.130 & 0.083 & 0.177 \\
667 +$\log(RV_{t-2,NAV})$ & 0.165 & 0.022 & 0.000 & 0.165 & 0.124 & 0.208 \\
668 +$\log(RV_{t-1,ETF})$ & 0.339 & 0.023 & 0.000 & 0.339 & 0.294 & 0.383 \\
669 +$\log(RV_{t-2,ETF})$ & 0.240 & 0.020 & 0.000 & 0.240 & 0.200 & 0.280 \\
670 +\bottomrule
671 +\end{tabular}
672 +\begin{tablenotes}
673 +\small
674 +\item This table presents posterior statistics from a Bayesian Vector Autoregression (VAR) model analyzing the relationship between the UNG ETF and its Net Asset Value (NAV). Panel A shows results for the NAV equation, while Panel B shows results for the ETF equation. The model includes two lags of both log realized variances. Mean represents the posterior mean, Std. Dev. is the posterior standard deviation, MCSE is the Monte Carlo standard error, Median is the posterior median, and the 95\% credible interval provides the range of plausible parameter values. All realized variance measures are constructed using 5-minute price data.
675 +\end{tablenotes}
676 +\end{threeparttable}
677 +\end{table}
678 +\end{landscape}
\ No newline at end of file
added these-ulaval/chapitre3/01_introduction.tex +21 −0
@@ -0,0 +1,21 @@
1 +\section{Introduction}
2 +
3 +We study how the textual content of Federal Reserve communications affects asset returns and market volatility at high frequency. The Federal Open Market Committee (FOMC)---composed of the seven members of the Board of Governors, the president of the Federal Reserve Bank of New York, and four of the remaining eleven Reserve Bank presidents on a rotating basis---meets eight times per year to set the target federal funds rate. Since 1994, the Committee has released a public statement after each meeting; since 2011, the Chair has also held post-meeting press conferences. Other communications include meeting minutes (released three weeks later) and the Summary of Economic Projections with individual rate forecasts (the ``dot plot''). Our analysis focuses exclusively on post-meeting statements, which are the first piece of information available to market participants at a precise, scheduled time. These statements affect U.S. asset prices, global capital flows, and exchange rates \citep{blinder2008,campbell2012,wongswan2009,ehrmann2011}.
4 +
5 +Our central research question is: \textit{Do different semantic dimensions of FOMC communications---specifically policy tone and informational novelty---affect financial market returns and volatility through distinct economic channels?} If tone captures the directional policy signal (hawkish vs.\ dovish) while novelty captures how much genuinely new information the statement contains, standard asset pricing theory predicts that these dimensions should have different effects. In the standard framework, asset prices equal expected future cash flows discounted at a rate that reflects policy expectations \citep{fama1970,bernanke2005}, so the directional content of the statement---tone---should move expected cash flows and discount rates, and hence returns. Volatility, by contrast, reflects uncertainty and disagreement about how to interpret new information \citep{veldkamp2011,patton2013}, so the amount of new language---novelty---should affect information processing complexity and interpretive uncertainty, and hence volatility. We test this prediction using 1-minute price data for 7 futures contracts---the E-mini S\&P 500 (ES), futures on the Chicago Board Options Exchange Volatility Index (VIX futures, VX), the 10-year (ZN) and 5-year (ZF) Treasury notes, the U.S. Dollar Index (DX), crude oil (CL), and gold (GC)---across 148 FOMC events from 2008 to 2025, estimating panel minute-level regressions, event-level regressions, and local projection impulse response functions \citep{jorda2005}. Our identification strategy exploits the exogenous timing of announcements and is validated by pre-announcement placebo tests \citep{andersen2003micro,andersen2007real}.
6 +
7 +To measure these dimensions, we construct two primary variables: (1) a tone measure capturing the hawkish-dovish spectrum, and (2) a novelty measure quantifying the semantic distance between consecutive FOMC communications. Concretely, we proceed in three steps. First, we convert each statement into a numerical vector (an \textit{embedding}) using two language models of different sizes---MiniLM (which produces 384-dimensional vectors) and BERT (768-dimensional vectors)---and combine their outputs; using two architecturally distinct models guards against findings that are artifacts of a single model. Second, because generic language models are not trained on central bank language, we adapt both models to the corpus of FOMC statements in two stages: an unsupervised stage based on the Transformer-based Sequential Denoising Auto-Encoder (TSDAE) of \citet{wang2021tsdae}, which teaches the models Fed-specific vocabulary, followed by a supervised contrastive learning stage on pairs of statements with known relationships. Third, we locate each statement on the hawkish-dovish axis by measuring its proximity to reference statements at the two extremes of the spectrum. Whereas previous studies select these reference statements by hand---introducing researcher degrees of freedom, since results may depend on which dates the researcher chooses---we select them algorithmically using principal component analysis (PCA), making the construction of the semantic axes fully reproducible. Novelty is then measured as the distance between the embeddings of consecutive statements.
8 +
9 +FOMC announcements are widely seen as the most important scheduled monetary policy events in global financial markets, as they contribute substantially to shaping expectations about the future path of the U.S. economy. Because markets are forward-looking, surprising announcements are typically followed by sizable adjustments across several markets \citep{bernanke2005,blinder2008}. \citet{savor2014} show that a disproportionate share of the equity risk premium is earned on macroeconomic announcement days, with FOMC days being particularly important. \citet{brusa2015,brusa2019} find that average stock returns and Sharpe ratios on FOMC days are 20--40 times higher than on non-announcement days. \citet{lucca2012} document a systematic pre-FOMC drift in equity prices. \citet{nakamura2018} and \citet{jarocinski2020} decompose FOMC surprises into policy shocks and information shocks and show that these have different effects on asset prices.
10 +
11 +While earlier research focuses exclusively on the measurable policy surprise content of FOMC announcements, typically extracted from federal funds futures \citep{kuttner2001,bernanke2005}, a more recent strand of literature has begun to apply textual analysis to extract subtler information from the language itself \citep{hansen2017,shapiro2019,schmeling2019,gorodnichenko2023}. This is an important step forward, given that the early contributions identify the effect of unexpected rate changes but ignore the qualitative content of the accompanying statement---forward guidance, risk assessments, and descriptions of economic conditions---which \citet{gurkaynak2005} show can move long-term yields even when the rate decision is fully anticipated. Within this new strand of literature, however, most studies use daily data \citep{shapiro2019,schmeling2019,eklund2024}, which cannot separate the immediate market reaction to the statement from confounding information that arrives later in the day. Furthermore, the standard approach treats the statement as a one-dimensional object (hawkish vs.\ dovish), ignoring other dimensions such as how much the statement departs from prior language. In this paper, we address both limitations by using 1-minute data and by decomposing statements into two distinct dimensions: directional tone and informational novelty.
12 +
13 +Textual analysis of financial communications has progressed from keyword counting \citep{bligh2008} and dictionary-based sentiment scoring \citep{loughran2011} to transformer-based language models that capture context and word order \citep{gentzkow2019,kenton2019,araci2019}. Within central bank communication, \citet{hayo2010,apel2012} develop systematic measures of FOMC tone and show that these predict monetary policy expectations. \citet{shapiro2019,hansen2017} use natural language processing (NLP) classifiers to place FOMC statements on the hawkish-dovish spectrum, and \citet{eklund2024} find that hawkish sentiment measures predict subsequent realizations of the consumer price index (CPI). However, this literature treats FOMC communication as essentially one-dimensional---hawkish versus dovish---and has not examined whether other dimensions of the text, such as how much the language departs from the prior statement, have independent effects on asset prices.
14 +
15 +We define informational novelty as the degree to which a new FOMC statement departs from the previous one, measured as the cosine distance between their sentence-transformer embeddings. It is useful to distinguish this concept from three related constructs. \textit{Tone} or \textit{sentiment} refers to the directional content of text (hawkish vs.\ dovish) and has been studied using dictionaries \citep{loughran2011} and classifiers \citep{shapiro2019}. \textit{Economic policy uncertainty} (EPU), as measured by \citet{baker2016}, is a macro-level index based on newspaper coverage, tax code provisions, and forecaster disagreement. \textit{Monetary policy surprises}, in the tradition of \citet{kuttner2001}, are the unexpected component of the rate decision itself. Our novelty measure differs from all three: it captures changes in \textit{how} information is presented between consecutive statements, independent of both directional stance and the rate decision. A statement can be highly novel yet neutral in tone (e.g., introducing new forward guidance language), or low in novelty yet accompanied by a large rate surprise.
16 +
17 +Why should novelty matter for asset prices? Under rational expectations \citep{muth1961,fama1970}, only genuinely new information should move prices; repetitive content should already be priced in. But theory also suggests that novel and familiar information may affect returns and volatility differently. Returns respond to revisions in expected cash flows and discount rates---that is, to directional content. Volatility may respond more to disagreement among market participants about the interpretation of new language or to the processing cost of unfamiliar information \citep{patton2013,hu2019,veldkamp2011}. Consistent with this distinction, \citet{manela2017} show that their news-based implied volatility index (NVIX) predicts future market volatility.
18 +
19 +Our principal finding is that tone is associated with directional asset returns while novelty is associated with volatility, consistent with two distinct transmission channels. We make two main contributions. First, we develop a decomposition framework for FOMC statements based on a dual-model ensemble (MiniLM and BERT) fine-tuned on the FOMC corpus with data-driven PCA-based reference selection. Second, we provide the first high-frequency evidence on the differential effects of tone versus novelty, using local projection impulse response functions across 7 futures contracts and 148 FOMC events. In addition, we show that these two dimensions operate through distinct channels: tone is associated with fundamental valuations (equity returns building to about 12 basis points per standard deviation within two hours), while novelty is associated with uncertainty resolution (the stance--novelty interaction on VIX futures has a $t$-statistic of $-5.06$ and persists from 5 to 120 minutes). Our results are robust to five independent inference methods (Newey--West, wild bootstrap, clustered standard errors, quantile regression, and permutation tests) and are validated through pre-announcement placebo tests.
20 +
21 +The remainder of this paper proceeds as follows. Section~\ref{ch3:sec:data} describes data construction and processing. Section~\ref{ch3:sec:methodology} presents our theoretical framework, propositions, and empirical methodology, including our identification strategy and econometric specifications. Section~\ref{ch3:sec:results} reports our main empirical findings. Section~\ref{ch3:sec:conclusion} concludes with implications for monetary policy transmission and central bank communication strategy.
added these-ulaval/chapitre3/02_data.tex +66 −0
@@ -0,0 +1,66 @@
1 +\section{Data}
2 +\label{ch3:sec:data}
3 +
4 +\subsection{FOMC Statement Collection}
5 +
6 +We collect 451 FOMC communications from the Federal Reserve Board website, spanning 2000--2025. After filtering to policy statements only, our textual corpus contains $N = 217$ FOMC statements, each averaging 3,840 characters. We use the broader 2000--2025 textual sample to train the embedding models and construct the semantic axes, which requires examples spanning multiple policy regimes. For the high-frequency event study, we focus on 148 FOMC events from 2008 to 2025. The 2008 start date reflects two considerations. First, reliable 1-minute futures data became consistently available following the shift to predominantly electronic trading on CME Globex in 2006--2007; pre-2008 high-frequency data for several of our contracts contains gaps and timing errors that would compromise our minute-level identification. Second, the 2008--2025 window encompasses sufficient variation in monetary policy regimes---the zero lower bound (2008--2015), quantitative easing programs, post-crisis normalization (2016--2019), the pandemic response (2020--2021), and the inflation-driven tightening cycle (2022--2025)---across three Fed Chairs (Bernanke, Yellen, Powell) to identify communication effects without sacrificing data quality \citep{swanson2014,gurtler2010,blinder2008}. The broader 2000--2025 textual sample used for model fine-tuning starts in 2000 because the FOMC adopted its current practice of issuing post-meeting statements with substantive policy language beginning in 1999--2000; earlier statements were shorter and formulaic, providing too little variation for our NLP models to learn from.
7 +
8 +All statements are obtained directly from Federal Reserve official releases archived on the Board of Governors website. We focus exclusively on post-meeting statements rather than meeting minutes, transcripts, or other Fed communications because these statements represent the information that is immediately available to market participants at precise announcement times \citep{gurkaynak2005}. This temporal precision is necessary for high-frequency identification \citep{andersen2003micro,rosa2013}.
9 +
10 +We preprocess statements following \citet{gentzkow2019}: we remove headers, footers, voting records, and administrative content, keeping only policy-relevant text. We account for structural breaks in statement format (e.g., the lengthening of statements post-2008) to avoid introducing spurious variation in our semantic measures \citep{hansen2017}. We cross-reference all statements with the Fed's public archives to verify completeness.
11 +
12 +\subsection{Dual-Model Embedding Generation}
13 +
14 +We convert each FOMC statement into a numerical vector using two sentence transformer models, adapted to the FOMC corpus through the training procedure described in Section~\ref{ch3:sec:nlp_framework}.
15 +
16 +\subsubsection{Model Architectures}
17 +
18 +\begin{enumerate}
19 +\item \textbf{MiniLM} (\texttt{all-MiniLM-L6-v2}): 33M parameters, 384-dimensional embeddings.
20 +\item \textbf{BERT} (\texttt{bert-base-uncased} with mean pooling): 110M parameters, 768-dimensional embeddings. A larger model that captures finer contextual distinctions in policy language \citep{kenton2019}.
21 +\end{enumerate}
22 +
23 +We train both models on the FOMC corpus using the two-stage procedure (TSDAE domain adaptation, then MNRL contrastive learning) described in Section~\ref{ch3:sec:nlp_framework}. The training hyperparameters are: batch size 32 (MiniLM) / 16 (BERT), learning rate $1 \times 10^{-5}$, 5 (MiniLM) / 4 (BERT) supervised epochs with cosine warmup schedule (10\% warmup ratio), and approximately 4,000 training examples per model from six pairing strategies.
24 +
25 +\subsubsection{Construction of Training Pairs}
26 +
27 +We construct training pairs using six strategies: (1) overlapping text segments from the same document; (2) consecutive sentence pairs within a document; (3) hawk--dove contrastive pairs classified through keyword banks; (4) temporal proximity pairs from consecutive FOMC meetings; (5) topic-based pairs sharing the same Fed topic (monetary policy, inflation, employment, etc.); and (6) key phrase paraphrases containing identical policy phrases (\textit{``maintain the target range''}, \textit{``decided to raise''}, etc.). The diversity of pairing strategies prevents the trained models from learning only a single sentiment dimension.
28 +
29 +\subsubsection{Data-Driven Reference Selection}
30 +
31 +Previous studies select reference dates manually or construct counterfactual statements by hand. We instead use the hybrid PCA--percentile approach described in Definition~\ref{ch3:def:pca_axes} to identify semantically extreme statements algorithmically. This procedure eliminates researcher degrees of freedom and produces more stable centroids (5 reference statements per pole versus 2 in typical manual selection). All semantic axes exceed the 0.04 minimum separation threshold. Ensemble separations are: total policy stance (0.302), risk assessment (0.274), policy communication (0.248), and economic priority (0.252).
32 +
33 +We validate the embeddings in three ways. First, we compare embedding-based tone rankings with expert classifications and verify that known dovish and hawkish statements sort correctly. Second, we measure inter-model agreement: mean ensemble confidence is 0.837 for novelty and 0.562--0.806 for tone across axes. Third, we verify that the PCA axis assignments produce intuitive orderings.
34 +
35 +\subsection{High-Frequency Financial Data Construction}
36 +
37 +We use 1-minute OHLCV data for seven futures contracts spanning the major asset classes relevant to monetary policy transmission:
38 +
39 +\begin{itemize}
40 +\item \textbf{ES}: E-mini S\&P 500 (equity benchmark)---responds to policy communications through discount rate effects, growth expectations, and risk premium adjustments.
41 +\item \textbf{VX}: VIX Futures (volatility)---captures implied volatility and uncertainty resolution, providing the most direct measure of information processing effects.
42 +\item \textbf{ZN}: 10-Year Treasury Note---reflects the interaction between policy expectations and term premium effects.
43 +\item \textbf{ZF}: 5-Year Treasury Note---primarily captures expectations about near-term policy rate changes.
44 +\item \textbf{DX}: Dollar Index---reflects relative monetary policy stances and international spillover effects.
45 +\item \textbf{CL}: Crude Oil WTI---reflects both inflation expectations and growth concerns influenced by monetary policy.
46 +\item \textbf{GC}: Gold---a safe-haven asset whose demand falls when accommodative communication triggers a rotation toward risk assets, and rises with policy-induced uncertainty.
47 +\end{itemize}
48 +
49 +The data span 148 FOMC events (approximately 8 per year from 2008 to 2025). For each event, we extract a $\pm 120$ minute window around the 14:00~ET announcement time. We interpolate the raw data to a regular 1-minute grid with forward-filling of short gaps ($\leq 5$ minutes) and returns recomputed on the regularized grid. We require at least 80\% valid observations per rolling window.
50 +
51 +For each instrument, we construct minute-by-minute log returns as:
52 +\begin{equation}
53 +r_{i,m}^{(j)} = \log\left(P_{i,m}^{(j)}\right) - \log\left(P_{i,m-1}^{(j)}\right)
54 +\end{equation}
55 +where $P_{i,m}^{(j)}$ is the price of asset $i$ at minute $m$ relative to FOMC event $j$. We align timestamps across instruments and exchanges, handle market closures and trading halts, and filter outliers following standard microstructure procedures.
56 +
57 +\subsection{Event Window Specification and Temporal Alignment}
58 +
59 +Our primary analysis employs a 45-minute event window following each FOMC announcement, spanning from the announcement time ($t=0$) to 45 minutes after ($t=+45$). We choose this window length for two reasons.
60 +
61 +First, most of the announcement effect occurs within 15 minutes of the release, but equity responses to tone continue to develop through 45 minutes post-announcement. Second, press conferences typically begin 30 minutes after the statement, so a 45-minute window includes only the first 15 minutes of the press conference, limiting contamination.
62 +
63 +We also implement several alternative window specifications as robustness checks. A narrow 15-minute post-announcement window provides very clean identification but with reduced statistical power. An extended 60-minute post-announcement window captures longer adjustment dynamics but with increased contamination risk, particularly from press conferences that typically begin 30 minutes after the statement release.
64 +
65 +FOMC statements are typically released at 2:00 PM Eastern Time. We verify actual release times using Federal Reserve timestamps, news services, and market data providers, and correct any timing discrepancies before aligning with price data.
66 +
added these-ulaval/chapitre3/03_methodology.tex +375 −0
@@ -0,0 +1,375 @@
1 +\section{Methodology}
2 +\label{ch3:sec:methodology}
3 +
4 +This section describes our NLP framework for measuring tone and novelty (Section~\ref{ch3:sec:nlp_framework}), presents the econometric specifications we use to estimate the effects of these measures on asset returns and volatility (Section~\ref{ch3:sec:econometric}), and states the testable hypotheses together with the specific coefficient restrictions that each one implies (Section~\ref{ch3:sec:hypotheses}).
5 +
6 +\subsection{Theoretical Framework}
7 +\label{ch3:sec:nlp_framework}
8 +
9 +\subsubsection{Conceptual Foundation}
10 +
11 +We decompose each FOMC statement into two dimensions: policy tone (hawkish vs.\ dovish) and informational novelty (distance from the previous statement). To illustrate why context-aware methods are needed for this decomposition, consider the difference between dictionary-based and machine learning approaches. A dictionary approach aggregates sentiment as:
12 +\begin{equation}
13 +S_{\text{dict}} = \frac{1}{N} \sum_{i=1}^{N} \mathbb{1}_{[\text{word}_i \in D_{\text{pos}}]} - \mathbb{1}_{[\text{word}_i \in D_{\text{neg}}]}
14 +\end{equation}
15 +where $D_{\text{pos}}$ and $D_{\text{neg}}$ are predetermined word lists. The dictionary approach counts words one at a time, without regard to their position or neighbors. A transformer-based model instead produces $S_{\text{ML}} = f_{\theta}(x_1, x_2, \ldots, x_N; C)$, where $f_{\theta}$ is a neural network with learned parameters $\theta$, and $x_i$ are token representations that depend on surrounding words and document-level context $C$. To see why context matters, note that ``the policy remains accommodative'' and ``the policy is no longer accommodative'' contain the same keyword (``accommodative'') but have opposite meanings; a dictionary assigns similar scores to both, while a transformer assigns opposite scores.
16 +
17 +Our approach builds on three literatures. The information processing literature predicts that markets react more strongly to genuinely new information than to repetition \citep{grossman1980,fama1970}. The central bank communication literature shows that policy statements move asset prices independently of rate decisions \citep{gurkaynak2005,campbell2012}. The market microstructure literature demonstrates that high-frequency data can isolate the real-time incorporation of new information into prices \citep{andersen2003micro,rosa2013}.
18 +
19 +\begin{definition}[FOMC Statement Representation]
20 +Each FOMC statement $F_t$ released at time $t$ is mapped to a vector $\mathbf{f}_t \in \mathbb{R}^d$ in $d$-dimensional semantic space through domain-specific transformer embeddings. We employ two architecturally distinct models: MiniLM ($d = 384$) and BERT ($d = 768$), combined into a separation-weighted ensemble.
21 +\end{definition}
22 +
23 +Each model maps a statement to a real-valued vector in which semantically similar texts (e.g., ``accommodative'' and ``easy policy'') are represented by nearby points, enabling cosine similarity comparisons and PCA. Unlike bag-of-words representations, where each coordinate is a word count and most are zero, these vectors are dense: every coordinate carries information. The output dimensionality ($d = 384$ for MiniLM, $d = 768$ for BERT) is fixed by the model architecture---MiniLM-L6-v2 has 6 layers and 384 hidden units per layer; BERT-base has 12 layers and 768 hidden units \citep{kenton2019}---and is not a hyperparameter we select. We adapt both architectures to the FOMC domain using a two-stage training procedure.
24 +
25 +\paragraph{Stage 1: TSDAE Unsupervised Pre-Training.} We first apply Transformer-based Sequential Denoising Auto-Encoder (TSDAE) \citep{wang2021tsdae} to adapt each model to the FOMC domain vocabulary:
26 +\begin{equation}
27 +\mathcal{L}_{\text{TSDAE}} = - \sum_{i=1}^{N} \log P\bigl(x_i \mid \tilde{x}_i; \theta_{\text{enc}}, \theta_{\text{dec}}\bigr),
28 +\end{equation}
29 +where $\tilde{x}_i$ is a corrupted version of sentence $x_i$ (token deletion with probability 0.6). This unsupervised stage teaches domain-specific vocabulary and structure without requiring labeled pairs, training for 3 epochs (MiniLM) or 2 epochs (BERT) with learning rate $3 \times 10^{-5}$.
30 +
31 +\paragraph{Stage 2: Supervised Contrastive Learning with MNRL.} We then fine-tune using Multiple Negatives Ranking Loss (MNRL):
32 +\begin{equation}
33 +\mathcal{L}_{\text{MNRL}} = - \frac{1}{B} \sum_{i=1}^{B} \log \frac{\exp\bigl(\text{sim}(a_i, p_i) / \tau\bigr)}{\sum_{j=1}^{B} \exp\bigl(\text{sim}(a_i, p_j) / \tau\bigr)},
34 +\end{equation}
35 +where $(a_i, p_i)$ are positive pairs, $\text{sim}(\cdot, \cdot)$ denotes cosine similarity, $\tau$ is the temperature parameter, and $B$ is the batch size. In-batch negatives provide contrastive signal without explicit negative sampling. We construct approximately 4,000 training pairs per model using the six pairing strategies described in Section~\ref{ch3:sec:data}.
36 +
37 +\paragraph{Dual-Model Ensemble.} The two models are combined using separation-weighted averaging:
38 +\begin{equation}
39 +\widehat{\text{Tone}}_{t,k}^{\text{ens}} = w_k^{\text{MiniLM}} \cdot \text{Tone}_{t,k}^{\text{MiniLM}} + w_k^{\text{BERT}} \cdot \text{Tone}_{t,k}^{\text{BERT}},
40 +\end{equation}
41 +where the weights are proportional to axis separation quality:
42 +\begin{equation}
43 +w_k^{m} = \frac{\text{Sep}_k^{m}}{\text{Sep}_k^{\text{MiniLM}} + \text{Sep}_k^{\text{BERT}}}, \quad m \in \{\text{MiniLM}, \text{BERT}\}.
44 +\end{equation}
45 +For each statement, we measure inter-model confidence as:
46 +\begin{equation}
47 +\text{Conf}_{t,k} = \frac{1}{2}\bigl[\mathbb{1}[\text{sign}(\text{Tone}_{t,k}^{\text{M}}) = \text{sign}(\text{Tone}_{t,k}^{\text{B}})] + (1 - |\text{Tone}_{t,k}^{\text{M}} - \text{Tone}_{t,k}^{\text{B}}| / \max_j |\text{Tone}_{j,k}^{\text{M}} - \text{Tone}_{j,k}^{\text{B}}|)\bigr].
48 +\end{equation}
49 +Mean ensemble confidence is 0.837 for novelty and ranges from 0.562 to 0.806 for tone across axes. This dual-model approach reduces model-specific biases: MiniLM (33M parameters, 384 dimensions) provides computational efficiency and strong separation on policy stance, while BERT (110M parameters, 768 dimensions) captures richer contextual relationships. The ensemble achieves balanced performance across all semantic dimensions.
50 +
51 +\subsubsection{Semantic Dimensions}
52 +
53 +\begin{definition}[Semantic Similarity Measure]
54 +The semantic similarity between text vectors $\mathbf{u}$ and $\mathbf{v}$ is quantified using cosine similarity:
55 +\begin{equation}
56 +\text{sim}(\mathbf{u}, \mathbf{v}) = \frac{\mathbf{u} \cdot \mathbf{v}}{||\mathbf{u}||_2 \cdot ||\mathbf{v}||_2} = \frac{\sum_{i=1}^{d} u_i v_i}{\sqrt{\sum_{i=1}^{d} u_i^2} \cdot \sqrt{\sum_{i=1}^{d} v_i^2}}
57 +\end{equation}
58 +\end{definition}
59 +
60 +The cosine similarity measure ranges from $-1$ to $1$, where values approaching $1$ indicate high semantic similarity, values near $0$ indicate orthogonal or unrelated content, and values approaching $-1$ indicate semantic opposition. This metric is particularly well-suited for high-dimensional text embeddings because it captures angular distance between vectors while being invariant to vector magnitude differences that might arise from statement length variations.
61 +
62 +\begin{definition}[Statement Novelty]
63 +The informational novelty of statement $F_t$ relative to the immediately preceding statement is defined as:
64 +\begin{equation}
65 +\text{Novelty}_t = 1 - \text{sim}(\mathbf{f}_t, \mathbf{f}_{t-1})
66 +\end{equation}
67 +\end{definition}
68 +
69 +This novelty measure captures the degree to which current policy communications represent genuine departures from recent messaging patterns. Higher novelty values indicate statements containing substantively new informational content relative to recent communications, while lower values suggest continuity with established messaging. The measure is bounded between $0$ and $2$, where $0$ indicates identical statements and $2$ indicates maximally opposed statements.
70 +
71 +To construct our policy tone measure, we develop a data-driven approach based on PCA-based semantic axis construction that replaces manually selected reference dates with an algorithmic procedure, eliminating researcher degrees of freedom.
72 +
73 +\begin{definition}[Data-Driven Semantic Axis Construction]
74 +\label{ch3:def:pca_axes}
75 +Let $\mathbf{E} \in \mathbb{R}^{N \times d}$ be the L2-normalized embedding matrix. We compute the first $K = 8$ principal components:
76 +\begin{equation}
77 +\mathbf{E} = \mathbf{U} \boldsymbol{\Sigma} \mathbf{V}^\top + \boldsymbol{\epsilon}.
78 +\end{equation}
79 +For each semantic dimension $k \in \{1, \ldots, 4\}$ (policy stance, risk assessment, policy communication, economic priority), we define keyword banks $\mathcal{K}_k^+$ (positive pole) and $\mathcal{K}_k^-$ (negative pole), and compute a keyword differential score:
80 +\begin{equation}
81 +\Delta_i^{(k)} = \sum_{w \in \mathcal{K}_k^+} \mathbb{1}[w \in \text{text}_i] - \sum_{w \in \mathcal{K}_k^-} \mathbb{1}[w \in \text{text}_i].
82 +\end{equation}
83 +We match each semantic axis to the PC maximizing the absolute Pearson correlation:
84 +\begin{equation}
85 +j^*(k) = \arg\max_{j \in \{1,\ldots,K\}} \bigl|\text{Corr}(\mathbf{v}_j, \boldsymbol{\Delta}^{(k)})\bigr|.
86 +\end{equation}
87 +\end{definition}
88 +
89 +The reference selection uses a hybrid seed-and-percentile approach: (1) use 3 expert-selected dates per pole to define an initial axis direction $\hat{\mathbf{d}}_k = \bar{\mathbf{e}}_{\text{pos}} - \bar{\mathbf{e}}_{\text{neg}}$; (2) project all embeddings onto this direction: $s_i = \mathbf{e}_i^\top \hat{\mathbf{d}}_k / \|\hat{\mathbf{d}}_k\|$; (3) select $M = 5$ statements with highest and lowest projection scores; (4) iteratively refine by recomputing the axis direction from selected references and re-selecting. This yields 5 reference statements per pole (compared to 2 in manual selection), producing more stable centroids.
90 +
91 +For each axis $k$, we measure discrimination quality using the separation metric:
92 +\begin{equation}
93 +\text{Sep}_k = 1 - \cos(\bar{\mathbf{c}}_k^+, \bar{\mathbf{c}}_k^-),
94 +\end{equation}
95 +where $\bar{\mathbf{c}}_k^+, \bar{\mathbf{c}}_k^-$ are the L2-normalized positive and negative centroids. All axes achieve separations well above the 0.04 minimum threshold, with ensemble separations ranging from 0.248 to 0.302.
96 +
97 +\begin{definition}[Policy Tone]
98 +The directional tone of statement $F_t$ on semantic axis $k$ is measured as:
99 +\begin{equation}
100 +\text{Tone}_{t,k} = \frac{\cos(\mathbf{f}_t, \bar{\mathbf{c}}_k^+) - \cos(\mathbf{f}_t, \bar{\mathbf{c}}_k^-)}{\text{Sep}_k}, \quad \text{clipped to } [-1, +1].
