Federal Reserve Communication and the COVID-19 Pandemic — Benchimol, Kazinnik and Saadon (September 2025)
Summary
A research paper titled Federal Reserve Communication and the COVID-19 Pandemic, by Jonathan Benchimol, Sophia Kazinnik and Yossi Saadon, dated September 2025, with a citation to the Manchester School, vol. 93(5). The paper examines Federal Reserve communication during the COVID-19 pandemic and compares it with the dot-com and global financial crises, using sentiment analysis, custom dictionaries and topic modeling. Its data set covers FFR announcements, FOMC minutes and Chairman speeches from 2000 to 2020, totaling 776 texts in Table 1. The authors report that Fed communication and policy actions were more reactive to the COVID-19 crisis than to previous crises, and that communicating about unconventional monetary policy has become a new normal since the Global Financial Crisis. The paper states it does not claim causality between communication and economic outcomes.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Cite this document
“Federal Reserve Communication and the COVID-19 Pandemic — Benchimol, Kazinnik and Saadon (September 2025).”
Original at arxiv.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/federal-reserve-communication-and-the-covid-19-pandemic-f498189b31/
, SHA-256 c3c6321aff3e2326c66588075f82bc42f35e463ea5d5b7767b78cb8603cb8062.Full text
Federal Reserve Communication
and the COVID-19 Pandemic*
Jonathan Benchimol, 1 Sophia Kazinnik 2 and Yossi Saadon 3
September 2025
Abstract
In this study, we examine the Federal Reserve’s communication strategies during
the COVID-19 pandemic, comparing them with communication during previous
periods of economic stress. Using specialized dictionaries tailored to COVID-19,
unconventional monetary policy (UMP), and financial stability, combined with
sentiment analysis and topic modeling techniques, we identify a distinct focus in
Fed communication during the pandemic on financial stability, market volatility,
social welfare, and UMP, characterized by notable contextual uncertainty.
Through comparative analysis, we juxtapose the Fed’s communication during
the COVID-19 crisis with its responses during the dot-com and global financial
crises, examining content, sentiment, and timing dimensions. Our findings reveal
that Fed communication and policy actions were more reactive to the COVID-19
crisis than to previous crises. Additionally, declining sentiment related to
financial stability in interest rate announcements and minutes anticipated
subsequent accommodative monetary policy decisions. We further document
that communicating about UMP has become the “new normal” for the Fed’s
Federal Open Market Committee meeting minutes and Chairman’s speeches
since the Global Financial Crisis, reflecting an institutional adaptation in
communication strategy following periods of economic distress. These findings
contribute to our understanding of how central bank communication evolves
during crises and how communication strategies adapt to exceptional economic
circumstances.
Keywords: Central bank communication, Unconventional monetary policy,
Financial stability, Text mining, COVID-19.
JEL Classification: C55, E44, E58, E63.
* This paper does not necessarily reflect the views of the Bank of Israel, the Federal Reserve Bank of
Richmond, or the Federal Reserve System. We thank Itamar Caspi (discussant) and participants at the
133rd American Economic Association (AEA/ASSA) annual meeting, the 2021 Central Bank Research
Association (CEBRA) annual meeting, and the Bank of Israel research seminars for their useful
comments. An extended version of this paper was published in Covid Economics, CEPR Press, issue 79,
pages 218-256, May 2021. The replication files for the methodology used in this paper are available at
doi.org/10.24433/CO.1132001.v1; Text data are available at github.com/JBenchimol/central-bank-texts
1 Bank of Israel, Research Department, Jerusalem, Israel. Corresponding author.
Email: jonathan.benchimol@boi.org.il
2 Stanford University, Digital Economy Lab, Stanford, CA, USA.
Email: kazinnik@stanford.edu
3 Bank of Israel, Research Department, Jerusalem, Israel.
Email: yosis@boi.org.il
1
Please cite this paper as:
Benchimol, J., Kazinnik, S., and Saadon, Y., 2025. Federal Reserve Communication
and the COVID-19 Pandemic. Manchester School, vol. 93(5), pages 464-484.
2
1. Introduction
In the wake of the COVID-19 pandemic, central banks globally committed to
undertaking “all necessary steps” to alleviate its economic impact, including lowering
interest rates to the zero lower bound (Carney, 2020). As the crisis unfolded in March
2020, the Federal Reserve (Fed) and numerous other central banks implemented robust
unconventional monetary policy (UMP) measures to surmount the constraints of
conventional monetary policy. Communication emerged as a pivotal tool through
which these policies were conveyed.
Central banks engage in communication across various topics through distinct
channels, guided by well-defined objectives (Hansen et al., 2019; Benchimol et al., 2020).
Their communication aims to inform (e.g., current and future policy objectives and
decisions), explain (e.g., past, present, and future economic outlooks and decisions),
and influence (e.g., current and future uncertainty and financial decisions) economic
agents. Typically disseminated in textual form (Haldane and McMahon, 2018), these
instances of communication underwent significant adaptations during the COVID-19
pandemic, a period that affected all sectors of the global economy (Chetty et al., 2024).
This descriptive study delves into the distinct communication strategies employed
by the Fed during the COVID-19 pandemic compared to past crises. We conduct a
comparative analysis of communication instances spanning the past two decades,
focusing on three critical periods: the dot-com crisis, the Global Financial Crisis (GFC),
and the COVID-19 pandemic. We use text analysis techniques to examine the content,
sentiment, and timing of the Fed’s communication across various channels, including
interest rate announcements, minutes, and speeches.
Our main contribution lies in revealing the more reactive nature of the Fed’s
communication and actions during the COVID-19 crisis relative to other crises. We
uncover significant differences in content, sentiment, and timing, particularly regarding
financial stability, market volatility, and UMP. The analysis gauges the extent and
intensity of these differences between and during crises. Additionally, we assess the
impact of Fed communication on equity volatility in the United States (US), and explore
the connection between (conventional) monetary policy and financial stability
sentiment (Born et al., 2014; Correa et al., 2021).
By offering a nuanced comparison and novel insights into the Fed’s crisis
communication strategies, this paper aims to inform future central bank practices
during periods of economic turmoil. We do not claim causality between communication
and economic outcomes, but we shed light on the Fed’s communication patterns and
their potential role in shaping expectations about policy responses. Our findings add to
the growing literature on central bank communication during crises, which emphasizes
the importance of timely and transparent communication in managing market
expectations (Blinder et al., 2008; Hansen and McMahon, 2016; Gardner et al., 2022).
Given the zero lower bound (ZLB) of the nominal interest rate, the Fed, like many
central banks, resorted to alternative channels, including communication, quantitative
3
easing (QE), balance sheet policies, lending facilities, forward guidance, and other
market operations and monetary measures (Bianchi et al., 2020; Guerrieri et al., 2022).
Our analysis reveals that Fed communications during the COVID-19 crisis were
characterized by uncertainty and heterogeneity, both temporally and across
communication types. We examine the efficacy of the Fed’s transparent communication
in supporting UMP measures amid the challenges posed by the COVID-19 pandemic
(Daly, 2020; Craig et al., 2021). Our results suggest that the Fed adeptly employed
communication during the COVID-19 crisis, reflecting an increased proficiency in
crisis-specific communication management.
Effective central bank communication necessitates clear and timely updates on
current and near-term policy actions in an environment fraught with economic
uncertainty. Our findings document the prevalence of positive financial stability
updates in monetary policy and financial market-related communications after the
GFC, and their relatively less negative tone during the COVID-19 crisis, attesting to this
shift. 4
Furthermore, this paper explores the nuanced differences in content, timing, and
sentiment of the Fed’s communications across crises, suggesting potential implications
for financial system developments (Nyman et al., 2021). Since the GFC, communication
on UMP has become a “new normal,” evident in major communication types such as
Federal Funds Rate (FFR) announcements, Federal Open Market Committee (FOMC)
minutes, and Fed Chairman speeches. The study also connects conventional monetary
policy and financial stability sentiment (FSS).
The subsequent sections of the paper are organized as follows. Section 2 details the
data and methodology, Section 3 presents sentiment analysis and topic modeling
results, Section 4 examines the Fed’s communications regarding UMP, Section 5
discusses the Fed’s early communications regarding the pandemic, Section 6 compares
the Fed’s conventional monetary policy to the FSS over the past two decades, Section 7
discusses policy implications, and Section 8 concludes.
2. Data and Methodology
2.1 Data
Our investigation centers on primary communications disseminated by the Fed to the
public, encompassing instances detailing monetary policy discussions (FFR
announcements and FOMC minutes) and speeches by the Fed Chairman from 2000 to
2020. The particulars of our dataset are concisely presented in Table 1.
To augment our analysis, we incorporate data pertinent to the pandemic—
specifically, the daily count of new COVID-19 cases—sourced from the COVID-19 Data
Repository maintained by the Johns Hopkins University Center for Systems Science and
4 To proxy for the degree of financial stability conveyed in a central bank communication, we calculate a
financial stability score for each relevant communication based on a word count of the terms that can also
be found in the financial stability dictionary (Correa et al., 2021).
4
Engineering (CSSE). 5 In addition, we incorporate daily market-based indicators,
including the S&P 500 equity index, the Chicago Board of Options Exchange Volatility
Index (VIX), the broad nominal effective exchange rate (NEER), and the nominal
interest rate (FFR), retrieved from Bloomberg.
Table 1. Descriptive Statistics: Federal Reserve Texts
No. Texts No. Words (average) Sample
FFR Announcements 181 400 2000–2020
FOMC Minutes 170 6809 2000–2020
Chairman Speeches 425 2931 2000–2020
Total 776 3213 2000–2020
Sources: The Federal Reserve Board of Governors and FederalReserve.gov archives.
2.2 Methodology
To quantify instances of Fed communication, we employ text-based measures assessing
uncertainty and sentiment, utilizing a range of custom dictionaries. The methodology
encompasses sentiment scoring, simple word-counting procedures, and topic modeling
techniques applied to the collection of texts.
First, we employ word counting procedures using distinct dictionaries: a finance-
specific sentiment dictionary (Loughran and McDonald, 2011; hereafter, LM), an FSS 6
dictionary (Correa et al., 2021), a UMP dictionary, and UMP and COVID-19 lexicons.
