04
From release architecture to public understanding.
Worcester College
Communication as Text Data.
Measurement, public attention, strategic disclosure, climate translation, and understandability.
From Release to Reception
- Measurement: make the missing middle observable.
- Architecture: ask how bundling changes attention.
- Unexpected content: isolate the first public reading.
- Public engagement: distinguish posting from exposure.
Communication becomes data only after the economic object, source, and empirical unit are fixed.
day4_00_opening.tex · line 7 · D4-F002
01
Central Banking · Lecture 4 · Worcester College
Part 01
Measure communication.
Text and digital traces reveal the missing middle
Measure the objectIdentify the mechanismRead the evidence
Standard Data Miss the Middle
\[\underbrace{i_t,\ \pi_t,\ y_t,\ r_t,\ q_t}
_{\text{standard macro-finance data}}
\qquad\Longrightarrow\qquad
\underbrace{\text{beliefs, expectations, prices}}
_{\text{final outcomes}}.\]
The empirical gap lies between the released decision and the observed outcome.
day4_01_measurement.tex · line 10 · D4-F004
Mechanism · read left to right
Transmission Runs Through People
\[\text{communication}
\longrightarrow
\text{attention}
\longrightarrow
\text{interpretation}
\longrightarrow
\text{belief updating}.\]
- A message cannot update a belief if it is not noticed.
- Exposure does not guarantee a shared interpretation.
- Text and digital traces make parts of this information environment observable.
Communication data measure the mechanism that endpoint data leave hidden.
day4_01_measurement.tex · line 30 · D4-F004
Conceptual illustration · not empirical evidence
Attention and Interpretation Fill the Gap
Teaching point
Reach is not impact: attention and interpretation sit between publication and outcomes.
Use the next slides for the documentary evidence.
day4_01_measurement.tex · line 54 · D4-F004 — pedagogical visual, no new empirical claim
Communication Expands the Information Set
\[\mathcal I_t
=
\mathcal I_t^{macro}
\cup\mathcal I_t^{market}
\cup\mathcal I_t^{communication}
\cup\mathcal I_t^{public}.\]
\[\mathcal I_t^{macro},\qquad \mathcal I_t^{market}.\]
\[\mathcal I_t^{communication},\qquad \mathcal I_t^{public}.\]
A policy message matters only after it enters an audience's information set.
day4_01_measurement.tex · line 65 · D4-F005
Economic object · define before estimating
Text Extends Policy Measurement
| Object | Standard data | Text or digital data |
|---|---|---|
| Policy decision | yes | yes |
| Public attention | rarely | yes |
| Public interpretation | rarely | yes |
| Disclosure choices | rarely | yes |
| Institutional admissibility | rarely | yes |
| Asset-price response | yes | yes |
The added value is access to reception, selection, and admissibility—not another policy-rate series.
day4_01_measurement.tex · line 90 · D4-F006
Match the Corpus to the Behaviour
| Source type | Examples | Economic object |
|---|---|---|
| Policy and official text | statements, minutes, speeches, transcripts, fiscal reports | priorities, guidance, disclosure |
| Firm disclosures | earnings calls, annual reports, filings, presentations | expectations, risk, investment plans |
| News and media | newspapers, newswires, analyst reports, blogs | information flow, sentiment, uncertainty |
| Social and public text | Twitter/X, Reddit, survey comments, reviews | interpretation, disagreement, beliefs |
| Attention traces | Google Trends, Wikipedia views, web traffic | search and reference behaviour |
Availability is not validity: source choice follows the economic object.
day4_01_measurement.tex · line 111 · D4-F007
Economic object · define before estimating
Rules Turn Sources into Variables
\[\text{raw source }s_i
\xrightarrow{\ \widehat\phi\ }
\widehat x_i
\xrightarrow{\ \text{econometric design}\ }
Y_i.\]
Unstructured material becomes evidence only through an explicit measurement rule.
day4_01_measurement.tex · line 136 · D4-F008
Interpretive boundary
Sources Reveal Different Stages
| Source | Measured object |
|---|---|
| Search or page views | attention and information acquisition |
| Official posts | engagement with public communication |
| Public text | interpretation, beliefs, disagreement |
| Institutional text | disclosure, omission, policy priorities |
No single source measures attention, interpretation, and disclosure at once.
day4_01_measurement.tex · line 158 · D4-F008
Economic object · define before estimating
One Document, Many Measurements
| Measurement | Variable | Question |
|---|---|---|
| Tone | hawkish/dovish score | how restrictive is the communication? |
| Topic | inflation, labour, risk | what is the discussion about? |
| Surprise | residualised tone | what was not predictable before release? |
| Coverage | statement–transcript gap | what was omitted from the focal release? |
| Linkage | mandate category | how does an issue become policy-relevant? |
The same minutes become a different variable when the research question changes.
day4_01_measurement.tex · line 177 · D4-F009
Economic object · define before estimating
The document is not the variable
\[\text{document}\neq\text{variable}.\]
Start with the quantity to be measured; only then select the text method.
day4_01_measurement.tex · line 197 · D4-F009
Economic object · define before estimating
Auditable Measurement Has a Workflow
\[\begin{aligned}
\text{source}
&\longrightarrow\text{cleaned corpus}
\longrightarrow\text{measurement rule}\\
&\longrightarrow\text{structured variable}
\longrightarrow\text{econometric design}.