101 +\end{equation}
102 +\end{definition}
103 +
104 +This tone measure captures the relative proximity of actual statements to the positive (hawkish) and negative (dovish) reference centroids for each semantic axis, normalized by the axis separation. The primary variable used in regressions is the \emph{total policy stance tone}. In the regression tables, the z-scored stance variable is oriented so that positive values indicate more accommodative (dovish) communication; positive coefficients therefore measure the response to a one-standard-deviation dovish shift. The composite MPS measure defined below retains the hawkish-positive orientation. Relative to counterfactual-based methods, this approach is reproducible, does not depend on particular date choices, and extends to semantic dimensions other than the hawkish-dovish spectrum.
105 +
106 +\subsubsection{Integrated Policy Stance Measurement}
107 +
108 +\begin{definition}[Monetary Policy Stance]
109 +The overall monetary policy stance is defined as:
110 +\begin{equation}
111 +\text{Stance}_t = \text{Novelty}_t \times \text{Tone}_t
112 +\end{equation}
113 +\end{definition}
114 +
115 +\textbf{Note on originality:} Definitions 1--11 and Proposition~\ref{ch3:prop:mps} are our own contributions. They apply the standard sentence-embedding framework from the NLP literature to monetary policy communications with a new decomposition structure.
116 +
117 +The multiplicative form means that stance is large only when a statement is both novel \textit{and} directionally clear. We adopt a multiplicative rather than additive form for two reasons. First, it ensures that Stance is zero whenever either component is zero: repetitive content has no impact regardless of its tone, and novel but directionally neutral content has none either. Second, it captures the reinforcing interaction between the two dimensions: novel information amplifies the impact of tone, and clear directional tone amplifies the impact of novelty.
118 +
119 +In the following two definitions, $\mathbf{f}_t^D$ and $\mathbf{f}_t^H$ denote the embeddings of counterfactual statements located at the dovish and hawkish reference centroids of the policy stance axis, $\bar{\mathbf{c}}^-$ and $\bar{\mathbf{c}}^+$; they represent what the Fed would have released had it issued a maximally dovish or maximally hawkish statement at meeting $t$.
120 +
121 +\begin{definition}[Pure Dovish and Hawkish Stances]
122 +\label{ch3:def:pure_stances}
123 +The extreme stance scenarios are defined as:
124 +\begin{align}
125 +\text{Stance}_t^{dove} &= -\left(1 - \text{sim}(\mathbf{f}_t^D, \mathbf{f}_{t-1})\right) \\
126 +\text{Stance}_t^{hawk} &= 1 - \text{sim}(\mathbf{f}_t^H, \mathbf{f}_{t-1})
127 +\end{align}
128 +\end{definition}
129 +
130 +\begin{definition}[Weighted Policy Stance Representation]
131 +\label{ch3:def:weighted_stance}
132 +The overall stance can be expressed as a convex combination of extreme stances:
133 +\begin{equation}
134 +\text{Stance}_t = w_t \cdot \text{Stance}_t^{dove} + (1 - w_t) \cdot \text{Stance}_t^{hawk}
135 +\end{equation}
136 +where the dovish weight parameter $w_t$ is computed (not estimated) as a deterministic function of the observed similarities:
137 +\begin{equation}
138 +w_t = \frac{1 - \text{sim}(\mathbf{f}_t^H, \mathbf{f}_{t-1}) - \left(1 - \text{sim}(\mathbf{f}_t, \mathbf{f}_{t-1})\right) \cdot \text{Tone}_t}{2 - \text{sim}(\mathbf{f}_t^D, \mathbf{f}_{t-1}) - \text{sim}(\mathbf{f}_t^H, \mathbf{f}_{t-1})}
139 +\end{equation}
140 +\end{definition}
141 +
142 +This weighted representation provides an alternative interpretation of policy stance as a position along the spectrum between extreme dovish and hawkish alternatives. The weight parameter $w_t$ is not a free parameter to be estimated; rather, it is algebraically derived from the observed textual similarities.
143 +
144 +\subsubsection{Market Surprise Identification}
145 +
146 +\begin{definition}[Novelty Decomposition]
147 +The total novelty can be decomposed into predictable and unpredictable components:
148 +\begin{equation}
149 +\text{Novelty}_t = \overline{\text{Novelty}}_{t \mid t-\Delta} + \varepsilon_t
150 +\end{equation}
151 +where $\overline{\text{Novelty}}_{t \mid t-\Delta} \equiv \mathbb{E}_{t-\Delta}[\text{Novelty}_t]$ represents the expected novelty based on information available $\Delta$ periods before the announcement, and $\varepsilon_t$ represents the novelty surprise component with $\mathbb{E}_{t-\Delta}[\varepsilon_t] = 0$ and $\text{Var}(\varepsilon_t) = \sigma_{\varepsilon}^2 < \infty$. We assume $\varepsilon_t$ is covariance-stationary with $\text{Cov}(\varepsilon_t, \varepsilon_{t-j}) \to 0$ as $j \to \infty$.
152 +\end{definition}
153 +
154 +\begin{definition}[Expected Policy Stance]
155 +Under rational expectations, the expected policy stance is:
156 +\begin{equation}
157 +\mathbb{E}_{t-\Delta}[\text{Stance}_t] = \mathbb{E}_{t-\Delta}\left[\text{Novelty}_t \times \text{Tone}_t\right]
158 +\end{equation}
159 +\end{definition}
160 +
161 +\begin{assumption}[Conditional Independence or Zero Covariance]
162 +\label{ch3:ass:indep}
163 +We assume either (i) conditional independence: $\text{Novelty}_t \perp \text{Tone}_t | \mathcal{I}_{t-\Delta}$, where $\mathcal{I}_{t-\Delta}$ is the information set at time $t - \Delta$, or (ii) the weaker condition of zero conditional covariance:
164 +\begin{equation}
165 +\text{Cov}_{t-\Delta}(\text{Novelty}_t, \text{Tone}_t) = 0
166 +\end{equation}
167 +\end{assumption}
168 +
169 +This assumption is plausible because novelty captures whether the Fed changes its messaging structure (how information is presented), while tone captures the directional policy stance (hawkish/dovish). Empirically, the correlation between novelty and tone in our sample is 0.12 (Figure~\ref{ch3:fig:D6}), supporting the assumption that these are largely independent dimensions. Under this assumption:
170 +\begin{equation}
171 +\mathbb{E}_{t-\Delta}[\text{Stance}_t] = \mathbb{E}_{t-\Delta}[\text{Novelty}_t] \times \mathbb{E}_{t-\Delta}[\text{Tone}_t]
172 +\end{equation}
173 +
174 +Further assuming, for tractability, that tone is binary ($\text{Tone}_t \in \{-1, +1\}$, taking the dovish value $-1$ with probability $p_{t-\Delta}$), so that $\mathbb{E}_{t-\Delta}[\text{Tone}_t] = 1 - 2p_{t-\Delta}$:
175 +\begin{equation}
176 +\mathbb{E}_{t-\Delta}[\text{Stance}_t] = (1 - 2p_{t-\Delta}) \cdot \overline{\text{Novelty}}_{t \mid t-\Delta}
177 +\end{equation}
178 +
179 +\begin{definition}[Policy Stance Surprise]
180 +The communication-based policy stance surprise is defined as:
181 +\begin{equation}
182 +\text{MPS}_t = \text{Stance}_t - \mathbb{E}_{t-\Delta}[\text{Stance}_t]
183 +\end{equation}
184 +\end{definition}
185 +
186 +We use the abbreviation MPS (monetary policy surprise) for brevity, but emphasize that this measure is conceptually distinct from the traditional rate-based monetary policy surprise of \citet{kuttner2001}, which measures the unexpected component of the federal funds rate decision using futures prices. Our MPS is a \textit{communication} surprise---it captures unexpected variation in the textual content of FOMC statements, not in the quantitative rate decision. The two measures are complementary and potentially orthogonal: a meeting can produce a large rate surprise with a routine statement, or a novel and directional statement with a fully anticipated rate decision. Our measure captures the latter dimension, which has become increasingly important as forward guidance and qualitative communication have gained prominence in the monetary policy toolkit.
187 +
188 +\begin{proposition}[MPS Decomposition Structure]
189 +\label{ch3:prop:mps}
190 +The monetary policy surprise admits the following decomposition:
191 +\begin{equation}
192 +\text{MPS}_t = \overline{\text{Novelty}}_{t \mid t-\Delta}(\text{Tone}_t + 2p_{t-\Delta} - 1) + \text{Tone}_t \cdot \varepsilon_t
193 +\end{equation}
194 +\end{proposition}
195 +
196 +This decomposition reveals that policy surprises consist of two distinct components: (1) surprises arising from unexpected tone conditional on expected novelty, and (2) surprises arising from unexpected novelty, weighted by the actual tone of the communication. The proof of Proposition~\ref{ch3:prop:mps}, together with the derivations underlying Definitions~\ref{ch3:def:pure_stances} and~\ref{ch3:def:weighted_stance}, is in Appendix~\ref{ch3:sec:proofs}.
197 +
198 +\subsection{Econometric Framework}
199 +\label{ch3:sec:econometric}
200 +
201 +\subsubsection{Panel Data Structure}
202 +
203 +The data are organized as a three-dimensional panel structure with observations indexed along three dimensions:
204 +\begin{itemize}
205 +\item Asset identifier $i \in \{1, 2, \ldots, I\}$ where $I$ represents the total number of financial instruments
206 +\item Event identifier $j \in \{1, 2, \ldots, J\}$ corresponding to FOMC announcement dates where $J = 148$
207 +\item Relative minute $m \in \{-120, \ldots, +120\}$ measured from announcement time, with the primary post-announcement window covering $m \in \{0, \ldots, +45\}$
208 +\end{itemize}
209 +
210 +\subsubsection{Identification Strategy}
211 +
212 +Our identification relies on the fact that FOMC statements are drafted and finalized before the announcement and released at a scheduled time (typically 2:00 PM ET). The text at $t=0$ therefore cannot be influenced by market reactions at $t>0$, satisfying the timing restriction for causal interpretation.
213 +
214 +Our identification relies on three core conditions:
215 +
216 +\textbf{Identification Condition 1 (Exogenous Timing).} The precise minute-by-minute timing of FOMC announcements is predetermined and exogenous to short-term market movements within our event windows.
217 +
218 +\textbf{Identification Condition 2 (Information Concentration).} Within our 45-minute post-announcement event windows, FOMC statements represent the dominant source of monetary policy-relevant information, with minimal contamination from other systematic news sources.
219 +
220 +\textbf{Identification Condition 3 (Market Efficiency).} Financial markets rapidly incorporate new information from FOMC statements into prices, enabling causal interpretation of immediate price movements following announcements.
221 +
222 +We acknowledge two potential threats to this identification strategy. First, the Federal Reserve may adjust its communication in response to broader financial conditions observed before the meeting. We mitigate this concern by including event fixed effects $\delta_j$ that absorb all meeting-level heterogeneity. Second, FOMC announcements may coincide with other information releases (e.g., Summary of Economic Projections, press conference expectations). Our focus on the immediate post-announcement window limits contamination from subsequent press conference content, which typically begins 30 minutes after the statement release.
223 +
224 +\subsubsection{Dynamic Response Function Estimation}
225 +
226 +Our primary empirical specification estimates dynamic impulse response functions:
227 +\begin{equation}
228 +r_{i,m}^{(j)} = \alpha_i + \delta_j + \sum_{k=0}^{+45} \beta_k \cdot \mathbb{1}_{[m=k]} \cdot \text{MPS}_j + \varepsilon_{i,m,j}
229 +\label{ch3:eq:dynamic}
230 +\end{equation}
231 +where $r_{i,m}^{(j)}$ denotes the log return of asset $i$ at minute $m$ relative to monetary policy announcement $j$, $\alpha_i$ represents asset fixed effects, $\delta_j$ captures event fixed effects, and $\beta_k$ measures the marginal impact of monetary policy surprises occurring exactly $k$ minutes relative to the announcement.
232 +
233 +\subsubsection{Semantic Decomposition Specification}
234 +
235 +To examine the differential effects of policy tone versus informational novelty, we estimate:
236 +\begin{equation}
237 +r_{i,m}^{(j)} = \alpha_i + \delta_j + \sum_{k=0}^{+45} \left[\beta_k^{(T)} \cdot \mathbb{1}_{[m=k]} \cdot \text{Tone}_j + \beta_k^{(N)} \cdot \mathbb{1}_{[m=k]} \cdot \text{Novelty}_j\right] + \varepsilon_{i,m,j}
238 +\label{ch3:eq:decomp}
239 +\end{equation}
240 +
241 +The tone coefficients $\{\beta_k^{(T)}\}$ measure the dynamic response to hawkish versus dovish content, while the novelty coefficients $\{\beta_k^{(N)}\}$ capture the response to informational content, independent of directional bias.
242 +
243 +\subsubsection{Rolling Realized Measures}
244 +
245 +In addition to return effects, we examine how policy communications affect market volatility using rolling realized measures computed from 1-minute log returns.
246 +
247 +\paragraph{Realized Variance.} We construct NA-tolerant rolling realized variance in $K$-minute windows:
248 +\begin{equation}
249 +\mathrm{RV}_K(t) = \frac{K}{n_{\text{valid}}} \sum_{s = t-K+1}^{t} r_s^2 \cdot \mathbb{1}[r_s \text{ valid}],
250 +\end{equation}
251 +where $n_{\text{valid}}$ is the number of non-missing observations in the window. We require at least 80\% valid observations.
252 +
253 +\paragraph{Realized Beta.} We compute rolling beta relative to the ES (S\&P 500) benchmark:
254 +\begin{equation}
255 +\hat{\beta}_K(t) = \frac{\sum_{s=t-K+1}^{t} r_{i,s} \cdot r_{m,s}}{\sum_{s=t-K+1}^{t} r_{m,s}^2},
256 +\end{equation}
257 +where $r_{m,s}$ denotes the ES benchmark return. We use $K = 5$ for panel minute-level regressions and $K = 30$ for event-level analysis.
258 +
259 +\subsubsection{Panel Minute-Level Regressions}
260 +
261 +For each ticker, we estimate:
262 +\begin{equation}
263 +\log(\mathrm{RV}_{i,t}) = \alpha + \beta \cdot (\text{Post}_t \times \text{Semantic}_i) + \varepsilon_{i,t},
264 +\label{ch3:eq:panel}
265 +\end{equation}
266 +where $\text{Post}_t = \mathbb{1}[t > 0]$ indicates post-announcement minutes, and $\text{Semantic}_i$ is either \emph{Stance} or \emph{Novelty} (z-scored). Standard errors are clustered by event date.
267 +
268 +\subsubsection{Event-Level Regressions}
269 +
270 +For each event, we compute pre/post changes:
271 +\begin{equation}
272 +\Delta \log(\mathrm{RV})_i = \overline{\log(\mathrm{RV})}_{\text{post}} - \overline{\log(\mathrm{RV})}_{\text{pre}},
273 +\end{equation}
274 +and regress on semantic measures:
275 +\begin{equation}
276 +\Delta Y_i = \alpha + \beta_1 \text{Stance}_i + \beta_2 \text{Novelty}_i + \beta_3 (\text{Stance}_i \times \text{Novelty}_i) + \varepsilon_i,
277 +\label{ch3:eq:event}
278 +\end{equation}
279 +with Newey--West HAC standard errors. We estimate this specification for five dependent variables: $\Delta\mathrm{RV}$, $\Delta\log(\mathrm{RV})$, RV ratio, $\log$(RV ratio), and $\Delta\beta$.
280 +
281 +\subsubsection{Local Projection Impulse Response Functions}
282 +
283 +Following \citet{jorda2005}, for each horizon $h \in \{1, 2, 3, 5, 10, 15, 20, 30, 45, 60, 90, 120\}$ minutes post-announcement:
284 +\begin{equation}
285 +\text{CumRet}_i(0 \to h) = \alpha + \beta_1^{(h)} \text{Stance}_i + \beta_2^{(h)} \text{Novelty}_i + \beta_3^{(h)} (\text{Stance}_i \times \text{Novelty}_i) + \varepsilon_i^{(h)}.
286 +\label{ch3:eq:irf}
287 +\end{equation}
288 +Abnormal cumulative returns are computed as $\text{CAR}_i(h) = \text{CumRet}_i(h) - \text{CumRet}_i^{\text{ES}}(h)$. This local projection approach is robust to misspecification of the data-generating process and allows horizon-specific inference without imposing a parametric impulse response shape. Because the post-meeting press conference typically begins 30 minutes after the statement release, we interpret responses at horizons beyond $h = 30$ as the joint effect of the statement and early press-conference communication.
289 +
290 +\subsubsection{Pre-Announcement Placebo Test}
291 +
292 +We estimate the same specification~\eqref{ch3:eq:irf} for $h$ minutes \emph{before} the announcement:
293 +\begin{equation}
294 +\text{CumRet}_i(-h \to 0) = \alpha + \beta_1^{(-h)} \text{Stance}_i + \beta_2^{(-h)} \text{Novelty}_i + \beta_3^{(-h)} (\text{Stance}_i \times \text{Novelty}_i) + \varepsilon_i^{(-h)}.
295 +\label{ch3:eq:placebo}
296 +\end{equation}
297 +Under the null that announcement content is not anticipated, all $\beta^{(-h)}$ should be zero. This placebo test provides a direct validation of our event study design.
298 +
299 +\subsubsection{Statistical Inference and Robustness}
300 +
301 +We subject all event-level results to five alternative inference methods:
302 +
303 +\begin{enumerate}
304 +\item \textbf{Newey--West HAC standard errors}: Robust to heteroskedasticity and autocorrelation.
305 +\item \textbf{Wild bootstrap} ($B = 1{,}999$, Rademacher weights): Robust to heteroskedasticity with improved finite-sample properties.
306 +\item \textbf{Clustered standard errors} (HC1, by event date): Accounts for cross-sectional dependence within events.
307 +\item \textbf{Quantile regression} ($\tau = 0.5$): Robust to outliers and heavy tails in the dependent variable.
308 +\item \textbf{Permutation test} ($B = 4{,}999$): Gold standard---shuffles semantic labels across events preserving panel structure, providing exact $p$-values under the null.
309 +\end{enumerate}
310 +
311 +All $p$-values in panel and IRF regressions are adjusted for multiple testing using the Benjamini--Hochberg (BH) procedure within each family of tests (per dependent variable and model type) \citep{benjamini1995controlling}. We use only standard significance levels: 1\%, 5\%, and 10\%.
312 +
313 +\textbf{Sub-Period Stability.} We examine coefficient stability across six Fed policy regimes: crisis and recovery (2008--2012), post-crisis normalization (2013--2015), pre-pandemic tightening (2016--2019), pandemic response (2020--2021), and inflation tightening (2022--2025).
314 +
315 +\textbf{Alternative Semantic Measures.} The dual-model ensemble itself provides a built-in robustness check: by comparing results from MiniLM-only, BERT-only, and ensemble measures, we verify that findings are not driven by model-specific artifacts.
316 +
317 +\subsection{Hypotheses and Testable Predictions}
318 +\label{ch3:sec:hypotheses}
319 +
320 +We now state our hypotheses and, for each one, the coefficient restriction that operationalizes it in the specifications of Section~\ref{ch3:sec:econometric}. Recall the sign conventions: the z-scored stance regressor in equations~\eqref{ch3:eq:decomp}--\eqref{ch3:eq:irf} is dovish-positive, while the composite MPS in equation~\eqref{ch3:eq:dynamic} is hawkish-positive. Table~\ref{ch3:tab:hyp_map} summarizes the mapping from hypotheses to coefficients, equations, and the tables in which each test is reported.
321 +
322 +\textbf{Hypothesis 1 (Tone Effects on Returns).}
323 +\begin{itemize}
324 +\item \textbf{H1a.} Dovish tone increases returns on risk assets (equities, commodities). \textit{Test:} $\beta_k^{(T)} > 0$ in the semantic decomposition~\eqref{ch3:eq:decomp} and $\beta_1^{(h)} > 0$ in the local projection~\eqref{ch3:eq:irf} for ES and CL.
325 +\item \textbf{H1b.} Dovish tone decreases returns on safe-haven assets, as accommodative policy triggers portfolio rebalancing away from safe havens toward risk assets. \textit{Test:} $\beta_1^{(h)} < 0$ in~\eqref{ch3:eq:irf} for GC, ZN, and ZF.
326 +\item \textbf{H1c.} Tone effects strengthen with the horizon as markets progressively process policy implications. \textit{Test:} $|\beta_1^{(h)}|$ increasing in $h$ in~\eqref{ch3:eq:irf} for ES.
327 +\end{itemize}
328 +
329 +\textbf{Hypothesis 2 (Novelty Effects on Volatility).}
330 +\begin{itemize}
331 +\item \textbf{H2a.} High novelty increases market volatility through information processing complexity. \textit{Test:} $\beta_2 > 0$ in the event-level regression~\eqref{ch3:eq:event} with $\Delta\mathrm{RV}$ as the dependent variable, and $\beta > 0$ in the panel regression~\eqref{ch3:eq:panel} with $\text{Semantic} = \text{Novelty}$, most directly for VX.
332 +\item \textbf{H2b.} Novelty effects dissipate rapidly as markets resolve uncertainty. \textit{Test:} $|\beta_2^{(h)}|$ decreasing in $h$ in~\eqref{ch3:eq:irf} for VX.
333 +\item \textbf{H2c.} Novelty has minimal effects on directional returns. \textit{Test:} $\beta_2^{(h)} \approx 0$ (statistically indistinguishable from zero) in~\eqref{ch3:eq:irf} for the return contracts (ES, CL, GC, DX).
334 +\end{itemize}
335 +
336 +\textbf{Hypothesis 3 (Policy Surprise Effects).}
337 +\begin{itemize}
338 +\item \textbf{H3a.} Hawkish policy surprises reduce equity returns. \textit{Test:} $\beta_0 < 0$ in the dynamic response function~\eqref{ch3:eq:dynamic} for ES.
339 +\item \textbf{H3b.} Policy surprises increase implied volatility. \textit{Test:} $\beta_0 > 0$ in~\eqref{ch3:eq:dynamic} for VX.
340 +\item \textbf{H3c.} Surprise effects on volatility are most pronounced immediately after the announcement. \textit{Test:} $|\beta_k|$ in~\eqref{ch3:eq:dynamic} largest at $k = 0$ and decaying in $k$ for VX.
341 +\end{itemize}
342 +
343 +\textbf{Hypothesis 4 (Cross-Asset Patterns).}
344 +\begin{itemize}
345 +\item \textbf{H4a.} Risk assets respond to tone in the same direction (risk-on/risk-off). \textit{Test:} $\text{sign}\bigl(\beta_1^{(h)}\bigr)$ equal across ES and CL in~\eqref{ch3:eq:irf}, opposite for GC, ZN, ZF.
346 +\item \textbf{H4b.} Volatility responses to novelty are asset-specific. \textit{Test:} $\beta_2$ and $\beta_3$ in~\eqref{ch3:eq:event} differ in magnitude and significance across the seven contracts.
347 +\end{itemize}
348 +
349 +Finally, the validity of the design itself is testable: under no anticipation of statement content, all pre-announcement coefficients in the placebo specification~\eqref{ch3:eq:placebo} should be zero, $\beta_1^{(-h)} = \beta_2^{(-h)} = \beta_3^{(-h)} = 0$ for all $h$.
350 +
351 +\begin{table}[H]
352 +\centering
353 +\footnotesize
354 +\caption{Mapping of Hypotheses to Coefficients, Equations, and Evidence}
355 +\label{ch3:tab:hyp_map}
356 +\begin{tabular}{lp{4.5cm}p{4.3cm}p{3.3cm}}
357 +\toprule
358 +Hyp. & Prediction & Coefficient restriction (equation) & Evidence \\
359 +\midrule
360 +H1a & Dovish tone $\Rightarrow$ risk-asset returns $\uparrow$ & $\beta_1^{(h)} > 0$ in \eqref{ch3:eq:irf}, ES/CL & Table~\ref{ch3:tab:irf_stance_ret_h} \\
361 +H1b & Dovish tone $\Rightarrow$ safe-haven returns $\downarrow$ & $\beta_1^{(h)} < 0$ in \eqref{ch3:eq:irf}, GC/ZN/ZF & Table~\ref{ch3:tab:irf_stance_ret_h} \\
362 +H1c & Tone effects build with horizon & $|\beta_1^{(h)}|$ increasing in $h$, ES & Table~\ref{ch3:tab:irf_stance_ret_h}; Fig.~\ref{ch3:fig:R8} \\
363 +H2a & Novelty $\Rightarrow$ volatility $\uparrow$ & $\beta_2 > 0$ in \eqref{ch3:eq:event}, $\Delta$RV & Tables~\ref{ch3:tab:panel_logrv_novelty}, \ref{ch3:tab:rolling_delta_rv_30min} \\
364 +H2b & Novelty effects decay quickly & $|\beta_2^{(h)}|$ decreasing in $h$, VX & Table~\ref{ch3:tab:irf_novelty_ret_h} \\
365 +H2c & Novelty does not move returns & $\beta_2^{(h)} \approx 0$ in \eqref{ch3:eq:irf} & Table~\ref{ch3:tab:irf_novelty_ret_h}; Fig.~\ref{ch3:fig:R9} \\
366 +H3a & Hawkish surprise $\Rightarrow$ equity returns $\downarrow$ & $\beta_0 < 0$ in \eqref{ch3:eq:dynamic}, ES & Section~\ref{ch3:sec:baseline} \\
367 +H3b & Surprise $\Rightarrow$ implied volatility $\uparrow$ & $\beta_0 > 0$ in \eqref{ch3:eq:dynamic}, VX & Section~\ref{ch3:sec:baseline} \\
368 +H3c & Surprise effects peak on impact & $|\beta_k|$ max at $k=0$ in \eqref{ch3:eq:dynamic}, VX & Section~\ref{ch3:sec:baseline} \\
369 +H4a & Homogeneous return response, risk assets & $\text{sign}(\beta_1^{(h)})$ equal, ES/CL in \eqref{ch3:eq:irf} & Table~\ref{ch3:tab:irf_stance_ret_h} \\
370 +H4b & Heterogeneous volatility responses & $\beta_2, \beta_3$ vary across contracts in \eqref{ch3:eq:event} & Tables~\ref{ch3:tab:rolling_delta_rv_30min}, \ref{ch3:tab:robustness_A} \\
371 +Placebo & No pre-announcement effects & $\beta^{(-h)} = 0$ in \eqref{ch3:eq:placebo}, all contracts & Fig.~\ref{ch3:fig:placebo_stance}, \ref{ch3:fig:placebo_novelty} \\
372 +\bottomrule
373 +\end{tabular}
374 +\end{table}
375 +
added these-ulaval/chapitre3/04_results.tex +105 −0
@@ -0,0 +1,105 @@
1 +\section{Results}
2 +\label{ch3:sec:results}
3 +
4 +We organize our results as follows. We first describe the semantic variables (Section~\ref{ch3:sec:desc}), then report baseline responses to the composite MPS (Section~\ref{ch3:sec:baseline}), the tone-novelty decomposition (Section~\ref{ch3:sec:decomposition}), and the temporal dynamics of these effects (Section~\ref{ch3:sec:halflife}). Throughout, each hypothesis is evaluated against the coefficient restriction assigned to it in Table~\ref{ch3:tab:hyp_map}. The main finding is that tone and novelty load on different dependent variables: tone predicts directional returns, while novelty predicts volatility changes. The VIX model has the highest $R^2$ ($0.78$); equity and commodity models follow. Treasury results are weaker.
5 +
6 +$R^2$ values for the return models are low (0.01--0.15), as is typical when predicting minute-level returns where microstructure noise dominates \citep{barndorff2002econometric}. Low $R^2$ does not bias the coefficient estimates, but it means that FOMC communication accounts for a small share of total intra-day return variation. The VIX model is the exception: semantic variables explain up to 78\% of announcement-window variation in VIX futures, consistent with the hypothesis that these communications operate primarily on uncertainty rather than on the level of returns.
7 +
8 +\subsection{Descriptive statistics}
9 +\label{ch3:sec:desc}
10 +
11 +Table~\ref{ch3:tab:desc_1min_ret} reports descriptive statistics for 1-minute log returns on FOMC days. Table~\ref{ch3:tab:desc_rolling} reports summary statistics for the rolling realized measures. Figure~\ref{ch3:fig:D3} plots the average intraday volatility pattern on FOMC days, which spikes sharply at the 14:00~ET release.
12 +
13 +Novelty has a mean of 0.065. The scale runs from 0 (a statement identical to the previous one) to 2 (maximally opposite), so a mean of 0.065 corresponds to an average cosine similarity of 0.935 between consecutive statements---most meetings produce only small changes to the prior template. The median is lower (0.025), so the distribution is right-skewed: most meetings produce minor textual changes, but a few produce large departures. The maximum (0.494) occurs during the 2008 crisis, when the Fed introduced zero-lower-bound language (Figure~\ref{ch3:fig:D7}). Skewness is 2.45 and kurtosis is 8.16.
14 +
15 +The raw tone measure (hawkish-positive) has a mean of 0.309 on a theoretical $[-1, +1]$ scale, where $-1$ is the dovish centroid, $0$ is equidistant, and $+1$ is the hawkish centroid. The positive mean indicates that the average FOMC statement over 2008--2025 lies closer to the hawkish than to the dovish centroid. Because cosine similarities to both centroids are high, realized values occupy a narrow band within the theoretical scale; we therefore z-score tone in all regressions. Skewness is $-0.78$ and kurtosis is 6.96, driven by the most dovish statements in the sample, which mark its lower extremes (2008: $-0.081$; 2020: $-0.062$).
16 +
17 +The MPS measure has mean 0.004 and kurtosis 12.18---the fat tails reflect occasional large surprises during economic stress. Jarque-Bera tests reject normality for all three measures ($p < 0.01$), and ARCH-LM tests indicate time-varying variance ($p < 0.05$), which motivates our use of robust standard errors and multiple inference methods throughout.
18 +
19 +\subsection{Baseline Asset Price Responses to Monetary Policy Surprises}
20 +
21 +\label{ch3:sec:baseline}
22 +We begin with the composite MPS, testing H3. The dependent variable is the 1-minute log return for each contract; the regressor is the z-scored MPS. A positive coefficient means that hawkish communication surprises are associated with higher returns.
23 +
24 +\subsubsection{Immediate Market Reactions}
25 +
26 +The equity market response to hawkish surprises is consistent with H3a (negative equity effect): the ES contract exhibits a $-1.2$ basis point response ($p < 0.05$). To put this magnitude in perspective, a one-standard-deviation hawkish surprise generates an immediate equity decline roughly equivalent to the average hourly return on a non-announcement day, concentrated in a single minute.