Second, we utilize sentiment scoring based on the LM dictionary to proxy for
sentiment and contextual uncertainty in the Fed’s communications. Multiple sentiment
scores and polarity indicators are built using general (NRC, SentiWords, Hu&Liu,
Jockers) and specialized (FSS, UMP) dictionaries. Following Loughran and McDonald
(2016), we incorporate valence shifters to capture nuances in sentiment, such as
negation (e.g., “not good”), amplification (e.g., “very good”), or de-amplification (e.g.,
“somewhat good”). This approach allows us to more accurately assess the sentiment
conveyed in Fed communications, recognizing that context significantly alters the
meaning and intensity of sentiment-laden words.
Third, we employ topic modeling for unsupervised machine learning analysis,
specifically the Latent Dirichlet Allocation (LDA) algorithm (Blei et al., 2003). This
technique allows us to extract and examine thematic content in the Fed’s
communications, comparing it to economic and financial developments. LDA treats
documents as random mixtures over latent topics, where each topic is characterized by
a distribution over words. This probabilistic approach enables us to identify the main
themes in Fed communications without imposing predefined categories.
5 The full dataset can be accessed at github.com/CSSEGISandData/COVID-19
6 Correa et al. (2021) capture movements in financial cycle indicators and sentiments conveyed in
financial stability reports.
5
The dictionary capturing the sentiment about UMP measures merges two
dictionaries: Erasmus and Hollander (2020), which focuses on forward guidance and
quantitative measures, and Henry (2008), which focuses more on the regulatory context,
structural attributes, and the dual informational-promotional role of earnings press
releases.
The lexicon that captures communications regarding the UMP measures (UMP
terms) collects words related to UMP policies from Fed communications using topic
modeling and Bag-of-Words techniques (Table 2).
Table 2. Unconventional Monetary Policy Lexicon
asset purchases depreciation pressure market disrupt risk premium
helicopter direct lending market functioning securities purchases
QE ELB monetary base stagflation
securities purchases foreign exchange reserve monetary stimulus support
balance sheet forward guidance money supply support liquidity
business support funding negative policy supporting corporat
credit facilit insolvency negative rate swap line
credit impair intervention NIRP unconventional
deferral lending facilit quantitative easing ZLB
deflation lower bound relaxing regulatory
Notes: This dictionary primarily includes words and their root forms extracted from Fed communications
using topic modeling and Bag-of-Words techniques (see Benchimol et al., 2022).
The lexicon that captures communications regarding COVID-19 compiles relevant
keywords related to the pandemic, identifying virus-related content (word frequency)
in the Fed’s communications (Table 3). By counting the number of COVID-19-related
words, we can estimate the attention paid to the pandemic at each point in time.
Table 3. COVID-19 Lexicon
acute elderly infect pandemic severe acute
cases emergency infection pneumonia sickness
confin epidem infection rate quarantine spreading
contagio epidemic lockdown relief syndrom
corona hcov mask reproduction rate testing
coronavirus health medical respirator vaccin
covid hospital morbid respiratory virus
death hubei morbidity rate sars wave
disabilit human mortal sars cov wuhan
disease illness ncov sarscov
disorder inception rate outbreak sars-cov
Notes: This dictionary includes words and root words related to the COVID-19 that appeared in both
media sources (e.g., Google Trends search queries) and recent Fed communications (Bag-of-Words).
6
For sentiment analysis, the LM dictionary, designed for assessing sentiment and
uncertainty in financial text and widely used in the finance and economics literature
(Loughran and McDonald, 2016), is complemented by commonly used sentiment
dictionaries in text mining literature (Jockers, NRC, Hu&Liu). Valence shifters are
incorporated to capture nuances in sentiment. Sentiment scores are calculated using a
formula that accounts for the positive and negative words in the corresponding
dictionaries for the text, such that:
𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑡𝑡
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑡𝑡 = , (1)
𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑡𝑡 + 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡
where 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑡𝑡 (𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 ) is the total number of words classified as positive
(negative) in the corresponding dictionary for text t.
Two types of sentiment indicators based on the LM dictionary are constructed: a
standard score measure and a polarity measure that includes neutral, positive, negative,
very positive, or very negative sentiment, considering the words preceding and
following the target word.
3. Results
3.1 Sentiment Analysis
This section presents sentiment scores, providing insights into the Fed’s response to the
evolving developments during the COVID-19 pandemic, marked by a notable decline
in sentiment scores and an increase in uncertainty-related words in the first quarter of
2020.
Figure 1 shows sentiment scores within FFR announcements, revealing a downturn
from January to April 2020, coinciding with the onset of the COVID-19 crisis. The
depicted sentiment trends reflect an overall pessimistic view during the early months
of the pandemic, but indicate a more optimistic sentiment by the third quarter of 2020
(Panels A1, A3, A6, and A8). This timeframe aligns with heightened contextual
uncertainty (Panel A2), denoting the number of words indicative of uncertainty scaled
by text length.
Utilizing the LM dictionary, the polarity indicator (Panel A3) reveals a decline but
exhibits a more optimistic sentiment for 2020:Q3. FFR announcements serve as a
summary of the current state of the economy and monetary policy decisions. Figure 1
shows that the impact of the shock that started in January 2020 persisted until April
2020. The sharp decline in sentiment observed in 2020:Q1-Q2 coincides with the rise in
contextual uncertainty that lasted from 2020:Q1-Q3. These trends mirror the global
spread of COVID-19 and its anticipated spillover effects on the US economy.
Figure 1 presents a comparative analysis of sentiment indicators, revealing nuanced
dynamics in the polarity indices derived from Jockers, NRC, SentiWords, Hu&Liu, and
our LM-based polarity index. Notably, the Jockers, NRC, and SentiWords indices
appear less informative as they are based on general interest dictionaries, thus ignoring
7
the specifics of economics and finance, while the Hu&Liu polarity index exhibits similar
dynamics to our LM-based polarity index.
Figure 1. Sentiment Scores
Panel A: FFR Announcements Panel B: FOMC Minutes
Panel C: Chairman Speeches Panel D: Main Fed Communications
Notes: Solid black lines represent sentiment score values. The shaded area represents the COVID-19 crisis.
The broad-coverage polarity index SentiWords captures a noteworthy surge in
sentiment within FFR announcements from the onset of the COVID-19 crisis. This
suggests a potential shift in the Fed’s communication strategy compared to other
crises, 7 possibly reflecting a deliberate attempt to convey confidence amid
unprecedented economic uncertainty.
The FSS has experienced a decline into negative territory since 2019:Q4, indicating a
prevalence of negative financial stability-related words in FFR announcements during
this period, which preceded the actual outbreak of COVID-19 in the US. This suggests
7 The SentiWords polarity index declines sharply for the GFC and dot-com crises. The full sample results
are available upon request. See also Benchimol et al. (2020).
8
that concerns about financial stability were emerging even before the pandemic began
to significantly impact the US economy.
Before the COVID-19 crisis, the sentiment related to UMP was notably positive,
reflecting the Fed’s plan to gradually reduce the size of its balance sheet. However, with
the onset of the COVID-19 crisis, emergency policies were accompanied by
communications with a more negative tone, contributing to a surge in financial
uncertainty (Section 6).
Figure 2 shows a comparative sentiment analysis across the entire sample covering
the dot-com, GFC, and COVID-19 crises. The figure presents sentiment trajectories
during the three major economic crises of the past two decades, allowing us to compare
how the Fed’s communication sentiment evolved during each period of economic
stress.
Figure 2. A Tale of Three Crises: Sentiment.
0.2
9.0
7.0 0.1
5.0 0
3.0
-0.1
1.0
-0.2
-1.0
-0.3
-3.0
-5.0 -0.4
Financial Stability Sentiment UMP Sentiment Loughran (score) Hu & Liu
Notes: The vertical shaded areas represent the NBER recession periods. Solid lines represent sentiment
scores computed from FFR announcements. The different colored lines represent different sentiment
measures as indicated in the legend.
The figure presents a more stable sentiment trajectory during the COVID-19 crisis
compared to the sharp declines seen in previous crises, suggesting a different
communicative approach by the Fed during the recent pandemic. This could reflect the
Fed’s learning from previous crises and its enhanced ability to manage communication
during periods of economic stress. It could also indicate that the Fed was attempting to
project greater stability and confidence despite the severity of the economic shock
caused by the pandemic.
Figure 2 shows that the FSS experienced a marked deterioration before the GFC and
the dot-com crisis, contrasting with its significantly positive status preceding the onset
of the COVID-19 pandemic. In the context of sentiment polarity indices, the Hu&Liu
9
index improved from the GFC to the COVID-19 crisis, with a less pronounced decline
during the latter than during the former. In contrast, LM’s sentiment index did not
exhibit improvement between crises, but it experienced a milder decline during the
COVID-19 crisis than in previous crises. Furthermore, the volatility of these indicators
was less conspicuous during the COVID-19 crisis than during the GFC and dot-com
crises. While Figure 2 provides insights into the dynamics of the three crises, it also
vividly depicts the distinct trajectories of Fed communication policies across these
periods.
Figure 1 (Panel B) illustrates sentiment scores in FOMC minutes, with an increasing
sentiment trend during the early phase of the COVID-19 crisis. This positive trend may
reflect the Fed’s strategic use of reassuring language to mitigate economic concerns
during this period. The dynamics of nearly all sentiment indicators show a sharp
deterioration in 2020:Q2, a trend more pronounced for the FOMC minutes than for the
FFR announcements.
As a possible indication of the disparity between the depiction of the ongoing
economic scenario in FFR announcements and the in-depth discussions and provisional
solutions to the COVID-19 crisis found in the minutes, the LM score and polarity indices
underwent a pronounced decline in sentiment related to financial uncertainty from
January to April 2020.
The sentiment score related to UMP declined until 2020:Q2, followed by a sharp
upturn leading to a positive trajectory. This shift aligns with the more positive language
observed in FOMC minutes concerning UMP measures undertaken during the COVID-
19 crisis. Figure 1 (Panel B8) also highlights that policymakers perceived a significant
financial stability risk in 2020:Q2, according to FOMC minutes, albeit less severe than
during the GFC. 8
The SentiWords index, known for its extensive coverage, reveals an intriguing
pattern of increasing sentiment from the onset of the COVID-19 crisis. This trend may
be attributed to the Fed’s communication strategy, employing more positive language
to reassure and pacify economic agents. In contrast, the SentiWords indicator sharply
declined to historically low levels during the GFC, suggesting a different approach to
crisis communication during the COVID-19 pandemic. 9
Figure 1 (Panel C) provides sentiment indicators for the Fed Chairman’s official
speeches, showing greater volatility but a milder decline in sentiment than in the FOMC
minutes and FFR announcements. This suggests that the Chairman’s speeches may
have focused more on managing expectations during the pandemic. The flexibility of
speeches compared to the straightforward and more supervised content of the minutes
and announcements explains this result.