\end{aligned}\]
Measurement is not preprocessing; it is an explicit empirical stage.
day4_01_measurement.tex · line 213 · D4-F010
Interpretive boundary
Each Stage Has Its Own Failure
| Stage | What happens |
|---|---|
| Source | collect posts, speeches, minutes, transcripts, searches, or views |
| Cleaning | remove retweets, bots, false matches, duplicates, and missing metadata |
| Measurement | classify tone, topic, interpretation, linkage, or omitted content |
| Aggregation | move to event, meeting, institution-year, or release-window units |
| Validation | compare human labels, test placebos, and propagate uncertainty |
A credible variable records how raw material changed at every stage.
day4_01_measurement.tex · line 233 · D4-F010
Three questions organise the empirical cases
\[\text{communication}
\longrightarrow
\text{measured variable}
\longrightarrow
\text{economic mechanism}.\]
| Research question | Observed source | Economic object |
|---|---|---|
| Who pays attention? | searches, page views, public and official posts | attention and engagement |
| How is the message interpreted? | public text around announcements | direction, disagreement, narrative |
| What becomes part of the record? | statements, minutes, transcripts, speeches | disclosure and admissibility |
The lecture moves from reception to interpretation and then to institutional selection.
day4_01_measurement.tex · line 253 · D4-F011
02
Central Banking · Lecture 4 · Worcester College
Part 02
Architecture shapes attention.
One richer event can still create fewer opportunities to notice
Measure the objectIdentify the mechanismRead the evidence
Institutional change · August 2015
Warsh Replaced Sequence with Bundle
Before
Several release dates
Decision → forecast report + press conference → minutes + vote
Several smaller attention waves
After
One full-release bundle
Decision + minutes + vote + forecast + press conference
One larger focal occasion
The reform changed release architecture, not the existence of policy information.
day4_02_architecture.tex · line 10 · D4-F013
Economic mechanism · two margins move at once
Bundling Moves Two Attention Margins
Intensive margin
A richer event
More content arrives together, raising salience on the bundled day.
Attention per event ↑
Extensive margin
Fewer occasions
One forecast-round opportunity disappears after bundling.
Attention occasions ↓
A larger event can coexist with less attention over the whole forecast round.
day4_02_architecture.tex · line 38 · D4-F014
Conceptual illustration · not empirical evidence
A Bigger Release Can Mean Less Attention
Teaching point
Bundling raises attention per event but removes an attention occasion.
Use the next slides for the documentary evidence.
day4_02_architecture.tex · line 63 · D4-F014 — pedagogical visual, no new empirical claim
Interactive question · which attention margin dominates?
Net Attention: Gain Minus Lost Occasion
\[\Delta A_q=\text{salience gain}-\text{lost occasion}\]
Net attention · hypothetical units
+25
The actual sign must come from the event and forecast-round evidence.
Communication architecture determines both the size and number of attention triggers.
day4_02_architecture.tex · line 75 · D4-F014
Interpretive boundary
Three Traces, Three Attention Behaviours
| Trace | Series | Behaviour measured |
|---|---|---|
| Twitter/X | log unique users posting original BoE/MPC tweets | real-time public discussion |
| Google Trends | UK index for BoE, interest rates, and inflation | active information acquisition |
| Wikipedia | log views for BoE and monetary-policy pages | reference and background checking |
The outcome is revealed attention: discussion, search, and reference behaviour.
day4_02_architecture.tex · line 96 · D4-F015
Abnormal Attention Uses a Baseline
\[\underbrace{A_{e,p}}_{\text{abnormal attention}}
=
\underbrace{Y_{t(e),p}}_{\text{event-day activity}}
-
\underbrace{\bar Y^{\,base}_{e,p}}_{\text{matched baseline}}.\]
The design explains an event spike, not raw platform volume.
day4_02_architecture.tex · line 122 · D4-F016
Match Weekdays; Exclude Nearby Events
| Object | Construction |
|---|---|
| Event activity \(Y_{t(e),p}\) | log users, search index, or log page views on the event date |
| Baseline \(\bar Y^{base}_{e,p}\) | same-platform mean on matching non-event weekdays in the prior 30 days |
| Exclusion | remove days near other MPC events from the baseline |
| Outcome | event-specific abnormal attention \(A_{e,p}\) |
Matching the attention environment separates an announcement spike from ordinary weekly rhythms.
day4_02_architecture.tex · line 144 · D4-F016
Interpretive boundary
Event and Round Analyses Differ
| Unit | Pre-reform | Post-reform |
|---|---|---|
| Event | decision day or Inflation Report day | full-release day or standard meeting |
| Forecast round | decision day plus Inflation Report day | one full-release day |
- Event level: is the bundled day more salient?
- Forecast round: does one bundled day replace attention from two occasions?
The unit of analysis determines whether the intensive or extensive margin is visible.
day4_02_architecture.tex · line 166 · D4-F017
Quarterly Attention Sums Event Spikes
\[CumLogAtt_q
=
\sum_{e\in\mathcal E_q}A_{e,TW}.\]
Digital traces let the researcher count attention by event before aggregating it.
day4_02_architecture.tex · line 190 · D4-F017
Documentary evidence · source-preserving figure
Twitter Peaks on Decision Day
Across \(h=-3,\ldots,+3\), event-day log unique users spike and attention decays sharply after the decision.
day4_02_architecture.tex · line 207 · D4-F018
Documentary evidence · source-preserving figure
Google Search Peaks on Decision Day
The event-day distribution shifts upward relative to adjacent days, consistent with active information acquisition around MPC announcements.
day4_02_architecture.tex · line 219 · D4-F019
Documentary evidence · source-preserving figure
Wikipedia Views Peak on Decision Day
Reference and background-checking activity rises on the MPC event day and recedes afterward.
day4_02_architecture.tex · line 231 · D4-F020
Documentary evidence · source-preserving figure
Quarterly Twitter Attention Shifts Left
Pre: \(N=18\), mean \(2.966\), s.d. \(0.631\). Post: \(N=25\), mean \(1.742\), s.d. \(0.466\); the post mean is below the pre minimum \(1.945\).
day4_02_architecture.tex · line 243 · D4-F021
Documentary evidence · source-preserving figure
The Reform Shifts Levels, Not Trends
Separate OLS trends and 95% bands show no statistically significant within-regime trend; the pattern is consistent with a break at implementation.