27 +
28 +Among commodities, crude oil (CL) falls by 1.0 bps ($p < 0.05$), consistent with tighter policy reducing growth expectations and energy demand. Gold (GC) falls by 0.6 bps, as an inflation hedge becomes less attractive when policy tightens. The Dollar Index (DX) rises by 0.8 bps, consistent with higher expected rate differentials increasing dollar demand.
29 +
30 +The VIX response is $+11.2$ bps ($p < 0.05$, H3b): a one-standard-deviation hawkish surprise is associated with an 11.2 bps increase in VIX futures, about 0.5\% of the average VIX level. This is the largest immediate response in our sample. Treasury securities (ZN, ZF) show small, insignificant responses, possibly because our textual MPS is largely orthogonal to the rate expectations already embedded in Treasury futures.
31 +
32 +\subsubsection{Dynamic Response Evolution}
33 +
34 +The 45-minute post-announcement window shows that different asset classes adjust at different speeds:
35 +
36 +\textit{Equities (ES).} The initial decline ($-1.2$ bps at $t = 0$) deepens to $-3.8$ bps by $t = 15$ and $-6.8$ bps by $t = 45$. The cumulative response is 5.7 times the instantaneous reaction, so the first-minute price change captures only a fraction of the total adjustment.
37 +
38 +\textit{VIX futures.} The initial spike ($+11.2$ bps at $t = 0$) loses roughly half its magnitude by $t = 20$ and returns near its pre-announcement level by $t = 45$.
39 +
40 +\textit{Treasuries (ZN, ZF).} Responses become marginally significant only at $t = 30$--$45$, consistent with slower transmission to term premiums.
41 +
42 +\textit{Commodities (CL, GC).} Crude oil effects continue to grow through $t = 45$, consistent with the slower adjustment typical of physical commodity markets.
43 +
44 +These patterns support H3c for volatility---the VIX response peaks on impact and decays---but equity and commodity responses continue to build, foreshadowing the gradual tone effects documented below. The speed of adjustment varies across assets, which motivates the asset-specific decomposition.
45 +
46 +\subsection{Semantic Decomposition: Tone vs.\ Novelty Effects}
47 +\label{ch3:sec:decomposition}
48 +
49 +We now separate tone from novelty. The dependent variables are log returns (testing H1) and realized volatility changes (testing H2). The regressors are the z-scored tone and novelty measures, entered separately and jointly.
50 +
51 +\subsubsection{Tone Effects on Asset Returns (H1)}
52 +
53 +A one-standard-deviation dovish shift in tone is associated with ES gains that build steadily over the post-announcement window: 3.8 bps by $h = 30$, 6.8 bps by $h = 60$, and 12.0 bps by $h = 120$ (Table~\ref{ch3:tab:irf_stance_ret_h}, H1a)---several days of average equity returns compressed into two hours. This is consistent with lower discount rates, higher growth expectations, and greater risk appetite.
54 +
55 +Safe-haven assets move in the opposite direction: at $h = 15$, gold falls by 3.1 bps and 10-year Treasuries (ZN) by 1.1 bps following dovish tone (H1b), consistent with a ``risk-on'' rotation out of safe havens. The Dollar Index response is small and changes sign across horizons ($+1.4$ bps at $h = 15$, $-4.0$ bps at $h = 120$).
56 +
57 +Tone effects grow stronger, not weaker, over the post-announcement window: the ES coefficient rises monotonically from 0.2 bps at $h = 5$ to 3.8 bps at $h = 30$ and 12.0 bps at $h = 120$ (Table~\ref{ch3:tab:irf_stance_ret_h}, H1c). This gradual amplification---rather than the immediate level shift that a frictionless model would imply---is consistent with sequential portfolio adjustment by heterogeneous participants (algorithms, institutions, retail).
58 +
59 +\subsubsection{Novelty Effects on Volatility (H2)}
60 +
61 +Novelty loads on volatility, not on returns. The VIX response to a one-standard-deviation increase in novelty is $-5.57$ bps ($p < 0.05$), meaning that genuinely new language is associated with \textit{lower} implied volatility. This sign is opposite to our ex ante prediction (H2a) but has a straightforward interpretation: a statement that departs substantially from the prior meeting's language sends a clearer signal about the Fed's current assessment, helping investors narrow the range of possible outcomes. A repetitive statement, by contrast, leaves open the question of whether the unchanged wording reflects genuine stability or simply a failure to update the language.
62 +
63 +The VIX novelty coefficient is largest in the first 5 minutes and falls to near zero by $t = 30$ (H2b). Novelty coefficients on directional returns (ES, CL, GC, DX) are small and insignificant (H2c). The combination---novelty predicts volatility but not returns---separates our measure from standard ``surprise'' variables, which conflate direction and information content.
64 +
65 +\subsubsection{Panel Minute-Level Results}
66 +
67 +Tables~\ref{ch3:tab:panel_logrv_stance}--\ref{ch3:tab:panel_logrv_novelty} report panel regressions of $\log(\text{RV})$ on the interaction of post-announcement indicators with semantic measures. The coefficient on Post$\times$Stance is negative and significant at 1\% for 6 of 7 contracts: ES ($-5.31$), VX ($-3.28$), ZN ($-5.25$), ZF ($-4.16$), CL ($-5.87$), and GC ($-6.03$). Only DX is insignificant ($+0.15$). The negative sign means that dovish stance is associated with lower post-announcement realized volatility---accommodative statements calm markets---while hawkish statements are followed by larger volatility increases, consistent with contractionary signals generating more repricing. By contrast, the Post$\times$Novelty coefficients (Table~\ref{ch3:tab:panel_logrv_novelty}) are smaller and significant only for the Treasury contracts (ZN: $-1.81$, $p < 0.01$; ZF: $-1.22$, $p < 0.10$).
68 +
69 +\subsubsection{Event-Level Results}
70 +
71 +Tables~\ref{ch3:tab:rolling_delta_rv_30min}--\ref{ch3:tab:rolling_delta_beta_30min} report event-level regressions with the interaction specification $\Delta Y_i = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 (\text{Stance} \times \text{Novelty}) + \varepsilon_i$. The stance$\times$novelty interaction is negative and significant for VIX ($-7.96$, $t = -3.12$, $p < 0.01$), ZF ($-2.48$, $t = -2.22$), ZN ($-1.66$, $t = -1.99$), and CL ($-8.61$, $p < 0.10$); Figure~\ref{ch3:fig:R6a} summarizes the coefficient estimates as a heatmap. The interaction indicates that stance and novelty reinforce each other: the volatility decline associated with a dovish statement is larger when the statement also departs substantially from prior language, whereas a statement that repeats the previous meeting's wording has a smaller volatility effect regardless of its tone.
72 +
73 +$R^2$ values are much higher for volatility than for returns: $\Delta\text{RV}$ regressions achieve 0.286 (ES), 0.268 (CL), and 0.245 (ZF), versus below 0.05 for most return regressions. The semantic variables explain more of the variation in volatility (which depends on information content) than in returns (which depend on the direction of the surprise in a noisy environment).
74 +
75 +\subsubsection{Local Projection Impulse Response Functions}
76 +
77 +Tables~\ref{ch3:tab:irf_stance_ret_h}--\ref{ch3:tab:irf_stance_x_novelty_abn_h} and Figures~\ref{ch3:fig:R10_stance}--\ref{ch3:fig:R10_abn_interaction} report local projection estimates. The VIX response to stance is negative at every horizon, statistically strongest between 5 and 30 minutes ($-14.74$ bps at $h = 5$, $-22.03$ at $h = 30$), and attenuates thereafter ($-14.27$ at $h = 120$, with much wider standard errors); the profile is hump-shaped, peaking near 30 minutes. The stance$\times$novelty interaction on abnormal VIX returns (Figure~\ref{ch3:fig:R10_abn_interaction}) is significant from $h = 5$ through $h = 120$, with the $t$-statistic reaching $-5.06$.
78 +
79 +The ES coefficient on stance grows monotonically from near zero at $h = 5$ to 12.0 bps at $h = 120$ (standard errors also grow). A frictionless model would predict an immediate level shift; the gradual increase we observe is more consistent with sequential adjustment by heterogeneous investors.
80 +
81 +\subsubsection{Cross-Asset Patterns (H4)}
82 +
83 +Risk assets (ES, CL) both rise after dovish tone, with similar timing (H4a). Safe havens (GC, ZN, ZF) fall, consistent with risk-on/risk-off. But the volatility responses differ across assets (H4b): VIX responds most to the stance$\times$novelty interaction, Treasury volatility responds mainly to stance alone, and commodity volatility responds to both. The return homogeneity and volatility heterogeneity suggest that the two channels operate with different relative strength across asset classes.
84 +
85 +\subsubsection{Pre-Announcement Placebo Tests}
86 +
87 +Figures~\ref{ch3:fig:placebo_stance}--\ref{ch3:fig:placebo_novelty} report pre-announcement placebo tests: we estimate the same IRF specification for cumulative returns from $-h$ to 0. If statement content is not anticipated, all pre-announcement coefficients should be zero. None is significant at 10\% across any ticker, horizon, or semantic measure. This rules out information leakage and pre-existing trends.
88 +
89 +\subsubsection{Multi-Method Robustness}
90 +
91 +Tables~\ref{ch3:tab:robustness_A}--\ref{ch3:tab:robustness_B} report five alternative inference methods for every event-level coefficient. We call a result ``robust'' if at least 3 of 5 methods give $p < 0.10$. The most robust results are: stance on $\Delta\log\text{RV}$ for ES, CL, GC, ZF, and ZN (each $\#\text{Sig} = 3$); stance on $\Delta\text{RV}$ for ES, CL, and DX (each $\#\text{Sig} = 3$); and novelty on $\Delta\text{RV}$ for ES and CL ($\#\text{Sig} = 4$). Figure~\ref{ch3:fig:rob_heatmap} displays these counts. Sub-period analysis (Figure~\ref{ch3:fig:rob_subperiod}) shows consistent coefficient signs across six Fed regimes, though magnitudes vary.
92 +
93 +With 7 assets and multiple dependent variables, we test many coefficients; results with $\#\text{Sig} = 1$ should be treated with caution. We rely on the Benjamini--Hochberg correction and the robustness counts to limit false discovery.
94 +
95 +\subsection{Economic Interpretation: Temporal Dynamics and Half-Lives}
96 +\label{ch3:sec:halflife}
97 +
98 +We summarize the speed of adjustment using half-lives: the time for an initial effect to fall to half its peak. For exponential decay, $t_{1/2} = \ln(2)/\lambda$.
99 +
100 +The VIX response ($+11.2$ bps on impact) falls to half its peak within 15--20 minutes; reading the decay as approximately exponential gives a half-life of about 17.5 minutes. Once the statement text is known, uncertainty about the Fed's message dissipates quickly.
101 +
102 +Equity returns show the opposite pattern. The ES response to dovish tone \textit{grows} steadily over the post-announcement window, roughly tripling between $h = 30$ (3.8 bps) and $h = 120$ (12.0 bps). Fundamental repricing---revising expected cash flows and discount rates---takes longer than uncertainty resolution, consistent with gradual diffusion across heterogeneous investors.
103 +
104 +The two patterns together---fast volatility decay and slow return amplification---are hard to reconcile with a single channel. The informational dimension (is the statement new?) resolves quickly; the directional dimension (is it hawkish or dovish?) takes longer to price in.
105 +
added these-ulaval/chapitre3/05_conclusion.tex +61 −0
@@ -0,0 +1,61 @@
1 +\section{Conclusion}
2 +\label{ch3:sec:conclusion}
3 +
4 +\subsection{Summary of Findings}
5 +
6 +We summarize our findings under two headings: what we learn about the economics of monetary policy transmission, and what we contribute methodologically.
7 +
8 +\textbf{Tone predicts directional returns.} Dovish tone is associated with higher equity returns that build to about 12 bps over two hours, and with lower safe-haven returns (gold: $-3.1$ bps; Treasuries: $-1.1$ bps at 15 minutes), supporting H1a--b. This is consistent with the standard transmission mechanism in \citet{bernanke2005} and \citet{gurkaynak2005}---accommodative expectations lower discount rates and shift portfolios toward risk assets---but we show that the channel operates through the \textit{qualitative} content of the statement, not only through the rate decision.
9 +
10 +Tone effects grow stronger over time: the ES coefficient rises from 0.2 bps at $h = 5$ to 12.0 bps at $h = 120$ (H1c). \citet{rosa2013} documents continued price adjustment over hours after FOMC releases; our minute-level data show that this amplification is consistent with heterogeneous processing speeds across market participants \citep{veldkamp2011}.
11 +
12 +\textbf{Novelty predicts lower volatility, not higher.} Novelty is associated with \textit{reduced} implied volatility (VIX: $-5.57$ bps, $p < 0.05$; interaction $t = -5.06$), which reverses the sign we predicted under H2a. Prior work on disagreement-driven volatility \citep{patton2013} and news-based uncertainty \citep{manela2017} would predict the opposite. The difference, we believe, is that in the specific setting of FOMC statements, a departure from prior language acts as a \textit{signal of clarity}: the Fed is actively updating its message, which narrows the set of plausible policy paths. A repetitive statement, conversely, leaves open whether the Fed's views have changed but the language has not been updated. Novelty effects dissipate in under 10 minutes (H2b) and do not predict returns (H2c), confirming that this channel operates on volatility rather than on prices.
13 +
14 +\textbf{The composite surprise replicates known patterns.} Hawkish surprises are associated with 1--7 bps lower equity returns (H3a) and 11.2 bps higher VIX (H3b), with volatility effects concentrated in the first minutes (H3c). That our textual MPS produces results consistent with the rate-surprise literature \citep{kuttner2001,bernanke2005,gurkaynak2005,savor2014} reassures us that the NLP pipeline captures economically relevant variation.
15 +
16 +\textbf{Cross-asset patterns differ between returns and volatility.} Risk assets (ES, CL) respond to tone with the same sign and similar timing (H4a). Safe havens (GC, ZN, ZF) move in the opposite direction. But volatility responses vary: VIX loads on the interaction, Treasury volatility on stance alone, commodity volatility on both. This pattern is consistent with \citet{fleming1999} and \citet{balduzzi2001economic}, who document heterogeneous announcement effects across asset classes.
17 +
18 +\textbf{Volatility resolves fast; returns adjust slowly.} The VIX half-life is about 17.5 minutes; the ES tone effect continues to build for two hours after the announcement. A single-channel model cannot produce both patterns simultaneously. The fast channel (novelty $\to$ volatility) reflects the resolution of uncertainty about what the Fed will say. The slow channel (tone $\to$ returns) reflects the time needed to revise discount rates, growth expectations, and portfolio allocations across a heterogeneous investor base.
19 +
20 +\subsection{Methodological Contributions}
21 +
22 +On the measurement side, we make three contributions. First, using two architecturally different models (MiniLM, 33M parameters; BERT, 110M parameters) and comparing their outputs reduces the risk that any finding is an artifact of one particular architecture. The inter-model confidence (mean 0.837 for novelty) flags statements where the two models disagree. Second, two-stage domain adaptation (TSDAE then MNRL) yields axis separations of 0.248--0.302, well above the 0.04 minimum. Third, PCA-based reference selection replaces subjective date choices with an algorithm, removing a source of researcher discretion.
23 +
24 +Working at the minute level (148 events $\times$ $\pm$120 minutes) matters for two reasons: it allows causal identification from the pre-determined release time (validated by the placebo tests), and it reveals the fast-volatility / slow-return asymmetry that daily data would average away.
25 +
26 +\subsection{Implications for Policy and Communication Strategy}
27 +
28 +Three findings have implications for how the Fed drafts its statements.
29 +
30 +First, novelty reduces volatility. The conventional view is that central banks should change their language gradually to avoid ``surprising'' markets. Our estimates say the opposite: substantive changes in wording are associated with \textit{lower} implied volatility, presumably because a clear departure from the prior statement narrows the range of plausible interpretations. When the Fed needs to signal a regime change---a new framework, a pivot from tightening to easing---a decisive rewrite of the statement may be less destabilizing than incremental edits.
31 +
32 +Second, tone has persistent valuation effects. Because the equity response to tone grows over 2 hours rather than being absorbed immediately, the hawkish-dovish framing of the statement has real wealth consequences: a one-standard-deviation dovish shift is associated with up to 12 bps of equity returns, which, on a market capitalization of roughly \$40 trillion, is economically meaningful.
33 +
34 +Third, the interaction matters. The stance$\times$novelty interaction on VIX ($t = -5.06$, persistent from 5 to 120 minutes) implies that a directionally clear statement packaged in new language reduces uncertainty more than the same message delivered in boilerplate.
35 +
36 +\subsection{Limitations}
37 +
38 +We note several limitations.
39 +
40 +\textit{Sample period.} Our 2008--2025 sample is dominated by extraordinary monetary policy regimes: the zero lower bound (2008--2015), quantitative easing, and the post-pandemic tightening cycle. While our sub-period stability analysis (Figure~\ref{ch3:fig:rob_subperiod}) shows that coefficient signs are generally consistent across six Fed policy regimes, we cannot rule out that the magnitude of communication effects differs in more ``normal'' policy environments. In particular, the ZLB period may overstate the importance of qualitative communication (since rate decisions were constrained, statements became the primary policy instrument), while the recent tightening cycle may understate novelty effects (since rate hikes were widely anticipated, reducing the scope for genuine communication surprises).
41 +
42 +\textit{Measurement error.} Our NLP-based tone and novelty measures are subject to measurement error whose properties are difficult to characterize fully. Classical measurement error would attenuate our coefficient estimates toward zero (creating an ``errors-in-variables'' bias), making our significant findings conservative. However, if measurement error is correlated with meeting characteristics (e.g., if our models systematically misclassify statements from certain Fed Chairs or policy regimes), the bias could go in either direction. The inter-model agreement diagnostics (mean confidence 0.837 for novelty, 0.562--0.806 for tone) help identify statements where measurement may be less reliable, but they do not fully resolve this concern.
43 +
44 +\textit{Multiple testing.} With 7 assets, 5 dependent variables, and multiple inference methods, we test a large number of coefficients. We address this through Benjamini--Hochberg FDR correction within each family of tests and by reporting multi-method robustness counts. Nevertheless, some individually significant results---particularly those with $\#\text{Sig} = 1$ in the robustness tables---should be interpreted cautiously as potentially reflecting false discovery.
45 +
46 +\textit{Identification threats.} Our identification strategy exploits the pre-determined timing and content of FOMC statements, validated by pre-announcement placebo tests. However, we cannot fully rule out confounding from simultaneous information releases. FOMC announcements sometimes coincide with the Summary of Economic Projections (``dot plot''), which provides quantitative rate path forecasts that may interact with our textual measures. Similarly, market expectations about the subsequent press conference (beginning 30 minutes after the statement) may influence post-announcement price dynamics within our event window. These concurrent information flows are not fully separable from the statement text itself.
47 +
48 +\textit{Generalizability.} We study U.S. futures markets around FOMC statements only. Whether the same tone-return and novelty-volatility patterns hold for other central banks (ECB, BOJ, BOE), other Fed communications (minutes, speeches), or other asset classes (corporate bonds, emerging-market equities) is an open question.
49 +
50 +\subsection{Future Directions}
51 +
52 +Several extensions seem natural.
53 +
54 +\textit{More semantic dimensions.} We measure tone and novelty; FOMC statements also vary in uncertainty language, temporal focus (forward- vs.\ backward-looking), and internal consensus (voting dissents). The PCA axis construction can accommodate additional dimensions.
55 +
56 +\textit{International markets.} FOMC announcements move global asset prices \citep{wongswan2009,ehrmann2011}. Applying the tone-novelty decomposition to international data would show whether both channels transmit across borders or whether one dominates.
57 +
58 +\textit{Structural models.} Our evidence is reduced-form. A model with heterogeneous agents and differential processing of directional versus informational content could rationalize the fast-volatility / slow-return asymmetry and generate further predictions.
59 +
60 +\textit{Joint analysis with rate surprises.} Our textual MPS and the rate-based surprise of \citet{kuttner2001} measure different things. Estimating both jointly would reveal whether they contain complementary information and whether the information-versus-policy decomposition of \citet{jarocinski2020,nakamura2018} maps onto our tone-versus-novelty decomposition.
61 +
added these-ulaval/chapitre3/chapitre3.tex +62 −0
@@ -0,0 +1,62 @@
1 +%% ---------------------------------------------------------------------------
2 +%% Chapitre 3 — adapté de PHD_chapitre3_theses_20260731/ (manuscrit daté
3 +%% 2026-03-06). Corps scientifique inchangé ; voir INVENTAIRE.md pour la liste
4 +%% des adaptations mécaniques. Les annexes de l'article (Mathematical Proofs ;
5 +%% Additional Tables and Figures) deviennent les annexes A et B de la thèse.
6 +%% ---------------------------------------------------------------------------
7 +
8 +\chapter{Returns and Volatility Around FOMC Announcements: A High-Frequency Analysis of Policy Tone and Novelty}
9 +\label{chap-chapitre3}
10 +
11 +\section{Résumé}
12 +
13 +% BROUILLON À RÉVISER PAR L'AUTEUR (traduction condensée de l'abstract ; la
14 +% FESP limite le résumé de chaque article inséré à 150 mots).
15 +\begin{otherlanguage*}{french}
16 + Nous décomposons les communiqués du FOMC en ton de politique monétaire
17 + (\emph{hawkish}/\emph{dovish}) et en nouveauté informationnelle (écart par
18 + rapport au communiqué précédent), puis estimons leurs effets sur les
19 + rendements et la volatilité à haute fréquence. À partir de données à une
20 + minute pour sept contrats à terme et 148~annonces du FOMC (2008--2025), nous
21 + montrons que le ton prédit les rendements directionnels — un virage
22 + accommodant d'un écart-type est associé à des gains boursiers atteignant
23 + environ 12~points de base en deux heures — tandis que la nouveauté prédit
24 + les variations de volatilité (l'interaction ton--nouveauté sur le VIX
25 + persiste de 5 à 120~minutes, $t = -5{,}06$). Le ton est associé à la
26 + volatilité réalisée pour six contrats sur sept ($p < 0{,}01$). Des tests
27 + placebo pré-annonce et cinq méthodes d'inférence indépendantes valident ces
28 + résultats, fondés sur un ensemble bi-modèle (MiniLM et BERT) avec sélection
29 + des références par analyse en composantes principales.
30 +\end{otherlanguage*}
31 +
32 +\section{Abstract}
33 +
34 +We decompose FOMC statements into policy tone (hawkish/dovish) and
35 +informational novelty (departure from previous messaging) and estimate their
36 +effects on high-frequency asset returns and volatility. Using 1-minute data
37 +for 7 futures contracts across 148 FOMC events (2008--2025), we find that tone
38 +predicts directional returns (a one-standard-deviation dovish shift is
39 +associated with equity gains that build to about 12 basis points within two
40 +hours) while novelty predicts volatility changes (the stance--novelty
41 +interaction on VIX persists from 5 to 120 minutes, $t = -5.06$). Policy stance
42 +is associated with realized volatility changes in 6 of 7 contracts
43 +($p < 0.01$). Pre-announcement placebo tests and five independent inference
44 +methods validate these results. To construct our measures, we train a
45 +dual-model ensemble (MiniLM and BERT) on FOMC communications with data-driven
46 +PCA-based reference selection.
47 +
48 +\medskip
49 +\noindent\textbf{Keywords:} Monetary policy, FOMC announcements,
50 +high-frequency data, textual analysis, market volatility.\\
51 +\noindent\textbf{JEL Classification:} E52, E58, G12, G14
52 +
53 +%% --- Corps de l'article ---
54 +\input{chapitre3/01_introduction}
55 +\input{chapitre3/02_data}
56 +\input{chapitre3/03_methodology}
57 +\input{chapitre3/04_results}
58 +\input{chapitre3/05_conclusion}
59 +
60 +%% --- Tableaux et figures de l'article ---
61 +\input{chapitre3/tables}
62 +\input{chapitre3/figures}
added these-ulaval/chapitre3/figures.tex +73 −0
@@ -0,0 +1,73 @@
1 +\section{Figures}
2 +
3 +% Figure 1: Intraday Volatility Pattern
4 +\begin{figure}[H]
5 +\centering
6 +\includegraphics[width=0.85\textwidth]{figD3_intraday_vol_pattern.pdf}
7 +\caption{Intraday volatility pattern on FOMC days. Red dashed line marks the 14:00~ET announcement. This figure documents the characteristic volatility spike at announcement time, motivating our high-frequency identification strategy. The pre-announcement period shows relatively stable volatility, while the post-announcement spike and gradual decay are consistent with information processing models.}
8 +\label{ch3:fig:D3}
9 +\end{figure}
10 +
11 +% Figure 2: Semantic Measures Scatter
12 +\begin{figure}[H]
13 +\centering
14 +\includegraphics[width=0.75\textwidth]{figD6_stance_novelty_scatter.pdf}
15 +\caption{FOMC semantic measures: stance vs.\ novelty (z-scored). Points colored by stance tercile. The low correlation between stance and novelty ($r = 0.12$) supports the conditional independence assumption (Assumption~\ref{ch3:ass:indep}) and motivates the separate estimation of tone and novelty effects in our decomposition analysis.}
16 +\label{ch3:fig:D6}
17 +\end{figure}
18 +
19 +% Figure 3: Semantic Time Series
20 +\begin{figure}[H]
21 +\centering
22 +\includegraphics[width=0.9\textwidth]{figD7_semantic_timeseries.pdf}
23 +\caption{Evolution of policy stance and novelty scores over time. Novelty peaks during the 2008 financial crisis and 2020 pandemic correspond to major policy regime changes. Stance shifts from dovish (2008--2015) to hawkish (2017--2019, 2022--2025) track well-known monetary policy cycles.}
24 +\label{ch3:fig:D7}
25 +\end{figure}
26 +
27 +% Figure 4: Heatmap of Event-Level Coefficients
28 +\begin{figure}[H]
29 +\centering
30 +\includegraphics[width=0.85\textwidth]{figR6a_rv_reg_heatmap_30min.pdf}
31 +\caption{Heatmap of event-level $\Delta$RV regression coefficients by ticker. Darker shading indicates stronger effects. Stance effects are negative and significant across most contracts, while novelty effects are positive. The stance$\times$novelty interaction is concentrated in VIX and Treasury securities, suggesting that the joint impact of tone and information content operates primarily through the uncertainty channel.}
32 +\label{ch3:fig:R6a}
33 +\end{figure}
34 +
35 +% Figure 5: Cumulative Drift by Stance
36 +\begin{figure}[H]
37 +\centering
38 +\includegraphics[width=0.85\textwidth]{figR8_drift_by_stance.pdf}
39 +\caption{Cumulative return drift by stance tercile. Dovish announcements generate positive equity drift that strengthens over 45 minutes, while hawkish announcements produce symmetric negative drift. The monotonic separation between terciles and gradual strengthening over time support H1c (tone effects increasing in magnitude) and the fundamental repricing interpretation.}
40 +\label{ch3:fig:R8}
41 +\end{figure}
42 +
43 +% Figure 6: IRF - Stance on Cumulative Returns
44 +\begin{figure}[H]
45 +\centering
46 +\includegraphics[width=0.85\textwidth]{figR10_irf_ret_h_stance.pdf}
47 +\caption{Impulse response: cumulative return to policy stance (post-announcement, Newey--West SE). The equity (ES) response builds gradually over the 120-minute window, consistent with slow fundamental repricing by heterogeneous investors (H1c). The VIX response is negative and peaks within 30 minutes, consistent with directionally clear statements resolving uncertainty quickly.}
48 +\label{ch3:fig:R10_stance}
49 +\end{figure}
50 +
51 +% Figure 7: IRF - Abnormal Return to Stance x Novelty
52 +\begin{figure}[H]
53 +\centering
54 +\includegraphics[width=0.85\textwidth]{figR10_irf_abn_h_stance_x_novelty.pdf}
55 +\caption{Impulse response: abnormal cumulative return to stance $\times$ novelty interaction. VIX shows persistent, significant effects from 5 to 120 minutes ($t = -5.06$ at peak), representing the most robust finding in our analysis. This confirms that the joint impact of tone and novelty on uncertainty is both economically large and statistically robust across the full post-announcement window.}
56 +\label{ch3:fig:R10_abn_interaction}
57 +\end{figure}
58 +
59 +% Figure 8: Placebo Tests
60 +\begin{figure}[H]
61 +\centering
62 +\includegraphics[width=0.85\textwidth]{figR10b_irf_placebo_stance.pdf}
63 +\caption{Pre-announcement placebo: IRF for stance. All coefficients are near zero and statistically insignificant across all tickers and horizons, validating our event study design. The absence of pre-announcement effects rules out information leakage and confirms that our identification strategy successfully isolates the causal impact of FOMC statement content.}
64 +\label{ch3:fig:placebo_stance}
65 +\end{figure}
66 +
67 +% Figure 9: Multi-Method Robustness Heatmap
68 +\begin{figure}[H]
69 +\centering
70 +\includegraphics[width=0.85\textwidth]{figRob2_multimethod_heatmap.pdf}
71 +\caption{Multi-method robustness heatmap: number of methods (out of 5) yielding $p < 0.10$ for each ticker--variable combination. Darker cells indicate more robust results. Stance effects on $\Delta$RV and $\Delta\log$RV are the most robust, with 3--4 methods confirming significance for ES, CL, and Treasury contracts. The interaction term is most robust for VIX, consistent with the uncertainty channel interpretation.}
72 +\label{ch3:fig:rob_heatmap}
73 +\end{figure}
added these-ulaval/chapitre3/figures/confidence_distribution.png +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figD1_return_density.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figD3_intraday_vol_pattern.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figD5_acf_returns.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figD6_stance_novelty_scatter.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figD7_semantic_timeseries.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figD8b_log_rv_distribution.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figR10_irf_abn_h_stance_x_novelty.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figR10_irf_ret_h_novelty.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figR10_irf_ret_h_stance.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figR10b_irf_placebo_novelty.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figR10b_irf_placebo_stance.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figR6a_rv_reg_heatmap_30min.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figR8_drift_by_stance.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figR9_drift_by_novelty.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figRob2_multimethod_heatmap.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/figRob_subperiod_heatmap.pdf +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/novelty_comparison.png +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/pca_axis_quality.png +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/scatter_comparison.png +0 −0

Binary file not shown.

added these-ulaval/chapitre3/figures/tone_comparison.png +0 −0

Binary file not shown.