8 The full sample results are available upon request. See also Benchimol et al. (2020).
9
The volatility in sentiments conveyed by Fed Chairman’s speeches compared to FFR announcements and FOMC
minutes is discussed in Benchimol et al. (2020).
10
Despite a worsening economic situation and declining sentiment from February 2020
onward, there is a concurrent decrease in contextual uncertainty. This apparent
contradiction might reflect the Chairman’s efforts to reduce perceived uncertainty
through clear communication about the Fed’s policy responses, even as the economic
outlook deteriorated.
However, the relatively small sample size of Chairman speeches during the COVID-
19 crisis may affect the volatility of sentiment indicators presented in Figure 1 (Panel
C), making them less erratic than those for the GFC.
Similar to FFR announcements and FOMC minutes, the expansive scope of the
SentiWords dictionary captures the distinct communication policy employed by the
Fed during the COVID-19 crisis, evident at least until 2020:Q4. The LM dictionary also
reveals a decline in the use of uncertainty-related words in Chairman speeches
following the COVID-19 outbreak in China, potentially reflecting a deliberate
communication strategy to project confidence and reduce market uncertainty.
Figure 1 (Panel D) presents a consolidated view of the Fed’s communication
strategies during critical periods by aggregating FFR announcements, FOMC minutes,
and Fed chair speeches into a single corpus for sentiment analysis. This comprehensive
approach allows us to identify overall trends in the Fed’s communication sentiment
across different channels.
In Figure 1 (Panels C and D), the sentiment scores show a marked decrease in
uncertainty sentiment from January to April 2020. This period correlates with the Fed’s
implementation of liquidity measures to stabilize the financial market. Figure 1
suggests that the Fed’s communication was consistently tailored across various types,
such as FFR announcements, FOMC minutes, and Chairman speeches, to address the
crisis effectively.
By consolidating all communication types within our dataset, this comprehensive
overview of the Fed’s communications reveals that the substantial decline in
uncertainty sentiment was primarily driven by minutes and speeches. The FSS also
exhibited a notable decline from January to April 2020, correlating with liquidity
measures taken by the Fed to stabilize market volatility and uncertainty, as elaborated
in Section 6.
The findings presented in Figures 1 and 2 consistently lead to a compelling
conclusion: During the COVID-19 crisis, the Fed conveyed specific sentiment patterns
across all communication types—FFR announcements, FOMC minutes, and Chairman
speeches. Moreover, these results underscore the distinct sentiments prevailing during
the COVID-19 crisis compared to the GFC, suggesting an evolution in the Fed’s crisis
communication approach over time.
This section delves into the noteworthy sentiment deterioration experienced from
January to April 2020. Subsequently, sentiment rebounded into positive territory, with
certain measures surpassing pre-crisis levels. By examining the pronounced temporal
variations in sentiment, we observe significant changes in the Fed’s communication
11
patterns during the early stages of the pandemic, followed by a shift toward more
positive sentiment as the Fed implemented its policy responses.
Statistical evidence confirms milder sentiment deterioration during COVID-19. The
average magnitude of sentiment breaks was 0.157 for COVID-19 compared to 0.213 for
the GFC, with statistically significant differences. 10 In addition, break points in
sentiment occurred approximately a month earlier in the COVID-19 crisis than in
previous crises, suggesting more rapid Fed communication adaptation. The analysis
reveals that break magnitudes differ systematically across communication channels,
with FSS showing larger breaks in FFR announcements (0.217) than in speeches (0.156).
Post-break sentiment trajectories exhibit distinct patterns across crises, with COVID-19
showing more rapid mean reversion, supporting the hypothesis of enhanced crisis
management.
3.2 Topic Modeling
This section employs topic modeling to elucidate the underlying themes influencing
Fed communication. The analysis, facilitated by the Latent Dirichlet Allocation (LDA)
algorithm—a widely recognized unsupervised machine learning technique—identifies
and quantifies prevalent themes in the text corpus, offering insights into the Fed’s real-
time assessments of economic and financial risks.
The examination reveals that, during the initial stages of the COVID-19 pandemic,
themes related to policy intervention assumed prominence, overshadowing discussions
on inflation expectations and financial stability. This focus on policy intervention in the
Fed’s communication during the COVID-19 pandemic serves as a distinctive feature
compared to the communication strategies employed during the GFC.
Emphasizing that the topics were extracted directly from the text sample, it is crucial
to note that the employed topic modeling methodology does not rely on predetermined
dictionaries. In contrast to sentiment analysis, this approach represents a less
supervised method for discerning word-topic linkages within texts, allowing for the
emergence of natural thematic patterns in communications.
The topics extracted from FFR announcements over two decades are visualized in
Figure 3, highlighting that policy interventions became a more dominant topic during
the COVID-19 crisis, while there was less focus on inflation expectations. Compared to
other crises, the discourse on policy interventions garnered increased prominence
during the COVID-19 crisis.
Figure 3 illustrates a decline in the significance of the topic of inflation expectations
during the COVID-19 crisis, juxtaposed with a heightened importance of the topic of
economic growth, encompassing considerations and concerns related to economic
growth.
Moreover, Figure 3 offers insights into the relative impact of each crisis on the Fed’s
communications through FFR announcements. While the dot-com crisis minimally
10
T-statistic: 2.76; p value : 0.009.
12
influenced the topics conveyed to the public in FFR announcements, both the GFC and
the COVID-19 crises significantly shaped the topics conveyed. This suggests that the
severity and nature of these latter crises required more substantial shifts in
communication focus. During the European debt crisis, financial stability became a
much more prominent topic in the Fed's communications than during the dot‐com
crisis.
Figure 3. A Tale of Three Crises: Topics
Notes: Bars represent the topic probability computed from FFR announcements. For clarity and
robustness, we restrict attention to the six most frequently discussed topics. Benchimol et al. (2020)
present the topic analysis and words constituting each topic used to generate this figure. The pink bars
represent “Policy Intervention,” yellow bars represent “Interest Rate,” purple bars represent “Economic
Growth,” green bars represent “Inflation,” blue bars represent “Financial Markets,” and orange bars
represent “Inflation Expectations.”
Figure 4 (Panel A) shows the results of topic analysis using FFR announcements,
indicating a significant increase in discussions about policy interventions following the
COVID-19 outbreak. This suggests a proactive communicative stance by the Fed in
response to the emerging crisis, with a clear emphasis on communicating its policy
actions to market participants and the public.
The onset of the COVID-19 outbreak coincided with a notable decline in the topic
probability related to inflation expectations in FFR announcements. This aligns with the
prevailing monetary policy considerations during that period when the focus shifted
prominently to policy intervention and addressing the immediate economic fallout
from the pandemic.
13
Additionally, there was a smaller decline in discussions concerning inflation,
coupled with an increase in discussions about economic growth within FFR
announcements during the COVID-19 crisis. This shift in focus reflects the Fed’s
priorities during the pandemic, with greater emphasis on supporting economic activity
and less concern about immediate inflationary pressures.
Figure 4. Topic Analysis
Panel A: FFR Announcements Panel B: FOMC Minutes
Panel C: Chairman Speeches Panel D: Main Fed Communications
Notes: For clarity and robustness, we restrict attention to the six most frequently discussed topics, except
for Panel D, considering a broader set of texts. Words constituting each topic are available upon request.
The observed escalation in the topic probability of policy intervention in FFR
announcements led to a corresponding reduction in the topic probability related to
inflation expectations and, to a lesser extent, to inflation. Despite one of the Fed’s
primary objectives being the stabilization of prices, this finding suggests that FFR
announcements became less tethered to inflation concerns following the COVID-19
outbreak. Furthermore, the slight increase in the topic probability associated with
14
economic growth after the COVID-19 outbreak in China indicates the Fed’s
apprehension that the pandemic threatened economic growth.
Figure 4 (Panel B) presents results for the topic analysis of FOMC minutes,
demonstrating an increased focus on policy interventions and financial markets,
indicating the committee’s heightened attention to these areas during the COVID-19
pandemic.
Notably, akin to FFR announcements, the FOMC minutes also exhibit an influence
of the policy intervention topic, despite the predominant focus on interest rates. The
probability of discussing inflation in the FOMC minutes declined following the COVID-
19 outbreak, while there was an upsurge in discussions related to policy intervention
and financial markets.
The thematic content conveyed by the FOMC minutes underscores a significant
augmentation in the coverage of policy intervention and the foreign economy.
Intriguingly, the expansion of coverage concerning the foreign economy had
commenced even before the onset of the COVID-19 outbreak, while coverage of topics
such as inflation and interest rates had already begun to decline. This suggests the Fed
was monitoring international developments and potential spillovers before the
pandemic fully emerged in the US.
Figure 4 (Panel C) provides a topic analysis of speeches by the Chairman of the Fed,
noting a shift in focus toward social welfare concerns after the COVID-19 outbreak,
which aligns with broader societal issues in the US, such as education and inequality.
This observation highlights that Chairman speeches often address matters unrelated to
the Fed’s primary objective of price stabilization, delving into areas like education,
healthcare, and development economics, encompassing family and labor markets. 11
The divergence in supervision and audience focus between speeches, intended for
broader audiences, and traditional FFR announcements and FOMC minutes, which
concentrate more on inflation and output growth, may explain the heightened
discussions surrounding social welfare issues. A noticeable uptick in discussions
related to economic policy is also observed following June 2020. While this could be
attributed to COVID-19 spillovers, the influence of the approaching US election cannot
be dismissed. Economic policy considerations had already commanded attention in
most Chairman speeches before the COVID-19 outbreak.