day4_02_architecture.tex · line 255 · D4-F022
Documentary evidence · source-preserving figure
Surprises Raise Visibility, Not Persistence
High shock means absolute UKMPD surprise above the sample median. Across 117 MPC events, the larger impact response still decays rapidly.
day4_02_architecture.tex · line 267 · D4-F023
Empirical result · comparison and scope
Bundling Raises Salience, Lowers Attention
\[\text{more attention per event}
\quad\not\Rightarrow\quad
\text{more attention over the round}.\]
| Object | Finding | Interpretation |
|---|---|---|
| Single event | full-release days attract more discussion and search | the remaining event is more salient |
| Forecast round | cumulative Twitter attention falls about \(41\%\) | the lost second occasion dominates |
Communication design changes both salience and the number of opportunities to notice.
day4_02_architecture.tex · line 279 · D4-F024
Super Thursday creates an attention trade-off
\[\text{one larger public occasion}
\quad\text{versus}\quad
\text{two separate attention triggers}.\]
01
Event salience
02
Round exposure
A bigger event-day spike does not replace repeated exposure.
day4_02_architecture.tex · line 306 · D4-F025
Mechanism · two design margins
Architecture Changes Topics and Timing
Communicationarchitecture
=
Topic compositionwhat is bundled
+
Timinghow often attention is triggered
Composition: richer releases carry more forecast and outlook discussion.
Timing: bundling creates a sharper event but removes a separate occasion.
The communication technology changes both content per event and the number of attention opportunities.
day4_02_architecture.tex · line 331 · D4-F026
Interpretive boundary
Attention is not belief or understanding
\[\text{attention trace}
\neq
\text{beliefs}
\neq
\text{understanding}.\]
| Limit | Interpretation |
|---|---|
| Selected users | Twitter/X is a trace, not the whole public |
| Normalised searches | Google Trends measures relative search intensity |
| Bundled reform | timing, minutes, and event frequency changed together |
The credible claim is narrower: architecture changes exposure opportunities.
day4_02_architecture.tex · line 362 · D4-F027
03
Central Banking · Lecture 4 · Worcester College
Part 03
Unexpected words shift readings.
Interpretation can move without approval
Measure the objectIdentify the mechanismRead the evidence
Measure the First Directional Reading
\[\begin{aligned}
\text{BoE communication bundle}
&\longrightarrow \text{attentive-public interpretation}\\
&\longrightarrow \text{possible wider diffusion}.
\end{aligned}\]
| Object | Design choice |
|---|---|
| Setting | scheduled post-2015 MPC bundles |
| Public text | BoE-related Twitter/X discourse around release |
| Window | post-release discourse relative to the pre-release baseline |
| Target | directional interpretation, not volume or generic sentiment |
The outcome is the first public reading—not representative household beliefs.
day4_03_unexpected.tex · line 10 · D4-F029
Interpretive boundary
Classify Direction, Not Sentiment
\[\widehat L_i^{direction}
\in
\{\text{hawkish},\text{dovish},\text{neither}\}.\]
A positive or negative emotional tone is not automatically hawkish or dovish.
day4_03_unexpected.tex · line 41 · D4-F030
Interactive measurement · direction, not volume
PHI Nets Hawkish and Dovish Readings
\[\mathrm{PHI}=\frac{N^{hawk}-N^{dove}}{N^{all}}\]
PHI · all readings fixed at 100
+0.18
Neutral or non-directional posts remain in the denominator.
The index measures the direction of the public reading, not its volume.
day4_03_unexpected.tex · line 60 · D4-F030
Measure the Post-Release PHI Change
\[\Delta\mathrm{PHI}_m
=
\mathrm{PHI}_{m,0\text{--}24h}
-
\mathrm{PHI}_{m,pre48h}.\]
A meeting-level change is more informative than the raw post-release index.
day4_03_unexpected.tex · line 81 · D4-F030
Empirical result · comparison and scope
Unexpected Tone Removes Predictability
\[\operatorname{Tone}^{BoE}_m
=
\alpha+\Pi'I_m+\theta_{qoy}+u_m,
\qquad
UnexpectedTone_m=\widehat u_m.\]
| Term | Meaning |
|---|---|
| \(\operatorname{Tone}^{BoE}_m\) | hawk–dove tone of the minutes |
| \(I_m\) | pre-release OIS expectations, rate surprise, macro data, gilt and FX moves |
| \(UnexpectedTone_m\) | tone not predicted by that information set |
Residualised tone is the text analogue of a policy surprise.
day4_03_unexpected.tex · line 100 · D4-F031
Reduced-form transmission test · hold the rate surprise apart
Separate Language from Rate Surprises
\[\begin{aligned}
\Delta \mathrm{PHI}_m
&= \beta_0 + \beta_1\,UnexpectedTone_m + \beta_2\,RateSurprise_m \\
&\quad + \boldsymbol{\beta}'W_m + \theta_{qoy} + \varepsilon_m
\end{aligned}\]
β₁Language channelDoes an unexpected statement reading move the public reading?
β₂Rate-surprise channelSeparates the conventional policy surprise from language.
The coefficient is reduced-form: unexpected content may combine language with private central-bank information.
day4_03_unexpected.tex · line 129 · D4-F032
Measurement boundary · direction is not evaluation
A hawkish reading can be supportive or critical
Interpretation
“The Bank sounds hawkish.”
Direction: hawkish, dovish, or neither.
Evaluation
“That stance is right—or a mistake.”
Approval: supportive, critical, or neither.
\[Approval_{m,w}=\frac{N^{support}_{m,w}-N^{critical}_{m,w}}{N^{all}_{m,w}}\]
Persuasion requires approval to move, not only the directional reading.
day4_03_unexpected.tex · line 151 · D4-F033
Conceptual illustration · not empirical evidence
Direction Does Not Determine Approval
Teaching point
Hawkish versus dovish is a reading; support versus criticism is an evaluation.
