Monetary Policy in the Media Spotlight: Sentiments, Signals, and Economic Impact

Monetary Policy in the Media Spotlight: Sentiments, Signals, and Economic Impact

\pagenumbering{gobble}

\pagenumbering{arabic}

\oldsection{Introduction} \addcontentsline{apptoc}{section}{\numberline{\Alph{section}}Introduction}

Central banks devote a large share of their work to communicating with the public. The Bernanke quote is a deliberate exaggeration, but it captures a now-mainstream view: explaining current decisions, signaling future intentions, and shaping public expectations are integral to the conduct of monetary policy. A growing empirical literature documents systematic effects of central-bank language on interest rates, asset prices, and survey expectations.\footnote{See doh2022deciphering, silva2025text, sekkel2025money, aruoba2024identifying, alexopoulos2024more, gorodnichenko2023voice, cieslak2023policymakers, gardner2022words, shapiro2022taking, hansen2018transparency.} This work builds on a long-standing tradition emphasizing the role of narratives and communication in monetary policy fortinthesis, romer1989does.

Central-bank messages do not, however, reach the public directly. Survey evidence on both sides of the Atlantic shows that well over half of households learn about monetary policy through traditional and online news media, while only a small minority reads central-bank communications first-hand blinder2024communication. Figure (ref) illustrates this pattern for the euro area. The media therefore stand between the central bank and the public as an intermediate producer of signals.\footnote{coibion2022mp provide experimental evidence that this indirect news-media route is a measurable but partial channel of monetary-policy communication: households exposed to a USA Today summary of an FOMC meeting revise their inflation expectations, but only about half as strongly as those reading the FOMC statement directly.}

The media coverage contains pieces of the central bank's own communication, but also factual reporting of the macroeconomic events that the central bank is also reacting to, and editorial choices that depend on news values, journalistic conventions, and competition for attention nimark2014man, chahrour2021sectoral. As carroll2003macroeconomic shows, household expectations are formed in part by reading what the news contains, so a media coverage that selectively emphasizes some aspects of monetary policy and downplays others can shift the public's view of the policy stance even when the underlying actions are unchanged.

This composite nature makes media coverage an endogenous outcome of the same macroeconomic system that monetary policy operates on. Figure (ref) depicts the resulting structure. The central bank acts on its instruments, economic outcomes feed back into the next policy decision, and the central bank's own communication reaches the media alongside the flow of economic news. The media reshape these inputs through editorial choices into the narratives that households and firms read; agents form perceptions and adjust their consumption and investment decisions, which in turn affect economic outcomes. The diagram has two key implications. First, the media-narrative node has three labeled inputs so any indicator of media sentiment is, by construction, a composite of central-bank communication (Communication), current fundamentals (Economic News), and an editorial residual (Editorial). Second that residual is plausibly orthogonal to policy decisions and identifying how monetary policy responds to media sentiment requires an empirical strategy that respects this endogeneity, rather than treating sentiment as a passive informational signal.

To formalize the endogenous role of media narratives, we adopt the behavioral New-Keynesian framework of gabaix2020behavioral, in which a fixed attention parameter $M^h$ down-weights the response of current decisions to expected future variables, and we augment it along two dimensions. The central-bank reaction function includes a feedback to a media-sentiment index, and that index follows a law of motion that decomposes it explicitly into four blocks: a slow-moving autoregressive component, a fundamentals component, a communication component built from anticipated policy shocks, and a residual narrative innovation that captures editorial choices. Sentiment is therefore an endogenous equilibrium object: it does not enter the IS or Phillips-curve equations directly, but it interacts with monetary policy decisions through the central bank's reaction function and feeds back into the expectations-relevant macroeconomic dynamics through the equilibrium response of the policy rate. The model produces sharp implications for impulse responses to forward guidance.

We then turn to measurement. Using natural-language-processing tools applied to a corpus of more than 50,000 articles from major Canadian newspapers between 1977 and 2024, we construct media indicators of monetary-policy sentiment along four dimensions: topic, tone, temporal orientation, and uncertainty. We use dictionary methods, transformer-based encoder models (FinBERT and ModernFinBERT), and a structured generative-LLM framework (CBILA). We find that media sentiments affect consumer inflation and wage expectations, and improve the out-of-sample forecast precision of GDP growth and CPI inflation.

We then estimate forward-looking Taylor rules for the Bank of Canada, treating media sentiment as endogenous as the model dictates. Estimated by GMM with appropriate instruments, the dictionary, FinBERT and ModernFinBERT sentiment measures all load positively and significantly on the policy rate; the gap between OLS and GMM estimates confirms that simultaneity between sentiment and policy is a first-order concern. The CBILA generative-LLM tones fail standard IV validity tests because their pre-training data extend into the estimation sample, which contaminates the lagged-instrument exclusion restriction.

We then move from the partial-equilibrium evidence on the Taylor rule to the full dynamic transmission of monetary policy in a Bayesian structural VAR. Sign restrictions derived from the model identify three structural shocks: an anticipated monetary policy shock, an unanticipated monetary policy shock, and a narrative shock that captures autonomous variation in media coverage orthogonal to fundamentals and to central-bank communication. The unanticipated easing cuts the policy rate on impact and is followed by a hump-shaped expansion of output and prices; the anticipated easing produces a smaller impact response of the rate, a slow build-up of output that peaks near the announced implementation date, and a positive price response, consistent with the cognitive-discounting mechanism. The narrative shock contributes a non-trivial share of medium-horizon macroeconomic variance. A counterfactual that shuts down the dynamic feedback from media sentiment to the rest of the system attenuates the response of output and prices to the anticipated monetary policy shock, confirming that media sentiment operates as a quantitatively relevant propagation channel. Media coverage therefore matters both as a separate source of macroeconomic fluctuation and as a transmission mechanism for conventional monetary policy.

Two recent papers study the role of narratives in monetary policy. andre2026narratives measure the narratives US households use to interpret macroeconomic phenomena, show in randomized experiments that these narratives shape inflation expectations, and embed them in an otherwise standard New Keynesian model. kaminski2026narratives extract narratives from FOMC transcripts via an LLM-based directed-acyclic-graph approach and find that the transmission of monetary policy is strongly narrative dependent. Our focus is the media that mediate between the central bank and households: the news coverage that filters central-bank communication and economic events into the narratives households read and the central bank observes. The three papers therefore cover complementary nodes of monetary-policy transmission --- households, central bank, and the media in between.

The paper is structured as follows. Section (ref) introduces the behavioral New-Keynesian model with media as a narrative friction. Section (ref) constructs measures of media coverage about monetary policy, including the dictionary, FinBERT, and CBILA generative-AI frameworks. Section (ref) relates these measures to household expectations and evaluates the predictive content of the indicators for realized macroeconomic outcomes. Section (ref) estimates text-augmented Taylor rules. Section (ref) identifies the dynamic transmission of monetary policy and media sentiment in a Bayesian SVAR. Section (ref) concludes.

\oldsection{A Behavioral New-Keynesian Model with Media Narratives} \addcontentsline{apptoc}{section}{\numberline{\Alph{section}}A Behavioral New-Keynesian Model with Media Narratives}

We embed the media into a stylized New-Keynesian model with cognitive discounting in order to (i) discipline how narratives interact with monetary policy, and (ii) guide the empirical strategy of Sections (ref), (ref), (ref) and (ref). Two features of the model are central. First, following gabaix2020behavioral, both the Euler equation and the Phillips curve carry an attention parameter $M$ that downweights distant-future events, so that forward guidance does not pass through one-for-one. Second, the public's view of monetary policy is filtered by a media-sentiment variable $s_t$ that aggregates four objects: the central bank's own communication, current fundamentals, an autoregressive component, and a residual narrative innovation. The first three components are by construction correlated with the policy decision; only the last is plausibly independent. The model is therefore the structural counterpart of Figure (ref): the three labeled inputs of the media-narrative node (Communication, Economic News, Editorial) correspond to three of the four blocks of the sentiment law of motion introduced in Section (ref), and the central-bank reaction function closes the feedback loop in the diagram.

\oldsubsection{Households and firms} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Households and firms}

A representative household consumes a final good $C_t$, supplies labor $N_t$, and invests in nominal bonds $B_t$ that pay gross return $R_{t-1}$ between $t-1$ and $t$. Lifetime utility is

with period budget constraint $P_t C_t + B_t = W_t N_t + R_{t-1} B_{t-1} + \int \Pi_t(j)\, dj$. Here $\sigma$ and $\psi$ are the inverse intertemporal and Frisch elasticities, $\chi > 0$ is the labor-disutility weight, $\beta$ is the discount factor, $W_t$ is the nominal wage, and $\Pi_t(j)$ are firm $j$'s profits. A continuum of price-setting firms $j \in [0,1]$ produces differentiated varieties with linear technology $Y_t(j) = N_t(j)$ and faces Calvo pricing rigidities with reset probability $1-\theta$. The final good is a CES aggregator with elasticity $\epsilon$, yielding the standard demand schedules and the aggregate price index $P_t = \big( \int_0^1 P_t(j)^{1-\epsilon} dj \big)^{1/(1-\epsilon)}$.

\oldsubsection{Central bank with feedback to sentiment} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Central bank with feedback to sentiment}

The central bank sets the gross nominal rate $R_t$ following an inertial Taylor rule that, compared to the textbook specification, includes a feedback to media sentiment $S_t$:

The smoothing parameter is $\rho_r$. The elasticities $\phi_\pi$, $\phi_x$, and $\phi_s$ govern the response to inflation $\Pi_t$, the output gap $\tilde Y_t \equiv Y_t/Y_t^n$, and the sentiment index $S_t$, respectively. Variables without subscripts denote steady-state values. Two innovations drive the policy rate. The conventional surprise $\varepsilon^u_t$ is i.i.d. The anticipated shock $\varepsilon^a_{t-\tau}$ is announced $\tau$ periods in advance and is meant to capture explicit central-bank communication about a future change in the policy stance, that is, forward guidance.\footnote{The Bank of Canada has progressively expanded its set of communication tools and now publishes a summary of policy deliberations after each decision, starting in February 2023; see jain2023summaries for the institutional context.} The cognitive-discounting block below determines how strongly agents react to the announced component.