Figure 4 (Panel D) offers a comprehensive analysis of the Fed’s communication
strategies by combining FFR announcements, FOMC minutes, and Fed chair speeches
into a single corpus for topic modeling. Considering the substantial number of texts
analyzed and their distinct characteristics, 12 more topics are considered for the
11 The most frequently used words and word fragments (root words) in the context of the topic of social
welfare are communiti, economi, educ, work, develop, research, busi, job, peopl, help, opportun, import, and
family.
12 The Fed does not communicate similarly or on similar topics across its various channels, which include
FFR announcements, FOMC minutes, and speeches by the Fed chairman. Different topics are emphasized
and communicated in varying ways across these channels. This lack of consistency in communication
15
aggregate topic modeling. These encompass most of the topics outlined in Figures 3 and
4 (Panels A to C).
Notably, the monetary policy topic incorporates references to UMP and
unemployment, aligning with the Fed’s dual mandate. Despite this, the primary focus
on inflation expectations persists, a trend possibly reinforced by the challenges posed
by the COVID-19 crisis and concerns regarding long-term interest rates.
Figure 4 (Panel D) highlights three prominent topics that swiftly gained
prominence. Initially, there was an escalation in Fed communication concerning the
foreign economy as the COVID-19 pandemic extended beyond China to the global
stage (01-03/2020).
Subsequently, discussions on financial stability gained momentum amid growing
fears about the impact of the COVID-19 crisis on the financial system (03-05/2020).
Lastly, there was an upswing in Fed communication related to social welfare,
corresponding to the perceived necessity for additional relief plans from the
government and the Fed (04-07/2020). While discussions on conventional monetary
policy, particularly average inflation targeting, and UMP witnessed a decline at the
pandemic’s onset, they experienced an upward trajectory thereafter (06-10/2020).
In sum, disparities in the content and timing of Fed communication are evident
across the three discussed crises. Unlike the GFC and dot-com crises, the COVID-19
crisis prompted a shift in the focus of communications from discussions on inflation
expectations to considerations of policy intervention.
Notably, the topic of policy intervention took greater precedence in communication
during the COVID-19 crisis compared to the GFC and dot-com crises. This suggests that
policymakers not only implemented distinct policy interventions but also approached
the discussion of these interventions differently across the crises. The content,
sentiment, and timing of these communications were contingent on the nature of the
crisis, with the COVID-19 pandemic requiring a particularly rapid and extensive policy
response.
To quantify the differences in topic distributions observed across crises, we employ
two complementary metrics: (1) topic coherence scores using Normalized Pointwise
Mutual Information (NPMI) to evaluate semantic quality, and (2) Kullback-Leibler (KL)
divergence to measure cross-crisis differences in topic distributions.
For topic coherence, we calculate NPMI scores for each topic following the approach
of Röder et al. (2015):
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(𝑤𝑤𝑖𝑖 , 𝑤𝑤𝑗𝑗 ) = 𝑙𝑙𝑙𝑙𝑙𝑙(𝑃𝑃(𝑤𝑤𝑖𝑖 , 𝑤𝑤𝑗𝑗 )⁄𝑃𝑃(𝑤𝑤𝑖𝑖 )𝑃𝑃(𝑤𝑤𝑗𝑗 )…−𝑙𝑙𝑙𝑙𝑙𝑙(𝑃𝑃(𝑤𝑤𝑖𝑖 , 𝑤𝑤𝑗𝑗 )) (2)
where 𝑤𝑤𝑖𝑖 and 𝑤𝑤𝑗𝑗 are words in the same topic, 𝑃𝑃(𝑤𝑤𝑖𝑖 ) and 𝑃𝑃(𝑤𝑤𝑗𝑗 ) represent the probabilities
of individual words 𝑤𝑤𝑖𝑖 and 𝑤𝑤𝑗𝑗 occurring in the corpus, and 𝑃𝑃(𝑤𝑤𝑖𝑖 , 𝑤𝑤𝑗𝑗 ) represents the
approaches has implications when considering aggregating these texts regarding the number of topics to
assume.
16
probability of both words co-occurring in the corpus. Higher NPMI scores indicate
greater semantic coherence.
For cross-crisis comparisons, we calculate the KL divergence between topic
distributions:
𝐾𝐾𝐾𝐾(𝑃𝑃𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 || 𝑃𝑃𝐺𝐺𝐺𝐺𝐺𝐺 ) = ∑𝑖𝑖 𝑃𝑃𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑖𝑖) × 𝑙𝑙𝑙𝑙𝑙𝑙(𝑃𝑃𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑖𝑖)/𝑃𝑃𝐺𝐺𝐺𝐺𝐺𝐺 (𝑖𝑖)) (3)
where PCOVID(i) and PGFC(i) represent the probability of topic i during the COVID-19
and GFC periods, respectively. The Policy Intervention topic exhibits both the highest
coherence during COVID-19 (0.380) and the largest divergence from previous crises. 13
This constitutes statistical evidence for a significant shift in communication focus
during the pandemic, with more consistent and cohesive messaging about policy
actions than in previous crises. Similarly, the Inflation Expectations topic shows
significantly reduced prominence during COVID-19 compared to the dot-com crisis, 14
confirming the observation that monetary policy communications became less tethered
to inflation concerns during the pandemic.
4. Unconventional Monetary Policy
This section investigates the correlation between the Fed’s communication, its actions,
UMP, and the impact of the COVID-19 pandemic. Figure 5 juxtaposes UMP terms with
UMP measures as indicated by Fed balance sheet changes, providing insight into the
timing relationship between Fed communication about UMP and actual
implementation of such policies.
Figure 5 shows that the Fed communicated more extensively regarding UMP during
the GFC than during the COVID-19 crisis. However, the crucial element lies in the
timing. Unlike the GFC, where a four-month delay existed between UMP actions (July
2008) and UMP communication (November 2008), the Fed acted and communicated
swiftly during the COVID-19 crisis, ensuring better coordination between
communication and actions. This improved synchronization likely reflects lessons
learned from the GFC about the importance of clear and timely communication when
implementing unconventional policies.
Moreover, Figure 5 reveals that the Fed’s communications on UMP during the
COVID-19 crisis, based on a word count of terms from the UMP lexicon (Table 2), align
with effective UMP measures that influenced changes in the Fed’s balance sheet with
some lags. Unlike the GFC, when actions were implemented before communication,
during the COVID-19 crisis, actions followed communication, indicating a shift in the
coordination of responses. This suggests a more deliberate strategy of preparing
markets for policy changes through forward guidance during the pandemic.
Our UMP dictionary proves versatile in identifying periods beyond crises where
substantial UMP measures were undertaken to support the US economy. Each peak in
13
KL = 0.834; p value < 0.001.
14
KL = 0.673; p value < 0.01.
17
UMP measure-related communication subsequently impacted the Fed’s balance sheet
shortly after the communication event. Notably, communications in 2013–2014 signaled
the cessation of accommodative policies, aligning with the Fed’s balance sheet changes,
showcasing the UMP dictionary’s ability to capture tapering communication strategies.
Figure 5. Unconventional Monetary Policy Terms
Panel A: FFR Announcements Panel B: FOMC Minutes
7.0 8.0 5.0 8.0
4.5 7.0
6.0 7.0
4.0
6.0 6.0
5.0 3.5
5.0 3.0 5.0
4.0
4.0 2.5 4.0
3.0 2.0
3.0 3.0
2.0 1.5
2.0 2.0
1.0
1.0 1.0 1.0
0.5
0.0 0.0 0.0 0.0
Panel C: Chairman Speeches Panel D: Main Fed Communications
6 8.0 5.0 8.0
7.0 4.5 7.0
5
4.0
6.0 6.0
3.5
4
5.0 3.0 5.0
3 4.0 2.5 4.0
3.0 2.0 3.0
2 1.5
2.0 2.0
1.0
1
1.0 0.5 1.0
0 0.0 0.0 0.0
Notes: The gray shaded areas represent NBER recession periods. The blue shaded areas represent the
word-counting indicator based on the lexicon presented in Table 2 (left axis). The red shaded area
represents total assets (minus eliminations from consolidation) in trillions of US dollars in the Fed’s
balance sheet (right axis).
Source: Fed balance sheet data is collected from the Board of Governors of the Federal Reserve System.
In response to economic growth expectations, the Fed initiated a gradual reversal of
QE policies, known as “tapering,” in 2013. Fed Chairman Ben Bernanke’s
announcement signaled the reduction of monthly asset purchases if favorable economic
conditions persisted, illustrating how communication can be used to prepare markets
for policy normalization after periods of unconventional measures.
Panel B of Figure 5 demonstrates that FOMC minutes discussed UMP actions
somewhat earlier during the COVID-19 crisis than during the GFC. The frequency of
18
discussions on UMP parallels the tapering discussions conducted by the FOMC in 2013,
with an overall higher level observed in late 2020 than during the GFC. This suggests
an institutional learning process, with the Fed drawing on its GFC experience to more
quickly incorporate discussions of unconventional tools in its deliberations during the
COVID-19 crisis.
An intriguing distinction emerges between UMP discussions in FFR announcements
and FOMC minutes. Despite the latter being less scripted and more extensive than the
former, discussions on UMP in FOMC minutes were more pronounced prior to the
COVID-19 crisis than before the GFC or even during the initial phase of the GFC. This
observation is corroborated by a comparative analysis presented in panels A and B in
Figure 5, suggesting that UMP had become a more routine topic of discussion in FOMC
deliberations even before the pandemic struck.
In synthesizing discussions among monetary policy committee members, FOMC
minutes consistently incorporate more UMP terms, encompassing discussions and
controversies related to potential solutions or policy implementations, than FFR
announcements do. The frequent use of such terms dates back to the GFC, indicating a
structural shift in the composition and focus of monetary policy discussions. This
evolutionary pattern in the Fed’s discourse suggests an institutional learning process,
wherein unconventional policy tools have become progressively integrated into the
standard policy lexicon.
Panel C of Figure 5 provides a word count analysis of UMP-related terms in speeches
delivered by the Fed Chairman, revealing distinct communication patterns across
different crisis periods.
Figure 5 (Panel C) illustrates that significant communication shocks, such as those
observed during the GFC or the third round of monthly purchases of Treasury
securities and mortgage-backed securities in September 2012 (third QE), are closely
associated with the UMP content found in speeches by the Fed Chairman. Each
communication peak coincided with or preceded changes in the Fed’s balance sheet
dynamics. Peaks that did not impact the Fed’s balance sheet were typically linked to
forward guidance communications, highlighting the dual role of Chairman speeches in
both preparing markets for balance sheet operations and managing expectations about
the future path of policy.