Use the next slides for the documentary evidence.
day4_03_unexpected.tex · line 177 · D4-F033 — pedagogical visual, no new empirical claim
Interpretation Moves Within Six Hours
\[\Delta\mathrm{PHI}_m^{a\text{--}b}
=
\mathrm{PHI}_{m,a\text{--}b}
-
\mathrm{PHI}_{m,pre48h}.\]
| Window | Interpretation |
|---|---|
| \(0\text{--}6h\) | immediate reading before a full news cycle |
| \(6\text{--}24h\) | same-day reading after media and social circulation |
| \(24\text{--}48h\) | whether the interpretation survives into the next day |
The evidence is an attention burst, not a persistent shift in discussion.
day4_03_unexpected.tex · line 189 · D4-F034
Topic Shares Show What Travels
\[Share^k_{m,w}
=
\frac{N^k_{m,w}}{N^{all}_{m,w}},
\qquad
\Delta Share^k_m
=
Share^k_{m,post}-Share^k_{m,pre48h}.\]
Inflation is the clearest topic response: concrete price language travels more readily than technical policy language.
day4_03_unexpected.tex · line 217 · D4-F035
Interpretive boundary
Hawkish Content Shifts Interpretation
\[\text{interpretive channel operates}
\qquad\text{but}\qquad
\text{persuasive channel is not detected}.\]
| Margin | Finding |
|---|---|
| Interpretation | \(\Delta\operatorname{PHI}\) rises over 0–24 hours |
| Timing | strongest in the first six hours; fades by 48 hours |
| Approval | no comparable movement in support for the Bank |
| Topics | inflation discussion is the clearest supporting response |
The design studies public reading—not representative beliefs or trust.
day4_03_unexpected.tex · line 240 · D4-F036
Documentary evidence · source-preserving figure
Tone Groups Diverge at Release
The distance opens sharply at the release and then gradually closes.
day4_03_unexpected.tex · line 267 · D4-F037
Documentary evidence · source-preserving figure
Bundling Sharpens the Same-Day Response
The post-2015 bundled sample shows a clearer interpretation response immediately after release, with attenuation across later windows.
day4_03_unexpected.tex · line 278 · D4-F038
Documentary evidence · source-preserving figure
Inflation Travels Most Clearly
The public does not absorb every component equally. Price and cost-of-living language is the most direct bridge to everyday experience.
day4_03_unexpected.tex · line 290 · D4-F039
04
Central Banking · Lecture 4 · Worcester College
Part 04
Publication is not exposure.
Official posts compete for voluntary attention
Measure the objectIdentify the mechanismRead the evidence
Conceptual illustration · not empirical evidence
Publication does not guarantee exposure
Teaching point
A public document cannot change beliefs unless the audience encounters it.
Use the next slides for the documentary evidence.
day4_04_public.tex · line 10 · D4-F040--D4-F041 — pedagogical visual, no new empirical claim
Exposure Starts with Encounter
\[\text{conditional on the Bank posting, which messages attract interaction?}\]
| Object | Measurement |
|---|---|
| Source | complete official @bankofengland record, July 2011–July 2022 |
| Unit | individual official tweet |
| Outcome | likes, retweets, replies, and quote tweets |
| Covariates | format, timing, reply/original mode, readability, length |
Exposure is a prior margin: a message must win attention before it can inform or persuade.
day4_04_public.tex · line 22 · D4-F041
Compare Counts in Similar Environments
\[\mathbb E[Y_i\mid X_i]=\exp(X_i'\beta),
\qquad
Y_i=Likes_i+Retweets_i+Replies_i+Quotes_i.\]
| Comparison | Purpose |
|---|---|
| All tweets | raw feature–engagement association |
| Same broad period | absorb follower growth and slow platform changes |
| Same week, day, hour | compare more similar attention environments |
These are conditional associations. The Bank chooses format and timing; neither is randomly assigned.
day4_04_public.tex · line 50 · D4-F042
Design Margins Predict Engagement
| Margin | Finding | Reading |
|---|---|---|
| Visual format | photo and video posts attract much more interaction | presentation helps the message compete in the feed |
| Policy timing | MPC-day posts attract more engagement | policy is already salient |
| Macro state | the MPC-day premium rises in the high-inflation period | public attention is state-dependent |
The largest engagement margins come from format, timing, and context—not tiny wording changes.
day4_04_public.tex · line 80 · D4-F043
Interpretive boundary · visible interaction is only one gate
Reach, Dialogue, Readability Differ
ReachOriginal posts broadcast widely.
DialogueReplies serve conversation, not raw reach.
ReadabilityEasier syntax is associated with interaction.
Engagement ≠ comprehension ≠ trust ≠ approval
Engagement is not comprehension, trust, approval, or expectation updating.
day4_04_public.tex · line 102 · D4-F043
Documentary evidence · source-preserving figure
Posting Rises Across Regimes
Monthly volume trends upward, with visible changes around governorship, the pandemic, and platform-resolution milestones.
day4_04_public.tex · line 125 · D4-F044
Documentary evidence · source-preserving figure
Thursday Has the Most Posts
The weekday distribution is uneven: Thursday has the largest count, while weekend posting is sparse.
day4_04_public.tex · line 138 · D4-F045
Documentary evidence · source-preserving figure
Original Posts Dominate; Replies Grow
Original posts remain the principal mode; replies become a more visible part of the account's output in later years.
day4_04_public.tex · line 150 · D4-F046
Documentary evidence · source-preserving figure
Posting Hour Shapes Engagement
The hour profile is highly uneven, consistent with a changing attention environment across the day.
day4_04_public.tex · line 162 · D4-F047
Documentary evidence · source-preserving figure
Engagement Rates Are Volatile
Likes, retweets, replies, and quotes do not move in lockstep; aggregate engagement can conceal which interaction margin changed.
day4_04_public.tex · line 174 · D4-F048
Documentary evidence · source-preserving figure
Hashtags Use a Narrow Vocabulary
#BoE is dominant, followed by a much smaller group of recurring campaign, institutional, and topical tags.
day4_04_public.tex · line 186 · D4-F049
Documentary evidence · source-preserving figure
Public Language Centres on Rates
Bank, England, rate, UK, and inflation anchor the vocabulary, alongside people, work, money, policy, and economy.
day4_04_public.tex · line 198 · D4-F050
05
Central Banking · Lecture 4 · Worcester College
Part 05
Disclosure unfolds in stages.