\oldsubsection{Cognitive discounting and equilibrium} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Cognitive discounting and equilibrium}

For the news media of Figure (ref) to play a substantive role in the framework, agents cannot be fully attentive to every signal that the central bank and the macroeconomy generate. If they were, fully rational agents would extract the same information from the underlying communication and fundamentals as journalists do, and the media node would be mechanically redundant. We follow gabaix2020behavioral and add a friction on agents' attention to distant signals; the next subsection then interprets the news media as the institution that produces the summaries on which the inattentive public relies.

We adopt the reduced-form behavioral specification of gabaix2020behavioral, in which agents' perceived law of motion of the state vector $\mathbf{X}_t$. This is operationalized by downweighted future variables with an attention parameter $M^h \in [0,1]$ for households (with $M^h = 1$ recovering full rationality), as well as $M^f$ for firms making pricing decisions. We do not re-derive the linearization here; following Gabaix's derivation, the equilibrium around the zero-inflation steady state takes the cognitively discounted New-Keynesian form

where lowercase letters denote log deviations and $\kappa$ takes its standard New-Keynesian value. We treat $M^h$ and $M^f$ as exogenous attention parameters, in line with gabaix2020behavioral. The Taylor rule ((ref)) is itself a reduced-form structural specification: the loading $\phi_s$ on sentiment is taken as a parameter of the rule rather than derived from a central-bank optimisation problem.

\oldsubsection{Media as a narrative friction} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Media as a narrative friction}

Cognitive discounting reduces the response of current decisions to expected future variables, but it does not by itself say who summarizes those distant signals for an inattentive public. The news media are the natural candidate, and the link between the two frictions is tight: limited attention is what gives media narratives traction, and media narratives are what an inattentive public actually reads briand2024inflation. The channel through which the central bank reaches agents is therefore not direct but is mediated by journalists, who can amplify, distort, or filter the original signal nimark2014man,chahrour2021sectoral,carroll2003macroeconomic. Sentiment $s_t$ is an intermediate variable, not a primitive: it summarizes how the public reads both the current state of the economy and the central bank's stance toward the future. Closely related, andre2026narratives embed household narratives as subjective causal models of inflation in an otherwise standard New-Keynesian framework and show that they affect aggregate equilibrium outcomes. Our framework can be read as a media-side counterpart: the narratives are produced by an intermediate institution (journalists) rather than held directly by agents, and they enter both the law of motion of $s_t$ and the central bank's reaction function.

Figure (ref) illustrates the editorial dimension on three Canadian articles published around inflation-policy turning points. The same underlying policy environment is rendered as a war on inflation that generates uncertainty, as an obstacle that complicates the central bank's task, or as a return to the target range, depending on the journalistic angle. The factual content of each article overlaps substantially with the macroeconomic news the central bank itself reads, but the headline tone, framing, and word choice are not pinned down by that content alone. The variation that is left after conditioning on observable fundamentals and on the central bank's communication is what we call the editorial component, and it is the empirical counterpart of the residual narrative innovation $\varepsilon^s_t$ in the law of motion below.

We model the law of motion of media sentiment as

treating $s_t$ as an aggregate of the press tone read by households and firms. This decomposes $s_t$ into four blocks: an autoregressive component capturing slow-moving narrative momentum, a fundamentals component (output gap and inflation up to horizon $\tau$), a communication component built from the announced policy shocks $\{\varepsilon^a_{t-\tau+k}\}_{k=0}^\tau$, and a residual term $\varepsilon^s_t$ that the model treats as orthogonal to the previous three blocks. Strictly speaking, $\varepsilon^s_t$ is the noise term of the LoM in the sense of the news and noise literature lorenzoni2009theory, blanchard2013news, angeletos2013sentiments; we label it a narrative shock as a substantive interpretation, since the orthogonal residual of a sentiment LoM is most naturally attributed to editorial choices and framing nimark2014man, chahrour2021sectoral. The orthogonality of $\varepsilon^s_t$ to fundamentals and communication is an identifying assumption, not an empirical claim about the data: it is the assumption that, jointly with the sign restrictions of Section (ref), lets the SVAR isolate the narrative shock as a separate structural object. The discount factor $M \in (0,1]$ in ((ref)) has the same interpretation as $M^h$ in ((ref))--((ref)), reflecting that agents read distant news with declining intensity. The fundamentals loadings $\lambda_x, \lambda_\pi > 0$ encode the assumption that the press reads good fundamentals (high $x_t, \pi_t$) as positive sentiment.\footnote{There are only demand-side shocks in the model.} The sign of the loading $\lambda_a$ on the announced policy shock is more subtle, because an announcement of a future easing ($\varepsilon^a < 0$) admits two readings.

Under $\lambda_a < 0$, the press reads the announcement at face value like households: an anticipated easing is read as positive news because the rate is going to fall. grigoli2026monetary provide direct survey evidence consistent with this reading. Using a randomized-information experiment they show that U.S. households interpret a higher federal funds rate as raising inflation and worsening economic conditions, a pattern that fits a cost-channel reading of monetary policy rather than the central-bank information effect of romer2000federal. Therefore, households read an anticipated rate cut as good news for the outlook, which is exactly the face-value reading that $\lambda_a < 0$ encodes on the media side. Both the fundamentals block and the communication block of ((ref)) then contribute positively to $s_0$.

Under $\lambda_a > 0$, the press attaches to the announcement a sign that is opposite to its substantive policy content, through a central bank information effect on the media side: journalists read through the literal cut to the implicit signal of incipient weakness that motivates it, so the announced easing is framed as bad news for the outlook romer2000federal, nakamura2018high. Under this case, the impact response of sentiment to an anticipated easing is positive, $s_0 > 0$, only when the fundamentals block of the LoM dominates the direct loading of the announcement, $\lambda_x x_0 + \lambda_\pi \pi_0 > \lambda_a |\varepsilon^a_0|$. We verify numerically that the inequality holds under our calibration. Even when it fails, the contemporaneous lean against the wind on $r_t$ can still hold under a weaker condition: the policy response to inflation outweighs the negative sentiment contribution to the rule, $\phi_\pi \pi_0 + \phi_x x_0 > \phi_s |s_0|$. With $\phi_\pi = 1.5$ and $\phi_s = 0.5$, the policy weight on $\pi_0$ is three times the weight on $s_0$, so this weaker condition is satisfied for substantially smaller values of $\pi_0$ than the dominance for $s_t$.

This decomposition is the structural counterpart of the diagram in Figure (ref), and it has direct implications for empirical work. Three of the four blocks of $s_t$ are mechanically correlated with the policy innovations in equation ((ref)): the autoregressive component is predetermined and feeds into $r_{t-1}$; the fundamentals component covaries with $r_t$ through the central bank's response to $\pi_t$ and $x_t$; the communication component is a function of the same announced shocks $\varepsilon^a$ that already enter the policy rule. Only the residual $\varepsilon^s_t$ is plausibly orthogonal to $\varepsilon^u_t$. Section (ref) returns to this decomposition when discussing the identification of $\phi_s$.

\oldsubsection{Impulse responses} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Impulse responses}

We solve the linearized system ((ref))-((ref)) together with the sentiment law of motion ((ref)) under the calibration in Table (ref), and trace out impulse responses to a one-standard-deviation anticipated expansionary shock to the policy rate, $\varepsilon^a_{t-4}$, under different combinations of the attention parameter $M^h$ and the sentiment loading $\phi_s$.

Figure (ref) traces impulse responses of output and inflation to the anticipated expansionary shock under three calibrations: $M^h = 1.0, \phi_s = 0$ (full attention, no sentiment feedback); $M^h = 0.8, \phi_s = 0$ (cognitive discounting, no sentiment feedback); and $M^h = 0.8, \phi_s = 0.5$ (cognitive discounting with active sentiment feedback). Three patterns are worth noting.

First, lowering household attention from $M^h = 1$ to $M^h = 0.8$ flattens and lengthens the responses of output and inflation, with the peak shifted from impact toward the announced implementation date $t = 4$: cognitive discounting is a propagation mechanism in itself, since forward-looking agents do not fully internalize the announced future cut at the moment it is revealed.

Second, when $\phi_s > 0$ the policy rate rises on impact, despite the announcement being for a future cut. The mechanism is a lean against the wind that runs through the sentiment law of motion and the reaction function. At $t = 0$ the announced cut has not yet materialized, so $r_0$ moves only through $(1-\rho_r)(\phi_\pi \pi_0 + \phi_x x_0 + \phi_s s_0)$. Forward-looking households and firms anticipate the future easing and bid up current $\pi_0$ and $x_0$ through the cognitive-discounting Euler equation and Phillips curve. The press loads on this incipient optimism through the fundamentals block of ((ref)), and $s_0$ rises under the dominance condition $\lambda_x x_0 + \lambda_\pi \pi_0 > \lambda_a |\varepsilon^a_0|$ discussed in Section (ref), which holds in our $M^h = 0.8$ calibration. The central bank reads this rise in $s_0$ as a signal of incipient overheating and tightens contemporaneously via $\phi_s s_0 > 0$. The net effect of the lean is to dampen the expansionary impact of the announcement: under $\phi_s = 0.5$ the responses of output and inflation lie inside the corresponding $\phi_s = 0$ benchmark, closer to the zero-line.

Third, were an independent narrative shock $\varepsilon^s_t$ to realize alongside the announcement, the policy rate would move in the same direction as $\varepsilon^s_t$: a positive narrative shock would reinforce the contractionary lean already at work, leading to a stronger rate hike on impact and a more muted real response; a negative narrative shock would produce the opposite. The model therefore predicts that media-driven optimism or pessimism, conditional on the same fundamental announcement, can amplify or dampen the macroeconomic effects of monetary policy.