Fed Chairman speeches emerge as the primary platform for discussing UMP terms,
serving as a critical channel for articulating complex policy innovations to a broader
audience. Examining the frequency of UMP terms in speeches delivered before and
after each crisis yields intriguing insights. Panel C of Figure 5 reveals an increase in the
frequency of UMP terms in speeches following the GFC, dot-com, and COVID-19 crises.
Notably, the frequency was more pronounced, and UMP terms occurred earlier during
the COVID-19 crisis than in other crises, suggesting a more rapid deployment of the
communication toolkit during the pandemic.
19
Panel D of Figure 5 presents a comprehensive UMP indicator for key Fed
communications, aggregating data from FFR announcements, FOMC minutes, and
Chairman speeches to provide a holistic view of UMP communication patterns.
The figure highlights distinctions in the timing of UMP communications and actions
between the COVID-19 and GFC periods. During the GFC, UMP actions preceded
comprehensive communication about these policies, creating a lag in market
understanding of the Fed’s strategy. In contrast, during the COVID-19 crisis,
communications about UMP measures were more synchronized with their
implementation, potentially enhancing policy effectiveness through clearer signaling
channels.
The emergence of a post-GFC “new normal” is of particular significance,
characterized by more frequent UMP communications and actions than during the pre-
GFC era. This sustained reliance on UMP tools suggests a fundamental transformation
in monetary policy implementation, where previously “unconventional” measures
have become standard components of the policy toolkit. This structural evolution
reflects the constraints imposed by the proximity to the zero lower bound and the need
for alternative policy instruments when conventional interest rate adjustments are
limited.
To formalize the lead-lag relationships observed in Figure 5, we implement Granger
causality tests within a Vector Autoregression (VAR) framework. For each
communication type and crisis period, we estimate:
𝑈𝑈𝑈𝑈𝑈𝑈 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑡𝑡 = 𝛼𝛼1 + ∑𝑘𝑘𝑖𝑖=1 𝛽𝛽𝑖𝑖1 𝑈𝑈𝑈𝑈𝑈𝑈 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑡𝑡−𝑖𝑖 + ∑𝑘𝑘𝑖𝑖=1 𝛾𝛾𝑖𝑖1 𝐹𝐹𝐹𝐹𝐹𝐹 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑡𝑡−𝑖𝑖 + 𝜀𝜀𝑡𝑡1 (4)
𝐹𝐹𝐹𝐹𝐹𝐹 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑡𝑡 = 𝛼𝛼2 + ∑𝑘𝑘𝑖𝑖=1 𝛽𝛽𝑖𝑖2 𝐹𝐹𝐹𝐹𝐹𝐹 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑡𝑡−𝑖𝑖 + ∑𝑘𝑘𝑖𝑖=1 𝛾𝛾𝑖𝑖2 𝑈𝑈𝑈𝑈𝑈𝑈 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑡𝑡−𝑖𝑖 + 𝜀𝜀𝑡𝑡2 (5)
where the optimal lag length k is determined using the Bayesian Information Criterion
(BIC). We test two null hypotheses:
(1) 𝐻𝐻0 : 𝛾𝛾12 = 𝛾𝛾22 = ⋯ = 𝛾𝛾𝑘𝑘2 = 0, which implies UMP communication does not Granger
cause balance sheet changes;
(2) 𝐻𝐻0 : 𝛾𝛾11 = 𝛾𝛾21 = ⋯ = 𝛾𝛾𝑘𝑘1 = 0, which implies balance sheet changes do not Granger
cause UMP communication.
We find statistical evidence for a fundamental shift in the Fed's communication
strategy between crises. During the GFC, the Fed followed an action-first approach,
where policy implementation preceded communication, as shown by the significant
Granger causality from the Fed Balance Sheet to UMP terms across all communication
channels. 15
In contrast, during the COVID-19 crisis, the Fed adopted a communication-first
strategy, where communication about potential measures preceded their
implementation. This is demonstrated by the significant Granger causality from UMP
15
F test: 8.23 to 12.36, p value < 0.01.
20
terms to Fed assets. 16 This shift may suggest institutional learning from the GFC
experience and a greater emphasis on forward guidance during the COVID-19 crisis.
The differentiation across communication channels is particularly noteworthy.
Chairman speeches showed the strongest forward-looking property during COVID-
19, 17 consistent with their role in signaling upcoming policy changes. FOMC minutes,
while still exhibiting the timing shift, showed a less dramatic pattern change, reflecting
their function as detailed records rather than active signaling tools.
In summary, communicating about QE and providing forward guidance (UMP) have
evolved into the “new normal” for the Fed since the GFC (Bernanke, 2020). This
transformation represents a significant institutional adaptation, whereby
communication has become increasingly integral to policy implementation, particularly
in low interest rate environments. Furthermore, the frequency of UMP terms remained
elevated during the COVID-19 crisis compared to preceding crises, indicating the
continued importance of these policy tools in the Fed’s response to economic
disruptions. The statistical significance of UMP-related Granger causality across all
communication types quantifies this “new normal.”
5. COVID-19 Pandemic
This section investigates the utilization of COVID-19 terms presented in Table 3. We
analyze these terms alongside UMP terms, contextual uncertainty terms, financial
volatility, and new COVID-19 cases to understand how the Fed’s communication
evolved during the pandemic and how it related to the progression of the health crisis.
Figure 6 delineates the distribution of COVID-19-related terms employed in the Fed’s
primary communications throughout 2020. The figure reveals a striking pattern: the
speeches of the Fed Chairman anticipated the surges in new COVID-19 cases. This
observation warrants careful interpretation, given the delayed commencement of
testing in the US compared to other countries. Nevertheless, the speeches exhibited a
forward-looking perspective, anticipating the potential impacts of the virus on the US
economy, even as it originated in China and before it significantly impacted US
economic activity.
Notably, the Fed Chairman’s speeches serve as a more timely and flexible
communication vehicle than FOMC minutes or FFR announcements, 18 allowing them
to address emerging issues more rapidly. Chairman speeches cover more on-demand
topics and are disseminated more quickly and informally (often broadcast live) than
other communication types, allowing for considerations related to public health,
politics, or foreign affairs that are less emphasized in other communication formats.
16
F test : 10.54 to 17.82, p value < 0.01.
17
F test: 17.82, p value < 0.001.
18 This difference arises from the distinct communication goals and review processes associated with each
text: FFR announcements (rigorously reviewed, formal statements), minutes (internal dialogues among
committee members), and Chairman speeches (intended for broader audience comprehension, not
necessarily focused on interest rate decisions). Our texts ignore questions from the public or journalists.
21
Figure 6 presents two types of COVID-19 waves: one based on new COVID-19 cases
from medical statistics and the other depicting the intensity of COVID-19-related terms
in the Fed’s communications, based on the COVID-19 lexicon presented in Table 3.
Visually, Fed communication about the virus appears to precede new case waves, 19
though we are cautious not to claim any causal relationship between these patterns.
Figure 6. COVID-19 and Fed Communication
4.5 120000
4
100000
3.5
3 80000
2.5
60000
2
1.5 40000
1
20000
0.5
0 0
Announcements Minutes Speeches New Cases
Notes: The shaded areas represent the word-counting indicator for each communication type based on
the COVID-19 lexicon presented in Table 3 (left axis). The dashed line represents the number of new
COVID-19 cases in the US (right axis).
Source: Johns Hopkins University Center for Systems Science and Engineering (CSSE).
The COVID-19 virus’s magnitude and severity were rapidly acknowledged and
communicated to the public by the Fed, primarily through FFR announcements and
Chairman speeches. FFR announcements utilized more COVID-19-related terms than
other communication types and contributed more to the initial communication wave
than speeches, albeit with a slight lag. The decline in the intensity of COVID-19 terms
in Chairman speeches in the second quarter of 2020 correlates with a decline in positive
sentiment, as reported in Figure 1. This decline aligns with the increased emphasis on
social welfare in the Chairman’s speeches during this period, as reported in Figure 4.
Consequently, both the topics and sentiments of the Chairman’s speeches were
influenced by the COVID-19 outbreak and its broader economic implications.
Interestingly, the decline in the frequency of COVID-19-related terms after 2020:Q3
explains the SentiWords sentiment increase in Figure 1 (Panels A7, B7, and D7), despite
the rise in new cases during that quarter. This pattern suggests that the Fed’s
communications adopted a more positive tone in the later phases of the pandemic,
19 Granger causality tests also confirm this finding, but given the few observations available, the results
are not reported and are available upon request.
22
potentially to bolster market and public confidence as the economic recovery began to
take shape.
In summary, the waves of Fed communications about the COVID-19 crisis
anticipated those of new COVID-19 cases. The Chairman’s speeches communicated
about the first wave of COVID-19 earlier than other communication types (FFR
announcements and FOMC minutes), affirming that speeches, being less scripted, allow
for quicker communication on emerging issues. This indicates the Fed’s early
recognition of the severity and magnitude of the COVID-19 pandemic and its related
economic implications.
Figure 7 compares new COVID-19 cases in the US with word-counting indicators
based on UMP and COVID-19 lexicons, providing insight into the relationship between
pandemic developments and the Fed’s policy communications.
Figure 7. COVID-19 and UMP Terms in Main Fed Communications
2 120000
1.8
100000
1.6
1.4
80000
1.2
1 60000
0.8
40000
0.6
0.4
20000
0.2
0 0
COVID-19 Terms UMP Terms VIX New Cases
Notes: The black line represents the word-counting indicator based on the COVID-19 lexicon presented
in Table 3. The dashed line represents the number of new COVID-19 cases in the US (right axis). The
green line represents the VIX (volatility index).
Sources: Bloomberg and Johns Hopkins University Center for Systems Science and Engineering (CSSE).
The figure also incorporates the VIX to capture financial market volatility during the
pandemic, allowing us to examine how the uncertainty created by the virus outbreak
related to the Fed’s communications about UMP and COVID-19.
In Figure 7, the timing of Fed communications about COVID-19 is compared to the
emergence of new COVID-19 cases and financial market volatility, showing that the
Fed’s use of COVID-19-related terms predated the first wave of cases in the US and
coincided with increased VIX values. UMP-related terms and actions in response to the
COVID-19 pandemic helped mitigate this volatility, suggesting a stabilizing effect of
the Fed’s policy interventions on financial markets.