A statement is focal; minutes reveal what the focal text compressed
Measure the objectIdentify the mechanismRead the evidence
Conceptual illustration · not empirical evidence
Minutes Restore Compressed Content
Teaching point
Minutes restore context that the focal statement leaves compressed.
Use the next slides for the documentary evidence.
day4_05_strategic.tex · line 10 · D4-F051--D4-F054 — pedagogical visual, no new empirical claim
Documentary evidence · source-preserving figure
Minutes Are the Second Public Layer
\[\mathcal I_t^{public}
=
\{\text{statement}_t,\text{minutes}_t,\ldots\}.\]
The statement announces the decision. The minutes later disclose the discussion, alternatives, risks, and views of voting and non-voting participants.
day4_05_strategic.tex · line 23 · D4-F052, D4-F053, D4-F054
Documentary evidence · source-preserving figure
One Meeting, Several Public Records
Read the release sequence as a disclosure architecture—not as repeated copies of one message.
day4_05_strategic.tex · line 51 · D4-F055
Documentary evidence · source-preserving figure
Minutes Greatly Expand the Word Count
Length creates room for nuance, but also raises the cost of finding the policy-relevant sentence.
day4_05_strategic.tex · line 67 · D4-F056
Release architecture · before and after 2004
The release lag changed after 2004
Before 2005
Meeting XStatement
6–8 weeks
MinutesNear meeting X+1
After 2004
Meeting XDecision + vote
exactly 3 weeks
MinutesVoting + non-voting views
The shorter lag increases timeliness, but minutes remain a separate information release.
day4_05_strategic.tex · line 79 · D4-F057, D4-F058
Case evidence · a delayed document changed the expected path
April Minutes Shifted the June Outlook
27 APRMeetingDecision and focal statement
18 MAYMinutesApril record released; dollar reaches a nine-week high
15 JUNNext decisionMarkets had brought forward the expected timing of a rise
A delayed document can revise the expected path of future policy.
day4_05_strategic.tex · line 115 · D4-F059
Documentary evidence · source-preserving figure
The press read the minutes as a June warning
The Fed's message was read as: pay attention; a June increase was being considered seriously.
The press transformed a long release into a focal policy-path signal.
day4_05_strategic.tex · line 142 · D4-F060
Case evidence · The Guardian
Press Outlets Shared the June Signal
A June rise was possible.
A long document was compressed into one forward-policy proposition.
Media compression amplified a common reading of the minutes.
day4_05_strategic.tex · line 165 · D4-F060
Case evidence · Reuters
Reuters carried the same June signal
A June move was firmly on the table.
A long document was compressed into one forward-policy proposition.
Media compression amplified a common reading of the minutes.
day4_05_strategic.tex · line 165 · D4-F060
Interpretive boundary
June Did Not Validate the Forecast
\[\text{minutes signal}
\longrightarrow
\text{earlier expected tightening}
\;\not\Rightarrow\;
\text{June rate increase}.\]
| What the episode shows | What it does not show |
|---|---|
| lagged minutes can contain path-relevant information and change positions | that the May 2016 reaction was typical across all minutes releases |
On 15 June, the Fed did not raise the target rate and reaffirmed the prevailing range.
day4_05_strategic.tex · line 194 · D4-F061
Measurement · compare release time with an ordinary afternoon
Measure Ten-Minute Volatility
Ordinary day
Minutes day
14:00 release
\[SD(r)=\sqrt{\frac{\sum_{t=1}^{T}(r_t-\bar r)^2}{T-1}}\]
Ask whether volatility rises exactly when the minutes arrive.
day4_05_strategic.tex · line 224 · D4-F062 · schematic, not source point estimates
Counterfactual intraday variability · standard deviation
Baseline: Ordinary Afternoons
EUR/USD
.04%
S&P 500
.14%
10-year yield
.04 bps
3-month yield
.71 bps
This is the normal 13:00–16:00 volatility benchmark—not an announcement effect.
day4_05_strategic.tex · line 266 · D4-F063
Minutes afternoons · standard deviation
Minutes Days Modestly Shift Volatility
EUR/USD
.06%
S&P 500
.15%
10-year yield
.05 bps
3-month yield
.48 bps
Minutes create an event, but an unconditional SD comparison is not identification.
day4_05_strategic.tex · line 291 · D4-F063
Statement afternoons · standard deviation
Statements Have the Most Volatility
EUR/USD
.11%
S&P 500
.23%
10-year yield
.10 bps
3-month yield
1.44 bps
The immediate focal release is most volatile; later minutes can still contain path news.
day4_05_strategic.tex · line 316 · D4-F063
Empirical result · comparison and scope
Regress Returns on Target and Path
\[\Delta r_{x,d,\tau}
=
\alpha+\beta_1TS_{d,\tau}
+\beta_2PS_{d,\tau}
+\varepsilon_{x,d,\tau}.\]
| Object | Meaning |
|---|---|
| \(\Delta r_{x,d,\tau}\) | return of asset \(x\) in the ten-minute interval at release time \(\tau\) on announcement day \(d\) |
| \(TS_{d,\tau}\) | target surprise |
| \(PS_{d,\tau}\) | path surprise |
| \(\varepsilon_{x,d,\tau}\) | residual return |
day4_05_strategic.tex · line 340 · D4-F064
Minutes response to the path surprise · coefficient
Minutes Load Most on the Path Factor
S&P 500
−.94***
EUR
1.18***
10-year yield
17.79***
3-month yield
5.62
The path coefficient is the recurring statistically significant minutes-day signal.