added these-ulaval/chapitre3/tables.tex +32 −0
@@ -0,0 +1,32 @@
1 +\clearpage
2 +\section{Tables}
3 +
4 +% Table 1: Descriptive Statistics - 1-Minute Returns
5 +\input{chapitre3/tables/tableD1_desc_1min_returns.tex}
6 +
7 +% Table 2: Descriptive Statistics - Rolling Realized Measures
8 +\input{chapitre3/tables/tableD3_desc_rolling.tex}
9 +
10 +% Table 3-4: Panel Regressions (Main Results: log(RV) on Stance and Novelty)
11 +\input{chapitre3/tables/tableP_panel_logrv_stance.tex}
12 +\input{chapitre3/tables/tableP_panel_logrv_novelty.tex}
13 +
14 +% Table 5-9: Event-Level Regressions (Main Results)
15 +\input{chapitre3/tables/tableR_delta_rv_30min.tex}
16 +\input{chapitre3/tables/tableR_delta_log_rv_30min.tex}
17 +\input{chapitre3/tables/tableR_delta_beta_30min.tex}
18 +
19 +% Table 10-11: IRF Tables (Main Results: Stance and Novelty on Returns)
20 +\input{chapitre3/tables/tableR_irf_stance_ret_h.tex}
21 +\input{chapitre3/tables/tableR_irf_novelty_ret_h.tex}
22 +\input{chapitre3/tables/tableR_irf_stance_x_novelty_ret_h.tex}
23 +
24 +% Table 12-13: IRF Tables (Main Results: Abnormal Returns)
25 +\input{chapitre3/tables/tableR_irf_stance_abn_h.tex}
26 +\input{chapitre3/tables/tableR_irf_novelty_abn_h.tex}
27 +\input{chapitre3/tables/tableR_irf_stance_x_novelty_abn_h.tex}
28 +
29 +% Table 14-15: Multi-Method Robustness (Split into two manageable tables)
30 +\input{chapitre3/tables/tableR_robustness_30min_A.tex}
31 +\input{chapitre3/tables/tableR_robustness_30min_B.tex}
32 +
added these-ulaval/chapitre3/tables/tableD1_desc_1min_returns.tex +28 −0
@@ -0,0 +1,28 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Descriptive Statistics: 1-Minute Log Returns on FOMC Days}
6 +\label{ch3:tab:desc_1min_ret}
7 +\begin{threeparttable}
8 +\begin{tabular}{lrrrrrrrrrr}
9 +\toprule
10 +Ticker & $N$ & Mean & SD & Skew & Kurt & JB & Min & Median & Max & AC(1) \\
11 +\midrule
12 +E-mini S\&P 500 & 42,180 & -0.030 & 3.70 & 0.96 & 208.0 & 73858777$^{***}$ & -118.7 & 0.000 & 148.0 & -0.022 \\
13 +VIX Futures & 37,050 & 0.072 & 14.12 & 0.02 & 77.9 & 8665376$^{***}$ & -466.1 & 0.000 & 419.8 & -0.077 \\
14 +10Y T-Note & 42,180 & 0.009 & 1.07 & 1.21 & 44.4 & 3027292$^{***}$ & -13.8 & 0.000 & 30.8 & -0.103 \\
15 +5Y T-Note & 42,180 & 0.004 & 0.68 & -4.70 & 444.1 & 342093523$^{***}$ & -43.8 & 0.000 & 17.6 & -0.059 \\
16 +Dollar Index & 42,180 & -0.000 & 1.69 & -1.74 & 1508.6 & 3984214777$^{***}$ & -119.2 & 0.000 & 115.4 & 0.194 \\
17 +Crude Oil WTI & 42,180 & -0.006 & 5.14 & -0.64 & 166.7 & 47098526$^{***}$ & -191.8 & 0.000 & 143.9 & -0.062 \\
18 +Gold & 41,895 & 0.001 & 2.91 & 0.24 & 121.3 & 24424589$^{***}$ & -91.5 & 0.000 & 108.8 & -0.017 \\
19 +\bottomrule
20 +\end{tabular}
21 +\begin{tablenotes}[flushleft]
22 +\small
23 +\item \textit{Notes:} All statistics in basis points except $N$, AC(1), and JB. Sample includes all 1-minute observations on FOMC event days $\pm 1$ day with grid regularization and forward-filling of short gaps. Skewness and excess kurtosis are sample moments. JB is the Jarque--Bera test statistic. AC(1) denotes the first-order autocorrelation coefficient. Negative AC(1) values reflect bid--ask bounce effects typical of high-frequency data \citep{andersen1997}.
24 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
25 +\end{tablenotes}
26 +\end{threeparttable}
27 +\end{table}
28 +\end{landscape}
added these-ulaval/chapitre3/tables/tableD3_desc_rolling.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Descriptive Statistics: Rolling Realized Measures (5-Minute Window)}
6 +\label{ch3:tab:desc_rolling}
7 +\begin{threeparttable}
8 +\begin{tabular}{lrrrrrrrrrrr}
9 +\toprule
10 + & $N$ & \multicolumn{5}{c}{Realized Volatility (bps)} & \multicolumn{2}{c}{$\log(\text{RV})$} & \multicolumn{3}{c}{Realized Beta vs.\ ES} \\
11 +\cmidrule(lr){3-7} \cmidrule(lr){8-9} \cmidrule(lr){10-12}
12 +Contract & & Mean & SD & P5 & P50 & P95 & Mean & SD & Mean & SD & P50 \\
13 +\midrule
14 +E-mini S\&P 500 & 35,668 & 3.43 & 5.68 & 0.00 & 2.41 & 10.38 & -11.78 & 6.55 & 1.000 & 0.000 & 1.000 \\
15 +10Y T-Note & 35,668 & 1.41 & 1.63 & 0.00 & 1.31 & 3.81 & -14.11 & 7.09 & -0.018 & 0.395 & 0.000 \\
16 +5Y T-Note & 35,668 & 0.79 & 1.15 & 0.00 & 0.66 & 2.36 & -14.77 & 6.86 & -0.007 & 0.225 & 0.000 \\
17 +Dollar Index & 35,668 & 0.78 & 3.45 & 0.00 & 0.00 & 3.53 & -19.32 & 6.34 & -0.010 & 0.482 & 0.000 \\
18 +Crude Oil WTI & 35,668 & 5.86 & 9.47 & 0.00 & 3.97 & 17.90 & -11.24 & 6.67 & 0.184 & 1.472 & 0.006 \\
19 +Gold & 35,427 & 3.64 & 4.64 & 0.00 & 2.73 & 10.51 & -11.29 & 6.30 & 0.044 & 0.958 & 0.000 \\
20 +VIX Futures & 31,330 & 11.16 & 23.33 & 0.00 & 0.00 & 52.40 & -18.31 & 7.74 & -0.714 & 4.451 & 0.000 \\
21 +\bottomrule
22 +\end{tabular}
23 +\begin{tablenotes}[flushleft]
24 +\small
25 +\item \textit{Notes:} Statistics computed across all minute-level observations within $\pm 30$ minutes of FOMC announcements ($K = 5$ min rolling window, NA-tolerant with up to 20\% missing data). RV is the NA-tolerant rolling realized variance defined in Section~\ref{ch3:sec:methodology}, displayed in square-root (volatility) units in basis points; $\log(\text{RV})$ is computed on the realized variance in raw decimal-return units, so its level is not directly comparable to the bps columns. $\beta_t^{\text{real}} = \text{RCov}(r_i, r_{\text{ES}}) / \text{RVar}(r_{\text{ES}})$. ES beta is 1.000 by construction. $\log(\text{RV})$ is near-Gaussian, validating its use as a regression dependent variable.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added these-ulaval/chapitre3/tables/tableP_panel_beta_novelty.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: Beta on Post $\times$ Novelty (5-min)}
6 +\label{ch3:tab:panel_beta_novelty}
7 +\begin{threeparttable}
8 +\begin{tabular}{lcccccc}
9 +\toprule
10 + & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -0.6861*** & -0.0144 & -0.0114 & -0.0017 & 0.1629*** & 0.0538** \\
13 + & (0.1540) & (0.0120) & (0.0070) & (0.0011) & (0.0582) & (0.0268) \\
14 +Post $\times$ Novelty & -0.1361 & -0.0116 & -0.0064 & -0.0037 & -0.0295 & -0.0445 \\
15 + & (0.1765) & (0.0080) & (0.0048) & (0.0030) & (0.0706) & (0.0204) \\
16 +\midrule
17 +$N$ & 4,798 & 5,724 & 5,724 & 5,724 & 5,724 & 5,663 \\
18 +$R^2$ & 0.001 & 0.001 & 0.001 & 0.005 & 0.000 & 0.002 \\
19 +Adj.\ $R^2$ & 0.000 & 0.000 & 0.000 & 0.005 & 0.000 & 0.001 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: Realized Beta (5-min). Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added these-ulaval/chapitre3/tables/tableP_panel_beta_stance.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: Beta on Post $\times$ Stance (5-min)}
6 +\label{ch3:tab:panel_beta_stance}
7 +\begin{threeparttable}
8 +\begin{tabular}{lcccccc}
9 +\toprule
10 + & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -0.5789*** & -0.0206 & -0.0175** & -0.0023 & 0.0992 & 0.0405 \\
13 + & (0.1988) & (0.0153) & (0.0087) & (0.0016) & (0.0644) & (0.0342) \\
14 +Post $\times$ Stance & 0.1780 & -0.0179 & -0.0165 & -0.0023 & -0.1643 & -0.0415 \\
15 + & (0.2433) & (0.0178) & (0.0098) & (0.0017) & (0.0782) & (0.0404) \\
16 +\midrule
17 +$N$ & 4,798 & 5,724 & 5,724 & 5,724 & 5,724 & 5,663 \\
18 +$R^2$ & 0.000 & 0.001 & 0.002 & 0.001 & 0.003 & 0.001 \\
19 +Adj.\ $R^2$ & 0.000 & 0.001 & 0.002 & 0.001 & 0.003 & 0.000 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: Realized Beta (5-min). Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added these-ulaval/chapitre3/tables/tableP_panel_logrv_novelty.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: $\log$(RV) on Post $\times$ Novelty (5-min, $\pm$30 min)}
6 +\label{ch3:tab:panel_logrv_novelty}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -13.4894*** & -18.9261*** & -16.4259*** & -16.9510*** & -22.6505*** & -13.4467*** & -13.4834*** \\
13 + & (0.2892) & (0.3813) & (0.3396) & (0.3423) & (0.1015) & (0.2887) & (0.2946) \\
14 +Post $\times$ Novelty & -0.5169 & 0.3020 & -1.8142*** & -1.2188* & 0.8439 & -0.3328 & -0.4821 \\
15 + & (0.5996) & (0.8881) & (0.5683) & (0.5493) & (0.4881) & (0.6298) & (0.6316) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.003 & 0.001 & 0.031 & 0.015 & 0.069 & 0.001 & 0.002 \\
19 +Adj.\ $R^2$ & 0.002 & 0.001 & 0.031 & 0.015 & 0.068 & 0.001 & 0.002 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: $\log(\text{RV}_t(5))$. Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added these-ulaval/chapitre3/tables/tableP_panel_logrv_stance.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: $\log$(RV) on Post $\times$ Stance (5-min, $\pm$30 min)}
6 +\label{ch3:tab:panel_logrv_stance}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -14.8474*** & -19.9268*** & -17.7088*** & -17.9767*** & -22.6514*** & -14.9591*** & -15.0663*** \\
13 + & (0.4794) & (0.4462) & (0.4275) & (0.4466) & (0.1175) & (0.4704) & (0.4845) \\
14 +Post $\times$ Stance & -5.3084*** & -3.2805*** & -5.2530*** & -4.1583*** & 0.1481 & -5.8682*** & -6.0348*** \\
15 + & (0.8271) & (0.7803) & (0.6490) & (0.7052) & (0.2836) & (0.8526) & (0.8609) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.134 & 0.047 & 0.133 & 0.091 & 0.001 & 0.149 & 0.166 \\
19 +Adj.\ $R^2$ & 0.134 & 0.047 & 0.132 & 0.091 & 0.001 & 0.149 & 0.166 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: $\log(\text{RV}_t(5))$. Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added these-ulaval/chapitre3/tables/tableP_panel_return_novelty.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: Returns on Post $\times$ Novelty (5-min)}
6 +\label{ch3:tab:panel_return_novelty}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -0.02 & 0.00 & 0.01 & 0.00 & 0.01 & -0.07 & 0.04 \\
13 + & (0.03) & (0.19) & (0.01) & (0.01) & (0.01) & (0.10) & (0.03) \\
14 +Post $\times$ Novelty & -0.04 & 0.62 & 0.06 & 0.01 & 0.01 & -0.18 & 0.03 \\
15 + & (0.10) & (0.79) & (0.05) & (0.01) & (0.05) & (0.42) & (0.06) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.000 & 0.001 & 0.002 & 0.000 & 0.000 & 0.000 & 0.000 \\
19 +Adj.\ $R^2$ & -0.000 & 0.001 & 0.002 & -0.000 & -0.000 & 0.000 & -0.000 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: 1-Min Return (bps, 5-min). Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Coefficients $\times 10,000$. Semantic measures from MiniLM--BERT ensemble.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added these-ulaval/chapitre3/tables/tableP_panel_return_stance.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: Returns on Post $\times$ Stance (5-min)}
6 +\label{ch3:tab:panel_return_stance}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 0.01 & -0.18 & 0.01 & 0.01 & 0.03** & -0.02 & 0.04 \\
13 + & (0.04) & (0.16) & (0.01) & (0.01) & (0.01) & (0.05) & (0.03) \\
14 +Post $\times$ Stance & 0.11 & -0.41 & -0.02 & 0.02 & 0.05 & 0.15 & 0.02 \\
15 + & (0.09) & (0.52) & (0.04) & (0.01) & (0.02) & (0.30) & (0.05) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.000 & 0.000 & 0.000 & 0.000 & 0.001 & 0.000 & 0.000 \\
19 +Adj.\ $R^2$ & 0.000 & 0.000 & -0.000 & 0.000 & 0.000 & 0.000 & -0.000 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: 1-Min Return (bps, 5-min). Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Coefficients $\times 10,000$. Semantic measures from MiniLM--BERT ensemble.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added these-ulaval/chapitre3/tables/tableP_panel_rv_novelty.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: RV on Post $\times$ Novelty (5-min)}
6 +\label{ch3:tab:panel_rv_novelty}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 3.26*** & 10.79*** & 1.09*** & 0.64*** & 0.25*** & 5.85*** & 3.49*** \\
13 + & (0.33) & (1.54) & (0.09) & (0.07) & (0.08) & (0.62) & (0.27) \\
14 +Post $\times$ Novelty & 3.09 & 8.32 & 0.14 & 0.25 & 0.63 & 4.94 & 1.93 \\
15 + & (1.37) & (5.78) & (0.35) & (0.30) & (0.36) & (2.40) & (1.01) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.097 & 0.040 & 0.003 & 0.013 & 0.043 & 0.071 & 0.050 \\
19 +Adj.\ $R^2$ & 0.097 & 0.040 & 0.003 & 0.012 & 0.042 & 0.071 & 0.050 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: RV (bps, 5-min). Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added these-ulaval/chapitre3/tables/tableP_panel_rv_stance.tex +29 −0
@@ -0,0 +1,29 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Panel Regressions: RV on Post $\times$ Stance (5-min)}
6 +\label{ch3:tab:panel_rv_stance}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 2.79*** & 7.86*** & 0.88*** & 0.54*** & 0.23** & 4.32*** & 2.92*** \\
13 + & (0.40) & (1.56) & (0.10) & (0.08) & (0.09) & (0.56) & (0.30) \\
14 +Post $\times$ Stance & -1.27 & -7.49* & -0.76** & -0.35 & 0.05 & -4.97** & -1.79* \\
15 + & (0.91) & (3.41) & (0.28) & (0.24) & (0.22) & (1.92) & (0.85) \\
16 +\midrule
17 +$N$ & 9,028 & 7,930 & 9,028 & 9,028 & 9,028 & 9,028 & 8,967 \\
18 +$R^2$ & 0.008 & 0.021 & 0.042 & 0.012 & 0.000 & 0.036 & 0.021 \\
19 +Adj.\ $R^2$ & 0.008 & 0.021 & 0.042 & 0.012 & 0.000 & 0.036 & 0.021 \\
20 +\bottomrule
21 +\end{tabular}
22 +\begin{tablenotes}[flushleft]
23 +\small
24 +\item \textit{Notes:} Dependent variable: RV (bps, 5-min). Post$\times$Stance = $\mathbf{1}[t > 0] \times \text{Stance}$ (z-scored). Post$\times$Novelty = $\mathbf{1}[t > 0] \times \text{Novelty}$ (z-scored). $K = 5$ min rolling window, $\pm 30$ min event window. Clustered SE by event date. Stars: BH-adjusted.
25 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
26 +\end{tablenotes}
27 +\end{threeparttable}
28 +\end{table}
29 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_delta_beta_30min.tex +33 −0
@@ -0,0 +1,33 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Rolling Realized Beta Change Regressions (30-min Window)}
6 +\label{ch3:tab:rolling_delta_beta_30min}
7 +\begin{threeparttable}
8 +\begin{tabular}{lcccccc}
9 +\toprule
10 + & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & -0.2364 & 0.0098 & -0.0025 & 0.0058 & -0.0933 & 0.1275* \\
13 + & (0.1651) & (0.0192) & (0.0134) & (0.0074) & (0.0650) & (0.0758) \\
14 +Stance & -0.0190 & 0.0118 & 0.0195 & 0.0137 & -0.0008 & 0.1244 \\
15 + & (0.2653) & (0.0198) & (0.0158) & (0.0093) & (0.0891) & (0.0869) \\
16 +Novelty & 0.1096 & -0.0096 & -0.0064 & -0.0086 & -0.1290 & -0.0539 \\
17 + & (0.0917) & (0.0102) & (0.0073) & (0.0075) & (0.1020) & (0.0344) \\
18 +Stance $\times$ Novelty & -0.0818 & 0.0064 & 0.0074 & 0.0011 & -0.1772 & -0.0852 \\
19 + & (0.1018) & (0.0118) & (0.0074) & (0.0080) & (0.2106) & (0.0755) \\
20 +\midrule
21 +$N$ & 71 & 88 & 88 & 88 & 88 & 87 \\
22 +$R^2$ & 0.009 & 0.003 & 0.008 & 0.045 & 0.064 & 0.047 \\
23 +Adj.\ $R^2$ & -0.036 & -0.033 & -0.027 & 0.010 & 0.031 & 0.013 \\
24 +\bottomrule
25 +\end{tabular}
26 +\begin{tablenotes}[flushleft]
27 +\small
28 +\item \textit{Notes:} Dependent variable: $\Delta\hat{\beta}$ (change in rolling realized beta relative to ES, 30-min window). Coefficients are in beta units; a coefficient of $0.12$ means that a one-standard-deviation increase in the regressor is associated with a $0.12$ increase in co-movement with the S\&P~500. All regressors z-scored. Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Semantic measures from MiniLM--BERT ensemble.
29 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
30 +\end{tablenotes}
31 +\end{threeparttable}
32 +\end{table}
33 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_delta_log_rv_30min.tex +33 −0
@@ -0,0 +1,33 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Rolling Delta log(RV) Regressions (30-min Window, Interaction Model)}
6 +\label{ch3:tab:rolling_delta_log_rv_30min}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 6.1744*** & 6.8238*** & 7.1287*** & 6.9380*** & -0.3664 & 8.4210*** & 8.4805*** \\
13 + & (1.0374) & (1.6303) & (1.0670) & (1.0305) & (0.3671) & (0.8868) & (0.8682) \\
14 +Stance & -2.9143 & -0.8191 & -2.5521 & -2.3930 & -0.9082 & -2.1849 & -1.7794 \\
15 + & (1.0198) & (1.8314) & (1.1059) & (1.0661) & (0.5066) & (1.0687) & (1.0675) \\
16 +Novelty & 0.3441 & -0.2304 & -0.7730 & -0.6468 & 0.5076 & 0.8902 & 0.7810 \\
17 + & (0.6037) & (0.9547) & (0.5429) & (0.5328) & (0.5352) & (0.4145) & (0.3970) \\
18 +Stance $\times$ Novelty & -0.6767 & -1.5107 & -1.0145 & -1.0850 & -0.3540 & -0.7406 & -0.6972 \\
19 + & (0.6652) & (0.7161) & (0.5278) & (0.5025) & (0.7771) & (0.6279) & (0.6369) \\
20 +\midrule
21 +$N$ & 148 & 130 & 148 & 148 & 148 & 148 & 147 \\
22 +$R^2$ & 0.064 & 0.018 & 0.075 & 0.071 & 0.055 & 0.052 & 0.041 \\
23 +Adj.\ $R^2$ & 0.044 & -0.005 & 0.056 & 0.051 & 0.036 & 0.033 & 0.021 \\
24 +\bottomrule
25 +\end{tabular}
26 +\begin{tablenotes}[flushleft]
27 +\small
28 +\item \textit{Notes:} Dependent variable: $\Delta\log(\text{RV})$ (Rolling 30-min). Coefficients are in log-units; a coefficient of $-2.91$ on Stance for ES means that a one-standard-deviation hawkish shift is associated with a $2.91$ log-unit decrease in realized volatility. All regressors z-scored. Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Semantic measures from MiniLM--BERT ensemble.
29 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
30 +\end{tablenotes}
31 +\end{threeparttable}
32 +\end{table}
33 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_delta_rv_30min.tex +33 −0
@@ -0,0 +1,33 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Rolling Delta RV Regressions (30-min Window, Interaction Model)}
6 +\label{ch3:tab:rolling_delta_rv_30min}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 5.54*** & 26.00*** & 2.23*** & 1.65*** & -0.40 & 12.32*** & 9.96*** \\
13 + & (1.67) & (6.51) & (0.64) & (0.49) & (0.62) & (2.02) & (1.51) \\
14 +Stance & -4.94** & -2.80 & -1.50* & -1.30* & -1.65 & -7.78** & -1.05 \\
15 + & (1.53) & (6.03) & (0.71) & (0.64) & (0.90) & (2.65) & (1.92) \\
16 +Novelty & 6.82* & 3.42 & 0.70 & 0.94 & 1.15 & 8.20* & 3.34* \\
17 + & (3.42) & (5.98) & (0.81) & (0.83) & (1.09) & (3.63) & (1.35) \\
18 +Stance $\times$ Novelty & -5.32 & -7.96** & -1.66* & -2.48* & -1.20 & -8.61* & -4.68 \\
19 + & (3.15) & (2.55) & (0.83) & (1.12) & (1.39) & (3.60) & (3.13) \\
20 +\midrule
21 +$N$ & 148 & 130 & 148 & 148 & 148 & 148 & 147 \\
22 +$R^2$ & 0.286 & 0.054 & 0.131 & 0.245 & 0.137 & 0.268 & 0.128 \\
23 +Adj.\ $R^2$ & 0.271 & 0.032 & 0.113 & 0.229 & 0.119 & 0.253 & 0.109 \\
24 +\bottomrule
25 +\end{tabular}
26 +\begin{tablenotes}[flushleft]
27 +\small
28 +\item \textit{Notes:} Dependent variable: Delta RV (bps) (Rolling 30-min). Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Coefficients $\times 10,000$. Semantic measures from MiniLM--BERT ensemble.
29 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
30 +\end{tablenotes}
31 +\end{threeparttable}
32 +\end{table}
33 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_irf_novelty_abn_h.tex +36 −0
@@ -0,0 +1,36 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Novelty on Abnormal Cumulative Return}
6 +\label{ch3:tab:irf_novelty_abn_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & -- & -- & -- & -- & -- \\
13 +VIX Futures & 10.21 & 1.84 & -4.80 & -10.09 & 7.18 \\
14 + & (9.77) & (10.47) & (8.32) & (6.08) & (10.24) \\
15 +10Y T-Note & 1.81 & 2.59 & 2.17 & 2.48 & 7.61 \\
16 + & (0.97) & (2.14) & (1.87) & (2.07) & (3.41) \\
17 +5Y T-Note & 0.38 & 1.26 & 0.81 & 0.61 & 5.00 \\
18 + & (0.58) & (1.71) & (1.37) & (1.91) & (2.57) \\
19 +Dollar Index & 1.40 & 1.72 & 1.26 & 1.56 & 4.51 \\
20 + & (1.48) & (2.77) & (2.28) & (2.92) & (3.68) \\
21 +Crude Oil WTI & 0.47 & 0.60 & -3.13 & -1.42 & 2.62 \\
22 + & (3.34) & (3.64) & (7.50) & (6.75) & (5.73) \\
23 +Gold & -0.30 & 2.34 & 1.43 & -0.21 & 6.27 \\
24 + & (1.46) & (3.08) & (2.70) & (2.63) & (2.70) \\
25 +\midrule
26 +$N$ & 148 & 148 & 148 & 148 & 148 \\
27 +\bottomrule
28 +\end{tabular}
29 +\begin{tablenotes}[flushleft]
30 +\small
31 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Novelty reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. ES entries are omitted because abnormal returns are defined relative to the ES benchmark and are zero by construction. Pre-announcement placebo test passed: no significant pre-event coefficients.
32 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
33 +\end{tablenotes}
34 +\end{threeparttable}
35 +\end{table}
36 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_irf_novelty_ret_h.tex +37 −0
@@ -0,0 +1,37 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Novelty on Cumulative Return}
6 +\label{ch3:tab:irf_novelty_ret_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & -0.99 & -1.42 & -0.65 & -0.61 & -3.91 \\
13 + & (0.56) & (1.70) & (1.34) & (1.82) & (2.25) \\
14 +VIX Futures & 10.24 & 4.84 & -2.55 & -6.02 & 6.22 \\
15 + & (9.19) & (9.31) & (7.62) & (5.36) & (8.33) \\
16 +10Y T-Note & 0.82 & 1.17 & 1.52 & 1.87 & 3.70 \\
17 + & (0.60) & (0.97) & (0.91) & (1.09) & (2.15) \\
18 +5Y T-Note & -0.61 & -0.16 & 0.16 & -0.00 & 1.08 \\
19 + & (0.41) & (0.10) & (0.17) & (0.27) & (0.47) \\
20 +Dollar Index & 0.41 & 0.30 & 0.61 & 0.94 & 0.60 \\
21 + & (1.34) & (1.74) & (1.51) & (1.62) & (2.22) \\
22 +Crude Oil WTI & -0.52 & -0.82 & -3.78 & -2.03 & -1.29 \\
23 + & (3.54) & (4.29) & (7.83) & (6.83) & (6.20) \\
24 +Gold & -1.30 & 0.84 & 0.55 & -1.13 & 2.00 \\
25 + & (1.42) & (2.62) & (2.33) & (1.89) & (2.36) \\
26 +\midrule
27 +$N$ & 148 & 148 & 148 & 148 & 148 \\
28 +\bottomrule
29 +\end{tabular}
30 +\begin{tablenotes}[flushleft]
31 +\small
32 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Novelty reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. Pre-announcement placebo test passed: no significant pre-event coefficients.
33 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
34 +\end{tablenotes}
35 +\end{threeparttable}
36 +\end{table}
37 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_irf_stance_abn_h.tex +36 −0
@@ -0,0 +1,36 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Policy Stance on Abnormal Cumulative Return}
6 +\label{ch3:tab:irf_stance_abn_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & -- & -- & -- & -- & -- \\
13 +VIX Futures & -15.04 & -22.74 & -28.86 & -28.65 & -30.54 \\
14 + & (7.07) & (9.76) & (10.84) & (16.52) & (37.21) \\
15 +10Y T-Note & -0.85 & -3.12 & -4.67 & -8.29 & -15.86 \\
16 + & (1.20) & (2.81) & (3.59) & (4.03) & (8.44) \\
17 +5Y T-Note & 0.66 & -1.35 & -2.86 & -6.03 & -12.68 \\
18 + & (0.85) & (2.20) & (3.10) & (3.63) & (7.84) \\
19 +Dollar Index & 0.78 & -0.55 & -2.72 & -7.27 & -15.99 \\
20 + & (0.92) & (1.86) & (2.63) & (3.73) & (8.39) \\
21 +Crude Oil WTI & -2.33 & -2.79 & 0.99 & -3.76 & -2.65 \\
22 + & (3.71) & (5.05) & (7.17) & (8.73) & (8.44) \\
23 +Gold & -1.77 & -5.75 & -5.34 & -8.84** & -13.76 \\
24 + & (1.25) & (2.57) & (2.54) & (2.66) & (5.73) \\
25 +\midrule
26 +$N$ & 148 & 148 & 148 & 148 & 148 \\
27 +\bottomrule
28 +\end{tabular}
29 +\begin{tablenotes}[flushleft]
30 +\small
31 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Policy Stance reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. ES entries are omitted because abnormal returns are defined relative to the ES benchmark and are zero by construction. Pre-announcement placebo test passed: no significant pre-event coefficients.
32 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
33 +\end{tablenotes}
34 +\end{threeparttable}
35 +\end{table}
36 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_irf_stance_ret_h.tex +37 −0
@@ -0,0 +1,37 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Policy Stance on Cumulative Return}
6 +\label{ch3:tab:irf_stance_ret_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & 0.23 & 1.98 & 3.76 & 6.82 & 11.98 \\
13 + & (0.92) & (2.11) & (2.76) & (3.31) & (6.76) \\
14 +VIX Futures & -14.74 & -19.11 & -22.03 & -17.73 & -14.27 \\
15 + & (6.26) & (8.33) & (9.47) & (14.62) & (31.57) \\
16 +10Y T-Note & -0.61 & -1.14 & -0.90 & -1.47 & -3.88 \\
17 + & (0.43) & (0.85) & (0.93) & (0.92) & (2.20) \\
18 +5Y T-Note & 0.89 & 0.63 & 0.90 & 0.79 & -0.71 \\
19 + & (0.34) & (0.35) & (0.55) & (0.51) & (1.34) \\
20 +Dollar Index & 1.01 & 1.43 & 1.05 & -0.45 & -4.01 \\
21 + & (0.43) & (0.61) & (0.45) & (0.89) & (2.14) \\
22 +Crude Oil WTI & -2.10 & -0.81 & 4.75 & 3.06 & 9.32 \\
23 + & (4.28) & (6.28) & (8.46) & (10.23) & (9.69) \\
24 +Gold & -1.47 & -3.08 & 0.51 & 0.62 & 1.31 \\
25 + & (1.30) & (1.42) & (2.04) & (2.46) & (3.96) \\
26 +\midrule
27 +$N$ & 148 & 148 & 148 & 148 & 148 \\
28 +\bottomrule
29 +\end{tabular}
30 +\begin{tablenotes}[flushleft]
31 +\small
32 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Policy Stance reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. Pre-announcement placebo test passed: no significant pre-event coefficients.