23
The VIX exhibited a remarkable surge with the onset of the COVID-19 outbreak in
China and several other countries, including the US. This spike in market volatility
reflected the unprecedented uncertainty created by the pandemic and its potential
economic consequences. Notably, mentions of COVID-19 in the Fed’s communications
preceded considerations of UMP and the subsequent waves of new cases in the US.
Between May and July 2020, the Fed extensively communicated about unconventional
monetary policies, and during this period, despite a significant increase in new COVID-
19 cases, the Fed’s communications and actions were associated with a moderation in
financial volatility. Following this timeframe, the heightened frequency of UMP-related
terms in the Fed’s communications as the pandemic worsened might have contributed
to stabilizing the VIX, suggesting that clear communication about policy interventions
helped reduce market uncertainty.
Figure 8 illustrates our COVID-19 and UMP word-counting indicators alongside the
LM dictionary of contextual uncertainty terms, allowing us to examine how pandemic-
related communications interacted with expressions of uncertainty in Fed
communications.
Figure 8. COVID-19, UMP, and Uncertainty in Main Fed Communications
70 120000
60 100000
50
80000
40
60000
30
40000
20
10 20000
0 0
COVID-19 Terms UMP Terms Uncertainty Terms New Cases
Notes: The dashed line represents the number of new COVID-19 cases in the US (right axis). The
contextual uncertainty indicator is the number of uncertainty terms, according to LM.
Source: Johns Hopkins University Center for Systems Science and Engineering (CSSE).
The uncertainty in Fed communication exhibits a notable correlation with the
number of UMP terms found in those communications, representing what we interpret
as the “uncertainty effect.” New COVID-19 cases, UMP, and uncertainty show
significant correlation in the Fed’s communications, particularly during the second half
of 2020. This correlation was not as prominent at the beginning of the COVID-19 sample
period, primarily because the sudden virus outbreak caught everyone by surprise,
24
increasing the frequency of uncertainty-related terms before others. The “uncertainty
effect” manifests during crisis periods, necessitating UMP to mitigate market and
economic uncertainty.
Figure 8 presents the proactive influence of uncertainty- and UMP-related terms in
the Fed’s communication concerning COVID-19. The escalation in the utilization of
uncertainty terms seems to precede rises in new virus cases. The correlation between
contextual uncertainty, defined by the LM dictionary, and UMP-related terms from the
UMP lexicon presented in Table 2 is notably positive at 0.44 for the weekly average
communications spanning from 2000 to 2020 (i.e., 1090 observations), indicating a
consistent relationship between expressions of uncertainty and discussions of
unconventional policy measures.
Figure 9 presents our COVID-19 and UMP word-counting indicators alongside the
FSS and the number of new COVID-19 cases in the US, allowing us to examine how
pandemic-related communications interacted with financial stability sentiment.
Figure 9. COVID-19, UMP, and Financial Stability in Main Fed Communications
4 120000
3 100000
2 80000
1 60000
0 40000
-1 20000
-2 0
COVID-19 Terms UMP Terms Financial Stability Sentiment New Cases
Notes: The black line represents the word-counting indicator based on the COVID-19 lexicon presented
in Table 3. The dashed line represents the number of new COVID-19 cases in the US (right axis). The
financial stability index is rescaled to match scale constraints.
Sources: Johns Hopkins University Center for Systems Science and Engineering (CSSE).
Except for the initial phase of the COVID-19 crisis, increases in sentiment associated
with FSS align with rises in new virus cases. The diminishing contribution of the
Chairman’s speeches partially contributes to the decline observed in the FSS toward the
end of the sample period.
Figure 9 shows that the decrease in FSS trails a few weeks behind the upsurges in
both COVID-19- and UMP-related terms in the Fed’s communications. This temporal
25
sequence aligns with theoretical expectations, as discussions and decisions regarding
financial stability typically unfold after shocks to financial stability become apparent.
The successive deteriorations in FSS preceding the increments in the number of new
COVID-19 cases may substantiate the anticipatory nature of the Fed’s discussions on
stabilization policies.
Following the GFC, Fed communication presaged effective UMP implementations,
with the timing and scale of those implementations exhibiting substantial disparities
across crises. Figure 9 shows that Fed communication during the COVID-19 crisis
similarly foreshadowed waves of new COVID-19 cases. The UMP terms were
strategically used to mitigate market volatility and the economic ramifications of
COVID-19.
To formally quantify the dynamic relationships between Fed communication
indicators and related economic variables during the COVID-19 crisis, we implement a
Vector Autoregression (VAR) framework. This approach allows us to establish
temporal precedence and measure the magnitude of relationships between
communication variables.
We estimate the following five-variable VAR model:
𝑌𝑌𝑡𝑡 = 𝐴𝐴0 + 𝐴𝐴1 𝑌𝑌𝑡𝑡−1 + ⋯ + 𝐴𝐴𝑝𝑝 𝑌𝑌𝑡𝑡−𝑝𝑝 + 𝜀𝜀𝑡𝑡 (6)
where 𝑌𝑌𝑡𝑡 = [𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑡𝑡 , 𝑈𝑈𝑈𝑈𝑈𝑈 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑡𝑡 , 𝐹𝐹𝐹𝐹𝐹𝐹𝑡𝑡 , 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑡𝑡 , 𝑉𝑉𝑉𝑉𝑉𝑉𝑡𝑡 ], representing
COVID-19 terms frequency, UMP terms frequency, financial stability sentiment,
COVID-19 case changes, and market volatility, respectively.
Prior to estimation, we conduct stationarity tests using both Augmented Dickey-
Fuller and KPSS tests. COVID-19 cases and VIX are first-differenced to ensure
stationarity. We select the optimal lag order using the Bayesian Information Criterion
to balance model fit with parsimony.
Table 4. VAR Estimates
Dependent variable Independent variable Coefficient Std error t-test p value
COVID-19 Cases Constant 0.281 0.027 10.2 0.00015
COVID-19 Cases COVID-19 Cases (t-1) 1.141 0.041 27.5 1.19E-06
FSS VIX (t-1) -0.833 0.152 -5.4 0.00278
FSS COVID-19 Cases (t-1) -0.549 0.152 -3.6 0.0154
FSS FSS (t-1) 0.669 0.21 3.1 0.0246
FSS COVID-19 Terms (t-1) 0.58 0.206 2.8 0.037
Table 4 presents significant coefficient estimates from our VAR model, showing how
each communication variable predicts subsequent changes in others. We find
statistically significant predictive relationships from COVID-19 terms to UMP terms,
indicating that discussions of the pandemic systematically preceded discussions of
unconventional monetary policy. Similarly, UMP terms positively predict subsequent
financial stability sentiment, suggesting that communication about unconventional
policy measures contributed to improved financial stability perceptions.
26
Our analysis further reveals that the contextual uncertainty articulated in the Fed’s
communications adeptly foreshadowed the waves of COVID-19. Ultimately, the dip in
sentiment associated with the Fed’s communications on financial stability generally
anticipated the surge in the number of new COVID-19 cases. These patterns suggest
that the Fed’s communications incorporated forward-looking assessments of both the
pandemic’s progress and its economic implications, reflecting the institution’s capacity
to integrate diverse sources of information into its policy deliberations and
communications.
6. Financial Stability
UMP and uncertainty-related terms play pivotal roles in influencing financial stability
and volatility. This section delves into examining FSS and contextual uncertainty,
explaining their respective dynamics concerning the conventional monetary policy
instrument (FFR) and financial market volatility as represented by the VIX.
Figure 10 compares FSS, UMP-related terms, and uncertainty-related terms extracted
from FFR announcements with financial market volatility, providing insight into how
the Fed’s communications about these issues relate to market conditions.
Figure 10. Financial Stability and FFR Announcements
9.0 70
7.0 60
5.0 50
3.0 40
1.0 30
-1.0 20
-3.0 10
Financial Stability Sentiment UMP Terms Uncertainty Terms FFR VIX
Notes: The gray shaded areas represent NBER recession periods. The right axis indicates the VIX level.
Figure 10 shows a reduction in the use of uncertainty-related terms during the
pandemic, particularly in comparison to previous crises. This pattern may indicate a
deliberate communicative strategy by the Fed to manage market perceptions and
reduce perceived uncertainty, even as the economic situation deteriorated.
The Fed appears to have strategically used fewer uncertainty-related terms in its FFR
announcements during the COVID-19 crisis than in previous years or crises. This
27
reduction in contextual uncertainty may suggest a communication strategy that aims to
mitigate market uncertainty and volatility, especially during periods of heightened
concern due to the pandemic. Unlike the dot-com and GFC scenarios, there was no
pronounced decline in the FSS during the COVID-19 crisis. This is attributable to the
unprecedented and rapidly evolving nature of the crisis and related early interventions,
leading to better-managed uncertainties through more coordinated policy responses.
During the dot-com and GFC crises, as well as the COVID-19 crisis, terms associated
with UMP played a pivotal role in curbing market volatility. In these critical periods,
FFR announcements detailing the implementation of UMP measures had a tangible
impact on reducing the VIX. Notably, FSS demonstrated a close association with the
FFR level, with a decline in FSS corresponding to a decrease in the FFR, suggesting a
consistent relationship between financial stability sentiment and conventional policy
adjustments.
Figure 11 mirrors the trends observed in Figure 10, with the focus shifted to FOMC
minutes instead of FFR announcements, allowing us to examine how the more detailed
policy discussions in minutes relate to financial stability and market conditions.
Figure 11. Financial Stability and FOMC Minutes
70
7.0
60
5.0 50
40
3.0
30
1.0
20
-1.0
10
-3.0 0
Financial Stability Sentiment UMP Terms Uncertainty Terms FFR VIX
Notes: The gray shaded areas represent NBER recession periods. The right axis indicates the VIX level.
There is a discernible uptick in the utilization of uncertainty-related terms in FOMC
minutes preceding each crisis. However, an intriguing pattern emerges in Figure 11,
akin to the observations in Figure 10, wherein there is a clear inclination to diminish the
use of uncertainty-related words during crisis periods. The VIX exhibits a negative
correlation with FSS, and both precede UMP terms. This chronological sequence
typically signals the implementation of UMP measures to stabilize markets, allaying
concerns about financial stability and subsequently reducing financial volatility.