day4_05_strategic.tex · line 369 · D4-F065
Statement response to the path surprise · coefficient
Statements Trigger Stronger Reactions
S&P 500
−2.82***
EUR
3.63***
10-year yield
59.81***
3-month yield
3.18
Focality strengthens the immediate response without making later disclosure redundant.
day4_05_strategic.tex · line 392 · D4-F065
Source values · 19 ten-minute observations in each series
At the ZLB, Minutes Raise Yield Volatility
Minutes dayNon-announcement day
The source comparison describes roughly \(2\times\) versus \(5\times\) release-day volatility.
day4_05_strategic.tex · line 415 · D4-F066
Documentary evidence · source-preserving figure
ZLB Dollar-Path News Was Salient
\[\begin{aligned}
\text{QE}
&\longrightarrow
\text{larger Fed balance sheet}\\
&\longrightarrow
\text{greater USD-news sensitivity}.
\end{aligned}\]
During the ZLB, the source describes a roughly fourfold balance-sheet expansion and a depreciating, globally influential dollar.
day4_05_strategic.tex · line 455 · D4-F067
Non-ZLB response to the minutes path surprise · coefficient
Minutes Path News Matters Beyond the ZLB
S&P 500
−.95***
EUR
.78**
10-year yield
12.87**
3-month yield
8.09
Path news matters outside the constrained-rate regime as well.
day4_05_strategic.tex · line 487 · D4-F068
ZLB response to the minutes path surprise · coefficient
ZLB Responses Shift to EUR and 10Y
S&P 500
−.61
EUR
2.53***
10-year yield
31.33***
3-month yield
−1.16
When the current rate is constrained, news about the future path becomes more valuable.
day4_05_strategic.tex · line 511 · D4-F068
Documentary evidence · source-preserving figure
At the ZLB: less surprise, more reaction
The size of the measured surprise and the size of the market response need not move together.
day4_05_strategic.tex · line 534 · D4-F069
Documentary evidence · source-preserving figure
Minutes and Statements Differ in Tone
Document tone is another dimension along which the delayed and focal releases can diverge.
day4_05_strategic.tex · line 547 · D4-F070
One Meeting Leaves Three Records
\[\text{same meeting}
\quad
\begin{cases}
\text{statement: immediate, short, highly focal},\\
\text{minutes: delayed, longer, less focal},\\
\text{transcript: delayed archival benchmark}.
\end{cases}\]
| Layer | Research use | Communication margin |
|---|---|---|
| Statement | focal public signal | selection into the short message |
| Minutes | later public disclosure | partial restoration of omitted content |
| Transcript | archival benchmark | meeting-level proposition set |
day4_05_strategic.tex · line 561 · D4-F071
Measurement · transcript-anchored omission
Measure what the statement leaves out
Transcriptfull meeting record
Propositionspolicy claims
Dimensionoutlook · path · risk · disagreement
Coverageappears in statement?
\[g_t^k=1-\frac{N_t^{k,\mathrm{covered}}}{N_t^{k,\mathrm{all}}}\]
0 means full coverage; 1 means complete omission.
day4_05_strategic.tex · line 589 · D4-F072
Economic object · define before estimating
Measure Omission by Policy Dimension
| Dimension \(k\) | What the transcript propositions describe |
|---|---|
| OUTLOOK | current conditions and the expected macroeconomic outlook |
| PATH | expected future policy path and policy inclination |
| RISK | risks, uncertainty, and contingencies around the outlook |
| DISAGREEMENT | conflicting views, reservations, and alternatives |
A single document-level word count cannot reveal selective compression.
day4_05_strategic.tex · line 618 · D4-F072
Statements omit disagreement most heavily
\[g^{DISAGREEMENT}>g^{RISK}>g^{PATH}>g^{OUTLOOK}.\]
| Dimension | Mean gap | Reading |
|---|---|---|
| OUTLOOK | .658 | most covered |
| PATH | .713 | selectively compressed |
| RISK | .782 | frequently omitted |
| DISAGREEMENT | .852 | most omitted |
day4_05_strategic.tex · line 639 · D4-F073
Later Channels Close Half the Gap
\[\overline g^{statement}=.745
\qquad\longrightarrow\qquad
\overline g^{statement+minutes}=.334.\]
Selective compression is not permanent secrecy. It is staged disclosure across channels with different timing and focality.
day4_05_strategic.tex · line 662 · D4-F073
Empirical result · comparison and scope
Omitted Outlook Predicts Repricing
\[\text{minutes repricing}
\;\leftarrow\;
\text{omitted outlook/path}
\times
\text{intermeeting activity news}.\]
| Outcome | Source result |
|---|---|
| Treasury and real yields | respond to the interaction |
| Equities and breakevens | no corresponding robust response |
| Content dimensions | outlook and path survive controls |
The evidence is reduced-form: omitted content may correlate with unobserved meeting information.
day4_05_strategic.tex · line 681 · D4-F074
Interact Outlook Gaps with Activity News
\[Y_t
=
\alpha
+\beta\bigl(g_t^{OUT}\times z_t^y\bigr)
+\gamma_1g_t^{OUT}
+\gamma_2z_t^y
+\varepsilon_t.\]
\(\beta>0\) means that activity news receives a larger market response when more meeting outlook content was left out of the statement.
day4_05_strategic.tex · line 712 · D4-F075
Interactive timeline · one meeting, three records
The communication record closes in stages
01StatementImmediate · focal · selective
02MinutesWeeks · additional context · partial closure
03TranscriptYears · archival · most complete
Selected layer
The statement optimises focality.
Its brevity makes the message salient, but selection means that discussion, risks, paths, and disagreement may remain outside the focal text.