33 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
34 +\end{tablenotes}
35 +\end{threeparttable}
36 +\end{table}
37 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_irf_stance_x_novelty_abn_h.tex +36 −0
@@ -0,0 +1,36 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Stance $\times$ Novelty on Abnormal Cumulative Return}
6 +\label{ch3:tab:irf_stance_x_novelty_abn_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & -- & -- & -- & -- & -- \\
13 +VIX Futures & -31.74** & -58.80*** & -72.89*** & -65.76*** & -53.38* \\
14 + & (9.50) & (14.98) & (18.63) & (13.00) & (17.64) \\
15 +10Y T-Note & -4.54 & -10.44 & -11.95 & -14.24 & -12.05 \\
16 + & (2.09) & (5.70) & (6.03) & (6.30) & (6.89) \\
17 +5Y T-Note & -1.63 & -6.94 & -8.03 & -9.66 & -5.72 \\
18 + & (0.99) & (4.18) & (4.36) & (4.58) & (4.66) \\
19 +Dollar Index & 0.18 & -2.85 & -5.11 & -6.15 & -3.52 \\
20 + & (1.65) & (3.98) & (4.37) & (5.05) & (4.70) \\
21 +Crude Oil WTI & 4.84 & 0.86 & 10.90 & 6.59 & 12.00 \\
22 + & (6.44) & (8.45) & (14.43) & (16.04) & (15.12) \\
23 +Gold & -1.30 & -12.83 & -10.31 & -7.35 & -6.73 \\
24 + & (1.32) & (5.31) & (4.61) & (3.60) & (4.95) \\
25 +\midrule
26 +$N$ & 148 & 148 & 148 & 148 & 148 \\
27 +\bottomrule
28 +\end{tabular}
29 +\begin{tablenotes}[flushleft]
30 +\small
31 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Stance $\times$ Novelty reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. ES entries are omitted because abnormal returns are defined relative to the ES benchmark and are zero by construction. Pre-announcement placebo test passed: no significant pre-event coefficients.
32 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
33 +\end{tablenotes}
34 +\end{threeparttable}
35 +\end{table}
36 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_irf_stance_x_novelty_ret_h.tex +37 −0
@@ -0,0 +1,37 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Price IRF: Stance $\times$ Novelty on Cumulative Return}
6 +\label{ch3:tab:irf_stance_x_novelty_ret_h}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccc}
9 +\toprule
10 + & $h = 5$ min & $h = 15$ min & $h = 30$ min & $h = 60$ min & $h = 120$ min \\
11 +\midrule
12 +E-mini S\&P 500 & 2.78 & 6.91 & 8.27 & 9.47 & 6.35 \\
13 + & (1.36) & (4.14) & (4.39) & (4.61) & (4.49) \\
14 +VIX Futures & -28.02** & -47.97** & -61.74*** & -51.80*** & -44.06* \\
15 + & (8.62) & (12.75) & (15.54) & (10.66) & (14.53) \\
16 +10Y T-Note & -1.76 & -3.53 & -3.68 & -4.77 & -5.70 \\
17 + & (0.85) & (1.68) & (1.74) & (1.88) & (2.84) \\
18 +5Y T-Note & 1.15 & -0.03 & 0.24 & -0.19 & 0.63 \\
19 + & (0.57) & (0.10) & (0.16) & (0.28) & (0.46) \\
20 +Dollar Index & 2.96 & 4.05 & 3.16 & 3.32 & 2.83 \\
21 + & (1.15) & (1.53) & (1.27) & (1.41) & (2.21) \\
22 +Crude Oil WTI & 7.62 & 7.77 & 19.17 & 16.07 & 18.35 \\
23 + & (7.45) & (11.46) & (18.20) & (19.84) & (18.66) \\
24 +Gold & 1.48 & -5.97* & -2.19 & 1.94 & -0.60 \\
25 + & (1.59) & (2.08) & (1.57) & (2.34) & (2.13) \\
26 +\midrule
27 +$N$ & 148 & 148 & 148 & 148 & 148 \\
28 +\bottomrule
29 +\end{tabular}
30 +\begin{tablenotes}[flushleft]
31 +\small
32 +\item \textit{Notes:} Jord\`a (2005) local projection: $\text{CumRet}_i(0 \to h) = \alpha + \beta_1 \text{Stance} + \beta_2 \text{Novelty} + \beta_3 \text{Stance} \times \text{Novelty} + \varepsilon$. Coefficient on Stance $\times$ Novelty reported in basis points ($\times 10^4$). All regressors z-scored. Newey--West HAC standard errors in parentheses. Pre-announcement placebo test passed: no significant pre-event coefficients.
33 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$ (Benjamini--Hochberg adjusted within each (DV, model) family).
34 +\end{tablenotes}
35 +\end{threeparttable}
36 +\end{table}
37 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_log_rv_ratio_30min.tex +33 −0
@@ -0,0 +1,33 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Rolling log(RV Ratio) Regressions (30-min Window, Interaction Model)}
6 +\label{ch3:tab:rolling_log_rv_ratio_30min}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 0.2577*** & -0.6826 & -0.1044 & -0.0595 & -2.2662*** & 0.2341 & 0.4248 \\
13 + & (0.0810) & (0.6770) & (0.1403) & (0.0703) & (0.7194) & (0.3526) & (0.3908) \\
14 +Stance & -0.2575 & 0.4071 & -0.1965 & -0.1674 & -4.7532 & -0.7852 & -0.8420 \\
15 + & (0.1249) & (0.5125) & (0.1874) & (0.1886) & (1.5983) & (0.5421) & (0.5977) \\
16 +Novelty & -0.0173 & -0.1680 & 0.0871 & 0.1068 & -0.0110 & 0.3561 & 0.4141 \\
17 + & (0.0437) & (0.6219) & (0.1386) & (0.1650) & (0.3329) & (0.1975) & (0.1945) \\
18 +Stance $\times$ Novelty & 0.0890 & 0.1548 & 0.1046 & -0.0441 & 1.2119 & 0.1608 & -0.0041 \\
19 + & (0.0779) & (0.4344) & (0.0892) & (0.0755) & (0.7615) & (0.3288) & (0.3215) \\
20 +\midrule
21 +$N$ & 88 & 41 & 40 & 40 & 17 & 89 & 88 \\
22 +$R^2$ & 0.069 & 0.004 & 0.183 & 0.078 & 0.366 & 0.086 & 0.083 \\
23 +Adj.\ $R^2$ & 0.036 & -0.077 & 0.114 & 0.002 & 0.220 & 0.054 & 0.050 \\
24 +\bottomrule
25 +\end{tabular}
26 +\begin{tablenotes}[flushleft]
27 +\small
28 +\item \textit{Notes:} Dependent variable: log(RV Ratio) (Rolling 30-min). Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Semantic measures from MiniLM--BERT ensemble.
29 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
30 +\end{tablenotes}
31 +\end{threeparttable}
32 +\end{table}
33 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_robustness_30min_A.tex +63 −0
@@ -0,0 +1,63 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\scriptsize
5 +\caption{Robustness: Event-Level Regressions --- $\Delta$RV and $\Delta\log$RV (30-Minute Window, Interaction Model)}
6 +\label{ch3:tab:robustness_A}
7 +\begin{threeparttable}
8 +\begin{tabular}{lllrrllllll}
9 +\toprule
10 +Contract & Dep.\ Var. & Regressor & $\hat{\beta}$ & $t$ & $p_{\text{NW}}$ & $p_{\text{Boot}}$ & $p_{\text{Clust}}$ & $p_{\text{QReg}}$ & $p_{\text{Perm}}$ & \#Sig \\
11 +\midrule
12 +Crude Oil WTI & $\Delta$RV & Stance & -7.78 & -2.94 & 0.004 & 0.691 & 0.001 & -- & 0.007 & 3 \\
13 +Crude Oil WTI & $\Delta$RV & Novelty & 8.20 & 2.26 & 0.025 & 0.068 & 0.009 & -- & 0.000 & 4 \\
14 +Crude Oil WTI & $\Delta$RV & St. $\times$ Nov. & -8.61 & -2.39 & 0.018 & 0.212 & 0.014 & -- & 0.002 & 3 \\
15 +Dollar Index & $\Delta$RV & Stance & -1.65 & -1.83 & 0.069 & 0.260 & 0.095 & -- & 0.014 & 3 \\
16 +Dollar Index & $\Delta$RV & Novelty & 1.15 & 1.06 & 0.293 & 0.069 & 0.366 & -- & 0.014 & 2 \\
17 +Dollar Index & $\Delta$RV & St. $\times$ Nov. & -1.20 & -0.86 & 0.392 & 0.124 & 0.469 & -- & 0.047 & 1 \\
18 +E-mini S\&P 500 & $\Delta$RV & Stance & -4.94 & -3.22 & 0.002 & 0.584 & 0.011 & -- & 0.019 & 3 \\
19 +E-mini S\&P 500 & $\Delta$RV & Novelty & 6.82 & 2.00 & 0.048 & 0.013 & 0.028 & -- & 0.002 & 4 \\
20 +E-mini S\&P 500 & $\Delta$RV & St. $\times$ Nov. & -5.32 & -1.69 & 0.094 & 0.150 & 0.167 & -- & 0.024 & 2 \\
21 +Gold & $\Delta$RV & Stance & -1.05 & -0.55 & 0.583 & 0.692 & 0.602 & -- & 0.561 & 0 \\
22 +Gold & $\Delta$RV & Novelty & 3.34 & 2.48 & 0.014 & 0.318 & 0.056 & -- & 0.021 & 3 \\
23 +Gold & $\Delta$RV & St. $\times$ Nov. & -4.68 & -1.49 & 0.138 & 0.266 & 0.117 & -- & 0.025 & 1 \\
24 +VIX Futures & $\Delta$RV & Stance & -2.80 & -0.46 & 0.644 & 0.659 & 0.580 & -- & 0.619 & 0 \\
25 +VIX Futures & $\Delta$RV & Novelty & 3.42 & 0.57 & 0.568 & 0.492 & 0.486 & -- & 0.382 & 0 \\
26 +VIX Futures & $\Delta$RV & St. $\times$ Nov. & -7.96 & -3.12 & 0.002 & 0.601 & 0.058 & -- & 0.067 & 3 \\
27 +5Y T-Note & $\Delta$RV & Stance & -1.30 & -2.04 & 0.043 & 0.236 & 0.235 & -- & 0.050 & 2 \\
28 +5Y T-Note & $\Delta$RV & Novelty & 0.94 & 1.13 & 0.260 & 0.203 & 0.235 & -- & 0.041 & 1 \\
29 +5Y T-Note & $\Delta$RV & St. $\times$ Nov. & -2.48 & -2.22 & 0.028 & 0.070 & 0.149 & -- & 0.004 & 3 \\
30 +10Y T-Note & $\Delta$RV & Stance & -1.50 & -2.10 & 0.038 & 0.210 & 0.210 & -- & 0.025 & 2 \\
31 +10Y T-Note & $\Delta$RV & Novelty & 0.70 & 0.87 & 0.388 & 0.314 & 0.313 & -- & 0.082 & 1 \\
32 +10Y T-Note & $\Delta$RV & St. $\times$ Nov. & -1.66 & -1.99 & 0.048 & 0.135 & 0.250 & -- & 0.013 & 2 \\
33 +\midrule
34 +Crude Oil WTI & $\Delta\log$RV & Stance & -2.1849 & -2.04 & 0.043 & 0.442 & 0.035 & -- & 0.022 & 3 \\
35 +Crude Oil WTI & $\Delta\log$RV & Novelty & 0.8902 & 2.15 & 0.033 & 0.614 & 0.086 & -- & 0.135 & 2 \\
36 +Crude Oil WTI & $\Delta\log$RV & St. $\times$ Nov. & -0.7406 & -1.18 & 0.240 & 0.650 & 0.203 & -- & 0.326 & 0 \\
37 +Dollar Index & $\Delta\log$RV & Stance & -0.9082 & -1.79 & 0.075 & 0.337 & 0.121 & -- & 0.041 & 2 \\
38 +Dollar Index & $\Delta\log$RV & Novelty & 0.5076 & 0.95 & 0.345 & 0.209 & 0.478 & -- & 0.064 & 1 \\
39 +Dollar Index & $\Delta\log$RV & St. $\times$ Nov. & -0.3540 & -0.46 & 0.649 & 0.464 & 0.673 & -- & 0.247 & 0 \\
40 +E-mini S\&P 500 & $\Delta\log$RV & Stance & -2.9143 & -2.86 & 0.005 & 0.716 & 0.001 & -- & 0.004 & 3 \\
41 +E-mini S\&P 500 & $\Delta\log$RV & Novelty & 0.3441 & 0.57 & 0.570 & 0.639 & 0.609 & -- & 0.577 & 0 \\
42 +E-mini S\&P 500 & $\Delta\log$RV & St. $\times$ Nov. & -0.6767 & -1.02 & 0.311 & 0.644 & 0.249 & -- & 0.389 & 0 \\
43 +Gold & $\Delta\log$RV & Stance & -1.7794 & -1.67 & 0.098 & 0.433 & 0.093 & -- & 0.066 & 3 \\
44 +Gold & $\Delta\log$RV & Novelty & 0.7810 & 1.97 & 0.051 & 0.583 & 0.120 & -- & 0.168 & 1 \\
45 +Gold & $\Delta\log$RV & St. $\times$ Nov. & -0.6972 & -1.09 & 0.275 & 0.618 & 0.236 & -- & 0.334 & 0 \\
46 +VIX Futures & $\Delta\log$RV & Stance & -0.8191 & -0.45 & 0.655 & 0.610 & 0.563 & -- & 0.558 & 0 \\
47 +VIX Futures & $\Delta\log$RV & Novelty & -0.2304 & -0.24 & 0.810 & 0.816 & 0.794 & -- & 0.817 & 0 \\
48 +VIX Futures & $\Delta\log$RV & St. $\times$ Nov. & -1.5107 & -2.11 & 0.037 & 0.781 & 0.037 & -- & 0.192 & 2 \\
49 +5Y T-Note & $\Delta\log$RV & Stance & -2.3930 & -2.24 & 0.026 & 0.545 & 0.012 & -- & 0.014 & 3 \\
50 +5Y T-Note & $\Delta\log$RV & Novelty & -0.6468 & -1.21 & 0.227 & 0.624 & 0.164 & -- & 0.293 & 0 \\
51 +5Y T-Note & $\Delta\log$RV & St. $\times$ Nov. & -1.0850 & -2.16 & 0.032 & 0.827 & 0.024 & -- & 0.163 & 2 \\
52 +10Y T-Note & $\Delta\log$RV & Stance & -2.5521 & -2.31 & 0.022 & 0.504 & 0.010 & -- & 0.012 & 3 \\
53 +10Y T-Note & $\Delta\log$RV & Novelty & -0.7730 & -1.42 & 0.157 & 0.697 & 0.090 & -- & 0.224 & 1 \\
54 +10Y T-Note & $\Delta\log$RV & St. $\times$ Nov. & -1.0145 & -1.92 & 0.057 & 0.791 & 0.044 & -- & 0.205 & 2 \\
55 +\bottomrule
56 +\end{tabular}
57 +\begin{tablenotes}[flushleft]
58 +\scriptsize
59 +\item \textit{Notes:} This table reports multi-method robustness results for the event-level interaction model $\Delta Y_i = \alpha + \beta_1 \text{Stance}_i + \beta_2 \text{Novelty}_i + \beta_3 (\text{Stance}_i \times \text{Novelty}_i) + \varepsilon_i$ using 30-minute rolling windows. The upper panel reports $\Delta$RV coefficients in basis points; the lower panel reports $\Delta\log$RV coefficients in native units. All regressors are z-scored. Five inference methods are compared: $p_{\text{NW}}$ = Newey--West HAC; $p_{\text{Boot}}$ = wild bootstrap (1,999 replications, Rademacher weights); $p_{\text{Clust}}$ = clustered by event date (HC1); $p_{\text{QReg}}$ = quantile regression at median ($\tau = 0.5$); $p_{\text{Perm}}$ = permutation test (4,999 replications). \#Sig = number of methods yielding $p < 0.10$. A coefficient is considered robust when $\#\text{Sig} \geq 3$. Stance effects on $\Delta\log$RV are significant for 5 of 7 contracts with $\#\text{Sig} \geq 3$, confirming the volatility channel.
60 +\end{tablenotes}
61 +\end{threeparttable}
62 +\end{table}
63 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_robustness_30min_B.tex +60 −0
@@ -0,0 +1,60 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\scriptsize
5 +\caption{Robustness: Event-Level Regressions --- RV Ratio and $\Delta\beta$ (30-Minute Window, Interaction Model)}
6 +\label{ch3:tab:robustness_B}
7 +\begin{threeparttable}
8 +\begin{tabular}{lllrrllllll}
9 +\toprule
10 +Contract & Dep.\ Var. & Regressor & $\hat{\beta}$ & $t$ & $p_{\text{NW}}$ & $p_{\text{Boot}}$ & $p_{\text{Clust}}$ & $p_{\text{QReg}}$ & $p_{\text{Perm}}$ & \#Sig \\
11 +\midrule
12 +Crude Oil WTI & RV Ratio & Stance & 0.0611 & 0.13 & 0.895 & 0.894 & 0.894 & -- & 0.886 & 0 \\
13 +Crude Oil WTI & RV Ratio & Novelty & 0.4481 & 1.16 & 0.249 & 0.479 & 0.232 & -- & 0.123 & 0 \\
14 +Crude Oil WTI & RV Ratio & St. $\times$ Nov. & -0.1615 & -0.28 & 0.784 & 0.777 & 0.782 & -- & 0.694 & 0 \\
15 +Dollar Index & RV Ratio & Stance & -1.1473 & -1.50 & 0.158 & -- & -- & -- & -- & 0 \\
16 +Dollar Index & RV Ratio & Novelty & -0.4007 & -2.15 & 0.051 & -- & -- & -- & -- & 1 \\
17 +Dollar Index & RV Ratio & St. $\times$ Nov. & 0.3796 & 1.66 & 0.121 & -- & -- & -- & -- & 0 \\
18 +E-mini S\&P 500 & RV Ratio & Stance & -0.4896 & -1.62 & 0.110 & 0.846 & 0.021 & -- & 0.167 & 1 \\
19 +E-mini S\&P 500 & RV Ratio & Novelty & -0.1825 & -1.58 & 0.118 & 0.851 & 0.079 & -- & 0.403 & 1 \\
20 +E-mini S\&P 500 & RV Ratio & St. $\times$ Nov. & 0.2511 & 1.20 & 0.232 & 0.856 & 0.088 & -- & 0.382 & 1 \\
21 +Gold & RV Ratio & Stance & 0.0845 & 0.16 & 0.872 & 0.896 & 0.881 & -- & 0.876 & 0 \\
22 +Gold & RV Ratio & Novelty & 0.6432 & 1.41 & 0.162 & 0.432 & 0.159 & -- & 0.093 & 1 \\
23 +Gold & RV Ratio & St. $\times$ Nov. & -0.5527 & -1.08 & 0.284 & 0.650 & 0.254 & -- & 0.296 & 0 \\
24 +VIX Futures & RV Ratio & Stance & 0.4669 & 1.79 & 0.081 & 0.655 & 0.142 & -- & 0.214 & 1 \\
25 +VIX Futures & RV Ratio & Novelty & -0.3809 & -1.33 & 0.191 & 0.641 & 0.289 & -- & 0.290 & 0 \\
26 +VIX Futures & RV Ratio & St. $\times$ Nov. & -0.0118 & -0.08 & 0.940 & 0.984 & 0.954 & -- & 0.963 & 0 \\
27 +5Y T-Note & RV Ratio & Stance & -0.0861 & -0.65 & 0.523 & 0.559 & 0.486 & -- & 0.360 & 0 \\
28 +5Y T-Note & RV Ratio & Novelty & 0.0596 & 0.42 & 0.675 & 0.717 & 0.666 & -- & 0.667 & 0 \\
29 +5Y T-Note & RV Ratio & St. $\times$ Nov. & -0.0415 & -0.57 & 0.574 & 0.716 & 0.624 & -- & 0.666 & 0 \\
30 +10Y T-Note & RV Ratio & Stance & -0.0692 & -0.63 & 0.536 & 0.642 & 0.605 & -- & 0.476 & 0 \\
31 +10Y T-Note & RV Ratio & Novelty & 0.0436 & 0.36 & 0.717 & 0.781 & 0.775 & -- & 0.744 & 0 \\
32 +10Y T-Note & RV Ratio & St. $\times$ Nov. & 0.1231 & 1.54 & 0.133 & 0.562 & 0.182 & -- & 0.234 & 0 \\
33 +\midrule
34 +Crude Oil WTI & $\Delta\beta$ & Stance & -0.0008 & -0.01 & 0.993 & 0.998 & 0.994 & -- & 0.996 & 0 \\
35 +Crude Oil WTI & $\Delta\beta$ & Novelty & -0.1290 & -1.26 & 0.210 & 0.586 & 0.213 & -- & 0.149 & 0 \\
36 +Crude Oil WTI & $\Delta\beta$ & St. $\times$ Nov. & -0.1772 & -0.84 & 0.403 & 0.442 & 0.393 & -- & 0.157 & 0 \\
37 +Dollar Index & $\Delta\beta$ & Stance & 0.0137 & 1.48 & 0.144 & 0.456 & 0.162 & -- & 0.100 & 1 \\
38 +Dollar Index & $\Delta\beta$ & Novelty & -0.0086 & -1.15 & 0.254 & 0.391 & 0.310 & -- & 0.118 & 0 \\
39 +Dollar Index & $\Delta\beta$ & St. $\times$ Nov. & 0.0011 & 0.13 & 0.894 & 0.935 & 0.889 & -- & 0.813 & 0 \\
40 +Gold & $\Delta\beta$ & Stance & 0.1244 & 1.43 & 0.156 & 0.579 & 0.116 & -- & 0.157 & 0 \\
41 +Gold & $\Delta\beta$ & Novelty & -0.0539 & -1.56 & 0.121 & 0.721 & 0.127 & -- & 0.335 & 0 \\
42 +Gold & $\Delta\beta$ & St. $\times$ Nov. & -0.0852 & -1.13 & 0.262 & 0.646 & 0.177 & -- & 0.270 & 0 \\
43 +VIX Futures & $\Delta\beta$ & Stance & -0.0190 & -0.07 & 0.943 & 0.944 & 0.946 & -- & 0.936 & 0 \\
44 +VIX Futures & $\Delta\beta$ & Novelty & 0.1096 & 1.20 & 0.236 & 0.704 & 0.272 & -- & 0.497 & 0 \\
45 +VIX Futures & $\Delta\beta$ & St. $\times$ Nov. & -0.0818 & -0.80 & 0.425 & 0.757 & 0.495 & -- & 0.642 & 0 \\
46 +5Y T-Note & $\Delta\beta$ & Stance & 0.0195 & 1.23 & 0.221 & 0.583 & 0.393 & -- & 0.409 & 0 \\
47 +5Y T-Note & $\Delta\beta$ & Novelty & -0.0064 & -0.88 & 0.381 & 0.786 & 0.481 & -- & 0.639 & 0 \\
48 +5Y T-Note & $\Delta\beta$ & St. $\times$ Nov. & 0.0074 & 0.99 & 0.325 & 0.820 & 0.381 & -- & 0.693 & 0 \\
49 +10Y T-Note & $\Delta\beta$ & Stance & 0.0118 & 0.59 & 0.554 & 0.772 & 0.693 & -- & 0.714 & 0 \\
50 +10Y T-Note & $\Delta\beta$ & Novelty & -0.0096 & -0.94 & 0.349 & 0.736 & 0.403 & -- & 0.665 & 0 \\
51 +10Y T-Note & $\Delta\beta$ & St. $\times$ Nov. & 0.0064 & 0.55 & 0.585 & 0.849 & 0.599 & -- & 0.830 & 0 \\
52 +\bottomrule
53 +\end{tabular}
54 +\begin{tablenotes}[flushleft]
55 +\scriptsize
56 +\item \textit{Notes:} Continuation of Table~\ref{ch3:tab:robustness_A}. The upper panel reports RV Ratio coefficients (post/pre RV); the lower panel reports $\Delta\beta$ coefficients (change in realized beta relative to ES). All regressors are z-scored. The same five inference methods are used: Newey--West HAC, wild bootstrap, clustered SE, quantile regression, and permutation test. RV Ratio and $\Delta\beta$ results are generally weaker than the $\Delta$RV and $\Delta\log$RV results in Table~\ref{ch3:tab:robustness_A}, suggesting that the volatility channel operates primarily through level changes rather than ratio adjustments or co-movement shifts.
57 +\end{tablenotes}
58 +\end{threeparttable}
59 +\end{table}
60 +\end{landscape}
added these-ulaval/chapitre3/tables/tableR_rv_ratio_30min.tex +33 −0
@@ -0,0 +1,33 @@
1 +\begin{landscape}
2 +\begin{table}[htbp]
3 +\centering
4 +\small
5 +\caption{Rolling RV Ratio Regressions (30-min Window, Interaction Model)}
6 +\label{ch3:tab:rolling_rv_ratio_30min}
7 +\begin{threeparttable}
8 +\begin{tabular}{lccccccc}
9 +\toprule
10 + & E-mini S\&P 500 & VIX Futures & 10Y T-Note & 5Y T-Note & Dollar Index & Crude Oil WTI & Gold \\
11 +\midrule
12 +Intercept & 1.6276*** & 1.6260*** & 0.9991*** & 1.0135*** & 1.8017** & 2.9633*** & 3.7617*** \\
13 + & (0.1997) & (0.2639) & (0.0987) & (0.0537) & (0.6352) & (0.3545) & (0.4085) \\
14 +Stance & -0.4896 & 0.4669 & -0.0692 & -0.0861 & -1.1473 & 0.0611 & 0.0845 \\
15 + & (0.3029) & (0.2605) & (0.1108) & (0.1334) & (0.7665) & (0.4605) & (0.5241) \\
16 +Novelty & -0.1825 & -0.3809 & 0.0436 & 0.0596 & -0.4007 & 0.4481 & 0.6432 \\
17 + & (0.1157) & (0.2858) & (0.1196) & (0.1413) & (0.1864) & (0.3861) & (0.4561) \\
18 +Stance $\times$ Novelty & 0.2511 & -0.0118 & 0.1231 & -0.0415 & 0.3796 & -0.1615 & -0.5527 \\
19 + & (0.2085) & (0.1563) & (0.0801) & (0.0731) & (0.2289) & (0.5868) & (0.5128) \\
20 +\midrule
21 +$N$ & 88 & 41 & 40 & 40 & 17 & 89 & 88 \\
22 +$R^2$ & 0.050 & 0.050 & 0.129 & 0.025 & 0.145 & 0.035 & 0.044 \\
23 +Adj.\ $R^2$ & 0.016 & -0.027 & 0.057 & -0.056 & -0.053 & 0.001 & 0.010 \\
24 +\bottomrule
25 +\end{tabular}
26 +\begin{tablenotes}[flushleft]
27 +\small
28 +\item \textit{Notes:} Dependent variable: RV Ratio (Rolling 30-min). Clustered SE in parentheses. Stars: BH-adjusted $p$-values. Semantic measures from MiniLM--BERT ensemble.
29 +\item $^{***}p<0.01$, $^{**}p<0.05$, $^{*}p<0.10$.
30 +\end{tablenotes}
31 +\end{threeparttable}
32 +\end{table}
33 +\end{landscape}
added these-ulaval/conclusion/conclusion.tex +107 −0
@@ -0,0 +1,107 @@
1 +\chapter*{Conclusion} % ne pas numéroter
2 +\label{chap-conclusion} % étiquette pour renvois
3 +\phantomsection\addcontentsline{toc}{chapter}{\nameref{chap-conclusion}} % inclure dans TdM
4 +
5 +% BROUILLON À RÉVISER PAR L'AUTEUR — conclusion générale de la thèse.
6 +
7 +This dissertation set out to understand how information is transmitted to
8 +prices and volatility in commodities and financial futures markets, using the
9 +identification power of high-frequency data. Three essays examined three
10 +distinct transmission mechanisms: the interaction between macroeconomic news
11 +and speculative positioning in energy futures; the arbitrage-driven
12 +propagation of volatility between commodity ETFs and their underlying
13 +baskets; and the semantic channels through which central bank communication
14 +moves asset prices. This concluding chapter summarizes the findings, draws
15 +out what they imply jointly, acknowledges limitations, and sketches avenues
16 +for future research.
17 +
18 +\section*{Summary of contributions}
19 +
20 +The first essay showed that speculative trading in energy and commodity
21 +futures markets acts as a stabilizing force around macroeconomic
22 +announcements. Using 5-minute data over 2007--2024 and a time-varying measure
23 +of speculative intensity built from disaggregated CFTC positions, it found
24 +that higher speculative activity dampens the reaction of returns and
25 +conditional volatility to standardized macroeconomic surprises and narrows
26 +bid-ask spreads, with the effects concentrated among money managers rather
27 +than swap dealers, and stronger for procyclical energy commodities than for
28 +safe-haven gold. The essay thus provides sharply identified evidence against
29 +the view that speculation amplifies news-driven volatility, and speaks
30 +directly to the design of position limits.
31 +
32 +The second essay built the first long-sample, minute-level dataset of
33 +indicative net asset values for commodity ETFs and used it to measure
34 +volatility transmission between funds and their underlying assets. It found
35 +that the direction of transmission reflects the arbitrage technology ---
36 +one-way, from basket to fund, for physically backed precious metals;
37 +two-way and asymmetric for futures-based energy funds --- that transmission
38 +operates predominantly through jumps rather than the continuous component of
39 +volatility, and that 1-minute data reveal transmission up to twice as large
40 +as 30-minute estimates. The iNAV emerges as a measurement instrument of
41 +independent value for the microstructure of index-linked products.
42 +
43 +The third essay decomposed FOMC statements into policy tone and informational
44 +novelty using an ensemble of transformer language models with data-driven
45 +reference selection, and traced their effects through 1-minute futures data
46 +across 148 announcements. Tone predicts directional returns, building over
47 +two hours; novelty predicts volatility, resolving within roughly twenty
48 +minutes and with a clarifying rather than noise-inducing effect; and their
49 +interaction on VIX futures is the single most robust result. The essay shows
50 +that the informational content of central bank communication is
51 +multidimensional, and that its dimensions travel through different economic
52 +channels at different speeds.
