01/
The information channel of monetary policy.
Trinity Term · 2026
Central Bank
Communication.
Words, expectations, and the anchoring of inflation in modern monetary policy.
02/
Lecture 2
Trinity Term · 2026
Measuring
expectations.
How economists infer beliefs from surveys, market prices, and central bank communications.
02 · Measuring expectations
From latent beliefs to observable prices, surveys, and event-window revisions
Core question
How do we infer expected inflation, expected rates, and expected policy paths before they are directly observed?
Measuring market expectations
Use traded prices to recover beliefs about future interest rates, inflation, exchange rates, and risk. The challenge is separating expectations from risk premia and liquidity premia.
Futures markets and financial assets
Fed funds and OIS futures price expected policy rates. Breakevens and inflation swaps price expected inflation. Bonds, FX, equities, and options reveal broader macro-financial expectations.
High-frequency data and identification
Compare prices immediately before and after central bank news. Narrow windows help isolate the expectation revision caused by the announcement from other macro news.
Lecture 2 builds the measurement toolkit: surveys provide direct stated beliefs, market prices provide real-time revealed beliefs, and high-frequency changes identify how communication updates those beliefs.
03/
Lecture 3
Trinity Term · 2026
Monetary policy
surprises.
Separating expected policy from unexpected policy news.
03 · Monetary policy surprises
Policy shocks are not policy changes; they are revisions relative to what markets expected
Definition
A monetary policy surprise is the change in expected policy, measured from asset prices in a narrow window around central bank news.
Actions vs words
"Do actions speak louder than words?" asks whether statements move markets beyond the policy-rate decision itself.
Short and long surprises
Near contracts capture target-rate news; longer maturities capture path, forward-guidance, and balance-sheet news.
Text-based surprises
Language can be measured directly: tone, topics, uncertainty, and hawkish or dovish shifts in statements and press conferences.
Principal components
PCA reduces many rate changes into a few factors: target, path, and longer-horizon policy news.
Event studies
Compare prices just before and after the announcement, so other news has little time to contaminate the measurement.
OLS response regressions
Estimate how asset prices react: $\Delta asset_t = \alpha + \beta\,surprise_t + u_t$.
Lecture 3 turns policy announcements into empirical shocks: identify the surprise, separate rate-path news from communication news, then estimate how markets respond.
04/
Lecture 4
Trinity Term · 2026
Surprise
transmission.
Measuring impacts across markets, countries, and horizons.
04 · Surprise transmission across markets and countries
Estimate how one identified policy surprise moves domestic, global, and emerging-market prices
Empirical object
Calculate transmission by regressing asset-price changes on the measured surprise: $\Delta y_{i,t,h}=\alpha_i+\beta_i\,MPsurprise_t+u_{i,t,h}$.
Domestic impact
Measure how domestic rates, equities, credit spreads, FX, and inflation compensation move around the announcement.
International impact
Use foreign asset returns in the same event window to estimate spillovers from the source central bank.
Other financial markets
Trace responses in equities, commodities, credit, volatility, exchange rates, and inflation-linked assets.
Domestic long-term rates
Decompose yield-curve effects into expected short rates, term premia, and balance-sheet or forward-guidance channels.
Global financial effects
Ask whether surprises shift global risk appetite, dollar funding conditions, capital flows, and cross-border risk premia.
Emerging markets
Compare countries by dollar exposure, exchange-rate regime, foreign debt, reserves, and central bank credibility.
International rates
Estimate pass-through from source-country surprises to foreign short rates, long rates, and forward curves.
Identification
Use narrow windows, controls, panels, and heterogeneity tests to separate policy news from unrelated market movements.
05/
Lecture 5
Trinity Term · 2026
Text in
macro-finance.
Turning language into data for monetary and financial economics.
05 · Text in macro-finance
Using data science to turn unstructured language into macro-financial variables
Text as data
Convert speeches, statements, news, social media, Google searches, and Wikipedia attention into structured measures of tone, topics, attention, and policy stance.
From text to variables
Clean documents, tokenize language, build dictionaries or embeddings, and aggregate words into document-level measures.
Sentiment and stance
Measure positive or negative tone, uncertainty, inflation concern, hawkishness, dovishness, and perceived policy restrictiveness.
Market reaction to words
Link wording choices to intraday changes in rates, equities, FX, credit spreads, and inflation compensation.
What is said, and not said
Topic models and word shares show what central banks talk about, what disappears, and which themes receive attention.
Public attention
Use social media, Google Trends, news rankings, and Wikipedia page views to measure what households and investors notice.
Climate communication
Measure what central banks say and do on climate change: risk language, financial-stability framing, supervision, and asset-purchase criteria.
Measurement, not programming: turning language and attention into macro-finance variables.
06/
Lecture 6
Trinity Term · 2026
Open-economy
macro-finance.
Interest parity, globalisation, and currency risk.
06 · Open-economy macro-finance
Exchange rates, interest-rate parity, and why currency returns are hard to rationalise
Core question
If capital is mobile, why do interest-rate differentials, exchange rates, and currency excess returns fail to line up cleanly?
Financial globalisation
Capital mobility links domestic rates, foreign rates, exchange rates, risk sharing, and international financial spillovers.