The model has three empirical implications. The sentiment law of motion gives $s_t$ a role in shifting household and firm expectations beyond their own persistence; the Taylor rule puts $\phi_s$ inside the policy rule as a structural loading whose consistent estimation has to handle the simultaneous determination of $s_t$ and $r_t$ through ((ref))--((ref)); and the impact responses derived above deliver sign restrictions that identify the three structural shocks $\{\varepsilon^u, \varepsilon^a, \varepsilon^s\}$ in a Bayesian SVAR.

\oldsection{Measuring sentiments and narratives about monetary policy} \addcontentsline{apptoc}{section}{\numberline{\Alph{section}}Measuring sentiments and narratives about monetary policy}

We extract signals about monetary policy narratives from news media using four features of natural language processing (NLP) techniques : (i) topic or narrative with Latent Dirichlet Allocation (LDA) developed by blei2003latent ; (ii) tone or sentiment with the dictionary method and large language models (LLMs), particularly FinBERT approach; (iii) time dimension with temporal tagging and (iv) uncertainty-adjusted topic attention leveraging word embedding (word2vec) as described by mikolov2013distributed.\footnote{See ash2023text for an overview of the usage of text in economic research.} We measure aggregated and topic-adjusted sentiment/uncertainty/tense from over 50,000 news articles published in major Canadian newspapers including the National Post, Calgary Herald, Edmonton Journal, Montreal Gazette, Ottawa Citizen, Regina Leader-Post, The Globe and Mail, Vancouver Sun, and the Victoria Times-Colonist from January 1977 to January 2024.\footnote{A parallel corpus of US press coverage of Canadian monetary policy from January 1980 onward is also available, totalling approximately 6,000 articles. The Wall Street Journal is the largest contributor with about 4,000 articles, followed by the New York Times, Washington Post, Los Angeles Times, and Chicago Tribune. We use it as a robustness sample given its substantially smaller volume than the Canadian corpus.} Our data, sourced from the ProQuest Database, is available at daily frequency and in real-time.

Sentiment and confidence have long been measured using qualitative data that draw on surveys asking people whether they think the economy is improving, staying the same, or getting worse. Such sentiment measures have been shown to have predictive power for macroeconomic outcomes, even when controlling for other factors. We consider model-free and model-based approaches to measure the tone or sentiment about the monetary policy sentences.

The model-free methods use dictionaries from gonzalez2021monetary. Model-based estimation of the tone of each document uses BERT (bidirectional encoder representations from transformers. BERT is a deep learning model introduced by devlin2018bert for natural language processing. It focuses on sequences of words rather than simply counting particular words in isolation, as in the economic policy uncertainty index. Pretrained on a corpus of more than 3.3 million words, with over 110 million parameters devlin2018bert, BERT avoids the subjective use of judgment in choosing the dictionary used to define, in our context, sentiment. Importantly, by using surrounding text, rather than simply reading from left to right, BERT aims to establish the context of the text and thus infer the meaning of language that might otherwise be ambiguous. To this day transformer models are considered the state-of-the-art NLP technique, having superseded all other language models.\footnote{Indicatively, GPT models, such as OpenAI’s ChatGPT, LLaMA, PaLM, and Claude are also based on the transformer paradigm. Also, see chu2024history for an overview.}

\oldsubsection{Dictionary and model-based sentiment} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Dictionary and model-based sentiment} We first measure the tone of the monetary-policy news corpus in a model-free manner, following the bag-of-words methodology of tetlock2007giving and loughran2011liability. We count the frequency of words appearing in a tone lexicon calibrated to central-bank vocabulary by gonzalez2021monetary and aggregate over the corpus at time $t$ as $$ \operatorname{Tone}_t^{\mathrm{Dictionary}}=\frac{n_t^{\text{Positive}}-n_t^{\text{Negative}}}{n_t^{\text{Words}}}, $$ where $n_t^{\text{Positive}}$ and $n_t^{\text{Negative}}$ are the counts of positive and negative tonal words at $t$, and $n_t^{\text{Words}}$ is the total word count. A higher value indicates a less negative sentiment in the news.

To reduce the subjective judgment involved in choosing a dictionary, we construct an alternative model-based measure of tone using FinBERT yang2020finbert, a variant of BERT pre-trained on financial texts.\footnote{The FinBERT weights are released by yang2020finbert and were estimated on a corpus of corporate filings, analyst reports and financial news running from $1990$ through $2014$, with the underlying BERT pre-training cut-off in October 2018. Both endpoints predate the bulk of our 1977--2024 sample on the right tail and predate or coincide with it elsewhere, so FinBERT scoring of articles in our corpus does not embed knowledge of post-publication macroeconomic outcomes. The look-ahead concern that we discuss for CBILA in Section (ref) therefore does not apply to FinBERT.} We apply FinBERT to the same corpus at the sentence level: each sentence is classified as positive, negative, or neutral, and we aggregate at time $t$ using the same ratio as in Section (ref) with sentences in place of words, $$ \text{Tone}_t^{\text{FinBERT}}=\frac{n_t^{\text{Positive}}-n_t^{\text{Negative}}}{n_t^{\text{Sentences}}}. $$

We plot the two standardized sentiments in Figure (ref) and note how the BERT-based estimates of tone appear to exhibit greater concordance with C.D. Howe recessions than those using the GT dictionary. As expected, media monetary policy sentiment falls in recessions.

However, Figure (ref) indicates that sentiment did not decline as sharply during the COVID-19 recession or the global financial crisis as it did following the 1980 recession when interest rates were significantly increased to combat high inflation.

\oldsubsection{Generative-AI sentiment: the CBILA framework} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Generative-AI sentiment: the CBILA framework}

The dictionary and FinBERT measures share a structural limitation: their scoring rule is fixed once and for all, either by the lexicon or by the pre-trained transformer weights. They do not exploit context beyond the document being scored. Recent generative large language models can, in principle, do better. They can read a sentence about monetary policy with attention to qualifiers, conditional statements, implicit forward guidance, and credibility cues, and they can return both a classification and an explanation shah2025words,gambacorta2024cb,hansen2024can, ash2023text. We exploit this capability to construct an alternative, semantically richer sentiment measure, which we use primarily to validate the FinBERT-based series across architectures.

We package the procedure into the Central Bank Intelligent Language Agent (CBILA). Each sentence in the corpus is passed through a structured prompt that instructs the model to act as a central-bank communication analyst and to return three outputs: a topic label among Monetary Policy, Inflation, Financial Stability, and similar policy-relevant categories; a temporal-orientation label among forward-looking, backward-looking, or neutral; and a sentiment score in $[-1,1]$ together with a categorical label among positive, negative, neutral, and irrelevant. The prompt also asks for a short rationale (generative models only), which we use only for diagnostic inspection. This reasoning-with-constraints design is intended to improve reproducibility and output stability, two attributes that matter for policy applications hansen2024can, gambacorta2024cb, gorodnichenko2023voice,ayivodji2023housing, korinek2023generative.

We apply the full CBILA pipeline to six instruction-tuned generative LLMs: DeepSeek-R1 Qwen 32B, Qwen 2.5 (7B and 32B), Gemma 27B, GPT-OSS-120B, and Llama 3.3 70B. For each sentence, these models produce all four CBILA outputs (topic label, temporal orientation, sentiment score and label, and rationale). FinBERT and ModernFinBERT are encoder-only sentence-level sentiment classifiers and cannot be run through the CBILA prompt pipeline; we include them in the cross-model comparison of Figures (ref)--(ref) as sentiment-only benchmarks, to assess the robustness of the sentiment dimension across architectures and training corpora. The rationale output is therefore available only for the six generative models and is used purely for diagnostic inspection. For the generative models, we evaluate both few-shot ([fs]) and zero-shot ([zs]) prompting configurations. The constrained structure of the CBILA prompt is designed to improve classification consistency and reduce stochastic variation across runs.

Figures (ref) and (ref) report the pairwise correlations of continuous sentiment scores and the Krippendorff $\alpha$ for nominal classifications, respectively. Three substantive findings stand out. First, modern instruction-tuned models agree strongly with each other across both few-shot and zero-shot configurations, with many pairwise correlations exceeding 0.7 and substantial inter-model agreement for the largest models according to Krippendorff $\alpha$. This overall agreement structure remains broadly stable across historical subperiods and alternative treatments of the irrelevant category. Second, agreement with the dictionary-based index is substantially lower, confirming that lexical methods capture only a subset of the semantic content that LLMs read. Third, this agreement holds across prompting configurations and across models trained in markedly different information environments, the U.S.-developed Gemma, GPT-OSS, and Llama families alongside the China-developed DeepSeek-R1 and Qwen families. This cross-corpus convergence is informative in its own right: cao2026foreign document substantial divergence between U.S. and Chinese LLMs on firm-level financial predictions, and the absence of comparable divergence in our setting indicates that sentence-level annotation of central-bank communication isolates a semantic signal that is robust to LLM training-corpus heterogeneity. The categorical decision pattern is also stable across architectures, with a dominant role for neutral and negative categories that reflects the informational and risk-focused tone of monetary-policy reporting. Additional robustness checks reported in Appendix Figures (ref) and (ref) show that the agreement structure remains qualitatively similar when excluding the irrelevant category from pairwise comparisons and across major monetary-policy subperiods.

These results validate CBILA as a measurement framework. They do, however, raise important concerns about the direct use of CBILA-based measures as regressors. The reason is look-ahead bias, a concern increasingly documented in the recent LLM-forecasting literature fariae2024ai,crane2025total,alam2026chatmacro,eliseev2026fake. The LLMs we use are pre-trained on text corpora that extend up to 2023 or 2024, so when we score an article published in, say, 1990, the model's internal representations may implicitly encode information about post-1990 macroeconomic outcomes. This contamination is structural. It is unambiguously problematic for the pseudo-out-of-sample forecasting exercise of Section (ref), where a CBILA-based predictor would inflate predictive performance through information unavailable to a real-time forecaster. It can also be problematic for causal regressions like the Taylor-rule estimation of Section (ref): the IV strategy with lagged macro instruments does not eliminate the bias if the leakage component of $s_t$ correlates with those instruments through the persistence of the macro process. We document the empirical relevance of this concern for the Taylor rule in Section (ref), and provide the formal analysis in Appendix (ref).