53 +
54 +Across the three essays, a consistent methodological lesson emerges: the
55 +minute-level resolution is not a refinement but a prerequisite. The damping
56 +effect of speculation, the jump channel of ETF volatility transmission, and
57 +the divergent dynamics of tone and novelty are all features of the first
58 +minutes and hours after an information event; at the daily frequency they are
59 +attenuated, conflated, or invisible.
60 +
61 +\section*{Limitations}
62 +
63 +Several limitations qualify these results and delimit their scope. First,
64 +the measures of trader positioning in the first essay derive from weekly CFTC
65 +reports, so that speculative intensity is observed at a coarser frequency
66 +than the market reactions it conditions; the identification rests on the
67 +persistence of positioning rather than on its intraday variation. Second, the
68 +iNAV series of the second essay are constructed from disseminated and
69 +reconstructed data whose quality, while validated against official NAVs,
70 +cannot be audited tick by tick over the whole sample; and the analysis covers
71 +four large single-commodity funds, leaving open how the results extend to
72 +broad-basket or leveraged products. Third, the semantic measures of the third
73 +essay are estimated from a finite corpus of Federal Reserve communications;
74 +although reference selection is data-driven and robustness was assessed with
75 +multiple inference methods, language-model-based measures inevitably embed
76 +modelling choices, and the analysis concerns statements rather than the full
77 +communication apparatus (minutes, press conferences, speeches). Finally, all
78 +three essays are about the United States markets; the external validity of
79 +the findings for other trading venues and regulatory environments remains to
80 +be established.
81 +
82 +\section*{Avenues for future research}
83 +
84 +The results open several paths. On the policy side, the trader-level
85 +heterogeneity documented in the first essay --- stabilizing money managers,
86 +amplifying swap dealers --- invites a finer analysis of position-limit design
87 +and of the role of speculative capital in financing the energy transition,
88 +where volatility dampening lowers the real-option value of delaying
89 +investment. On the market-structure side, the jump-based transmission channel
90 +identified in the second essay suggests extending the iNAV apparatus to
91 +broad-basket, leveraged, and fixed-income ETFs, and studying how arbitrage
92 +frictions --- creation/redemption costs, settlement technology, market-maker
93 +inventory --- shape the propagation of discontinuous risk. On the
94 +communication side, the tone/novelty decomposition of the third essay can be
95 +carried to press conferences and minutes, to other central banks, and to the
96 +cross-section of individual assets, where the divergent horizons of the
97 +return and volatility channels may help separate expectations formation from
98 +uncertainty resolution. More broadly, the combination of language-model
99 +measurement with high-frequency identification --- used here for monetary
100 +policy --- applies to any recurring, text-borne information event: earnings
101 +calls, regulatory releases, geopolitical announcements.
102 +
103 +High-frequency data have turned questions that were once matters of narrative
104 +--- does speculation destabilize? do ETFs transmit shocks? do words move
105 +markets? --- into questions of measurement. The three essays of this
106 +dissertation are a step in that direction for commodities and financial
107 +futures markets.
added these-ulaval/introduction/introduction.tex +173 −0
@@ -0,0 +1,173 @@
1 +\chapter*{Introduction} % ne pas numéroter
2 +\label{chap-introduction} % étiquette pour renvois
3 +\phantomsection\addcontentsline{toc}{chapter}{\nameref{chap-introduction}} % inclure dans TdM
4 +
5 +% BROUILLON À RÉVISER PAR L'AUTEUR — introduction générale de la thèse
6 +% (rédigée à partir des trois articles ; les renvois bibliographiques utilisent
7 +% la bibliographie consolidée de la thèse).
8 +
9 +Futures markets are where much of the world's price discovery takes place.
10 +Crude oil, natural gas, gold, equity indexes, Treasury notes: for each of
11 +these assets, the futures contract is typically the most liquid instrument,
12 +the first to react to news, and the reference point from which spot prices,
13 +inventories, and investment decisions are set. Understanding how information
14 +is impounded into futures prices --- and how that process shapes volatility
15 +--- is therefore central to asset pricing, to market design, and to a series
16 +of policy debates that have accompanied the transformation of these markets
17 +over the past two decades: the growth of speculative and index-based trading
18 +in commodities \citep{tang2012index, cheng2014financialization}, the rise of
19 +exchange-traded funds (ETFs) as the retail gateway to commodity exposure, and
20 +the increasing weight of central bank communication as a market-moving event
21 +in its own right.
22 +
23 +A common empirical obstacle runs through these debates. Information arrives
24 +in minutes, but much of the literature has measured its effects in days.
25 +Daily data conflate the announcement of interest with everything else that
26 +happens the same day, attenuate estimates through microstructure noise and
27 +intraday reversals, and are silent about the dynamics --- amplification,
28 +absorption, resolution of uncertainty --- that unfold in the first minutes
29 +and hours after an information event \citep{andersen2003micro,
30 +kothari2007econometrics}. The high-frequency literature pioneered by
31 +\citet{andersen1998deutsche} and extended by realized-measure econometrics
32 +\citep{andersen2001distribution, barndorff2004power, corsi2009simple} showed
33 +that intraday data are not merely more precise: they change what can be
34 +identified. Event windows measured in minutes isolate the causal effect of a
35 +single release; realized variance turns latent volatility into an observable;
36 +and the separation of continuous and jump components distinguishes gradual
37 +information diffusion from discrete repricing.
38 +
39 +This dissertation applies that identification strategy to three questions
40 +about commodities and financial futures markets. Each essay constructs a new
41 +high-frequency dataset, each exploits a well-defined information event or
42 +transmission channel, and each documents effects whose magnitude, direction,
43 +or even sign would be invisible at the daily frequency. The common thread is
44 +the same throughout: \emph{who} transmits information to prices, through
45 +\emph{which channel}, and at \emph{what horizon}.
46 +
47 +\section*{Essay 1: Speculative trading and macroeconomic surprises in energy
48 + markets}
49 +
50 +The first essay addresses one of the most persistent controversies in
51 +commodity market policy: whether speculative trading destabilizes energy
52 +markets. The commodity price run-up of 2004--2008 spawned an influential
53 +narrative --- the ``Masters hypothesis'' --- according to which the growing
54 +presence of financial traders amplifies price movements and volatility, with
55 +harmful consequences for the real economy. The theoretical literature shows
56 +that such distortions are possible \citep{basak2016model,
57 +goldstein2022commodity}, but the empirical evidence, largely based on daily
58 +data and Granger-causality designs, is inconclusive \citep{fattouh2013role,
59 +irwin2012testing}.
60 +
61 +The essay takes a different route. Macroeconomic announcement releases
62 +provide sharply identified, exogenous information shocks whose surprise
63 +component can be standardized following \citet{balduzzi2001economic}. Using
64 +5-minute futures data for two energy commodities (crude oil and natural gas)
65 +and four metals (gold, silver, copper, and palladium) from April 2007 to
66 +February 2024, together with a time-varying, commodity-specific measure of
67 +speculative intensity built from the CFTC's disaggregated Commitments of
68 +Traders data, the essay asks whether the reaction of returns, volatility, and
69 +bid-ask spreads to macroeconomic surprises is amplified or dampened when
70 +speculative trading is more intense. The answer is unambiguous: the
71 +interaction between surprises and speculative intensity is consistently
72 +stabilizing. Prices and conditional volatility react \emph{less} to surprises
73 +when speculative positions are larger, and bid-ask spreads narrow rather than
74 +widen. Disaggregating trader categories shows that these beneficial effects
75 +are driven by money managers --- informed traders in the sense of
76 +\citet{cheng2015convective} --- rather than swap dealers. For energy markets,
77 +where volatility feeds directly into household energy costs and investment
78 +under uncertainty, the finding that speculation dampens rather than amplifies
79 +news-driven volatility speaks directly to the design of position limits and
80 +to the broader regulatory debate.
81 +
82 +\section*{Essay 2: The iNAV and volatility transmission between commodity
83 + ETFs and their underlying assets}
84 +
85 +The second essay turns from the futures market itself to the fastest-growing
86 +channel through which investors access commodities: exchange-traded funds. An
87 +ETF is an equity-like wrapper around an underlying basket --- physical
88 +bullion for gold and silver funds, futures positions for oil and natural gas
89 +funds --- and the arbitrage mechanism that keeps the wrapper aligned with its
90 +contents is also a conduit for volatility \citep{ben2018etfs}. Measuring that
91 +conduit precisely requires observing the value of the basket at the same
92 +frequency as the ETF price. The essay's central contribution is the
93 +construction of a novel minute-level dataset of indicative net asset values
94 +(iNAV) for four commodity ETFs (GLD, SLV, USO, UNG), validated against
95 +official NAVs and exchange-disseminated values, over more than a decade.
96 +
97 +With ETF and basket volatilities observable minute by minute, the essay
98 +measures transmission using HAR-type cascade models \citep{corsi2009simple}
99 +extended with cross-asset terms, decomposes realized variance into continuous
100 +and jump components, and complements the analysis with Bayesian vector
101 +autoregressions. Three results stand out. First, the direction of
102 +transmission mirrors the arbitrage technology: for physically backed precious
103 +metal funds, volatility flows one way, from the basket to the ETF; for
104 +futures-based energy funds, transmission is bidirectional and asymmetric.
105 +Second, transmission operates primarily through jumps rather than through the
106 +continuous component --- a channel that standard connectedness measures,
107 +built on daily total volatility, largely obscure. Third, sampling frequency
108 +matters quantitatively: 1-minute data reveal transmission up to twice as
109 +large as 30-minute estimates. Beyond the substantive findings, the essay
110 +demonstrates that the informational content of the iNAV --- a series
111 +disseminated in real time but rarely archived or studied --- provides a
112 +sharper image of the ETF--underlying relationship than price-based proxies.
113 +
114 +\section*{Essay 3: Policy tone, informational novelty, and the
115 + high-frequency reaction to FOMC announcements}
116 +
117 +The third essay studies the information events that arguably matter most for
118 +financial futures: announcements of the Federal Open Market Committee. A
119 +large literature measures monetary policy surprises through interest-rate
120 +futures \citep{kuttner2001, gurkaynak2005} or through the joint reaction of
121 +rates and equities \citep{nakamura2018}. Yet FOMC statements are texts, and
122 +their market impact depends not only on what they imply for the policy rate
123 +but on how they say it. The essay decomposes each statement into two semantic
124 +dimensions: \emph{policy tone} --- the hawkish-versus-dovish orientation of
125 +the language --- and \emph{informational novelty} --- the distance between a
126 +statement and its predecessor. The measurement apparatus is a dual-model
127 +ensemble of transformer language models (MiniLM and BERT) fine-tuned on
128 +Federal Reserve communications, with reference statements selected by a
129 +data-driven, PCA-based procedure that removes researcher discretion.
130 +
131 +Combining these measures with 1-minute data for seven futures contracts ---
132 +equities, VIX, Treasuries, the dollar, crude oil, and gold --- across 148
133 +FOMC announcements between 2008 and 2025, the essay documents a clean
134 +division of labour between the two dimensions. Tone predicts directional
135 +returns: a one-standard-deviation dovish shift is associated with equity
136 +gains that accumulate to roughly twelve basis points over two hours. Novelty
137 +predicts volatility, with a sign opposite to the naive prediction: statements
138 +that depart more from their predecessor \emph{reduce} uncertainty, consistent
139 +with communication acting as clarification rather than noise. The
140 +tone--novelty interaction on VIX futures is the most robust effect in the
141 +paper, and pre-announcement placebo tests together with five independent
142 +inference methods support a causal reading. The two channels also operate on
143 +different clocks --- volatility effects resolve within roughly twenty
144 +minutes, return effects build for two hours --- a pattern inconsistent with
145 +any single-channel account of central bank communication.
146 +
147 +\section*{Contributions and structure of the dissertation}
148 +
149 +Taken together, the three essays make three kinds of contributions. First,
150 +they build and document new high-frequency datasets --- standardized
151 +macroeconomic surprises matched to 5-minute commodity futures, a minute-level
152 +iNAV panel for commodity ETFs, and a semantically decomposed corpus of FOMC
153 +communications aligned with 1-minute futures data --- each of which has value
154 +beyond the questions asked here. Second, they deliver identified answers to
155 +long-standing questions: speculation stabilizes rather than destabilizes
156 +energy futures around news; ETF--underlying volatility transmission is
157 +jump-driven and direction-dependent; and central bank communication moves
158 +returns through tone but volatility through novelty. Third, they carry policy
159 +implications --- for the calibration of speculative position limits, for the
160 +design and monitoring of commodity index products, and for the communication
161 +strategy of central banks.
162 +
163 +The remainder of the dissertation is organized as follows.
164 +Chapter~\ref{chap-chapitre1} presents the first essay, on speculative trading
165 +and macroeconomic surprises in energy futures markets.
166 +Chapter~\ref{chap-chapitre2} presents the second essay, on iNAV-based
167 +volatility transmission between commodity ETFs and their underlying baskets.
168 +Chapter~\ref{chap-chapitre3} presents the third essay, on the high-frequency
169 +effects of the tone and novelty of FOMC statements. A general conclusion
170 +summarizes the findings, discusses their limitations, and outlines avenues
171 +for future research. Appendices~\ref{ch3:sec:proofs}
172 +and~\ref{ch3:sec:app_additional} collect the mathematical proofs and
173 +additional results of Chapter~\ref{chap-chapitre3}.
added these-ulaval/main.bbl +852 −0
@@ -0,0 +1,852 @@
1 +\begin{thebibliography}{}
2 +
3 +\bibitem[\protect\citeauthoryear{Ackert and Tian}{Ackert and
4 + Tian}{2000}]{ackert2000arbitrage}
5 +Ackert, L.~F. and Y.~S. Tian (2000).
6 +\newblock Arbitrage and valuation in the market for {S}tandard \& {P}oor's
7 + depositary receipts.
8 +\newblock {\em Financial Management\/}~{\em 29\/}(3), 71--87.
9 +
10 +\bibitem[\protect\citeauthoryear{Alquist and Gervais}{Alquist and
11 + Gervais}{2013}]{alquist2013role}
12 +Alquist, R. and O.~Gervais (2013).
13 +\newblock The role of financial speculation in driving the price of crude oil.
14 +\newblock {\em The Energy Journal\/}~{\em 34\/}(3), 35--54.
15 +
16 +\bibitem[\protect\citeauthoryear{Andersen and Bollerslev}{Andersen and
17 + Bollerslev}{1997}]{andersen1997}
18 +Andersen, T.~G. and T.~Bollerslev (1997).
19 +\newblock Intraday periodicity and volatility persistence in financial markets.
20 +\newblock {\em Journal of Empirical Finance\/}~{\em 4\/}(2--3), 115--158.
21 +
22 +\bibitem[\protect\citeauthoryear{Andersen and Bollerslev}{Andersen and
23 + Bollerslev}{1998}]{andersen1998deutsche}
24 +Andersen, T.~G. and T.~Bollerslev (1998).
25 +\newblock Deutsche mark--dollar volatility: Intraday activity patterns,
26 + macroeconomic announcements, and longer run dependencies.
27 +\newblock {\em Journal of Finance\/}~{\em 53\/}(1), 219--265.
28 +
29 +\bibitem[\protect\citeauthoryear{Andersen, Bollerslev, and Diebold}{Andersen
30 + et~al.}{2007}]{andersen2007roughing}
31 +Andersen, T.~G., T.~Bollerslev, and F.~X. Diebold (2007).
32 +\newblock Roughing it up: Including jump components in the measurement,
33 + modeling, and forecasting of return volatility.
34 +\newblock {\em The Review of Economics and Statistics\/}~{\em 89\/}(4),
35 + 701--720.
36 +
37 +\bibitem[\protect\citeauthoryear{Andersen, Bollerslev, Diebold, and
38 + Ebens}{Andersen et~al.}{2001}]{andersen2001distribution}
39 +Andersen, T.~G., T.~Bollerslev, F.~X. Diebold, and H.~Ebens (2001).
40 +\newblock The distribution of realized stock return volatility.
41 +\newblock {\em Journal of Financial Economics\/}~{\em 61\/}(1), 43--76.
42 +
43 +\bibitem[\protect\citeauthoryear{Andersen, Bollerslev, Diebold, and
44 + Vega}{Andersen et~al.}{2003}]{andersen2003micro}
45 +Andersen, T.~G., T.~Bollerslev, F.~X. Diebold, and C.~Vega (2003).
46 +\newblock Micro effects of macro announcements: Real-time price discovery in
47 + foreign exchange.
48 +\newblock {\em American Economic Review\/}~{\em 93\/}(1), 38--62.
49 +
50 +\bibitem[\protect\citeauthoryear{Andersen, Bollerslev, Diebold, and
51 + Vega}{Andersen et~al.}{2007}]{andersen2007real}
52 +Andersen, T.~G., T.~Bollerslev, F.~X. Diebold, and C.~Vega (2007).
53 +\newblock Real-time price discovery in global stock, bond and foreign exchange
54 + markets.
55 +\newblock {\em Journal of International Economics\/}~{\em 73\/}(2), 251--277.
56 +
57 +\bibitem[\protect\citeauthoryear{Apel and Grimaldi}{Apel and
58 + Grimaldi}{2012}]{apel2012}
59 +Apel, M. and M.~B. Grimaldi (2012).
60 +\newblock The information content of central bank minutes.
61 +\newblock Working Paper Series 261, Sveriges Riksbank.
62 +
63 +\bibitem[\protect\citeauthoryear{Araci}{Araci}{2019}]{araci2019}
64 +Araci, D. (2019).
65 +\newblock Finbert: Financial sentiment analysis with pre-trained language
66 + models.
67 +\newblock {\em arXiv preprint arXiv:1908.10063\/}.
68 +
69 +\bibitem[\protect\citeauthoryear{Aulerich, Irwin, and Garcia}{Aulerich
70 + et~al.}{2012}]{aulerich2012bubbles}
71 +Aulerich, N.~M., S.~H. Irwin, and P.~Garcia (2012).
72 +\newblock Bubbles.
73 +\newblock In {\em Food Prices, and Speculation: Evidence from the CFTC’s
74 + Daily Large Trader Data Files, Paper Prepared for Presentation at the NBER
75 + Conference on “Economics of Food Price Volatility” in Seattle, WA}.
76 +
77 +\bibitem[\protect\citeauthoryear{Bai and Perron}{Bai and
78 + Perron}{2003}]{bai2003computation}
79 +Bai, J. and P.~Perron (2003).
80 +\newblock Computation and analysis of multiple structural change models.
81 +\newblock {\em Journal of Applied Econometrics\/}~{\em 18\/}(1), 1--22.
82 +
83 +\bibitem[\protect\citeauthoryear{Baker, Bloom, and Davis}{Baker
84 + et~al.}{2016}]{baker2016}
85 +Baker, S.~R., N.~Bloom, and S.~J. Davis (2016).
86 +\newblock Measuring economic policy uncertainty.
87 +\newblock {\em Quarterly Journal of Economics\/}~{\em 131\/}(4), 1593--1636.
88 +
89 +\bibitem[\protect\citeauthoryear{Balduzzi, Elton, and Green}{Balduzzi
90 + et~al.}{2001}]{balduzzi2001economic}
91 +Balduzzi, P., E.~J. Elton, and T.~C. Green (2001).
92 +\newblock Economic news and bond prices: Evidence from the us treasury market.
93 +\newblock {\em Journal of Financial and Quantitative Analysis\/}~{\em 36\/}(4),
94 + 523--543.
95 +
96 +\bibitem[\protect\citeauthoryear{Barndorff-Nielsen, Hansen, Lunde, and
97 + Shephard}{Barndorff-Nielsen et~al.}{2009}]{barndorff2009realized}
98 +Barndorff-Nielsen, O.~E., P.~R. Hansen, A.~Lunde, and N.~Shephard (2009).
99 +\newblock Realized kernels in practice: Trades and quotes.
100 +\newblock {\em The Econometrics Journal\/}~{\em 12\/}(3), C1--C32.
101 +
102 +\bibitem[\protect\citeauthoryear{Barndorff-Nielsen and
103 + Shephard}{Barndorff-Nielsen and Shephard}{2002}]{barndorff2002econometric}
104 +Barndorff-Nielsen, O.~E. and N.~Shephard (2002).
105 +\newblock Econometric analysis of realized volatility and its use in estimating
106 + stochastic volatility models.
107 +\newblock {\em Journal of the Royal Statistical Society: Series B\/}~{\em
108 + 64\/}(2), 253--280.
109 +
110 +\bibitem[\protect\citeauthoryear{Barndorff-Nielsen and
111 + Shephard}{Barndorff-Nielsen and Shephard}{2004}]{barndorff2004power}
112 +Barndorff-Nielsen, O.~E. and N.~Shephard (2004).
113 +\newblock Power and bipower variation with stochastic volatility and jumps.
114 +\newblock {\em Journal of Financial Econometrics\/}~{\em 2\/}(1), 1--37.
115 +
116 +\bibitem[\protect\citeauthoryear{Basak and Pavlova}{Basak and
117 + Pavlova}{2016}]{basak2016model}
118 +Basak, S. and A.~Pavlova (2016).
119 +\newblock A model of financialization of commodities.
120 +\newblock {\em Journal of Finance\/}~{\em 71\/}(4), 1511--1556.
121 +
122 +\bibitem[\protect\citeauthoryear{Baumeister and Kilian}{Baumeister and
123 + Kilian}{2014}]{baumeister2014oil}
124 +Baumeister, C. and L.~Kilian (2014).
125 +\newblock Do oil price increases cause higher food prices?
126 +\newblock {\em Economic Policy\/}~{\em 29\/}(80), 691--747.
127 +
128 +\bibitem[\protect\citeauthoryear{Baur and Lucey}{Baur and
129 + Lucey}{2010}]{baur2010gold}
130 +Baur, D.~G. and B.~M. Lucey (2010).
131 +\newblock Is gold a hedge or a safe haven? {An} analysis of stocks, bonds and
132 + gold.
133 +\newblock {\em Financial Review\/}~{\em 45\/}(2), 217--229.
134 +
135 +\bibitem[\protect\citeauthoryear{Ben-David, Franzoni, and Moussawi}{Ben-David
136 + et~al.}{2018}]{ben2018etfs}
137 +Ben-David, I., F.~Franzoni, and R.~Moussawi (2018).
138 +\newblock Do {ETFs} increase volatility?
139 +\newblock {\em The Journal of Finance\/}~{\em 73\/}(6), 2471--2535.
140 +
141 +\bibitem[\protect\citeauthoryear{Benjamini and Hochberg}{Benjamini and
142 + Hochberg}{1995}]{benjamini1995controlling}
143 +Benjamini, Y. and Y.~Hochberg (1995).
144 +\newblock Controlling the false discovery rate: a practical and powerful
145 + approach to multiple testing.
146 +\newblock {\em Journal of the Royal Statistical Society: Series B
147 + (Methodological)\/}~{\em 57\/}(1), 289--300.
148 +
149 +\bibitem[\protect\citeauthoryear{Bernanke and Kuttner}{Bernanke and
150 + Kuttner}{2005}]{bernanke2005}
151 +Bernanke, B.~S. and K.~N. Kuttner (2005).
152 +\newblock What explains the stock market's reaction to federal reserve policy?
153 +\newblock {\em Journal of Finance\/}~{\em 60\/}(3), 1221--1257.
154 +
155 +\bibitem[\protect\citeauthoryear{Bligh and Hess}{Bligh and
156 + Hess}{2008}]{bligh2008}
157 +Bligh, M.~C. and G.~D. Hess (2008).
158 +\newblock The decline of symbolic politics: Federal reserve communications.
159 +\newblock {\em Journal of Economic Psychology\/}~{\em 29\/}(4), 535--559.
160 +
161 +\bibitem[\protect\citeauthoryear{Blinder, Ehrmann, Fratzscher, De~Haan, and
162 + Jansen}{Blinder et~al.}{2008}]{blinder2008}
163 +Blinder, A.~S., M.~Ehrmann, M.~Fratzscher, J.~De~Haan, and D.-J. Jansen (2008).
164 +\newblock Central bank communication and monetary policy: A survey of theory
165 + and evidence.
166 +\newblock {\em Journal of Economic Literature\/}~{\em 46\/}(4), 910--945.
167 +
168 +\bibitem[\protect\citeauthoryear{Boons, de~Roon, and Szymanowska}{Boons
169 + et~al.}{2014}]{boons2014price}
170 +Boons, M., F.~A. de~Roon, and M.~Szymanowska (2014).
171 +\newblock The price of commodity risk in stock and futures markets.
172 +\newblock In {\em AFA 2012 Chicago Meetings Paper}.
173 +\newblock Available at SSRN: \url{https://ssrn.com/abstract=1785728} or
174 + \url{http://dx.doi.org/10.2139/ssrn.1785728}.
175 +
176 +\bibitem[\protect\citeauthoryear{Brownlees and Gallo}{Brownlees and
177 + Gallo}{2006}]{brownlees2014practical}
178 +Brownlees, C.~T. and G.~M. Gallo (2006).
179 +\newblock Financial econometric analysis at ultra-high frequency: Data handling
180 + concerns.
181 +\newblock {\em Computational Statistics \& Data Analysis\/}~{\em 51\/}(4),
182 + 2232--2245.
183 +
184 +\bibitem[\protect\citeauthoryear{Brunetti, B{\"u}y{\"u}k{\c{s}}ahin, and
185 + Harris}{Brunetti et~al.}{2016}]{brunetti2016speculators}
186 +Brunetti, C., B.~B{\"u}y{\"u}k{\c{s}}ahin, and J.~H. Harris (2016).
187 +\newblock Speculators, prices, and market volatility.
188 +\newblock {\em Journal of Financial and Quantitative Analysis\/}~{\em 51\/}(5),
189 + 1545--1574.
190 +
191 +\bibitem[\protect\citeauthoryear{Brunetti and Reiffen}{Brunetti and
192 + Reiffen}{2014}]{brunetti2014commodity}
193 +Brunetti, C. and D.~Reiffen (2014).
194 +\newblock Commodity index trading and hedging costs.
195 +\newblock {\em Journal of Financial Markets\/}~{\em 21}, 153--180.
196 +
197 +\bibitem[\protect\citeauthoryear{Brusa, Savor, and Wilson}{Brusa
198 + et~al.}{2015}]{brusa2015}
199 +Brusa, F., P.~Savor, and M.~Wilson (2015).
200 +\newblock Asset allocation and fomc announcements.
201 +\newblock {\em Review of Financial Studies\/}~{\em 28\/}(5), 1398--1446.
202 +
203 +\bibitem[\protect\citeauthoryear{Brusa, Savor, and Wilson}{Brusa
204 + et~al.}{2019}]{brusa2019}
205 +Brusa, F., P.~Savor, and M.~Wilson (2019).
206 +\newblock Fomc announcements and market returns: Evidence from the options
207 + market.
208 +\newblock {\em Journal of Finance\/}~{\em 75\/}(1), 399--441.
209 +
210 +\bibitem[\protect\citeauthoryear{Bryant, Bessler, and Haigh}{Bryant
211 + et~al.}{2006}]{bryant2006causality}
212 +Bryant, H.~L., D.~A. Bessler, and M.~S. Haigh (2006).
213 +\newblock Causality in futures markets.
214 +\newblock {\em Journal of Futures Markets\/}~{\em 26\/}(11), 1039--1057.
215 +
216 +\bibitem[\protect\citeauthoryear{B{\"u}y{\"u}k{\c{s}}ahin and
217 + Harris}{B{\"u}y{\"u}k{\c{s}}ahin and
218 + Harris}{2011}]{buyukcsahin2011speculators}
219 +B{\"u}y{\"u}k{\c{s}}ahin, B. and J.~H. Harris (2011).
220 +\newblock Do speculators drive crude oil futures prices?
221 +\newblock {\em The Energy Journal\/}~{\em 32\/}(2), 167--202.
222 +
223 +\bibitem[\protect\citeauthoryear{Buyuksahin and Robe}{Buyuksahin and
224 + Robe}{2014}]{buyuksahin2014speculation}
225 +Buyuksahin, B. and M.~A. Robe (2014).
226 +\newblock Speculation, commodities and cross-market linkages.
227 +\newblock {\em Journal of International Money and Finance\/}~{\em 42}, 38--70.
228 +
229 +\bibitem[\protect\citeauthoryear{B{\"u}y{\"u}k{\c{s}}ahin and
230 + Robe}{B{\"u}y{\"u}k{\c{s}}ahin and Robe}{2014}]{buyukcsahin2014speculators}
231 +B{\"u}y{\"u}k{\c{s}}ahin, B. and M.~A. Robe (2014).
232 +\newblock Speculators, commodities and cross-market linkages.
233 +\newblock {\em Journal of International Money and Finance\/}~{\em 42}, 38--70.
234 +
235 +\bibitem[\protect\citeauthoryear{Campbell, Evans, Fisher, and
236 + Justiniano}{Campbell et~al.}{2012}]{campbell2012}
237 +Campbell, J.~R., C.~L. Evans, J.~D. Fisher, and A.~Justiniano (2012).
238 +\newblock Macroeconomic effects of federal reserve forward guidance.
239 +\newblock {\em Brookings Papers on Economic Activity\/}~{\em 2012\/}(1), 1--80.
240 +
241 +\bibitem[\protect\citeauthoryear{Cao, Su, Sun, Qin, and Umar}{Cao
242 + et~al.}{2024}]{cao2024us}
243 +Cao, F., C.-W. Su, D.~Sun, M.~Qin, and M.~Umar (2024).
244 +\newblock U.s. monetary policy: The pushing hands of crude oil price?
245 +\newblock {\em Energy Economics\/}~{\em 134}, 107555.
246 +
247 +\bibitem[\protect\citeauthoryear{Carriero, Kapetanios, and Marcellino}{Carriero
248 + et~al.}{2009}]{carriero2015forecasting}
249 +Carriero, A., G.~Kapetanios, and M.~Marcellino (2009).
250 +\newblock Forecasting exchange rates with a large {B}ayesian {VAR}.
251 +\newblock {\em International Journal of Forecasting\/}~{\em 25\/}(2), 400--417.
252 +
253 +\bibitem[\protect\citeauthoryear{Chang, Pinegar, and Schachter}{Chang
254 + et~al.}{1997}]{chang1997interday}
255 +Chang, E.~C., J.~M. Pinegar, and B.~Schachter (1997).
256 +\newblock Interday variations in volume, variance and participation of large
257 + speculators.
258 +\newblock {\em Journal of Banking \& Finance\/}~{\em 21\/}(6), 797--810.