28
Given that the FOMC minutes offer intricate insights into the monetary policy
committee’s perspectives on the immediate policy stance and the economic outlook for
the US, they serve as an early conduit for conveying FSS and UMP terms compared to
FFR announcements. Notably, FFR increases align with elevated FSS levels, except for
the period from 2012 to 2015, when UMP communications and actions propelled the
FSS upwards, suggesting that unconventional policies can have distinct effects on
financial stability sentiment compared to conventional rate adjustments.
Figure 12 juxtaposes Fed Chairman speeches, FFR announcements, and financial
volatility, providing insight into how the Chairman’s communications relate to market
conditions and conventional policy measures.
Figure 12. Financial Stability and Monetary Policy in Fed Chairman’s Speeches
7.0 70
6.0
60
5.0
4.0 50
3.0
40
2.0
30
1.0
0.0 20
-1.0
10
-2.0
-3.0 0
Financial Stability Sentiment UMP Terms Uncertainty Terms FFR VIX
Notes: The gray shaded area represents NBER recession periods. The right axis indicates the VIX level.
The FSS expressed in speeches exhibits a lesser predictive capacity for future FFR
movements than announcements and minutes do. This may reflect the broader scope
and more varied objectives of speeches, which often address issues beyond immediate
policy concerns.
Chairman speeches may convey a negative FSS even when the FFR increases, a
phenomenon not typically observed in FFR announcements or FOMC minutes (refer to
Figures 13 and 14). This divergence could reflect the Chairman’s role in providing
context and nuance to policy decisions, including acknowledging risks and challenges
even as conventional policy tightens.
Drawing a comparison between two pivotal periods—the dot-com to the GFC period
(P1) and the GFC to the COVID-19 crisis period (P2)—provides valuable insights.
During P1, a scarcity of UMP-related terms in Fed Chairman speeches is observed. In
contrast, P2 marks the establishment of a “new normal,” marked by a noteworthy
correlation between UMP- and uncertainty-related terms, a connection absent in P1.
29
This correlation extends to the relationship between the VIX and uncertainty-related
terms, exhibiting a lesser degree of correlation in P1 than in P2, suggesting a structural
change in how uncertainty and unconventional policy are communicated and
perceived after the GFC.
Examining Figure 12 reveals a reduction in uncertainty-related terms in Fed
Chairman speeches during the COVID-19 crisis in comparison to previous years and
crises. However, this contrast is less pronounced in announcements and minutes. The
inherent variability in speeches can be attributed to their broad scope and varied
objectives, coupled with their somewhat less controlled nature than announcements
and minutes.
The communication pertaining to UMP transpired after volatility peaks in both the
GFC and COVID-19 crises. Figures 14 and 15 underscore the interventionist approach
adopted by the Fed, aligning with the central bank practices in many developed
countries. Following each FFR decrease, UMP communication, typically succeeded by
actions, serves as a compensatory measure for the central bank’s constraint in utilizing
the nominal interest rate—the primary policy instrument—at the ZLB.
Figure 13 amalgamates the three communication types employed by the Fed,
providing a comprehensive overview of how different communication channels
collectively relate to financial stability and market conditions.
Figure 13. Financial Stability and Monetary Policy
7.0 70
6.0
60
5.0
4.0
50
3.0
2.0 40
1.0
30
0.0
-1.0
20
-2.0
-3.0 10
FSS UMP Terms UMP Sentiment FFR VIX Uncertainty Terms
Notes: The gray shaded areas represent NBER recession periods. The right axis indicates the VIX and
uncertainty terms levels.
Figure 13 substantiates the assertion that, during crises, the Fed diminishes the
sentiment associated with UMP measures in tandem with a reduction in the frequency
of uncertainty-related terms in its communications. This coordinated approach may
30
reflect a strategy to balance policy transparency with the need to maintain market
confidence during periods of stress.
The pattern of declines in FSS generally anticipating increases in the VIX finds an
exception in the COVID-19 crisis. This aberration may be explained by the inherently
unpredictable nature of the COVID-19 crisis compared to the more foreseeable dot-com
and GFC crises, which had more conventional economic origins. The pandemic’s
unprecedented combination of supply and demand shocks, coupled with public health
dimensions, created a unique context for financial stability considerations.
A paradigm shift occurred in the aftermath of the GFC, defining a new normal for
the Fed’s communications. This shift was marked by an increasing reliance on
discussions centered around UMP tools, particularly forward guidance measures.
These UMP deliberations featured high levels of contextual uncertainty. However, the
established trajectory of this new normal encountered disruptions during the COVID-
19 crisis, prompting the Fed to adopt a communication strategy that entailed a reduced
utilization of uncertainty-related terms in their UMP communications.
While the preceding discussions predominantly delved into matters of financial
stability, it is imperative to acknowledge that the Fed’s communications equally
addressed concerns pertaining to economic stability. Figure 14 provides a
comprehensive overview of the word-counting indicators, as previously discussed,
specifically focusing on the exchange rate.
Figure 14. Financial Stability, NEER, and FFR Announcements
9.0 1.2
1.15
7.0
1.1
5.0
1.05
3.0 1
0.95
1.0
0.9
-1.0
0.85
-3.0 0.8
Financial Stability Sentiment UMP Terms Uncertainty Terms NEER
Notes: The gray shaded areas represent NBER recession periods. The NEER corresponds to the amount
of US dollars needed to purchase foreign currency (right axis). The FSS, UMP, and uncertainty terms are
related to FFR announcements.
Figure 14 illustrates a correlation between the NEER and the frequency of
uncertainty-related terms within FFR announcements. The figure suggests that UMP
communications and actions generally lead to a reduction in the NEER, although there
31
is an increase during the tapering period. This relationship between UMP
communications and exchange rate movements aligns with findings by Belke et al.
(2017), who analyze the influence of QE on exchange rates through interest rate
differentials.
Figure 15 establishes a connection between the sentiment indicators discussed earlier
and another metric of economic stability: the unemployment rate. It highlights an
almost consistently inverse relationship between sentiment and the unemployment
rate. Positive aggregate sentiment tends to coincide with a decline in the unemployment
rate, while a shift from positive to negative sentiment typically aligns with an increase
in unemployment. This relationship suggests that Fed communication sentiment may
reflect or anticipate labor market conditions, though we caution against inferring
causality given the complex dynamics involved, especially during the COVID-19 crisis.
Figure 15. Sentiment and Unemployment
4.0 14
3.0 12
2.0
10
1.0
8
0.0
6
-1.0
-2.0 4
-3.0 2
Average Sentiments Unemployment Rate
Notes: The gray shaded areas represent NBER recession periods. The right axis indicates the
unemployment rate levels. The average sentiment aggregate contains an equally weighted average of
sentiments according to LM (score and polarity), Hu&Liu (polarity), Jockers (polarity), NRC (polarity),
SentiWords (polarity), UMP (score), and financial stability (score) sentiments. 20
In contrast to previous crises, the COVID-19 crisis began with a rapid surge in the
US unemployment rate, soaring from 3.5% to 13% within a brief period (January to May
2020). Subsequently, the unemployment rate declined to lower levels after May 2020,
settling around 6%. The abrupt and short-lived shock to aggregate sentiment was even
briefer than the corresponding impact on the unemployment rate, suggesting a crisis-
20 To achieve a balanced aggregated indicator for each communication type, we weigh this average
sentiment aggregate for FFR announcements more than for FOMC minutes, which in turn are weighted
more than for Chairman speeches.
32
specific communication strategy geared towards conveying optimism during crises to
support economic confidence.
Figure 15 reaffirms the observation that the Fed adopted a communication strategy
aimed at promoting positive sentiments during the COVID-19 crisis, potentially to
counterbalance the extraordinary economic disruption caused by the pandemic and
support confidence in the recovery process.
In summary, we demonstrate that the FSS is communicated through FFR
announcements and FOMC minutes but is not as significantly expressed in speeches.
While rises in FSS typically herald increases in the FFR, this pattern is disrupted when
UMP terms are introduced into the Fed’s communications, often indicating actions
taken to enhance financial stability.
In addition, we have shown that the positive aggregated sentiment within the Fed’s
primary communications aligns with a decrease in unemployment. With few
exceptions, this relationship holds during periods of substantial UMP measures.
Concurrently, the NEER correlates with the degree of contextual uncertainty in the
Fed’s communications, highlighting the multifaceted relationships between
communication patterns, economic outcomes, and financial market variables.
7. Policy Implications
The Fed executed a more extensive array of unconventional monetary policies during
the COVID-19 pandemic than it did during the dot-com and GFC episodes. This rapid
deployment within a concise timeframe was necessitated by the sudden surge in
adverse economic shocks triggered by COVID-19 restrictions. The Fed’s proficiency in
crisis-specific communication and UMP tools, acquired during the dot-com and GFC
crises, likely played a role in comprehending and addressing the communication
challenges posed by the COVID-19 crisis. The success of these UMP measures hinged
on clear, transparent communications, and active engagement with financial markets
and the public.
Our study, employing both supervised and unsupervised learning methods,
illustrates that the Fed’s communications during the COVID-19 crisis significantly
deviated from those of preceding crises in terms of conveyed sentiments and
emphasized topics. Comparative analysis of the terms, sentiments, and topics within
the Fed’s communications related to COVID-19 and financial data confirms the
adoption of a distinct communication strategy during the COVID-19 crisis, differing
from strategies employed during the GFC and dot-com crises. We posit that these
refined communication tactics may enhance the Fed’s crisis management capabilities
and contribute to more effective policy implementation during periods of economic
stress.
Our analysis identifies this communication policy as one that instilled optimism in
the public during the height of the pandemic while addressing (and implementing)
unconventional monetary policies earlier in the crisis than observed in previous
33
instances (Figure 5, Panel D). The significance of these policies in mitigating risks and
uncertainties is suggested by Figures 10 and 11. Another critical observation
underscores the Fed’s forward-looking narrative and its adept use of communication to
convey a determined sentiment and rationalize UMP measures preceding each wave of
the virus or worsening financial conditions due to pandemic spillovers.