By contrast, the dictionary index is essentially free of look-ahead bias because the lexicon is fixed ex ante. The FinBERT measure is also largely safe, since the underlying model was trained on data ending well before the bulk of our sample. Both are therefore appropriate as the reference measures for the three empirical exercises that follow. In this paper, the CBILA series plays the complementary role of validating the qualitative content of the FinBERT signal across LLM architectures. A real-time-safe version of the CBILA pipeline would require temporally aligned language models, such as ChronoBERT/ChronoLLM in he2025chronologically, DatedGPT in yan2026datedgpt, or related temporal-alignment approaches. These frameworks aim to align the model’s effective information set to a target historical date through explicit knowledge cutoffs, temporally restricted training windows, or time-aware alignment procedures. In such a design, each article would be scored only by a model whose knowledge cutoff predates the article publication date, thereby preserving temporal consistency and substantially mitigating look-ahead contamination. This would make CBILA more suitable as a primary regressor in forecasting, expectations, and Taylor-rule exercises; we leave this extension for future work. An alternative route, used in stevanovic2026whosaw for inflation forecasting, is to design prompts that hold the training-leakage bias common across treatments and identify the variation of interest from cross-treatment differences. A parallel effort by the Bank of Canada uses a fine-tuned LLM to classify the tone of BoC communications as dovish, neutral, or hawkish, and documents that the tone of financial-sector commentary shifts in the direction of the tone the central bank uses wang2026sparks.

\oldsubsection{Temporal Focus and Forward-Looking Orientation} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Temporal Focus and Forward-Looking Orientation}

A second dimension of monetary-policy communication concerns its temporal orientation, whether narratives in the media emphasize past performance, present conditions, or expectations about the future. While the literature on central-bank communication has traditionally focused on the tone or sentiment of policy messages, much less attention has been devoted to their temporal horizon. Yet this aspect is crucial: the degree to which communication looks forward, rather than backward, determines how information about policy intentions is transmitted to markets and the public. Changes in the temporal balance of media narratives may thus signal shifts in how the public interprets policy credibility and the expected future stance of monetary policy.

To quantify this dimension, we employ a hybrid approach combining grammatical analysis and semantic filtering. First, we use spaCy’s linguistic model to identify the dominant verb tense of each sentence: past, present, or future. Second, we refine this classification using a curated dictionary of forward-looking expressions (e.g., will raise, is expected to cut, likely to remain). Third, we apply FinBERT-FLS, a fine-tuned BERT model trained to detect forward-looking statements (FLS) even when the grammatical form is not explicitly future. Each sentence is then assigned to one of four mutually exclusive categories (Past, Present, Future, or Other), thereby providing a detailed map of how policy discussions evolve over time (see ayivodji2023housing for methodological details).

Two patterns in the temporal decomposition deserve mention. First, present- and past-tense reporting dominate the coverage, but the forward-looking share is non-trivial and rises during regime transitions such as the adoption of inflation targeting in the early 1990s and the introduction of unconventional policies after 2008 (Appendix Figure (ref)). Second, the forward-looking share tracks the tenure of successive Bank of Canada governors who expanded communication and forward guidance, and falls during periods dominated by short-term operational adjustments (Appendix Figure (ref)). The temporal horizon of media coverage therefore responds to both macroeconomic regime shifts and institutional communication style.

Figure (ref) summarizes the correlations between the four temporal dimensions of sentiment (Full Sentiment, Past Focus, Present Focus, and Future Focus) over the full sample (1980–2024) and key subperiods. Adding this temporal layer reveals that correlations vary across policy regimes. Notably, the link between Full Sentiment and Future Focus weakens during high-uncertainty phases, suggesting that forward-looking narratives contain additional information not captured by aggregate tone. This distinction highlights that media optimism about current policy performance and confidence in future actions are conceptually distinct, and only the latter carries predictive power for the evolution of macroeconomic outcomes and policy rates.

Taken together, Figures (ref)–(ref) show that the temporal structure of monetary-policy communication is far from static. The forward-looking component fluctuates systematically with macroeconomic regimes and institutional practices, serving as a proxy for the public’s attention to the future stance of policy. By isolating this component, we capture a unique aspect of the monetary-policy transmission mechanism: how expectations are formed and reinforced through the media. The next section explores whether this forward-looking sentiment indeed predicts macroeconomic variables and the estimated parameters of a Taylor-type policy rule, thereby testing its informational and expectational content.

\oldsection{Empirical content of media sentiment} \addcontentsline{apptoc}{section}{\numberline{\Alph{section}}Empirical content of media sentiment}

The model of Section (ref) delivers a clean prediction about media sentiment and a more open empirical question. The clean prediction follows from the sentiment law of motion in equation ((ref)): movements in $s_t$ should shift the expectations of households and firms beyond what their own persistence already explains. We test this prediction directly in Section (ref) using the Bank of Canada household expectations survey. The more open question is whether the same media indicators carry information for realized macroeconomic outcomes. The model itself does not pin down the predictive content of $s_t$ for $\pi$ or $y$ over and above the conventional macro information set, but a coherent reading of the framework suggests that they should: if media coverage is informative enough for the central bank to read it (Section (ref)) and for the public to react to it, it should leave a footprint on subsequent realized outcomes. Section (ref) treats this as a separate empirical check, in line with the broader text-as-data forecasting literature.

\oldsubsection{Media sentiment and household expectations} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Media sentiment and household expectations}

The model of Section (ref) treats media narratives as the channel through which the central bank's communication and the state of fundamentals reach private agents. A direct empirical implication is that the sentiment indicators built in Section (ref) should help explain short- to medium-horizon expectations of macroeconomic variables, beyond what is already captured by the persistence of these expectations themselves. We test this implication on Canadian household survey data.

We use the Bank of Canada Canadian Survey of Consumer Expectations (CSCE), which provides quarterly measures of household expectations of inflation at multiple horizons, expected wage growth over the next 12 months, and perceptions of past wage growth. The estimation sample runs from 2014Q4 to 2023Q3, that is, 36 quarterly observations. For each expectation series $E_t z_{t,t+h}$ we estimate:

where $z \in \{\pi, W\}$ denotes inflation or wage growth, $h$ is the horizon in quarters, and $s_t$ is the FinBERT aggregate sentiment about Canadian monetary policy from Section (ref). The lag of the expectation absorbs the persistence and the slow-moving common trend in survey beliefs, so $\beta$ measures the additional information that current media sentiment carries about the path of household expectations. The CSCE backward-looking wage variable does not fit ((ref)) as a forward expectation; we report the corresponding regression with a one-period-lagged perception on the right-hand side as a perception equation, for comparability with the forward-looking columns.

}

\end{table} \fi

Table (ref) reports the estimates. The sentiment coefficient $\beta$ is positive at every horizon and significant at conventional levels for inflation expectations at the contemporaneous, one-year, and two-year horizons, for the forward 12-month expected wage growth and for the backward 12-month perception of wage growth. The estimate at the five-year horizon for inflation is similar in magnitude to the one-year estimate but loses significance, consistent with the view that media coverage maps to short- and medium-run macroeconomic outlooks rather than to long-run anchored expectations. After controlling for the persistence of beliefs, a one-standard-deviation rise in $s_t$ shifts measured expectations by roughly 0.13 to 0.27 percentage points.

These results corroborate the role assigned to $s_t$ in equation ((ref)). The systematic link between sentiment and survey expectations supports the interpretation of media narratives as an active component of the expectations-formation process. This complements the experimental evidence in andre2026narratives, who show that exogenously varying the narrative through which US households interpret inflation causally shifts their expectations of future inflation. The next subsection asks the complementary question of whether sentiment also helps predict realized macroeconomic outcomes, while Section (ref) examines whether the central bank's own behavior is consistent with this informational role.

\oldsubsection{Media sentiment as a predictor of realized output and inflation} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Media sentiment as a predictor of realized output and inflation}

Beyond the direct test of the model in Section (ref), we ask a complementary question: do the media indicators carry information for realized output and inflation, beyond what conventional macro predictors already deliver? The model of Section (ref) does not formally answer this question, but a positive finding would corroborate the broader informational content of media sentiment, which is a precondition for the central-bank behavior we estimate in Section (ref). We follow the text-as-data forecasting literature bybee2024business,ash2023text,ayivodji2023housing,ellingsen2022news,andre2026narratives and run pseudo-out-of-sample MIDAS forecasts of CPI inflation and real GDP growth.

We construct $244$ text-based predictors from the sentiment indicators of Section (ref), a topic decomposition of the corpus (twenty LDA topics), a media-based monetary policy uncertainty index, and tense-conditioned crosses of these dimensions. MIDAS regressions at horizons $h \in \{1, 2, 4, 6, 8\}$ quarters are estimated on the 1982Q1--2023Q3 sample (training 1982Q1--2002Q4, test 2003Q1--2023Q3) using the machine-learning competing models of goulet2022machine: penalised regressions (LASSO, Ridge, Elastic Net), tree-based methods (Random Forest, XGBoost), neural networks with one and three hidden layers, and three ensembles that aggregate them. Construction of the uncertainty index, the LDA topic decomposition, the full forecasting setup, the model-by-horizon RMSE tables, and the SHAP variable-importance analysis are reported in Appendix (ref).

Figure (ref) reports the relative RMSE of the all-models ensemble for four predictor sets, separately for GDP growth and CPI inflation. A value below one indicates an improvement over the random-walk benchmark.

Text-based predictors outperform the random-walk benchmark at most horizons. Across model families and predictor sets, relative RMSE for CPI inflation lies between $0.61$ and $0.96$ at horizons $h \in \{1,2,6,8\}$, with most cells significant under the Diebold-Mariano test, and is essentially flat or slightly above the benchmark at $h = 4$ (relative RMSE between $0.96$ and $1.06$, none significant); for real GDP growth the relative RMSE lies between $0.64$ and $0.77$ across all horizons (Appendix Tables (ref) and (ref)). Gains are largest when feature sets combine topics with tense and either sentiment (3T) or uncertainty (2TU); adding tense alone to topics yields little improvement. For CPI inflation the largest gains concentrate at the two- and six-quarter horizons, while for GDP growth the gains are more uniform across horizons. The SHAP variable-importance analysis in Appendix (ref) indicates that uncertainty-conditioned topics and their temporal variants (past, present, future) consistently appear among the top predictors, regardless of model family.