\(P\) is the focal public statement; adding channels \(C\), \(H\), and \(M\) progressively reduces the residual binary gap.
day4_05_strategic.tex · line 736 · D4-F076
Documentary evidence · source-preserving figure
Observed disclosure gap closure
\(P\) is the focal public statement; adding channels \(C\), \(H\), and \(M\) progressively reduces the residual binary gap.
day4_05_strategic.tex · line 736 · D4-F076
Documentary evidence · source-preserving figure
Policy Tools Carry Different Language
Counting a tool mention misses whether it is framed as active, contingent, constrained, or merely discussed.
day4_05_strategic.tex · line 749 · D4-F077
Documentary evidence · source-preserving figure
Frequency Is Only the First Layer
Frequency locates attention; context determines the policy meaning of that attention.
day4_05_strategic.tex · line 761 · D4-F078
06
Central Banking · Lecture 4 · Worcester College
Part 06
Climate enters the mandate.
Public salience and rate-setting deliberation are different communication layers
Measure the objectIdentify the mechanismRead the evidence
Documentary evidence · source-preserving figure
Climate Risk Reaches Several Channels
Physical and transition shocks can affect prices, output, financial institutions, and the transmission mechanism.
day4_06_climate.tex · line 10 · D4-F080
Documentary evidence · source-preserving figure
Risk Varies by Horizon and Severity
A long horizon does not remove a risk from the central-bank problem; it changes how the risk is monitored and translated.
day4_06_climate.tex · line 23 · D4-F081
Documentary evidence · source-preserving figure
Climate Shocks Can Threaten Stability
Asset repricing, collateral losses, credit exposures, and insurance gaps can move the issue inside a conventional stability mandate.
day4_06_climate.tex · line 36 · D4-F082
Policy design · allocation enters through relative financing conditions
Climate Tools Mix Support and Penalty
Brown-penalising
Raise relative cost
Example: coal-firm exclusion from the Norwegian sovereign wealth fund.
Green-supporting
Improve relative funding
Examples: BoJ climate loans, Bangladesh Bank green credit, ECB tilting.
Instrument choice exposes the boundary between risk management and allocation.
day4_06_climate.tex · line 48 · D4-F083
Documentary evidence · source-preserving figure
One Mandate Faces Many Demands
The answer cannot be read from the issue's importance alone. It depends on the price-stability, financial-stability, supervisory, and operational mandates of the institution.
day4_06_climate.tex · line 82 · D4-F084
Documentary evidence · source-preserving figure
Carney Made Climate a Policy Signal
The speech connected climate change to financial stability and to the mismatch between long-horizon risks and shorter institutional decision horizons.
day4_06_climate.tex · line 103 · D4-F085
Public Salience Can Precede Policy
\[\text{2015 public speech}
\;\not\Rightarrow\;
\text{immediate climate language in the MPC record}.\]
| Evidence | Count | Reading |
|---|---|---|
| MPC transcripts, 2015–2017 | 29 | more than 400,000 words |
| Climate references | 0 | absent from the sampled internal record |
| Brexit references | 449 | present in 22 meetings |
An issue may be salient before it enters monetary-policy deliberation.
day4_06_climate.tex · line 123 · D4-F086
Speech and Records Form the Corpus
| Dimension | Speeches | Policy records |
|---|---|---|
| Collected | 37,814 | 3,602 |
| Usable | 37,814 | 3,593 |
| Institutions | 136 | 26 |
| Period | 1986–2025 | 1995–2024 |
| Languages | predominantly English | 14 |
The two channels differ in scale, purpose, and institutional coverage.
day4_06_climate.tex · line 151 · D4-F087
Keywords Overstate Climate Evidence
| Measure | Speeches | Policy records |
|---|---|---|
| Climate candidates | 4,857 | 500 |
| Verified documents | 3,266 | 88 |
| Verified excerpts | 12,582 | 148 |
| False-positive rate | 32.8% | 82.4% |
| Common-sample mention | 5.5% | 2.4% |
| Common-sample word share | .565% | .079% |
Verification matters most in the policy record.
day4_06_climate.tex · line 175 · D4-F087
Policy Seriousness Has Three Layers
\[\text{public issue}
\longrightarrow
\text{monetary-policy record}
\longrightarrow
\text{policy-relevant language}.\]
01
Speeches
02
Official records
Mention is an entry condition; mandate linkage is the substantive test.
day4_06_climate.tex · line 200 · D4-F088
Keyword Searches Create False Positives
\[\{\text{oil, gas, coal, energy}\}
\not\equiv
\{\text{climate context}\}.\]
A dictionary can find candidates; it cannot determine the policy meaning.
day4_06_climate.tex · line 227 · D4-F089
Conceptual illustration · not empirical evidence
Context-Verified Mentions Pass the Gate
Teaching point
A keyword enters the evidence set only after context and mandate verification.
Use the next slides for the documentary evidence.
day4_06_climate.tex · line 249 · D4-F089 — pedagogical visual, no new empirical claim
Measurement · keywords are candidates, not evidence
Verification Creates Mandate Evidence
Keywordcandidate
Contextverified climate use
Mandateinflation · stability · tools
Policy codeor no linkage
88verified policy-record documents
148verified excerpts
Only context-verified, mandate-linked language counts as policy evidence.
day4_06_climate.tex · line 261 · D4-F089
Climate Enters Records Slowly
| Finding | Evidence |
|---|---|
| Lower official-record incidence | 5.5% of speeches versus 2.4% of records |
| Institutional non-entry | half of the 26 common institutions never mention climate in the policy record |
| Long translation delay | median speech-to-record lag: 18 years |
| Benchmark contrast | geopolitics maps quickly; climate slowly and selectively; digitalisation mostly outside rate setting |
Public discussion and official policy reasoning are not interchangeable corpora.
day4_06_climate.tex · line 289 · D4-F090
Translation · the record uses a narrower policy vocabulary
Language Narrows Toward the Mandate
Public speech
Broad issue language
Climate risk · transition · sustainable finance · long-horizon responsibility
Policy record
Mandate-specific vocabulary
Inflation · supply shocks · stability · collateral · asset purchases · supervision
\[\text{broad issue language}\longrightarrow\text{mandate-specific vocabulary}\]
Translation, not repetition, is the evidence of policy entry.