In summary, our findings demonstrate that communications explaining the adopted
policies and emergency programs succeeded in portraying them as effective tools
supporting economic recovery. The Fed employed a distinct communication strategy
tailored to the COVID-19 crisis, projecting reduced uncertainty and less negative
sentiment to the public while advocating for UMP measures to manage the crisis. 21 This
strategic approach aimed at conveying optimism without compromising transparency.
The Fed’s timely communications, coupled with decisive actions, effectively
contributed to stabilizing financial markets in conjunction with UMP and fiscal policy.
There are several potential reasons behind the Fed’s communication shifts over time.
Ball (1994) highlights the tension between transparency and time inconsistency,
suggesting that the Fed’s enhanced clarity might aim to strengthen its commitment to
long-term goals. Additionally, Bernanke et al. (1999) emphasize the role of managing
expectations, implying that the Fed’s communication could target anchoring inflation
expectations for greater policy effectiveness. Furthermore, Woodford (2003)
underscores the importance of central bank credibility, suggesting that the Fed’s
transparency might seek to bolster public trust in its policy decisions.
Moving beyond established theories, careful interpretation of the specific changes
implemented—such as the increased frequency of economic projections or the adoption
of “dot plots”—could reveal nuanced motivations. For instance, employing dot plots,
as analyzed by Wright (2012), might aim to enhance market understanding of
individual policymakers’ views and foster consensus building within the FOMC.
Finally, acknowledging the limitations of any single interpretation, future research, as
advocated by Orphanides and Williams (2005), could employ econometric techniques
to rigorously evaluate the impact of the Fed’s communication changes on market
expectations and economic outcomes, providing further clarity on the central bank’s
communication motives and their effectiveness.
8. Conclusion
This study provides a comprehensive examination of central bank communication over
the past two decades, with a particular focus on the COVID-19 pandemic. Our analysis
reveals distinct features in Fed communications during the COVID-19 crisis compared
to prior crises, including the GFC and the dot-com bubble. By correlating essential
communication terms, sentiments, and topics conveyed by the Fed to COVID-19 case
21
The significant Granger causality from UMP terms to Fed assets during COVID-19 across all communication
types (F = 10.54-17.82, p < 0.01) quantifies this distinct strategy.
34
numbers and financial data, we distinguish the adoption of a specific communication
strategy during the COVID-19 crisis.
Throughout the COVID-19 pandemic, Fed speeches highlighted topics related to
social welfare, diverging from the emphasis on policy interventions seen in FFR
announcements and minutes. Discussions around COVID-19 and UMP often delved
into market volatility, uncertainty, and financial stability. The sentiment of Fed
communication exhibited significant changes during the COVID-19 crisis compared to
the post-GFC period. Post-GFC, discussions on UMP became a “new normal” in the
Fed’s minutes and Chairman speeches. Additionally, we demonstrate that a negative
FSS typically precedes conventional monetary policy accommodation, with exceptions
during the ZLB period when conventional policy tools were constrained.
The COVID-19 crisis induced structural shifts in the content of Fed communication,
suggesting the implementation of a distinct communication policy compared to the dot-
com and GFC crises. This adaptive approach to crisis communication reflects the Fed’s
institutional learning and its evolving understanding of how communication can
enhance policy effectiveness during periods of economic stress.
Our research contributes to the growing literature on central bank communication
strategies (Hansen et al., 2018; Gardner et al., 2022; Bennani and Romelli, 2024) by
documenting specific changes in communication content, sentiment, and timing during
different crisis periods. The findings suggest that central banks can adapt their
communication approaches to the unique characteristics of different economic crises,
potentially enhancing policy effectiveness.
Future research avenues may explore the impact of these communication indicators
on changes in interest rates, market-based expectations, exchange rates, and various
asset prices within short time intervals (e.g., 30 minutes) surrounding policy
announcements, building on the work of Gürkaynak et al. (2020). Such analysis could
provide a more precise identification of the causal effects of specific communication
strategies on financial markets and economic expectations. Additionally, extending this
analysis to other major central banks would enable comparative studies of
communication strategies across different institutional contexts and economic
environments.
References
Ball, L. (1994). Credibility and Time Inconsistency. American Economic Review, 84(3), 161-
170.
Belke, A., D., Gros, and T. Osowski (2017). “The Effectiveness of the Fed’s Quantitative
Easing Policy: New Evidence Based on International Interest Rate Differentials,”
Journal of International Money and Finance 73(PB): 335-349.
35
Benchimol, J., S. Kazinnik, and Y. Saadon (2020). “Communication and Transparency
Through Central Bank Texts,” Paper presented at the 132nd Annual Meeting of the
American Economic Association, January 3–5, 2020, San Diego, CA, United States.
Benchimol, J., S. Kazinnik, and Y. Saadon (2022). “Text Mining Methodologies with R:
An Application to Central Bank Texts,” Machine Learning with Applications 8: 100286.
Bennani, H., & Romelli, D. (2024). “Exploring the informativeness and drivers of tone
during committee meetings: the case of the Federal Reserve.” Journal of International
Money and Finance 148: 103161.
Bernanke, B. S., Laubach, T., Mishkin, F. S., & Posen, A. S. (1999). Inflation Targeting:
Lessons from the International Experience. Princeton University Press.
Bernanke, B. S. (2020). “The New Tools of Monetary Policy,” American Economic Review
110(4): 943–983.
Bianchi, F., R. Faccini, and L. Melosi (2020). “Monetary and Fiscal Policies in Times of
Large Debt: Unity is Strength,” CEPR Discussion Paper #14720.
Blei, D. M., A. Y. Ng, and M. I. Jordan (2003). “Latent Dirichlet Allocation,” Journal of
Machine Learning Research 3: 993–1022.
Blinder, A. S., Ehrmann, M., Fratzscher, M., De Haan, J., & Jansen, D.-J. (2008). “Central
Bank Communication and Monetary Policy: A Survey of Theory and Evidence,”
Journal of Economic Literature 46(4): 910-945.
Born, B., Ehrmann, M., and Fratzscher, M., 2014. “Central Bank Communication on
Financial Stability,” Economic Journal, 124(577): 701-734.
Carney, M. (2020). “The Grand Unifying Theory (and Practice) of Macroprudential
Policy,” Speech at University College, London, on March 5, 2020, Bank of England.
Craig, B. R., T. Phelan, and J.-P. Siedlarek (2021). “Modeling Behavioral Responses to
COVID-19,” Economic Commentary Federal Reserve Bank of Cleveland 2021(05): 1–6.
Correa, R., K. Garud, J. M. Londono, and N. Mislang (2021). “Sentiment in Central
Banks’ Financial Stability Reports,” Review of Finance 25(1): 85–120.
Daly, M. C. (2020). “Is the Federal Reserve Contributing to Economic Inequality?”
FRBSF Economic Letter 2020(32): 1–7.
Erasmus, R. and H. Hollander (2020). “A Forward Guidance Indicator for the South
African Reserve Bank: Implementing a Text Analysis Algorithm,” Studies in
Economics and Econometrics 44(3): 41–72.
Chetty, R., J. N. Friedman, N. Hendren, M. Stepner, Opportunity Insights Team, C.
Baker, H. Barnhard, M. Bell. G. Bruich, T. Chelidze, L. Chu, W. Cineus, S. Devlin-Fol,
2024. “The Economic Impacts of COVID-19: Evidence from a New Public Database
Built Using Private Sector Data,” Quarterly Journal of Economics, 139(2): 829-889.
36
Gardner, B., Scotti, C., & Vega, C. (2022). “Words speak as loudly as actions: Central
bank communication and the response of equity prices to macroeconomic
announcements.” Journal of Econometrics, 231(2): 387-409.
Guerrieri, V., G. Lorenzoni, L. Straub, and I. Werning (2022). “Macroeconomic
Implications of COVID-19: Can Negative Supply Shocks Cause Demand
Shortages?,” American Economic Review 112(5): 1437-1474.
Gürkaynak, R. S., B. Kisacikoğlu, and J. H. Wright, 2020. “Missing Events in Event
Studies: Identifying the Effects of Partially Measured News Surprises,” American
Economic Review 110(12): 3871-3912.
Haldane, A. and M. McMahon (2018). “Central Bank Communications and the General
Public,” AEA Papers and Proceedings 108: 578–583.
Hansen, S., & McMahon, M. (2016). “Shocking language: Understanding the
macroeconomic effects of central bank communication.” Journal of International
Economics, 99: S114-S133.
Hansen, S., McMahon, M., & Prat, A. (2018). “Transparency and deliberation within the
FOMC: A computational linguistics approach.” Quarterly Journal of Economics, 133(2):
801-870.
Hansen, S., M. McMahon, and M. Tong (2019). “The Long-Run Information Effect of
Central Bank Communication,” Journal of Monetary Economics 108(C): 185–202.
Henry, E. (2008). “Are Investors Influenced by How Earnings Press Releases are
Written?” Journal of Business Communication 45(4): 363–407.
Loughran, T. and B. McDonald (2011). “When is a Liability not a Liability? Textual
Analysis, Dictionaries, and 10-Ks,” Journal of Finance 66(1): 35–65.
Loughran, T. and B. McDonald (2016). “Textual Analysis in Accounting and Finance: A
Survey,” Journal of Accounting Research 54(4): 1187–1230.
Nyman, R., S. Kapadia, and D. Tuckett (2021). “News and Narratives in Financial
Systems: Exploiting Big Data for Systemic Risk Assessment,” Journal of Economic
Dynamics and Control 127(C): 104119.
Orphanides, A., & Williams, J. C. (2005). “The Evolution of Monetary Policy
Communication,” Journal of Economic Perspectives, 19(4): 233-250.
Röder, M., Both, A., and Hinneburg, A. (2015). “Exploring the Space of Topic Coherence
Measures.” Proceedings of the Eighth ACM International Conference on Web Search and
Data Mining, 399-408.
Woodford, M. (2003). Interest and prices: Foundations of a theory of monetary policy.
Princeton University Press.
Wright, R. (2012). The Federal Open Market Committee’s new “dot plots”:
Communication and policy expectations. Economic Policy, 27(70), 329-365.
37
File and source
- File
- federal-reserve-communication-and-the-covid-19-pandemic.pdf
- Size
- 1,985,488 bytes
- SHA-256
- c3c6321aff3e2326c66588075f82bc42f35e463ea5d5b7767b78cb8603cb8062
- Original
- arxiv.org