\oldsection{The role for monetary policy: text-augmented Taylor rule} \addcontentsline{apptoc}{section}{\numberline{\Alph{section}}The role for monetary policy: text-augmented Taylor rule}

The behavioral New-Keynesian model of Section (ref) suggests that perceptions of the future stance of monetary policy, embedded in media coverage, can move expectations and economic outcomes through the cognitive-discounting channel. A natural empirical question is whether the central bank itself appears to internalize these perceptions when setting the policy rate. We address this question by estimating forward-looking monetary policy reaction functions for the Bank of Canada, augmented with the media-based sentiment indicators constructed in Section (ref).

\oldsubsection{Specification} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Specification}

Let $i_t$ denote the BoC overnight policy rate, $E_t \pi_{t+h_\pi}$ the BoC staff inflation forecast at horizon $h_\pi$, from the BoC Staff Economic Projections (SEP) database, $\pi^*_{t+h_\pi}$ the contemporaneous inflation target, and $E_t x_{t+h_x}$ the staff output-gap forecast at horizon $h_x$. Following pang2024boc, we set $h_\pi=4$ and $h_x=2$, and we allow the long-run real neutral rate to be time-varying through the proxy $r^*_t$, defined as the 10-year government bond yield net of the inflation nowcast orphanides2001monetary. The augmented inertial Taylor rule is

where $\rho \in (0,1)$ captures interest-rate smoothing, $\alpha$ is the loading on the time-varying neutral real rate, $c$ aggregates the time-invariant component of the neutral rate together with the constant of the rule, and $s_t$ is one of the standardized monetary-policy sentiment indicators built in Section (ref). We consider eight sentiment measures: (i) the dictionary-based aggregate sentiment (Dict); (ii) the FinBERT aggregate sentiment (BERT); (iii) the dictionary sentiment restricted to forward-looking sentences (Fut-Dict); the FinBERT sentiment by temporal orientation, broken down into (iv) past (Past), (v) present (Pres), and (vi) forward-looking (Fut-BERT); and the ModernFinBERT counterparts (vii) aggregate (\textit{M-BERT}) and (viii) forward-looking (\textit{Fut M-BERT}). ModernFinBERT is encoder-only with a pre-training cut-off that predates the bulk of the sample, so it carries no look-ahead bias by construction; the CBILA generative-LLM tones do, which is why we keep them separate in Appendix (ref). Specification ((ref)) without sentiment ($\delta = 0$) is reported as the \textit{Baseline} column.

As in pang2024boc, the estimation sample runs from 1991Q1, when the BoC adopted the inflation-targeting framework, to 2019Q4, the last quarter available at the SEP database, but also unaffected by the COVID-19 break in real-time data and forecasts. This delivers $T=110$ quarterly observations. All sentiment indicators are standardized prior to estimation, so that $\delta$ measures the basis-point response of the policy rate to a one standard-deviation movement in $s_t$.

\oldsubsection{Estimation strategy} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Estimation strategy}

pang2024boc, building on orphanides2001monetary, argue that OLS yields consistent estimates of forward-looking Taylor rules when the right-hand-side variables are constructed from real-time staff projections. By construction, those projections are produced under the assumption that the current policy rate stays unchanged, and are therefore orthogonal to the contemporaneous policy shock. The lagged interest rate is also predetermined, and the time-varying neutral rate is built from the 10-year bond yield, which is exogenous to current monetary policy decisions at quarterly frequency.

The same logic does not extend to media sentiment $s_t$. The model of Section (ref) makes clear that $s_t$ and $r_t$ are jointly determined at date $t$: through equation ((ref)), $s_t$ depends on the contemporaneous output gap and inflation, which in turn react to the contemporaneous policy decision; conversely, through equation ((ref)), $r_t$ depends on $s_t$. The simultaneity contaminates $\hat\delta_{\text{OLS}}$ in directions that depend on which monetary policy shock is at work and on which channel of $s_t$ dominates. An unanticipated tightening $\varepsilon^u_t > 0$ raises $r_t$ on impact (directly through the policy rule) and lowers $s_t$ through the fundamentals block, since output and inflation contract endogenously; the resulting negative covariance between $s_t$ and the policy innovation biases $\hat\delta_{\text{OLS}}$ downward. An anticipated tightening $\varepsilon^a_t > 0$, by contrast, does not enter the policy rule at $t$ directly; through the lean against the wind derived in Section (ref), $r_t$ falls slightly via $\phi_s s_t < 0$, and $s_t$ also falls because the fundamentals block dominates the communication block. The two co-moves generate a positive correlation that biases $\hat\delta_{\text{OLS}}$ upward, but by a small amount given the small magnitude of the contemporaneous lean. The net direction of the OLS bias therefore depends on the relative magnitude of the two channels and on the mix of MP shocks in the sample.

To address this simultaneity, we re-estimate ((ref)) by efficient two-step GMM in the spirit of clarida1998monetary,clarida2000monetary. The lagged interest rate, the staff forecasts and the time-varying neutral rate are taken as predetermined, while $s_t$ is treated as endogenous. Identification of $\delta$ requires instruments that are correlated with $s_t$ but uncorrelated with the contemporaneous policy innovation; we do not require the instruments to project $s_t$ onto its narrative-innovation block, only to be predetermined with respect to the date-$t$ shock. Our base instrument set follows the standard CGG configuration in the Canadian Taylor-rule literature: lags 2 to 5 of the policy rate together with lags 1 to 4 of the staff inflation gap and the staff output gap, giving 12 excluded instruments and 11 overidentifying restrictions. Appendix (ref) additionally reports estimates under a recent-lag set (lags 2 to 3 of the policy rate, lags 1 to 2 of the forecasts) and a distant-lag set (lags 4 to 5 of the policy rate, lags 3 to 4 of the forecasts); for the eight sentiment series of Section (ref) both alternatives deliver $\hat\delta$ quantitatively similar to the base set, but the picture differs for the CBILA generative-LLM tones discussed there. Instrument relevance is assessed with the first-stage $F$ on $s_t$ against the $F > 10$ benchmark of staiger1997instrumental. Instrument exogeneity is assessed with Hansen's $J$ test of overidentifying restrictions, computed using a block wild Rademacher bootstrap ($B=500$ resamples, non-overlapping block length $L=4$ matching the HAC bandwidth) rather than the asymptotic $\chi^{2}$ approximation; we make this choice because the bootstrap test of the GMM overidentifying restrictions admits a higher-order asymptotic refinement over the $\chi^{2}$ critical values hallhorowitz1996, and the wild scheme is robust to heteroskedasticity of unknown form in IV regressions davidsonmackinnon2010.

A more substantive alternative replaces the lagged levels of the policy rate and forecasts with one-step revisions of the staff projections themselves: $E_{t-1}(\pi_{t+4}) - E_{t-2}(\pi_{t+4})$ for inflation, $E_{t-1}(x_{t+2}) - E_{t-2}(x_{t+2})$ for the output gap, and lag 2 of the policy rate. This forecast-revision set isolates the new information that the BoC staff incorporated between $t-2$ and $t-1$. It differs from the lagged-level instruments in two ways. First, revisions are a short-window flow rather than a stock of past information, so they capture the recent updates of the staff's view that are most likely to drive sentiment innovations. Second, because they are differences of two pre-period forecasts, they purge the slow-moving common trends in inflation and activity that link the lagged levels of $i_{t-2},\ldots,i_{t-5}$ to the structural shocks at $t$. This gives a much smaller instrument set (3 instruments, 2 overidentifying restrictions), which sharpens the orthogonality argument at the cost of statistical power.

\oldsubsection{Results} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Results}

Table (ref) reports the OLS estimates (Panel A) and the GMM estimates with the base instrument set (Panel B) over 1991Q1 to 2019Q4, with Newey-West HAC standard errors ($L=4$). The reference specification without sentiment (Column 1) reproduces a textbook inertial Taylor rule for the BoC: high smoothing ($\rho=0.89$), a positive loading on the time-varying neutral rate ($\alpha=0.12$), a significant response to the inflation gap ($\gamma=0.28$) and to the output gap ($\beta=0.11$). These magnitudes line up with the post-1991 estimates in pang2024boc.

Adding sentiment yields a clear pattern. Under OLS, only the dictionary-based aggregate (Dict) and the dictionary forward-looking measure (Fut-Dict) enter significantly. The FinBERT measures and their ModernFinBERT counterparts are individually small under OLS and indistinguishable from zero. Once sentiment is treated as endogenous and instrumented (Panel B), all eight sentiment indicators load positively on the policy rate with point estimates between $0.27$ and $0.46$ and significance at the 1% level. First-stage $F$-statistics are large in every column, well above the Staiger--Stock benchmark, and the block wild bootstrap Hansen $J$ does not reject overidentification in any column.

Once $s_t$ is included and instrumented, the loading on the inflation gap remains positive and significant in most columns, while the output-gap coefficient becomes uniformly insignificant, echoing pang2024boc's point that the BoC's reaction function is dominated by inflation once forward-looking information is properly accounted for. The smoothing coefficient $\rho$ stays close to its baseline value, so sentiment captures a distinct dimension of the rule rather than an alternative source of inertia. The ModernFinBERT columns (M-BERT and Fut M-BERT) deliver $\hat\delta$ in the same range as FinBERT, with similar significance levels and $J$ $p$-values. This is the expected outcome of moving from FinBERT to a more recent encoder-only architecture that still predates the bulk of the sample.