day4_06_climate.tex · line 312 · D4-F091
Empirical result · comparison and scope
Mandate Linkage Rises in Official Records
\[\text{policy seriousness}
\approx
\text{appearance in mandate-relevant language}.\]
| Climate excerpt code | Speeches | Policy records |
|---|---|---|
| No mandate linkage | 72.2% | 33.8% |
| Inflation linkage | 6.3% | 33.1% |
| Operational linkage | 4.5% | 25.7% |
This operational definition measures translation into the mandate; it does not prove that policy is optimal or sufficient.
day4_06_climate.tex · line 337 · D4-F091
Documentary evidence · source-preserving figure
Climate Timing Differs Across Channels
Public speech typically moves first; official policy records follow with a slower and smaller response.
day4_06_climate.tex · line 365 · D4-F092
Documentary evidence · source-preserving figure
Leadership Shifts Climate Communication
The source figure isolates a visible change around Lagarde's ECB presidency; leadership is one possible accelerator of public signalling.
day4_06_climate.tex · line 378 · D4-F093
Documentary evidence · source-preserving figure
Commitment language evolves over time
The content question is not only whether climate appears, but how strongly the institution commits itself when it appears.
day4_06_climate.tex · line 392 · D4-F094
Documentary evidence · source-preserving figure
Commitment Is Stronger in Records
Mean commitment: speeches \(1.48\), minutes \(1.53\); Mann–Whitney \(p=.029\).
day4_06_climate.tex · line 404 · D4-F095
Documentary evidence · source-preserving figure
Official Climate Mentions Lag Speeches
The mean speech-to-record lag is \(14.3\) years; the source chart shows especially long delays for institutions such as the Fed and ECB.
day4_06_climate.tex · line 416 · D4-F096
07
Central Banking · Lecture 4 · Worcester College
Part 07
Understanding is a constraint.
Flesch–Kincaid measures sentence and word difficulty—not comprehension
Measure the objectIdentify the mechanismRead the evidence
Documentary evidence · source-preserving figure
FOMC speech readability varies over time
The source chart tracks Flesch–Kincaid grade level across FOMC speeches; a higher score indicates more years of schooling implied by the text's sentence and syllable structure.
day4_07_readability.tex · line 10 · D4-F098
Interactive diagnostic · what changes the score?
Grade Level Is a Mechanical Statistic
\[FK=0.39\left(\frac{words}{sentences}\right)+11.8\left(\frac{syllables}{words}\right)-15.59\]
Mechanical grade-level score
11.3
The score does not observe policy logic, prior knowledge, audience fit, or actual comprehension.
Readable syntax helps, but readability is not understanding.
day4_07_readability.tex · line 26 · D4-F098
Conceptual illustration · not empirical evidence
Readability Is a Tool, Not Understanding
Teaching point
A lower grade-level score does not establish comprehension.
Use the next slides for the documentary evidence.
day4_07_readability.tex · line 55 · D4-F098 — pedagogical visual, no new empirical claim
Documentary evidence · source-preserving figure
Document Type Changes Reading Burden
A statement, minutes, transcript, and speech perform different institutional jobs; comparing their grade levels should preserve that functional difference.
day4_07_readability.tex · line 67 · D4-F099
Documentary evidence · source-preserving figure
Communicator and Channel Both Matter
Chair communication and press-mediated communication are selected and edited under different constraints, so a grade-level gap is descriptive rather than causal.
The press transformed a long release into a focal policy-path signal.
day4_07_readability.tex · line 81 · D4-F100
Interpretive boundary
Use readability as a diagnostic, not a verdict
\[\text{lower }FK
\;\not\Rightarrow\;
\text{greater understanding}.\]
| Question | FK can help | FK cannot answer |
|---|---|---|
| Syntax | Are sentences and words mechanically demanding? | Is the policy logic conceptually clear? |
| Comparison | Do channels or speakers differ descriptively? | Did the channel cause understanding? |
| Design | Where should an editor inspect complexity? | Did households, markets, and specialists infer the same message? |
The communication objective is comprehension, not merely a lower score.
day4_07_readability.tex · line 95 · D4-F098, D4-F099, D4-F100
Section synthesis · selection today, restoration later
Disclosure Unfolds Across Time
Statementfocal selection
3 weeks
Minutespartial gap closure
Later channelsarchival record
Receptionprices + public reading
Later can be newOmission is selective.745 → .334 after minutes
A short focal statement and a complete public record are different objectives.
day4_08_synthesis_sources.tex · line 6 · D4-F051--D4-F078
Section synthesis · public importance is not yet policy entry
Climate Policy Requires Translation
Public saliencespeech and agenda
Official recorddeliberative entry
Mandate linkageinflation · stability · operations
\[\text{public salience}\longrightarrow\text{official record}\longrightarrow\text{inflation, stability, or operational linkage}\]
Policy seriousness appears in mandate-relevant language, not keyword volume alone.
day4_08_synthesis_sources.tex · line 34 · D4-F079--D4-F096
Three Constraints Shape Reception
\[\text{attention}
\not\equiv
\text{understanding}
\not\equiv
\text{persuasion}.\]
| Constraint | Empirical object | Correct interpretation |
|---|---|---|
| Selection | statement gap | what the focal text omits |
| Translation | mandate linkage | how an issue becomes policy relevant |
| Readability | FK grade level | mechanical syntactic burden |
Market regressions are reduced-form; text scores are descriptive; official mentions must be context-verified.
day4_08_synthesis_sources.tex · line 61 · D4-F051--D4-F100
A
Central Banking · Lecture 4 · Worcester College
Appendix
Complete source tables.
Projected slides show the coefficient that answers the question; the complete estimates remain here.
End · Lecture 4
Communication as Data.
Fatih Kansoy
Central Banking · Lecture 4
04/