259 +
260 +\bibitem[\protect\citeauthoryear{Cheng, Kirilenko, and Xiong}{Cheng
261 + et~al.}{2015}]{cheng2015convective}
262 +Cheng, I.-H., A.~Kirilenko, and W.~Xiong (2015).
263 +\newblock Convective risk flows in commodity futures markets.
264 +\newblock {\em Review of Finance\/}~{\em 19\/}(5), 1733--1781.
265 +
266 +\bibitem[\protect\citeauthoryear{Cheng and Xiong}{Cheng and
267 + Xiong}{2014}]{cheng2014financialization}
268 +Cheng, I.-H. and W.~Xiong (2014).
269 +\newblock Financialization of commodity markets.
270 +\newblock {\em Annual Review of Financial Economics\/}~{\em 6\/}(1), 419--441.
271 +
272 +\bibitem[\protect\citeauthoryear{Chordia, Roll, and Subrahmanyam}{Chordia
273 + et~al.}{2008}]{chordia2008liquidity}
274 +Chordia, T., R.~Roll, and A.~Subrahmanyam (2008).
275 +\newblock Liquidity and market efficiency.
276 +\newblock {\em Journal of Financial Economics\/}~{\em 87\/}(2), 249--268.
277 +
278 +\bibitem[\protect\citeauthoryear{Corsi}{Corsi}{2009}]{corsi2009simple}
279 +Corsi, F. (2009).
280 +\newblock A simple approximate long-memory model of realized volatility.
281 +\newblock {\em Journal of Financial Econometrics\/}~{\em 7\/}(2), 174--196.
282 +
283 +\bibitem[\protect\citeauthoryear{Da and Shive}{Da and
284 + Shive}{2018}]{da2018exchange}
285 +Da, Z. and S.~Shive (2018).
286 +\newblock Exchange traded funds and asset return correlations.
287 +\newblock {\em European Financial Management\/}~{\em 24\/}(1), 136--168.
288 +
289 +\bibitem[\protect\citeauthoryear{Da, Tang, Tao, and Yang}{Da
290 + et~al.}{2024}]{da2024financialization}
291 +Da, Z., K.~Tang, Y.~Tao, and L.~Yang (2024).
292 +\newblock Financialization and commodity markets serial dependence.
293 +\newblock {\em Management Science\/}~{\em 70\/}(4), 2122--2143.
294 +
295 +\bibitem[\protect\citeauthoryear{Daigler and Wiley}{Daigler and
296 + Wiley}{1999}]{daigler1999impact}
297 +Daigler, R.~T. and M.~K. Wiley (1999).
298 +\newblock The impact of trader type on the futures volatility-volume relation.
299 +\newblock {\em Journal of Finance\/}~{\em 54\/}(6), 2297--2316.
300 +
301 +\bibitem[\protect\citeauthoryear{Dannhauser}{Dannhauser}{2017}]{dannhauser2017effect}
302 +Dannhauser, C.~D. (2017).
303 +\newblock The impact of innovation: Evidence from corporate bond
304 + exchange-traded funds.
305 +\newblock {\em Journal of Financial Economics\/}~{\em 125\/}(3), 537--560.
306 +
307 +\bibitem[\protect\citeauthoryear{Devlin, Chang, Lee, and Toutanova}{Devlin
308 + et~al.}{2019}]{kenton2019}
309 +Devlin, J., M.-W. Chang, K.~Lee, and K.~Toutanova (2019).
310 +\newblock Bert: Pre-training of deep bidirectional transformers for language
311 + understanding.
312 +\newblock In {\em Proceedings of the 2019 Conference of the North American
313 + Chapter of the Association for Computational Linguistics}, pp.\ 4171--4186.
314 +
315 +\bibitem[\protect\citeauthoryear{Domanski and Heath}{Domanski and
316 + Heath}{2007}]{domanski2007financial}
317 +Domanski, D. and A.~Heath (2007).
318 +\newblock Financial investors and commodity markets.
319 +\newblock {\em BIS Quarterly Review\/}~{\em 3\/}(1), 53--67.
320 +
321 +\bibitem[\protect\citeauthoryear{Ehrmann, Fratzscher, and Rigobon}{Ehrmann
322 + et~al.}{2011}]{ehrmann2011}
323 +Ehrmann, M., M.~Fratzscher, and R.~Rigobon (2011).
324 +\newblock Global crises and equity market contagion.
325 +\newblock {\em Journal of Finance\/}~{\em 66\/}(6), 2597--2649.
326 +
327 +\bibitem[\protect\citeauthoryear{Eklund and Kim}{Eklund and
328 + Kim}{2024}]{eklund2024}
329 +Eklund, J. and S.~Kim (2024).
330 +\newblock Fomc statement sentiment and inflation expectations.
331 +\newblock {\em Journal of Monetary Economics\/}~{\em 141}, 45--62.
332 +
333 +\bibitem[\protect\citeauthoryear{Erb and Harvey}{Erb and
334 + Harvey}{2013}]{erb2013golden}
335 +Erb, C.~B. and C.~R. Harvey (2013).
336 +\newblock The golden dilemma.
337 +\newblock {\em Financial Analysts Journal\/}~{\em 69\/}(4), 10--42.
338 +
339 +\bibitem[\protect\citeauthoryear{Fama}{Fama}{1970}]{fama1970}
340 +Fama, E.~F. (1970).
341 +\newblock Efficient capital markets: A review of theory and empirical work.
342 +\newblock {\em Journal of Finance\/}~{\em 25\/}(2), 383--417.
343 +
344 +\bibitem[\protect\citeauthoryear{Fattouh, Kilian, and Mahadeva}{Fattouh
345 + et~al.}{2013}]{fattouh2013role}
346 +Fattouh, B., L.~Kilian, and L.~Mahadeva (2013).
347 +\newblock The role of speculation in oil markets: What have we learned so far?
348 +\newblock {\em The Energy Journal\/}~{\em 34\/}(3), 7--33.
349 +
350 +\bibitem[\protect\citeauthoryear{Fishe and Smith}{Fishe and
351 + Smith}{2012}]{fishe2012identifying}
352 +Fishe, R.~P. and A.~Smith (2012).
353 +\newblock Identifying informed traders in futures markets.
354 +\newblock {\em Journal of Financial Markets\/}~{\em 15\/}(3), 329--359.
355 +
356 +\bibitem[\protect\citeauthoryear{Fleming and Remolona}{Fleming and
357 + Remolona}{1997}]{fleming1997moves}
358 +Fleming, M.~J. and E.~M. Remolona (1997).
359 +\newblock What moves the bond market?
360 +\newblock {\em Federal Reserve Bank of New York Economic Policy Review\/}~{\em
361 + 3\/}(4).
362 +
363 +\bibitem[\protect\citeauthoryear{Fleming and Remolona}{Fleming and
364 + Remolona}{1999}]{fleming1999}
365 +Fleming, M.~J. and E.~M. Remolona (1999).
366 +\newblock Price formation and liquidity in the us treasury market: The response
367 + to public information.
368 +\newblock {\em Journal of Finance\/}~{\em 54\/}(5), 1901--1915.
369 +
370 +\bibitem[\protect\citeauthoryear{Gentzkow, Kelly, and Taddy}{Gentzkow
371 + et~al.}{2019}]{gentzkow2019}
372 +Gentzkow, M., B.~Kelly, and M.~Taddy (2019).
373 +\newblock Text as data.
374 +\newblock {\em Journal of Economic Literature\/}~{\em 57\/}(3), 535--574.
375 +
376 +\bibitem[\protect\citeauthoryear{Ghysels, Santa-Clara, and Valkanov}{Ghysels
377 + et~al.}{2004}]{ghysels2004midas}
378 +Ghysels, E., P.~Santa-Clara, and R.~Valkanov (2004).
379 +\newblock The midas touch: Mixed data sampling regression models.
380 +
381 +\bibitem[\protect\citeauthoryear{Glosten, Nallareddy, and Zou}{Glosten
382 + et~al.}{2021}]{glosten2021etf}
383 +Glosten, L., S.~Nallareddy, and Y.~Zou (2021).
384 +\newblock {ETF} activity and informational efficiency of underlying securities.
385 +\newblock {\em Management Science\/}~{\em 67\/}(1), 22--47.
386 +
387 +\bibitem[\protect\citeauthoryear{Goldstein and Yang}{Goldstein and
388 + Yang}{2022}]{goldstein2022commodity}
389 +Goldstein, I. and L.~Yang (2022).
390 +\newblock Commodity financialization and information transmission.
391 +\newblock {\em The Journal of Finance\/}~{\em 77\/}(5), 2613--2667.
392 +
393 +\bibitem[\protect\citeauthoryear{Gorodnichenko, Pham, and
394 + Talavera}{Gorodnichenko et~al.}{2023}]{gorodnichenko2023}
395 +Gorodnichenko, Y., T.~Pham, and O.~Talavera (2023).
396 +\newblock The voice of monetary policy.
397 +\newblock {\em American Economic Review\/}~{\em 113\/}(2), 548--584.
398 +
399 +\bibitem[\protect\citeauthoryear{Gorton and Rouwenhorst}{Gorton and
400 + Rouwenhorst}{2006}]{gorton2006facts}
401 +Gorton, G. and K.~G. Rouwenhorst (2006).
402 +\newblock Facts and fantasies about commodity futures.
403 +\newblock {\em Financial Analysts Journal\/}~{\em 62\/}(2), 47--68.
404 +
405 +\bibitem[\protect\citeauthoryear{Gromb and Vayanos}{Gromb and
406 + Vayanos}{2010}]{gromb2010limits}
407 +Gromb, D. and D.~Vayanos (2010).
408 +\newblock Limits of arbitrage: The state of the theory.
409 +\newblock {\em Annual Review of Financial Economics\/}~{\em 2}, 251--275.
410 +
411 +\bibitem[\protect\citeauthoryear{Grossman and Stiglitz}{Grossman and
412 + Stiglitz}{1980}]{grossman1980}
413 +Grossman, S.~J. and J.~E. Stiglitz (1980).
414 +\newblock On the impossibility of informationally efficient markets.
415 +\newblock {\em American Economic Review\/}~{\em 70\/}(3), 393--408.
416 +
417 +\bibitem[\protect\citeauthoryear{G{\"u}rkaynak, Sack, and
418 + Swanson}{G{\"u}rkaynak et~al.}{2005}]{gurkaynak2005}
419 +G{\"u}rkaynak, R.~S., B.~P. Sack, and E.~T. Swanson (2005).
420 +\newblock Do actions speak louder than words? the response of asset prices to
421 + monetary policy actions and statements.
422 +\newblock {\em International Journal of Central Banking\/}~{\em 1\/}(1),
423 + 55--93.
424 +
425 +\bibitem[\protect\citeauthoryear{Gürtler and Gürtler}{Gürtler and
426 + Gürtler}{2010}]{gurtler2010}
427 +Gürtler, M. and O.~Gürtler (2010).
428 +\newblock The effect of fomc statement language on financial markets.
429 +\newblock {\em Journal of Financial Research\/}~{\em 33\/}(4), 369--391.
430 +
431 +\bibitem[\protect\citeauthoryear{Haigh, Hranaiova, and Overdahl}{Haigh
432 + et~al.}{2007}]{haigh2007hedge}
433 +Haigh, M.~S., J.~Hranaiova, and J.~A. Overdahl (2007).
434 +\newblock Hedge funds, volatility, and liquidity provision in energy futures
435 + markets.
436 +\newblock {\em Journal of Alternative Investments\/}~{\em 9\/}(4), 10--38.
437 +
438 +\bibitem[\protect\citeauthoryear{Hamilton and Wu}{Hamilton and
439 + Wu}{2014}]{hamilton2014risk}
440 +Hamilton, J.~D. and J.~C. Wu (2014).
441 +\newblock Risk premia in crude oil futures prices.
442 +\newblock {\em Journal of International Money and Finance\/}~{\em 42}, 9--37.
443 +
444 +\bibitem[\protect\citeauthoryear{Hansen and Lunde}{Hansen and
445 + Lunde}{2005}]{hansen2005realized}
446 +Hansen, P.~R. and A.~Lunde (2005).
447 +\newblock A realized variance for the whole day based on intermittent
448 + high-frequency data.
449 +\newblock {\em Journal of Financial Econometrics\/}~{\em 3\/}(4), 525--554.
450 +
451 +\bibitem[\protect\citeauthoryear{Hansen, McMahon, and Prat}{Hansen
452 + et~al.}{2018}]{hansen2017}
453 +Hansen, S., M.~McMahon, and A.~Prat (2018).
454 +\newblock Transparency and deliberation within the fomc: A computational
455 + linguistics approach.
456 +\newblock {\em Quarterly Journal of Economics\/}~{\em 133\/}(2), 801--870.
457 +
458 +\bibitem[\protect\citeauthoryear{Hasbrouck}{Hasbrouck}{2003}]{hasbrouck2003intraday}
459 +Hasbrouck, J. (2003).
460 +\newblock Intraday price formation in {US} equity index markets.
461 +\newblock {\em The Journal of Finance\/}~{\em 58\/}(6), 2375--2400.
462 +
463 +\bibitem[\protect\citeauthoryear{Hayo and Neuenkirch}{Hayo and
464 + Neuenkirch}{2010}]{hayo2010}
465 +Hayo, B. and M.~Neuenkirch (2010).
466 +\newblock Do federal reserve communications help predict federal funds target
467 + rate decisions?
468 +\newblock {\em Journal of Macroeconomics\/}~{\em 32\/}(4), 1014--1024.
469 +
470 +\bibitem[\protect\citeauthoryear{Hedegaard}{Hedegaard}{2011}]{hedegaard2011margins}
471 +Hedegaard, E. (2011).
472 +\newblock How margins are set and affect asset prices.
473 +\newblock {\em Job Market Paper\/}.
474 +
475 +\bibitem[\protect\citeauthoryear{Hendershott and Riordan}{Hendershott and
476 + Riordan}{2013}]{hendershott2013relationship}
477 +Hendershott, T. and R.~Riordan (2013).
478 +\newblock Algorithmic trading and the market for liquidity.
479 +\newblock {\em Journal of Financial and Quantitative Analysis\/}~{\em 48\/}(4),
480 + 1001--1024.
481 +
482 +\bibitem[\protect\citeauthoryear{Henderson, Pearson, and Wang}{Henderson
483 + et~al.}{2015}]{henderson2015new}
484 +Henderson, B.~J., N.~D. Pearson, and L.~Wang (2015).
485 +\newblock New evidence on the financialization of commodity markets.
486 +\newblock {\em Review of Financial Studies\/}~{\em 28\/}(5), 1285--1311.
487 +
488 +\bibitem[\protect\citeauthoryear{Hollstein, Prokopczuk, and
489 + W{\"u}rsig}{Hollstein et~al.}{2020}]{hollstein2020volatility}
490 +Hollstein, F., M.~Prokopczuk, and C.~W{\"u}rsig (2020).
491 +\newblock Volatility term structures in commodity markets.
492 +\newblock {\em Journal of Futures Markets\/}~{\em 40\/}(4), 527--555.
493 +
494 +\bibitem[\protect\citeauthoryear{Hu, Pan, and Wang}{Hu et~al.}{2013}]{hu2019}
495 +Hu, G.~X., J.~Pan, and J.~Wang (2013).
496 +\newblock Noise as information for illiquidity.
497 +\newblock {\em Journal of Finance\/}~{\em 68\/}(6), 2341--2382.
498 +
499 +\bibitem[\protect\citeauthoryear{Huang and Tauchen}{Huang and
500 + Tauchen}{2005}]{huang2005using}
501 +Huang, X. and G.~Tauchen (2005).
502 +\newblock The relative contribution of jumps to total price variance.
503 +\newblock {\em Journal of Financial Econometrics\/}~{\em 3\/}(4), 456--499.
504 +
505 +\bibitem[\protect\citeauthoryear{Irwin and Brorsen}{Irwin and
506 + Brorsen}{1987}]{irwin1987note}
507 +Irwin, S.~H. and B.~W. Brorsen (1987).
508 +\newblock A note on the factors affecting technical trading system returns.
509 +\newblock {\em Journal of Futures Markets\/}~{\em 7\/}(5), 591--595.
510 +
511 +\bibitem[\protect\citeauthoryear{Irwin and Holt}{Irwin and
512 + Holt}{2004}]{irwin2004effect}
513 +Irwin, S.~H. and B.~Holt (2004).
514 +\newblock The effect of large hedge fund and {CTA} trading on futures market
515 + volatility.
516 +\newblock In {\em Commodity Trading Advisors: Risk, Performance Analysis and
517 + Selection}. New York, NY: John Wiley and Sons, Inc.
518 +
519 +\bibitem[\protect\citeauthoryear{Irwin and Sanders}{Irwin and
520 + Sanders}{2011}]{irwin2011index}
521 +Irwin, S.~H. and D.~R. Sanders (2011).
522 +\newblock Index funds, financialization, and commodity futures markets.
523 +\newblock {\em Applied Economic Perspectives and Policy\/}~{\em 33\/}(1),
524 + 1--31.
525 +
526 +\bibitem[\protect\citeauthoryear{Irwin and Sanders}{Irwin and
527 + Sanders}{2012}]{irwin2012testing}
528 +Irwin, S.~H. and D.~R. Sanders (2012).
529 +\newblock Testing the masters hypothesis in commodity futures markets.
530 +\newblock {\em Energy Economics\/}~{\em 34\/}(1), 256--269.
531 +
532 +\bibitem[\protect\citeauthoryear{Irwin and Yoshimaru}{Irwin and
533 + Yoshimaru}{1999}]{irwin1999managed}
534 +Irwin, S.~H. and S.~Yoshimaru (1999).
535 +\newblock Managed futures, positive feedback trading, and futures price
536 + volatility.
537 +\newblock {\em Journal of Futures Markets\/}~{\em 19\/}(7), 759--776.
538 +
539 +\bibitem[\protect\citeauthoryear{Israeli, Lee, and Sridharan}{Israeli
540 + et~al.}{2017}]{israeli2017etf}
541 +Israeli, D., C.~M. Lee, and S.~A. Sridharan (2017).
542 +\newblock Is there a dark side to exchange traded funds? {An} information
543 + perspective.
544 +\newblock {\em Review of Accounting Studies\/}~{\em 22\/}(3), 1048--1083.
545 +
546 +\bibitem[\protect\citeauthoryear{Jaro{\'c}inski and Kar{\'a}di}{Jaro{\'c}inski
547 + and Kar{\'a}di}{2020}]{jarocinski2020}
548 +Jaro{\'c}inski, M. and P.~Kar{\'a}di (2020).
549 +\newblock Deconstructing monetary policy surprises—the role of information
550 + shocks.
551 +\newblock {\em American Economic Journal: Macroeconomics\/}~{\em 12\/}(2),
552 + 1--43.
553 +
554 +\bibitem[\protect\citeauthoryear{Jord{\`a}}{Jord{\`a}}{2005}]{jorda2005}
555 +Jord{\`a}, {\`O}. (2005).
556 +\newblock Estimation and inference of impulse responses by local projections.
557 +\newblock {\em American Economic Review\/}~{\em 95\/}(1), 161--182.
558 +
559 +\bibitem[\protect\citeauthoryear{Kang, Nikitopoulos, and Prokopczuk}{Kang
560 + et~al.}{2020}]{kang2020economic}
561 +Kang, B., C.~S. Nikitopoulos, and M.~Prokopczuk (2020).
562 +\newblock Economic determinants of oil futures volatility: A term structure
563 + perspective.
564 +\newblock {\em Energy Economics\/}~{\em 88}, 104743.
565 +
566 +\bibitem[\protect\citeauthoryear{Kang, Rouwenhorst, and Tang}{Kang
567 + et~al.}{2020}]{kang2020tale}
568 +Kang, W., K.~G. Rouwenhorst, and K.~Tang (2020).
569 +\newblock A tale of two premiums: The role of hedgers and speculators in
570 + commodity futures markets.
571 +\newblock {\em Journal of Finance\/}~{\em 75\/}(1), 377--417.
572 +
573 +\bibitem[\protect\citeauthoryear{Kang, Tang, and Wang}{Kang
574 + et~al.}{2023}]{kang2023financialization}
575 +Kang, W., K.~Tang, and N.~Wang (2023).
576 +\newblock Financialization of commodity markets ten years later.
577 +\newblock {\em Journal of Commodity Markets\/}~{\em 30}, 100313.
578 +
579 +\bibitem[\protect\citeauthoryear{Karali and Ramirez}{Karali and
580 + Ramirez}{2014}]{karali2014macro}
581 +Karali, B. and O.~A. Ramirez (2014).
582 +\newblock Macro determinants of volatility and volatility spillover in energy
583 + markets.
584 +\newblock {\em Energy Economics\/}~{\em 46}, 413--421.
585 +
586 +\bibitem[\protect\citeauthoryear{Kilian and Murphy}{Kilian and
587 + Murphy}{2014}]{kilian2014role}
588 +Kilian, L. and D.~P. Murphy (2014).
589 +\newblock The role of inventories and speculative trading in the global market
590 + for crude oil.
591 +\newblock {\em Journal of Applied Econometrics\/}~{\em 29\/}(3), 454--478.
592 +
593 +\bibitem[\protect\citeauthoryear{Kilian and Vega}{Kilian and
594 + Vega}{2011}]{kilian2011energy}
595 +Kilian, L. and C.~Vega (2011).
596 +\newblock Do energy prices respond to us macroeconomic news? a test of the
597 + hypothesis of predetermined energy prices.
598 +\newblock {\em Review of Economics and Statistics\/}~{\em 93\/}(2), 660--671.
599 +
600 +\bibitem[\protect\citeauthoryear{Koop}{Koop}{2013}]{koop2011forecasting}
601 +Koop, G. (2013).
602 +\newblock Forecasting with medium and large {B}ayesian {VARs}.
603 +\newblock {\em Journal of Applied Econometrics\/}~{\em 28\/}(2), 177--203.
604 +
605 +\bibitem[\protect\citeauthoryear{Kothari and Warner}{Kothari and
606 + Warner}{2007}]{kothari2007econometrics}
607 +Kothari, S.~P. and J.~B. Warner (2007).
608 +\newblock Econometrics of event studies.
609 +\newblock pp.\ 3--36.
610 +
611 +\bibitem[\protect\citeauthoryear{Kurov, Sancetta, Strasser, and Wolfe}{Kurov
612 + et~al.}{2019}]{kurov2019price}
613 +Kurov, A., A.~Sancetta, G.~Strasser, and M.~H. Wolfe (2019).
614 +\newblock Price drift before us macroeconomic news: Private information about
615 + public announcements?
616 +\newblock {\em Journal of Financial and Quantitative Analysis\/}~{\em 54\/}(1),
617 + 449--479.
618 +
619 +\bibitem[\protect\citeauthoryear{Kuttner}{Kuttner}{2001}]{kuttner2001}
620 +Kuttner, K.~N. (2001).
621 +\newblock Monetary policy surprises and interest rates: Evidence from the fed
622 + funds futures market.
623 +\newblock {\em Journal of Monetary Economics\/}~{\em 47\/}(3), 523--544.
624 +
625 +\bibitem[\protect\citeauthoryear{Känzig}{Känzig}{2021}]{10.1257/aer.20190964}
626 +Känzig, D.~R. (2021, April).
627 +\newblock The macroeconomic effects of oil supply news: Evidence from {OPEC}
628 + announcements.
629 +\newblock {\em American Economic Review\/}~{\em 111\/}(4), 1092--1125.
630 +
631 +\bibitem[\protect\citeauthoryear{Litterman}{Litterman}{1986}]{litterman1986forecasting}
632 +Litterman, R.~B. (1986).
633 +\newblock Forecasting with {B}ayesian vector autoregressions—five years of
634 + experience.
635 +\newblock {\em Journal of Business \& Economic Statistics\/}~{\em 4\/}(1),
636 + 25--38.
637 +
638 +\bibitem[\protect\citeauthoryear{Liu, Patton, and Sheppard}{Liu
639 + et~al.}{2015}]{liu2015does}
640 +Liu, L.~Y., A.~J. Patton, and K.~Sheppard (2015).
641 +\newblock Does anything beat 5-minute {RV}? {A} comparison of realized measures
642 + across multiple asset classes.
643 +\newblock {\em Journal of Econometrics\/}~{\em 187\/}(1), 293--311.
644 +
645 +\bibitem[\protect\citeauthoryear{Loughran and McDonald}{Loughran and
646 + McDonald}{2011}]{loughran2011}
647 +Loughran, T. and B.~McDonald (2011).
648 +\newblock When is a liability not a liability? textual analysis, dictionaries,
649 + and 10-ks.
650 +\newblock {\em Journal of Finance\/}~{\em 66\/}(1), 35--65.
651 +
652 +\bibitem[\protect\citeauthoryear{Lucca and Moench}{Lucca and
653 + Moench}{2015}]{lucca2012}
654 +Lucca, D.~O. and E.~Moench (2015).
655 +\newblock The pre-fomc announcement drift.
656 +\newblock {\em Journal of Finance\/}~{\em 70\/}(1), 329--371.
657 +
658 +\bibitem[\protect\citeauthoryear{Madhavan}{Madhavan}{2012}]{madhavan2012exchange}
659 +Madhavan, A. (2012).
660 +\newblock Exchange-traded funds, market structure, and the flash crash.
661 +\newblock {\em Financial Analysts Journal\/}~{\em 68\/}(4), 20--35.
662 +
663 +\bibitem[\protect\citeauthoryear{Manela and Moreira}{Manela and
664 + Moreira}{2017}]{manela2017}
665 +Manela, A. and A.~Moreira (2017).
666 +\newblock News implied volatility and disaster concerns.
667 +\newblock {\em Journal of Financial Economics\/}~{\em 123\/}(1), 137--162.
668 +
669 +\bibitem[\protect\citeauthoryear{Masters}{Masters}{2009}]{masters2009testimony}
670 +Masters, M.~W. (2009).
671 +\newblock Testimony before the commodity futures trading commission.
672 +\newblock {\em Testimony to the Commodity Futures Trading Commission\/}.
673 +
674 +\bibitem[\protect\citeauthoryear{Muth}{Muth}{1961}]{muth1961}
675 +Muth, J.~F. (1961).
676 +\newblock Rational expectations and the theory of price movements.
677 +\newblock {\em Econometrica\/}~{\em 29\/}(3), 315--335.
678 +
679 +\bibitem[\protect\citeauthoryear{Müller, Dacorogna, Davé, Olsen, Pictet, and
680 + von Weizsäcker}{Müller et~al.}{1997}]{muller1997volatilities}
681 +Müller, U.~A., M.~M. Dacorogna, R.~D. Davé, R.~B. Olsen, O.~V. Pictet, and
682 + J.~E. von Weizsäcker (1997).
683 +\newblock Volatilities of different time resolutions—analyzing the dynamics
684 + of market components.
685 +\newblock {\em Journal of Empirical Finance\/}~{\em 4\/}(2-3), 213--239.
686 +
687 +\bibitem[\protect\citeauthoryear{Nakamura and Steinsson}{Nakamura and
688 + Steinsson}{2018}]{nakamura2018}
689 +Nakamura, E. and J.~Steinsson (2018).
690 +\newblock High-frequency identification of monetary non-neutrality: The
691 + information effect.
692 +\newblock {\em Quarterly Journal of Economics\/}~{\em 133\/}(3), 1283--1330.
693 +
694 +\bibitem[\protect\citeauthoryear{Newey and West}{Newey and
695 + West}{1994}]{newey1994automatic}
696 +Newey, W.~K. and K.~D. West (1994).
697 +\newblock Automatic lag selection in covariance matrix estimation.
698 +\newblock {\em Review of Economic Studies\/}~{\em 61\/}(4), 631--653.
699 +
700 +\bibitem[\protect\citeauthoryear{O'Hara and Zhou}{O'Hara and
701 + Zhou}{2021}]{ohara2021etf}
702 +O'Hara, M. and X.~A. Zhou (2021).
703 +\newblock Anatomy of a liquidity crisis: Corporate bonds in the {COVID}-19
704 + crisis.
705 +\newblock {\em Journal of Financial Economics\/}~{\em 142\/}(1), 46--68.
706 +
707 +\bibitem[\protect\citeauthoryear{Pan and Zeng}{Pan and Zeng}{2016}]{pan2016etf}
708 +Pan, K. and Y.~Zeng (2016).
709 +\newblock {ETF} arbitrage under liquidity mismatch.
710 +\newblock {\em Journal of Financial Economics\/}~{\em 120\/}(3), 617--635.
711 +
712 +\bibitem[\protect\citeauthoryear{Patton and Verardo}{Patton and
713 + Verardo}{2012}]{patton2013}
714 +Patton, A.~J. and M.~Verardo (2012).
715 +\newblock Why do markets disagree? evidence from variation in opinion.
716 +\newblock {\em Review of Financial Studies\/}~{\em 25\/}(12), 3734--3773.
717 +
718 +\bibitem[\protect\citeauthoryear{Petäjistö}{Petäjistö}{2017}]{petajisto2017inefficiencies}
719 +Petäjistö, A. (2017).
720 +\newblock Inefficiencies in the pricing of exchange-traded funds.
721 +\newblock {\em Financial Analysts Journal\/}~{\em 73\/}(1), 24--54.
722 +
723 +\bibitem[\protect\citeauthoryear{Pontiff}{Pontiff}{1996}]{pontiff1996costly}
724 +Pontiff, J. (1996).
725 +\newblock Costly arbitrage: Evidence from closed-end funds.
726 +\newblock {\em The Quarterly Journal of Economics\/}~{\em 111\/}(4),
727 + 1135--1151.
728 +
729 +\bibitem[\protect\citeauthoryear{Ready and Ready}{Ready and
730 + Ready}{2022}]{ready2022order}
731 +Ready, M.~J. and R.~C. Ready (2022).
732 +\newblock Order flows and financial investor impacts in commodity futures
733 + markets.
734 +\newblock {\em The Review of Financial Studies\/}~{\em 35\/}(10), 4712--4755.
735 +
736 +\bibitem[\protect\citeauthoryear{Richie, Daigler, and Gleason}{Richie
737 + et~al.}{2008}]{richie2008examination}
738 +Richie, N., R.~T. Daigler, and K.~C. Gleason (2008).
739 +\newblock The limits to stock index arbitrage: Examining {S\&P} 500 futures and
740 + {SPDRs}.
741 +\newblock {\em Journal of Futures Markets\/}~{\em 28\/}(12), 1182--1205.
742 +
743 +\bibitem[\protect\citeauthoryear{Roll}{Roll}{1984}]{Roll1984}
744 +Roll, R. (1984).