We then re-estimate the GMM specification with the forecast-revision instruments. Table (ref) delivers the same qualitative message as Panel B of Table (ref): all eight sentiment indicators load positively and significantly. Point estimates are uniformly larger under the forecast-revision instruments and standard errors are wider, reflecting the smaller instrument set. First-stage $F$-statistics remain large, and the block wild bootstrap $J$ does not reject in any column. That two conceptually different instrument sets yield consistent estimates supports interpreting the loading on $s_t$ as a genuine response of the BoC to media sentiment, not an artefact of any particular instrument choice.

These results show that media sentiment about Canadian monetary policy enters the BoC's estimated reaction function with a positive and economically meaningful loading, providing a partial-equilibrium counterpart to the behavioral New-Keynesian model of Section (ref): the central bank's behavior is correlated with the public's perception of the policy stance, beyond what is encoded in standard inflation and activity forecasts, and is consistent with the cognitive-discounting channel through which media sentiment attenuates or amplifies forward guidance. The gap between OLS and GMM estimates, particularly for the FinBERT- and ModernFinBERT-based measures, indicates that simultaneity between sentiment and policy decisions is a first-order concern; treating sentiment as predetermined understates its role in the policy rule.

We also estimated the same specifications with the CBILA generative-LLM tones described in Section (ref). These series carry potential look-ahead bias because the underlying models were pre-trained on text that extends into and beyond our 1991--2019 sample. Appendix (ref) documents that the CBILA-based $\hat\delta$ estimates are smaller and less stable across instrument sets than the FinBERT- and ModernFinBERT-based estimates, that the close-vs-distant pattern in the IV estimate diverges sharply for the CBILA series while remaining flat for the no-leakage series, and that the block wild bootstrap Hansen $J$ test rejects overidentification for several CBILA series at conventional levels. We provide a formal account of the underlying IV identification failure in the same appendix. The findings in this section therefore rest on the no-leakage series; the CBILA evidence supports the construction of an alternative central-bank communication index but is not currently suitable as a primary regressor in causal regressions of this type.

\oldsection{The Role of Media Sentiments for Monetary Policy Transmission} \addcontentsline{apptoc}{section}{\numberline{\Alph{section}}The Role of Media Sentiments for Monetary Policy Transmission}

In this section we use the model of Section (ref) to derive the sign restrictions that identify three structural shocks in a Bayesian SVAR: an unanticipated monetary policy shock $\varepsilon^u$, an anticipated monetary policy shock $\varepsilon^a$, and a narrative shock $\varepsilon^s$.

\oldsubsection{Sign restrictions implied by the behavioral NK model} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Sign restrictions implied by the behavioral NK model}

We trace the impact response of $(r_t, s_t, x_t, \pi_t)$ and the $h$-period-ahead expectations, $E_t[r_{t+h}]$, $E_t[x_{t+h}]$, and $E_t[\pi_{t+h}]$, to each of the three shocks under the expansionary normalisation: unanticipated easing $\varepsilon^u_0 < 0$, anticipated easing $\varepsilon^a_0 < 0$, and positive narrative $\varepsilon^s_0 > 0$. To keep the algebra transparent we work with the simplified version of the system ((ref))--((ref)) in which $\rho_s = 0$, $\tau = 1$, and the loadings of the sentiment law of motion ((ref)) are normalised to $\lambda_x = \lambda_\pi = \lambda_a = 1$, with all endogenous variables at their steady state at $t = -1$. We retain the Taylor smoothing parameter $\rho_r$, which generates the empirically relevant persistence in the policy rate. Under these simplifications the policy rule and sentiment law of motion at $t = 0$ become

where the $\rho_r r_{-1}$ and $\varepsilon^a_{-1}$ terms have been eliminated by the steady-state assumption at $t = -1$. The IS and Phillips equations ((ref))--((ref)) are unchanged.

For an unanticipated easing, $\varepsilon^u_0 < 0$ enters the Taylor rule additively and pushes the policy rate down on impact. Forward-looking households anticipate the lower rate path (smoothed by $\rho_r$) and bid up current output through the cognitively discounted IS curve; firms raise current inflation through the Phillips curve. With $x_0, \pi_0 > 0$, the LoM ((ref)) gives $s_0 = x_0 + \pi_0 > 0$. The Taylor-rule bracket $\phi_\pi \pi_0 + \phi_x x_0 + \phi_s s_0$ is therefore positive and partially leans against the direct cut, but $\varepsilon^u_0$ dominates and $r_0 < 0$. Taylor smoothing then transmits the easing forward: the rate follows an AR(1) with persistence $\alpha \in (0, 1)$ (determined by an endogenous fixed point that adjusts $\rho_r$ for the bracket's feedback), and the forward expectations of $r$, $x$ and $\pi$ inherit the impact-period signs and decay geometrically at rate $\alpha$. The full impact pattern is $r_0 < 0$, $x_0 > 0$, $\pi_0 > 0$, $s_0 > 0$, with $E_0[r_h] < 0$, $E_0[x_h] > 0$, $E_0[\pi_h] > 0$ for all $h \ge 0$. The detailed derivation is in Appendix (ref).

For an anticipated easing, $\varepsilon^a_0 < 0$ enters the LoM ((ref)) at $t = 0$ but the policy rule ((ref)) only at $t = 1$, since it lags by $\tau = 1$. The system has a recursive structure. From $t \ge 2$ on, no further shocks hit and the system reverts to the autonomous stable mode of the unanticipated case, $r_t = \alpha r_{t-1}$ and $z_t = G r_{t-1}$. At $t = 1$, the announcement enters the Taylor rule additively and acts structurally like a one-period rate innovation: $r_1 = \alpha r_0 + \beta_g \varepsilon^a_0$ with the same $\beta_g$ as in the unanticipated case, and $z_1 = G r_0 + g \varepsilon^a_0$ with the same impact loadings $g$. At $t = 0$, the contemporaneous IS, Phillips, Taylor and LoM equations are solved jointly with these one-step-ahead expectations as forward inputs, and with the $\phi_s \varepsilon^a_0$ contribution to the rate that the LoM passes through. Two implications follow. First, the rate leans against the wind on impact, $r_0 > 0$, because forward-looking households and firms bid up current $x_0, \pi_0$ in anticipation of the future cut and the central bank reads the implied rise in sentiment as an incipient overheating signal. Second, the sign of $s_0$ depends on whether the fundamentals block of the LoM dominates the direct loading of the announcement,

which we adopt as an identifying assumption and verify numerically (Appendix (ref)). Under this assumption, $s_0 > 0$. The inequality is not implied by standard NK calibrations: with $M^h \to 0$ the cognitive-discount damping shrinks $|x_0|, |\pi_0|$ to zero while the direct effect $|\varepsilon^a_0|$ is unchanged, so it fails and $s_0$ flips sign. Under the alternative case $\lambda_a < 0$ of Section (ref), $s_0 > 0$ holds unconditionally. The full impact pattern is $r_0 > 0,\; x_0 > 0,\; \pi_0 > 0,\; s_0 > 0$, with $E_0[r_h] < 0,\; E_0[x_h] > 0,\; E_0[\pi_h] > 0$ for $h \ge 1$. The contemporaneous reduced-form policy rule

makes the lean explicit: the first two terms aggregate the fundamentals-driven response (boosted by the sentiment loading $\phi_s$), and the third is the small direct sentiment-pass-through of $\varepsilon^a_0$. In the standard rational-expectations NK model the lean exists only through the small endogenous moves in $\pi_0, x_0$; here it acquires an additional driver, $\phi_s s_0 > 0$, which follows from $\phi_s > 0$ in the policy rule ((ref)) and is supported by the GMM estimates of Section (ref). The detailed derivation is in Appendix (ref).

For a positive narrative shock, $\varepsilon^s_0 > 0$ enters only the LoM ((ref)) at $t = 0$ and is absent from the IS, Phillips and Taylor equations at every date. Substituting ((ref)) into the Taylor rule ((ref)) with $\varepsilon^u_0 = \varepsilon^a_0 = \varepsilon^a_{-1} = 0$ gives

with $\widetilde\phi_\pi, \widetilde\phi_x$ as in ((ref)). The narrative innovation enters the rate additively through $\phi_s\, \varepsilon^s_0$, and the lean against the wind operates entirely through this sentiment channel since the standard Taylor inputs $\pi_0, x_0$ are themselves driven endogenously by $r_0$ in the contemporaneous block. Under $\phi_s > 0$ the central bank reads the autonomous rise in sentiment as an incipient overheating signal and tightens: $r_0 > 0$. The IS--NKPC block then contracts current fundamentals, $x_0 < 0$ and $\pi_0 < 0$. Substituting back into ((ref)) gives $s_0 > 0$ because the direct innovation $\varepsilon^s_0$ dominates the equilibrium contraction $|x_0 + \pi_0|$; under the calibration of Table (ref) the contraction is on the order of one-sixth of $\varepsilon^s_0$. The forward expectations follow the same geometric decay as the unanticipated case but with the opposite sign on $r_0$. The full impact pattern is $r_0 > 0,\; x_0 < 0,\; \pi_0 < 0,\; s_0 > 0$, with $E_0[r_h] > 0,\; E_0[x_h] < 0,\; E_0[\pi_h] < 0$ for $h \ge 1$. The detailed derivation is in Appendix (ref).

Table (ref) shows the sign restrictions used to identify shocks in the SVAR. We retain only the restrictions on $r_t, s_t$ and on the three forward expectations; the contemporaneous responses of $x_t$ and $\pi_t$ are left unrestricted in the SVAR because they are model predictions of small magnitude (cognitive discounting damps them by $(M^h M^f)^\tau$ at the $\tau = 4$ horizon used in the empirics) and imposing their sign would discard rotations that the data would otherwise accept.

Anticipated and unanticipated MP shocks generate qualitatively similar paths for the realised macro variables, and the sign restrictions of Table (ref) differ between the two only on the contemporaneous policy rate. Following damico2023, we condition the rotation on the joint behaviour of survey expectations and realised macro variables. For each shock $j$ and each candidate rotation $B^d$, the SVAR delivers two objects from the same draw: the impact response of the staff expectation, $B^d_{E,j}$, and the four-period-ahead IRF of the corresponding realised variable, $\widetilde B^d_{4,j}$. Under rational expectations the two coincide, and the shock-by-shock importance weight

penalises rotations under which they diverge. The hyperparameter $\delta$ controls the tightness of the penalty: $\delta \to 0$ approaches an indicator on exact equality (strict rationality), $\delta \to \infty$ removes it altogether. The weight is most informative for the $\varepsilon^u/\varepsilon^a$ pair: under an unanticipated easing $E[r]$ tracks the immediate fall in the realised rate, while under an anticipated easing the rate rises on impact through the lean but $E[r]$ is already low because the announced cut is priced in; the weight rewards rotations consistent with this divergence. The narrative shock is identifiable on the sign restrictions alone, since its forward-expectation block ($+,-,-$) differs from each MP shock on three sign moments.

Two substantive differences distinguish our identification from theirs. First, the contemporaneous lean against the wind on the policy rate has an additional driver in our model: in their setup the lean operates only through the standard Taylor channel $\phi_\pi \pi_0 + \phi_x x_0$, while we add the sentiment loading $\phi_s s_0$. Second, the friction that damps the forecast response to anticipated shocks differs: a signal-to-noise filter on the announcement in their model (a credibility friction), cognitive discounting on the perceived path in ours (a salience friction). Appendix (ref) reports the technical comparison.

\oldsubsection{Empirical implementation} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Empirical implementation}

We estimate a quarterly Bayesian VAR(2) on Canadian data over 1986Q4--2019Q4.\footnote{The start date is fixed by the data: 1986Q4 is the first vintage of the BoC staff projections database champagne2020bocsep for which the four-quarter-ahead horizon ($h=4$) is available, with target year 1987Q4. The publicly described start of the database is 1987Q1, which corresponds to the first forward target ($h=1$) of the 1986Q4 vintage; the vintage itself contains a 1986Q4 nowcast and forecasts for 1987Q1--Q4. With $p=2$ lags in the BVAR, the effective sample for the identified shocks runs from 1987Q2 to 2019Q4. The sample starts five years before the inflation-targeting regime adopted in 1991Q1, which is the start date of the Taylor-rule estimation in Section (ref). The longer SVAR window exploits the additional identifying variation available before 1991Q1: it contains Black Monday, the bond-market crisis and the inflation-targeting transition itself.} The vector of endogenous variables is

where the first three entries are the BoC staff projections at horizon $h=4$ (policy rate, log real GDP, log CPI), constructed from the BoC staff projections; $r_t$ is the BoC overnight rate; $\log Y_t$, $\log P_t$, $\log H_t$ are real GDP, headline CPI (seasonally adjusted via STL, cleveland1990stl) and total hours worked from the large macroeconomic database of fortin2022large; $i^{1\text{-}3y}_t$ is the average yield on Government of Canada bonds with 1--3 years to maturity; and $s_t$ is the standardised FinBERT aggregate sentiment of Section (ref), identical to the regressor in column 3 of Table (ref).\footnote{All log variables are multiplied by $100$ so that impulse responses read in percent, the policy rate and the bond yield are expressed in percentage points, and the staff projections of real GDP and CPI are in percentage-point deviations from their respective trends ($100 \times$ four-quarter log changes, level form). The sentiment series is standardised to mean zero and unit variance over the estimation sample.}

We consider the BoC staff projections rather than from private-sector survey forecasts (e.g., Consensus Economics Canada or BlueChip Economic Indicators) for three reasons. First, the staff projections are the information set the BoC actually consults when setting the policy rate, and so they are the natural empirical counterpart of the expectation block of the structural Taylor rule ((ref)). Second, the staff projections cover the full SVAR sample 1986Q4--2019Q4 at quarterly frequency with a consistent methodology, while Canadian private-sector survey forecasts typically start later and report annual averages rather than quarterly horizons, requiring an interpolation that introduces measurement noise into the survey-vs-VAR comparison underlying the loose-rationality weight. Third, staff projections sit upstream of the BoC's communication: they feed into the policy decision and into the Monetary Policy Report rather than reacting to them. Private-sector forecasts, by contrast, are produced after parsing BoC communication, press coverage and competing forecasts; they are downstream of both the communication channel of $s_t$'s law of motion and of the media-sentiment series itself. Using staff projections in the loose-rationality weight therefore avoids conditioning the identification of $\varepsilon^a$ and $\varepsilon^s$ on an external measure that is itself partly endogenous to those shocks.

We estimate the VAR under a Normal-Inverse-Wishart conjugate prior with Zellner-style shrinkage of the autoregressive coefficient matrix toward the OLS estimate, and sample $D_{\text{post}} = 500$ posterior draws of $(\Phi, \Sigma)$. Per posterior draw we generate $N_{\text{rot}} = 8000$ random orthogonal rotations using the QR algorithm of rubio2010structural, and retain those for which one rotation column matches each of the three sign patterns of Table (ref), weighted by the loose-rationality importance weight ((ref)) introduced in Section (ref). We set the penalty hyperparameter at $\delta = 0.5$, the central calibration of damico2023.\footnote{The three weights $(w_{\varepsilon^a}, w_{\varepsilon^u}, w_{\varepsilon^s})$ enter the posterior moments shock-by-shock. The procedure yields $3{,}179$ accepted rotations, with effective sample sizes near $3{,}177$ for each of the three identified shocks. Algorithmic details---prior covariance, exact sign-test, loose-rationality weight formula including the convention used to construct $\widetilde B^d_{4,j}$ for the policy-rate expectation, full diagnostics---are reported in Appendix (ref).}

\oldsubsection{Impulse responses and the narrative shock} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Impulse responses and the narrative shock}

Figures (ref)--(ref) report the impulse responses of the nine variables to one-standard-deviation realisations of the three identified shocks. In each figure the first row plots the average of $B^d_{E,j}$ and $\widetilde B^d_{4,j}$ defined in ((ref)); the second row plots $B^d_{E,j}$ alone, that is the impulse responses of the staff projections $E^S_t[r_{t+4}], E^S_t[\log Y_{t+4}], E^S_t[\log P_{t+4}]$. The third row contains the realised macro variables $r_t, \log Y_t, \log P_t$, and the fourth row contains $\log H_t$, the 1--3y bond yield, and the sentiment index $s_t$. Solid lines are weighted posterior medians; shaded bands are 68% credible bands across rotations.

The anticipated easing shock raises the current rate by approximately 6 basis points on impact, then lowers it to a trough 6 basis points below baseline around $h = 4$ when the announced cut materialises: the lean against the wind derived in ((ref)). Log GDP responds with a hump that builds slowly, reaches 0.20% above baseline by $h \approx 8$ and peaks at 0.26% around $h \approx 16$, while log CPI rises by 0.12% on impact and decays slowly. Agents lower their expected rate and raise their expected output and inflation paths. The sentiment variable rises with the announcement and reverts toward zero after the implementation.

The unanticipated easing shock cuts the current rate directly by 21 basis points on impact, gradually reverting to zero, with macro responses roughly 1.5 to 1.7 times larger than under the anticipated shock at the GDP peak and at the rate trough. The sentiment variable rises through the fundamentals channel of ((ref)) once $x_t$ and $\pi_t$ have responded. Together the two MP impulse responses reproduce the canonical pattern of damico2023: anticipated and unanticipated shocks are economically distinct, with cognitive discounting dampening the response of the current rate to anticipated announcements while leaving the medium-term macro response broadly comparable.

A positive media-sentiment innovation $\varepsilon^s_t > 0$ raises $s_t$ by $0.15$ standard deviations on impact. The Bank of Canada tightens the current rate by $17$ basis points, log GDP falls by $0.12\%$, log CPI by $0.11\%$, and log hours by $0.04\%$. Expected GDP falls by $0.16$ percentage points, expected CPI by $0.11$ percentage points, and the expected policy rate and the $1$--$3$ year government yield both rise by $0.12$ percentage points. The implied impact ratio $r_0/s_0 \approx 1.1$ percentage points per standard deviation of sentiment is roughly three times larger than the GMM estimate $\hat\delta$ of Section (ref), which is expected: the SVAR contemporaneous response includes the indirect feedbacks through the simultaneous response of all other variables in $Y_t$, whereas $\hat\delta$ in ((ref)) is a partial regression coefficient that holds the other right-hand-side variables fixed.

As assumed in Section (ref), the narrative shock is interpreted as an autonomous movement in media sentiment unrelated to either macroeconomic fundamentals or monetary-policy announcements. The SVAR results confirm this prior interpretation: a positive narrative innovation induces a contemporaneous tightening through the policy rule, and generates contractionary responses of output and inflation similar to those of a standard monetary-policy tightening, while remaining separately identified from both anticipated and unanticipated monetary-policy shocks through the opposite sign response of $s_t$. The narrative shock is therefore consistent with the idea that autonomous changes in media coverage can affect the economy indirectly through the central bank's reaction function, rather than through changes in the underlying fundamentals themselves. This mechanism is related to the Fed information effect literature romer2000federal, nakamura2018high, except that the narrative shock does not reveal information about fundamentals; instead, it shifts the policy response conditional on unchanged fundamentals. bauer2023alternative argue empirically that much of the apparent Fed information effect disappears once systematic policy responses are controlled for; our narrative shock can be interpreted as one possible source of the residual co-movement between media narratives and realised policy decisions.