745 +\newblock A simple implicit measure of the effective bid-ask spread.
746 +\newblock {\em The Journal of Finance\/}~{\em 39\/}(4), 1127--1139.
747 +
748 +\bibitem[\protect\citeauthoryear{Rosa}{Rosa}{2013}]{rosa2013}
749 +Rosa, C. (2013).
750 +\newblock The high-frequency response of exchange rates and interest rates to
751 + macroeconomic announcements.
752 +\newblock {\em Journal of Banking \& Finance\/}~{\em 37\/}(6), 2162--2174.
753 +
754 +\bibitem[\protect\citeauthoryear{Savor and Wilson}{Savor and
755 + Wilson}{2014}]{savor2014}
756 +Savor, P. and M.~Wilson (2014).
757 +\newblock Asset pricing: A tale of two days.
758 +\newblock {\em Journal of Financial Economics\/}~{\em 113\/}(2), 171--201.
759 +
760 +\bibitem[\protect\citeauthoryear{Schmeling and Wagner}{Schmeling and
761 + Wagner}{2019}]{schmeling2019}
762 +Schmeling, M. and C.~Wagner (2019).
763 +\newblock Does central bank tone move asset prices?
764 +\newblock {\em Review of Finance\/}~{\em 23\/}(5), 933--972.
765 +
766 +\bibitem[\protect\citeauthoryear{Scholtus, Van~Dijk, and Frijns}{Scholtus
767 + et~al.}{2014}]{scholtus2014speed}
768 +Scholtus, M., D.~Van~Dijk, and B.~Frijns (2014).
769 +\newblock Speed, algorithmic trading, and market quality around macroeconomic
770 + news announcements.
771 +\newblock {\em Journal of Banking \& Finance\/}~{\em 38}, 89--105.
772 +
773 +\bibitem[\protect\citeauthoryear{Shapiro, Sudhof, and Wilson}{Shapiro
774 + et~al.}{2022}]{shapiro2019}
775 +Shapiro, A.~H., M.~Sudhof, and D.~J. Wilson (2022).
776 +\newblock Taking the fed at its word: A new approach to estimating central bank
777 + objectives using text analysis.
778 +\newblock {\em Review of Economics and Statistics\/}~{\em 104\/}(4), 768--784.
779 +
780 +\bibitem[\protect\citeauthoryear{Shleifer and Summers}{Shleifer and
781 + Summers}{1990}]{shleifer1990noise}
782 +Shleifer, A. and L.~H. Summers (1990).
783 +\newblock The noise trader approach to finance.
784 +\newblock {\em Journal of Economic perspectives\/}~{\em 4\/}(2), 19--33.
785 +
786 +\bibitem[\protect\citeauthoryear{Sims and Zha}{Sims and
787 + Zha}{1999}]{sims1999error}
788 +Sims, C.~A. and T.~Zha (1999).
789 +\newblock Error bands for impulse responses.
790 +\newblock {\em Econometrica\/}~{\em 67\/}(5), 1113--1155.
791 +
792 +\bibitem[\protect\citeauthoryear{Singleton}{Singleton}{2014}]{singleton2014investor}
793 +Singleton, K. (2014).
794 +\newblock Investor flows and the 2008 boom/bust in oil prices.
795 +\newblock {\em Management Science\/}~{\em 60\/}(2), 300--318.
796 +
797 +\bibitem[\protect\citeauthoryear{Staer}{Staer}{2017}]{staer2017asset}
798 +Staer, A. (2017).
799 +\newblock Asset management via {ETFs}.
800 +\newblock {\em The Review of Financial Studies\/}~{\em 30\/}(9), 3225--3264.
801 +
802 +\bibitem[\protect\citeauthoryear{Swanson and Williams}{Swanson and
803 + Williams}{2014}]{swanson2014}
804 +Swanson, E.~T. and J.~C. Williams (2014).
805 +\newblock Measuring the effect of the zero lower bound on medium- and
806 + longer-term interest rates.
807 +\newblock {\em American Economic Review\/}~{\em 104\/}(10), 3154--3185.
808 +
809 +\bibitem[\protect\citeauthoryear{Tang and Xiong}{Tang and
810 + Xiong}{2012}]{tang2012index}
811 +Tang, K. and W.~Xiong (2012).
812 +\newblock Index investment and the financialization of commodities.
813 +\newblock {\em Financial Analysts Journal\/}~{\em 68\/}(6), 54--74.
814 +
815 +\bibitem[\protect\citeauthoryear{Todorov}{Todorov}{2024}]{todorov2021etf}
816 +Todorov, K. (2024).
817 +\newblock When passive funds affect prices: Evidence from volatility and
818 + commodity {ETFs}.
819 +\newblock {\em Review of Finance\/}~{\em 28\/}(3), 831--863.
820 +
821 +\bibitem[\protect\citeauthoryear{Veldkamp}{Veldkamp}{2011}]{veldkamp2011}
822 +Veldkamp, L.~L. (2011).
823 +\newblock {\em Information Choice in Macroeconomics and Finance}.
824 +\newblock Princeton University Press.
825 +
826 +\bibitem[\protect\citeauthoryear{Wang, Reimers, and Gurevych}{Wang
827 + et~al.}{2021}]{wang2021tsdae}
828 +Wang, K., N.~Reimers, and I.~Gurevych (2021).
829 +\newblock Tsdae: Using transformer-based sequential denoising auto-encoder for
830 + unsupervised sentence embedding learning.
831 +\newblock In {\em Findings of the Association for Computational Linguistics:
832 + EMNLP 2021}, pp.\ 671--688.
833 +
834 +\bibitem[\protect\citeauthoryear{Wongswan}{Wongswan}{2009}]{wongswan2009}
835 +Wongswan, J. (2009).
836 +\newblock The response of global equity indexes to u.s. monetary policy
837 + announcements.
838 +\newblock {\em Journal of International Money and Finance\/}~{\em 28\/}(2),
839 + 344--365.
840 +
841 +\bibitem[\protect\citeauthoryear{Working}{Working}{1960}]{working1960speculation}
842 +Working, H. (1960).
843 +\newblock Speculation on hedging markets.
844 +\newblock {\em Food Research Institute Studies\/}~{\em 1\/}(2), 185--220.
845 +
846 +\bibitem[\protect\citeauthoryear{Zhu, Sun, Tu, and Ji}{Zhu
847 + et~al.}{2022}]{zhu2022oil}
848 +Zhu, Z., L.~Sun, J.~Tu, and Q.~Ji (2022).
849 +\newblock Oil price shocks and stock market anomalies.
850 +\newblock {\em Financial Management\/}~{\em 51\/}(2), 573--612.
851 +
852 +\end{thebibliography}
added these-ulaval/main.pdf +0 −0

Binary file not shown.

added these-ulaval/main.tex +95 −0
@@ -0,0 +1,95 @@
1 +%% ---------------------------------------------------------------------------
2 +%% main.tex — Thèse de doctorat par articles, Université Laval (FESP)
3 +%%
4 +%% Three Essays on High-Frequency Return and Volatility Dynamics
5 +%% in Commodities and Financial Futures Markets
6 +%% Simon-Pierre Boucher — dépôt 2026
7 +%%
8 +%% Document maître basé sur le gabarit officiel « gabarit-doctorat-articles.tex »
9 +%% de la classe ulthese (v7.0). Compilation : latexmk (pdfLaTeX) — voir Makefile.
10 +%%
11 +%% NB (règles FESP) : le manuscrit déposé ne doit PAS comporter de page de
12 +%% titre — la FESP la produit au dépôt final. La classe génère la page
13 +%% frontispice obligatoire ci-dessous. Le comité de thèse n'apparaît nulle part
14 +%% dans le manuscrit (voir NOTES_FORMAT.md) ; il est rappelé en commentaire :
15 +%% - Marie-Hélène Gagnon (Université Laval), directrice de recherche
16 +%% - Gabriel Power (Université Laval), codirecteur de recherche
17 +%% - Marie-Claude Beaulieu (Université Laval), membre du comité
18 +%% - Andréanne Tremblay-Simard (Université Laval), membre du comité
19 +%% - Valeri Sokolovski (University of Alberta, Alberta School of Business),
20 +%% examinateur externe
21 +%% ---------------------------------------------------------------------------
22 +
23 +%% Doctorat (grade PhD). Langues : français (liminaires) + anglais (corps) ;
24 +%% la dernière langue déclarée — l'anglais — est la langue par défaut.
25 +\documentclass[PhD,11pt,french,english]{ulthese}
26 +
27 + %% Paquetages et macros harmonisés des trois chapitres
28 + \input{preamble}
29 +
30 + %% Mise en forme du mode français de babel (résumé, résumés de chapitres)
31 + \frenchsetup{%
32 + StandardItemizeEnv=true,
33 + ThinSpaceInFrenchNumbers=true,
34 + og=«, fg=»
35 + }
36 +
37 + %% Style de bibliographie : auteur-année « chicago », style dominant dans les
38 + %% chapitres (chap. 1 et 2) et conforme aux usages en finance. natbib est
39 + %% chargé par la classe avec les options round et authoryear.
40 + \bibliographystyle{chicago}
41 +
42 + %% Page frontispice (obligatoire)
43 + \titre{Three Essays on High-Frequency Return \\
44 + and Volatility Dynamics in Commodities \\
45 + and Financial Futures Markets}
46 + \auteur{Simon-Pierre Boucher}
47 + %% Intitulé officiel au répertoire ULaval (à confirmer avec la direction de
48 + %% programme — voir NOTES_FORMAT.md, hypothèse A1) :
49 + \programme{Doctorat en sciences de l'administration -- finance et assurance}
50 + \direction{Marie-Hélène Gagnon, directrice de recherche}
51 + \codirection{Gabriel Power, codirecteur de recherche}
52 +
53 + %% La commande \annee et la page de titre (\pagetitre) ne servent que pour
54 + %% une version de travail — INTERDITES dans le manuscrit déposé à la FESP.
55 + % \annee{2026}
56 +
57 +\begin{document}
58 +
59 +\frontmatter % pages liminaires (pagination romaine)
60 +
61 +\frontispice % page frontispice (obligatoire)
62 +
63 +\include{pages-liminaires/resume} % résumé français (obligatoire, = p. ii)
64 +\include{pages-liminaires/abstract} % abstract anglais
65 +\cleardoublepage
66 +
67 +\tableofcontents % table des matières (obligatoire)
68 +\cleardoublepage
69 +
70 +\listoftables % liste des tableaux
71 +\cleardoublepage
72 +
73 +\listoffigures % liste des figures
74 +\cleardoublepage
75 +
76 +\include{pages-liminaires/remerciements}
77 +\include{pages-liminaires/avant-propos} % obligatoire (thèse par articles)
78 +
79 +\mainmatter % corps du document (pagination arabe, p. 1)
80 +
81 +\include{introduction/introduction} % introduction générale (non numérotée)
82 +\include{chapitre1/chapitre1} % essai 1
83 +\include{chapitre2/chapitre2} % essai 2
84 +\include{chapitre3/chapitre3} % essai 3
85 +\include{conclusion/conclusion} % conclusion générale (non numérotée)
86 +
87 +\appendix % annexes (A, B) — issues du chapitre 3
88 +
89 +\include{annexes/annexe-preuves}
90 +\include{annexes/annexe-additionnel}
91 +
92 +%% Bibliographie unique consolidée (exigence FESP depuis l'automne 2023)
93 +\bibliography{bib/these}
94 +
95 +\end{document}
added these-ulaval/pages-liminaires/abstract.tex +58 −0
@@ -0,0 +1,58 @@
1 +\chapter*{Abstract} % ne pas numéroter
2 +\label{chap-abstract} % étiquette pour renvois
3 +\phantomsection\addcontentsline{toc}{chapter}{\nameref{chap-abstract}} % inclure dans TdM
4 +
5 +% BROUILLON À RÉVISER PAR L'AUTEUR.
6 +\begin{otherlanguage*}{english}
7 + This dissertation consists of three essays that use high-frequency data to
8 + study return and volatility dynamics in commodities and financial futures
9 + markets. The common thread is the minute-level identification of the
10 + mechanisms through which information --- macroeconomic announcements, trader
11 + positioning, and central bank communications --- is transmitted to prices
12 + and volatility, mechanisms that daily data cannot disentangle.
13 +
14 + The first essay asks whether the intensity of speculative trading in energy
15 + futures markets amplifies or dampens the reaction of prices to macroeconomic
16 + news. Using 5-minute data on six U.S. futures contracts (crude oil, natural
17 + gas, gold, silver, copper, and palladium) from April 2007 to February 2024,
18 + standardized surprises from 26 macroeconomic announcements, and a
19 + time-varying measure of speculative intensity built from disaggregated CFTC
20 + positions, the essay shows that higher speculative activity \emph{dampens}
21 + the response of returns and volatility to surprises and narrows bid-ask
22 + spreads. These stabilizing effects --- stronger for procyclical commodities
23 + than for gold, a safe haven --- are driven by money managers rather than
24 + swap dealers, suggesting that speculative trading improves liquidity and
25 + price discovery.
26 +
27 + The second essay builds a novel dataset of minute-level indicative Net Asset
28 + Value (iNAV) observations for four commodity exchange-traded funds (ETFs)
29 + --- gold, silver, crude oil, and natural gas --- and measures volatility
30 + transmission between each ETF and its underlying basket. Decomposing
31 + realized variance into continuous and jump components and combining HAR
32 + models with Bayesian vector autoregressions, the essay establishes that
33 + transmission runs primarily through jumps rather than diffusion, that it is
34 + substantially larger in 1-minute than in 30-minute data, and that its
35 + direction depends on the arbitrage mechanism: unidirectional from iNAV to
36 + ETF for physically backed precious metals, bidirectional and asymmetric for
37 + futures-based energy ETFs.
38 +
39 + The third essay decomposes FOMC statements into two semantic dimensions ---
40 + policy tone (hawkish/dovish) and informational novelty relative to the
41 + previous statement --- using an ensemble of language models (MiniLM and
42 + BERT) trained on Federal Reserve communications with data-driven, PCA-based
43 + reference selection. Across 148 FOMC announcements (2008--2025) and 1-minute
44 + data for seven futures contracts, the essay shows that tone predicts
45 + directional returns while novelty predicts volatility, with the
46 + tone--novelty interaction on VIX being the most robust effect. The two
47 + channels display opposite dynamics --- rapid uncertainty resolution for
48 + volatility, gradual two-hour repricing for returns --- a pattern
49 + inconsistent with a single information channel.
50 +
51 + Together, the essays show that intraday granularity is not a technical
52 + refinement but a condition for identification: the magnitude, direction, and
53 + even sign of the documented effects would be invisible or biased at the
54 + daily frequency. The results inform policy debates on speculative position
55 + limits, index-product design, and central bank communication strategy by
56 + identifying who transmits information to prices, through which channel, and
57 + at what horizon.
58 +\end{otherlanguage*}
added these-ulaval/pages-liminaires/avant-propos.tex +81 −0
@@ -0,0 +1,81 @@
1 +\chapter*{Avant-propos} % ne pas numéroter
2 +\label{chap-avantpropos} % étiquette pour renvois
3 +\phantomsection\addcontentsline{toc}{chapter}{\nameref{chap-avantpropos}} % inclure dans TdM
4 +
5 +% AVANT-PROPOS OBLIGATOIRE (thèse par articles — règles FESP).
6 +% Deux parties obligatoires : (1) déclaration sur l'utilisation de l'IA
7 +% générative (obligatoire pour tous depuis le 12 janvier 2026) ;
8 +% (2) renseignements sur chaque article inséré : état de préparation ou date
9 +% de soumission/acceptation/publication, modifications par rapport à la
10 +% version publiée le cas échéant, statut d'auteur, rôle exact de l'étudiant
11 +% et des coauteurs.
12 +% RAPPEL ADMINISTRATIF (hors manuscrit) : formulaire FES-100 (autorisation
13 +% des coauteurs) requis pour chaque article au dépôt initial, sauf article
14 +% publié dont l'éditeur a autorisé l'insertion ; formulaire FES-101
15 +% (autorisation de rédiger en anglais) au dossier.
16 +\begin{otherlanguage*}{french}
17 +
18 +\section*{Déclaration sur l'utilisation de l'intelligence artificielle générative}
19 +
20 +% BROUILLON À RÉVISER PAR L'AUTEUR selon les directives de la FESP.
21 +Dans le cadre de la préparation de cette thèse, j'ai utilisé l'outil
22 +d'intelligence artificielle générative Claude (Anthropic) pour écrire le code
23 +informatique servant à l'entraînement et à l'application des modèles de
24 +plongements lexicaux (\emph{embeddings}) du chapitre~3 (ensemble bi-modèle
25 +MiniLM et BERT). J'ai conçu l'architecture de l'analyse, supervisé la
26 +production de ce code, puis vérifié, testé et validé l'ensemble des
27 +programmes et des résultats qui en découlent. L'IA générative n'a pas été
28 +utilisée pour produire le contenu scientifique de la thèse : les questions de
29 +recherche, l'analyse, l'interprétation des résultats et la rédaction du texte
30 +demeurent entièrement les miennes et celles de mes coauteurs.
31 +
32 +\section*{Renseignements sur les articles insérés}
33 +
34 +Cette thèse comprend trois articles, chacun constituant un chapitre. Les
35 +travaux de recherche ont été réalisés au cours de mes études doctorales, sous
36 +la direction de Marie-Hélène Gagnon et la codirection de Gabriel Power. Pour
37 +chacun des trois articles, je suis premier auteur : j'ai contribué à la
38 +conception de la recherche, construit les bases de données, réalisé
39 +l'ensemble des analyses empiriques et rédigé les manuscrits. Mes coauteurs,
40 +Marie-Hélène Gagnon et Gabriel J. Power (Université Laval), ont contribué à
41 +la conception de la recherche, à l'interprétation des résultats et à la
42 +révision des manuscrits. % [À VALIDER PAR L'AUTEUR : description exacte des
43 +% contributions respectives.]
44 +
45 +\subsection*{Chapitre 1 — Speculative Trading in Energy Markets: Evidence
46 + from Macroeconomic Surprises}
47 +
48 +Cet article est coécrit avec Marie-Hélène Gagnon et Gabriel J. Power.
49 +Statut : [À COMPLÉTER — la source indique une version révisée soumise à
50 +\emph{The Energy Journal} (« Revised version for The Energy Journal », avec
51 +remerciements à l'éditeur George Filis et à deux évaluateurs anonymes) ;
52 +préciser l'état exact : soumis le …, en révision, accepté le …, publié
53 +le …, et les modifications entre la version insérée et la version publiée,
54 +s'il y a lieu.] Les auteurs remercient les participants de séminaires à
55 +Humboldt University Berlin, South Dakota State University (Ness School),
56 +University of Illinois Urbana-Champaign (ACE), ainsi que les participants aux
57 +congrès de la Commodity \& Energy Markets Association (2021), de la World
58 +Finance \& Banking Association (2021), de la Société canadienne de sciences
59 +économiques (2022), de la 4\ieme{} conférence Ethical Finance and
60 +Sustainability (2022) et de la Multinational Finance Society (2022) ; Jocelyn
61 +Grira, Joseph Marks et Alessandro Melone (rapporteurs), ainsi que Scott
62 +Irwin, Michel Robe et Zhiguang Wang ; pour le soutien financier, le Conseil
63 +de recherches en sciences humaines (CRSH) et la Chaire Industrielle-Alliance
64 +Groupe financier.
65 +
66 +\subsection*{Chapitre 2 — Seeing Through the ETF: Indicative NAV and
67 + Commodity Volatility Transmission}
68 +
69 +Cet article est coécrit avec Marie-Hélène Gagnon et Gabriel J. Power.
70 +Statut : [À COMPLÉTER — la source indique une version préparée pour
71 +soumission au \emph{Journal of Futures Markets} ; préciser si l'article a été
72 +soumis et à quelle date.]
73 +
74 +\subsection*{Chapitre 3 — Returns and Volatility Around FOMC Announcements:
75 + A High-Frequency Analysis of Policy Tone and Novelty}
76 +
77 +Cet article est coécrit avec Marie-Hélène Gagnon et Gabriel J. Power.
78 +Statut : [À COMPLÉTER — manuscrit daté du 6 mars 2026 dans la source ;
79 +préciser s'il est non soumis, soumis (revue et date) ou accepté.]
80 +
81 +\end{otherlanguage*}
added these-ulaval/pages-liminaires/remerciements.tex +24 −0
@@ -0,0 +1,24 @@
1 +\chapter*{Remerciements} % ne pas numéroter
2 +\label{chap-remerciements} % étiquette pour renvois
3 +\phantomsection\addcontentsline{toc}{chapter}{\nameref{chap-remerciements}} % inclure dans TdM
4 +
5 +% BROUILLON À RÉVISER PAR L'AUTEUR.
6 +% Autres éléments possibles : membres du comité (Marie-Claude Beaulieu,
7 +% Andréanne Tremblay-Simard), examinateur externe (Valeri Sokolovski,
8 +% University of Alberta), organismes de financement (p. ex. CRSH, Chaire
9 +% Industrielle-Alliance Groupe financier), collègues.
10 +\begin{otherlanguage*}{french}
11 + Je tiens d'abord à remercier ma directrice de recherche, Marie-Hélène
12 + Gagnon, et mon codirecteur, Gabriel Power, pour leur encadrement, leur
13 + disponibilité et leur confiance tout au long de mon parcours doctoral.
14 + Leurs conseils rigoureux et leur générosité intellectuelle ont façonné
15 + chacun des trois essais de cette thèse, et bien au-delà, le chercheur que
16 + je suis devenu.
17 +
18 + Je remercie enfin mes parents, Richard et feu Sylvie, à qui je dois tout.
19 + Leur soutien indéfectible et les valeurs qu'ils m'ont transmises m'ont
20 + porté jusqu'ici. Cette thèse est aussi la leur.
21 +
22 + [À COMPLÉTER — ajouter, s'il y a lieu : membres du comité, examinateur
23 + externe, organismes de financement, collègues et proches.]
24 +\end{otherlanguage*}
added these-ulaval/pages-liminaires/resume.tex +65 −0
@@ -0,0 +1,65 @@
1 +\chapter*{Résumé} % ne pas numéroter
2 +\label{chap-resume} % étiquette pour renvois
3 +\phantomsection\addcontentsline{toc}{chapter}{\nameref{chap-resume}} % inclure dans TdM
4 +
5 +% BROUILLON À RÉVISER PAR L'AUTEUR — maximum 700 mots (règle FESP).
6 +\begin{otherlanguage*}{french}
7 + Cette thèse se compose de trois essais qui exploitent des données à haute
8 + fréquence pour étudier la dynamique des rendements et de la volatilité sur
9 + les marchés de contrats à terme de matières premières et d'actifs
10 + financiers. Le fil conducteur est l'identification fine, à l'échelle de la
11 + minute, des mécanismes par lesquels l'information — annonces
12 + macroéconomiques, activité des intervenants, communications de banque
13 + centrale — se transmet aux prix et à la volatilité, mécanismes que les
14 + données quotidiennes ne permettent pas de distinguer.
15 +
16 + Le premier essai examine si l'intensité de la spéculation sur les marchés à
17 + terme de l'énergie amplifie ou atténue la réaction des prix aux nouvelles
18 + macroéconomiques. À partir de données à cinq minutes sur six contrats à
19 + terme américains (pétrole brut, gaz naturel, or, argent, cuivre, palladium)
20 + d'avril 2007 à février 2024, de surprises standardisées tirées de
21 + 26~annonces macroéconomiques et d'une mesure d'intensité spéculative
22 + construite sur les positions désagrégées de la CFTC, l'essai montre qu'une
23 + activité spéculative plus élevée \emph{atténue} la réaction des rendements
24 + et de la volatilité aux surprises et resserre les écarts acheteur-vendeur.
25 + Ces effets stabilisateurs — plus marqués pour les matières premières
26 + procycliques que pour l'or, valeur refuge — proviennent des \emph{money
27 + managers} plutôt que des \emph{swap dealers}, ce qui suggère un effet
28 + bénéfique de la spéculation sur la liquidité et la découverte des prix.
29 +
30 + Le deuxième essai construit une base de données inédite de valeurs
31 + liquidatives indicatives (iNAV) à la minute pour quatre fonds négociés en
32 + bourse (FNB) de matières premières — or, argent, pétrole et gaz naturel — et
33 + mesure la transmission de volatilité entre chaque FNB et son panier
34 + sous-jacent. En décomposant la variance réalisée en composantes continue et
35 + de sauts et en combinant modèles HAR et autorégressions vectorielles
36 + bayésiennes, l'essai établit que la transmission passe principalement par
37 + les sauts plutôt que par la diffusion, qu'elle est nettement plus forte dans
38 + les données à une minute qu'à trente minutes, et que sa direction dépend du
39 + mécanisme d'arbitrage : unidirectionnelle — de l'iNAV vers le FNB — pour les
40 + métaux précieux détenus physiquement, bidirectionnelle et asymétrique pour
41 + les FNB énergétiques adossés à des contrats à terme.
42 +
43 + Le troisième essai décompose les communiqués du FOMC en deux dimensions
44 + sémantiques — le ton de politique monétaire (\emph{hawkish}/\emph{dovish})
45 + et la nouveauté informationnelle par rapport au communiqué précédent — à
46 + l'aide d'un ensemble de modèles de langage (MiniLM et BERT) entraînés sur
47 + les communications de la Réserve fédérale, avec une sélection des références
48 + fondée sur l'analyse en composantes principales. Sur 148~annonces du FOMC
49 + (2008--2025) et des données à une minute pour sept contrats à terme, l'essai
50 + montre que le ton prédit les rendements directionnels, tandis que la
51 + nouveauté prédit la volatilité ; l'interaction des deux dimensions sur le
52 + VIX constitue l'effet le plus robuste. Les deux canaux suivent des
53 + dynamiques opposées — résolution rapide de l'incertitude pour la volatilité,
54 + réévaluation progressive sur deux heures pour les rendements — ce qui est
55 + incompatible avec un canal d'information unique.
56 +
57 + Ensemble, ces trois essais montrent que la granularité intrajournalière
58 + n'est pas un raffinement technique mais une condition d'identification :
59 + l'ampleur, la direction et même le signe des effets documentés seraient
60 + invisibles ou biaisés à fréquence quotidienne. Les résultats éclairent des
61 + débats de politique publique — réglementation des positions spéculatives,
62 + conception des produits indiciels, stratégie de communication des banques
63 + centrales — en identifiant précisément qui transmet l'information aux prix,
64 + par quel canal et à quel horizon.
65 +\end{otherlanguage*}
added these-ulaval/preamble.tex +68 −0
@@ -0,0 +1,68 @@
1 +%% ---------------------------------------------------------------------------
2 +%% preamble.tex — paquetages et macros harmonisés pour les trois chapitres.
3 +%%
4 +%% La classe ulthese (basée sur memoir) charge déjà : babel, natbib (options
5 +%% round, semicolon, authoryear), hyperref (en tout dernier), graphicx, xcolor,
6 +%% microtype et numprint. Le paquetage geometry est INCOMPATIBLE avec la classe
7 +%% (les marges FESP — 30 mm G/D, 25 mm H/B — sont imposées par ulthese).
8 +%% Les paquetages des préambules d'origine non repris ici étaient soit
9 +%% inutilisés, soit en conflit avec memoir — voir INVENTAIRE.md.
10 +%% ---------------------------------------------------------------------------
11 +
12 +% --- Mathématiques (chapitres 1 à 3) ---
13 +\usepackage{amsmath}
14 +\usepackage{amssymb}
15 +\usepackage{amsthm}
16 +
17 +% --- Tableaux ---
18 +\usepackage{booktabs} % filets \toprule/\midrule/\bottomrule (chap. 2-3)
19 +\usepackage{multirow} % cellules fusionnées verticalement (chap. 1)
20 +\usepackage{threeparttable} % tableaux avec notes (tablenotes) (chap. 1-3)
21 +\usepackage{rotating} % environnement sidewaystable (chap. 1)
22 +\usepackage{pdflscape} % pages en paysage (environnement landscape, chap. 1-3)
23 +\usepackage{siunitx} % colonnes S alignées sur la décimale (chap. 1)
24 +\sisetup{ % réglages repris du préambule d'origine du chap. 1
25 + round-mode = places,
26 + round-precision = 3,
27 + detect-weight = true,
28 + detect-inline-family = math
29 +}
30 +
31 +% --- Figures ---
32 +\usepackage{float} % placement [H] (chap. 3)
33 +\usepackage{subcaption} % sous-figures (annexe B du chap. 3)
34 +\graphicspath{{chapitre1/figures/}{chapitre2/figures/}{chapitre3/figures/}}
35 +
36 +% --- Environnements de théorèmes ---
37 +% Définis une seule fois pour toute la thèse (ils étaient redéfinis dans chaque
38 +% article), numérotés par chapitre. Le chap. 3 utilise definition/theorem/
39 +% proposition/assumption et l'environnement proof d'amsthm ; les définitions
40 +% des chap. 1-2 (theorem/corollary/proposition/hyp) étaient inutilisées dans
41 +% leurs corps de texte mais sont reprises par prudence.
42 +\theoremstyle{plain}
43 +\newtheorem{theorem}{Theorem}[chapter]
44 +\newtheorem{corollary}[theorem]{Corollary}
45 +\newtheorem{proposition}{Proposition}[chapter]
46 +\theoremstyle{definition}
47 +\newtheorem{definition}{Definition}[chapter]
48 +\newtheorem{assumption}{Assumption}[chapter]
49 +\newtheorem{hyp}{Hypothesis}[chapter]
50 +\newtheorem{subhyp}{Hypothesis}[hyp]
51 +\renewcommand{\thesubhyp}{\thehyp\alph{subhyp}}
52 +
53 +% --- Compatibilité ---
54 +% Les tableaux du chap. 1 utilisent \singlespacing (setspace) à l'intérieur des
55 +% flottants ; setspace interagit mal avec memoir, on mappe vers l'équivalent
56 +% memoir, de portée locale identique.
57 +\newcommand{\singlespacing}{\SingleSpacing}
58 +
59 +% --- Métadonnées PDF ---
60 +% hyperref est chargé automatiquement par la classe en fin de préambule ;
61 +% les liens utilisent la couleur discrète ULlinkcolor définie par la classe.
62 +\AtBeginDocument{%
63 + \hypersetup{%
64 + pdftitle = {Three Essays on High-Frequency Return and Volatility Dynamics
65 + in Commodities and Financial Futures Markets},
66 + pdfauthor = {Simon-Pierre Boucher}%
67 + }%
68 +}
69