Figure (ref) reports the median posterior time series of the three identified shocks together with 68% credible bands. The anticipated MP series shows positive spikes around 1992 and 2002--2003, periods in which staff projections embedded a meaningful tightening path (post-inflation-targeting adoption and post-deflation-scare reflation), and negative spikes in 1998Q1 and 2000Q2 that align with the forward-looking dovish turn around the Asian crisis and the pre-dot-com easing cycle. The 2008Q4 positive spike at the effective lower bound replicates the damico2023 finding that, in a horizon-four forecast, the announced future policy was less accommodative than agents had been pricing in. The unanticipated MP series displays the volatility cluster of 1987--1995 (Black Monday surprise easing, Mexican peso surprise tightenings, bond-market crisis), the negative spike in 2008Q2 (BoC's pre-emptive cuts in the early phase of the GFC), and the negative episode of 2011Q2 (BoC's hold during the European sovereign debt crisis). The narrative series has its largest negative episode in 1987Q4 (Black Monday narrative pessimism), positive spikes in 1998Q2 (post-Asian-crisis recovery narrative), 2006Q3 (pre-crisis optimism), 2009Q1 (post-trough recovery narrative) and 2017Q3 (narrative around the BoC's resumption of hikes), and a negative spike in 2019Q2 that captures end-of-cycle pessimism. The largest narrative episodes do not coincide with the largest MP episodes, which is the visual counterpart of the orthogonality result reported in Section (ref).

\paragraph{External validation.} Two concerns require attention. The first is the relation of $\varepsilon^a$ and $\varepsilon^u$ to the standard Canadian monetary surprise of champagne2018identifying, hereafter CS, which we expect to preserve the anticipated--unanticipated distinction. The second is the orthogonality of the narrative shock $\varepsilon^s$ to standard non-monetary sources of macroeconomic fluctuation: aggregate uncertainty, news-based policy uncertainty, financial-market volatility, and oil-price movements.

The leftmost column Figure (ref) reports the relation between our two monetary-policy shocks and the CS surprise shock. The two correlations have similar magnitude but opposite sign, which is consistent with the SVAR separating two empirically distinct components of monetary policy that the univariate CS series aggregates into a single measure. The bottom row of Figure (ref) reports the relation between the narrative shock and four standard non-monetary alternatives. None of these correlations is statistically significant, suggesting that the narrative shock is not capturing aggregate uncertainty, news-based policy uncertainty, financial-market volatility, or oil-price movements.\footnote{The correlations of $\varepsilon^a$ and $\varepsilon^u$ with the WTI change are $-0.24$ and $-0.25$, marginally significant. The SVAR includes the staff projections $E^S_t[r_{t+4}], E^S_t[\log Y_{t+4}], E^S_t[\log P_{t+4}]$, which embed the staff's expected output and inflation paths and therefore the role assigned to anticipated oil-price movements in those paths. The structural Taylor rule of Section (ref) maps these expectations into the systematic component of the policy rate; the part of that mapping driven by anticipated oil movements is absorbed by the projection block of the SVAR. The residual correlation of $-0.25$ may capture the part of contemporaneous oil-price movements that lies outside the staff information set at the projection date.}

In addition, we compare the macroeconomic transmission of $\varepsilon^{CS}$, $\varepsilon^a$ and $\varepsilon^u$ using jorda2005estimation local projections under a single specification:

where $\varepsilon_t = \{\varepsilon^{CS}, \varepsilon^a,\varepsilon^u \}$, with $x_{t-1} = (\log\mathrm{GDP}^{CA}, \log\mathrm{GDP}^{US}, \log\mathrm{WTI})_{t-1}$, following the implementation of moran2025chocs. Figure (ref) reports the responses of log real GDP and log CPI for the three shocks; for the Champagne--Sekkel CPI we additionally report the response on the longer 1981Q2--2019Q4 sample, which extends back into the pre-IT disinflation. All three shocks generate contractionary responses of output and prices, but the timing and persistence differ substantially across identification schemes. The Champagne--Sekkel shock produces the largest and most persistent decline in GDP, while the unanticipated SVAR shock $\varepsilon^u$ generates a more immediate but comparatively stable contraction. By contrast, the anticipated shock $\varepsilon^a$ has more muted effects on real activity. For prices, the responses to $\varepsilon^a$ and $\varepsilon^u$ are negative throughout, whereas the Champagne--Sekkel shock displays a short-horizon price puzzle on the post-IT sample that disappears once the pre-IT disinflation period is included.

These differences are informative. The Champagne--Sekkel series is identified from high-frequency market surprises and therefore combines multiple components of monetary-policy news into a single measure. By contrast, the SVAR explicitly separates anticipated and unanticipated monetary-policy shocks while controlling for the narrative channel through the inclusion of media sentiment. The comparison therefore suggests that part of the heterogeneity in the local-projection responses may reflect the interaction between policy surprises, forward-guidance effects, and media-driven interpretations of policy decisions. In particular, the decomposition helps isolate the component of policy communication that operates through anticipated future policy paths from the residual contemporaneous surprise component, while conditioning on the central bank's reaction to media narratives. The resulting responses are consistent with the idea that some of the co-movement typically attributed to monetary-policy shocks may instead reflect information or narrative effects embedded in broad market-based surprise measures.

\oldsubsection{Media sentiment as a transmission channel for monetary policy} \addcontentsline{apptoc}{subsection}{\numberline{\Alph{section}.\arabic{subsection}}Media sentiment as a transmission channel for monetary policy}

Beyond contributing as a separate structural shock, $s_t$ may also act as a transmission channel for the two MP shocks. Table (ref) reports the FEVD shares of the three identified shocks for each variable across horizons. Since only three of the nine structural shocks are identified, the residual share attributed to the remaining rotations is mechanically large and should not be interpreted structurally; the relevant comparison is therefore the relative contribution of the identified shocks. Two patterns emerge. First, the realised policy rate loads more heavily on the unanticipated monetary-policy shock than on the anticipated component, while the narrative shock also contributes meaningfully to policy-rate fluctuations. Second, the macro block attributes a persistent share of output, prices, and hours fluctuations to the narrative shock, with the contribution increasing over longer horizons. Narrative shocks therefore appear to be a non-negligible source of macroeconomic variation in the system.

The historical decomposition (Figure (ref)) is consistent with the IRF pattern implied by the model, in which negative narrative innovations contribute positively to macroeconomic activity while positive narrative innovations are contractionary through the policy rule. The 1987Q4 episode is associated with positive contributions of narrative shocks to GDP and hours during the subsequent expansion, whereas the cumulative effect of the positive narrative shocks of the mid-2000s contributes negatively to activity during the 2008--2010 period, with a noticeable effect on GDP at the trough of the recession. The 2017--2019 tightening cycle around the Bank of Canada's rate-hike resumption also appears clearly in both media sentiment and the realised policy rate.

To evaluate the impact of media sentiments for the monetary policy transmission we apply the channel-shutdown counterfactual of bernanke1997systematic, adapted by barsky2011news to news shocks. The BGW exercise evaluates how the transmission of monetary-policy shocks changes once the lagged reduced-form feedback from $s_t$ to the remaining variables is removed, while preserving the identified contemporaneous impact matrix and the internal dynamics of $s_t$ itself. The counterfactual therefore does not alter the structural identification of the shocks or the contemporaneous policy reaction; it isolates the dynamic propagation channel operating through media sentiment. Comparing the baseline and counterfactual impulse responses then measures the extent to which media narratives amplify or attenuate the transmission of monetary-policy shocks over time.

Figure (ref) reports the GDP and CPI responses to an anticipated expansionary monetary-policy shock in the baseline SVAR (solid blue) and in the counterfactual that shuts down the lagged feedback of $s_t$ on the other variables (dashed red). These are the empirical counterparts of the two macro variables on which Figure (ref) reports the predictions of the structural model. In the structural IRFs of Section (ref), an anticipated easing produces a slow-building expansion of output and inflation that peaks near the announced implementation date and operates partly through media-sentiment feedback under $\phi_s > 0$. The SVAR responses match this pattern: log $Y_t$ rises gradually, peaks around $h = 14$ at $0.26\%$ in the baseline, and the log $P_t$ response is positive on impact and stays positive throughout the response horizon. Shutting down the sentiment channel attenuates both responses --- the GDP peak falls to $0.16\%$ and the CPI response shifts down in the medium term --- without eliminating the expansion. The empirical pattern therefore confirms the prediction of Figure (ref): media sentiment is a quantitatively non-trivial propagation channel for monetary-policy transmission to real activity and prices.

\oldsection{Conclusion} \addcontentsline{apptoc}{section}{\numberline{\Alph{section}}Conclusion}

In this paper we embed the media into a behavioral New-Keynesian model with cognitive discounting. Media sentiment $s_t$ enters the central bank's reaction function and follows a law of motion that decomposes coverage into persistence, fundamentals, central-bank communication, and a residual narrative component. The first three blocks make $s_t$ an endogenous outcome of the system; only the last is plausibly orthogonal to monetary policy. The structure pins down the identification problem in empirical estimation.

We construct media indicators of monetary-policy sentiment from a corpus of more than 50,000 articles published in major Canadian newspapers between 1977 and 2024. The indicators span four dimensions --- topic, tone, temporal orientation, and uncertainty --- and rely on three classes of tools: dictionary methods, transformer-based encoder models (FinBERT and ModernFinBERT), and a structured generative-LLM framework (CBILA). Encoder-only models are the workhorse for the historical sample because their pre-training cutoff predates the bulk of the data; the CBILA series carry potential look-ahead bias and are kept separate.

Media sentiment shifts household inflation and wage expectations beyond their own persistence and improves out-of-sample forecasts of CPI inflation and real GDP growth. Once $s_t$ is treated as endogenous and estimated by GMM with appropriate instruments, the encoder-based sentiment measures enter the Bank of Canada's estimated Taylor rule with a positive and significant loading; the OLS--GMM gap confirms that simultaneity between sentiment and policy is a first-order concern. The CBILA generative-LLM tones fail standard IV validity checks because of their look-ahead bias.

A Bayesian SVAR identified by sign restrictions derived from the model isolates an anticipated MP shock, an unanticipated MP shock, and a narrative shock. The unanticipated easing cuts the policy rate on impact and is followed by a hump-shaped expansion of GDP and prices, while the anticipated easing produces a smaller impact response of the rate, a slow build-up of GDP that peaks near the announced implementation date, and a positive CPI response, consistent with the cognitive-discounting mechanism. The narrative shock contributes a non-trivial share of medium-horizon macroeconomic variance. A counterfactual also confirms that media sentiment is a quantitatively relevant propagation channel for monetary-policy transmission.

\begingroup \setstretch{1.0} \endgroup