Covered interest-rate parity
CIP is the no-arbitrage benchmark using spot FX, forward FX, and safe interest rates across currencies.
Offshore-onshore markets
Market segmentation, regulation, collateral, and funding constraints explain why the same currency can trade at different prices.
Uncovered interest parity
UIP says high-interest-rate currencies should be expected to depreciate enough to offset the yield advantage.
Carry trade as a test
Borrow low-rate currencies and buy high-rate currencies. Persistent profits are evidence against simple UIP.
Forward premium puzzle
Currencies with high interest rates often appreciate rather than depreciate, so returns appear predictable.
The lecture moves from arbitrage logic to equilibrium risk: when parity fails, the question is whether the gap is a friction, a premium, or a puzzle.
07/
Lecture 7
Trinity Term · 2026
Capital controls
and policy games.
Optimal restrictions, large economies, strategic behaviour, and retaliation.
07 · Capital controls and strategic behaviour
When restrictions on cross-border finance become macroeconomic policy instruments
Core question
Can a country improve domestic stability by taxing, limiting, or redirecting capital flows, and what happens when other countries respond?
Basics
Capital controls are taxes, limits, or administrative rules on inflows and outflows of foreign capital.
Model logic
Controls can correct externalities when private borrowing or lending creates crisis risk, exchange-rate pressure, or financial fragility.
Quantitative restrictions
Instead of changing prices through taxes, governments may cap quantities: who can borrow, lend, invest, or convert currency.
Optimal controls
The policy problem is to trade off risk reduction against lost financing, lower efficiency, and distortions in allocation.
Large economies
A large country's controls move world interest rates, exchange rates, and capital flows, so the policy creates spillovers.
Strategic behaviour
When each country chooses controls for itself, the global outcome may involve policy competition, leakage, and retaliation.
The lecture asks when capital controls are stabilising macroprudential tools, when they export costs to others, and when retaliation turns domestic policy into an international game.
Course logistics · Presentation
Week 8 practical assignment · 25% of the course mark
Format
A group presentation in the final class of the term.
Timing
20 minutes
The presentation is followed by questions.
Group size
3-5 people
Each group presents only in front of the lecturer.
Slides
Required
Use PowerPoint, LaTeX Beamer, or equivalent professional-quality slides.
Option A
High-effort, high-return: apply methods from high-frequency analysis or finance-meets-data-science to a current research question.
Option B
Standard-effort, standard-return: choose a topic from the course reading list and critically evaluate it.
Course logistics · Final exam
Two-hour final exam · 75% of the course mark
Weight
75%
Duration
2 hours
Choice
3 of 5
Exam instruction
Students answer three questions out of five in a two-hour final exam.
Presentations and the final exam are the two assessed components of the course: 25% presentation and 75% final exam.
01/A
Section A
Trinity Term · 2026
Central bank
communication
fundamentals.
Definitions, audiences, channels, and why words are part of monetary policy.
Ben S. Bernanke
Federal Reserve Chair, 2006–2014 · Nobel Laureate, 2022
Epigraph
Monetary policy is 98% talk and only 2% action.
Ben S. Bernanke
· on the practice of central banking, 2015
· Link to source
The policy rate is one number, set eight times a year. Markets price thousands of forward rates off the language used to justify it. The language is the policy.
Mario Draghi
President of the ECB, 2011–2019 · London, 26 July 2012
A speech that re-priced the eurozone
“Within our mandate, the ECB is ready to do whatever it takes to preserve the euro.”
“And believe me, it will be enough.”
The market response — three sentences, three years
24 JUL 2012
Italy 10y6.6%
Spain 10y7.6%
EUR/USD1.21
12 SEP 2012
Italy 10y5.1%
Spain 10y5.7%
EUR/USD1.29
13 DEC 2012
Italy 10y4.6%
Spain 10y5.5%
EUR/USD1.31
20 DEC 2013
Italy 10y4.1%
Spain 10y4.1%
EUR/USD1.37
The Outright Monetary Transactions (OMT) programme was never deployed. Words alone repriced sovereign risk because they were credible and state-contingent. Communication is policy when the framework supports it.
Decomposing a stressed sovereign yield
Term-structure components of an n-period bond under stress
yield ≈ expected policy path + term premium + sovereign stress premium
$$y_t^{(n)} \;\approx\; \tfrac{1}{n}\sum_{j=0}^{n-1} E_t[i_{t+j}] \;+\; TP_t^{(n)} \;+\; SP_t$$
Expected policy path
$\boldsymbol{\tfrac{1}{n}\sum_{j=0}^{n-1} E_t[i_{t+j}]}$ — the average expected short rate over the bond's life. The index j runs over each future period to maturity.
Term premium
$\boldsymbol{TP_t^{(n)}}$ — compensation required by the marginal investor for holding a long-maturity claim rather than rolling short rates.
Sovereign stress premium
$\boldsymbol{SP_t}$ — issuer-specific compensation: default risk, illiquidity, and (in a currency union) redenomination risk.
The July 2012 intervention operated through $\boldsymbol{SP_t}$, not $\boldsymbol{E_t[i_{t+j}]}$: it collapsed the market-implied probability of euro break-up. Routine forward guidance, by contrast, works through $\boldsymbol{E_t[i_{t+j}]}$. Communication is a regular instrument of monetary policy — not a collection of famous speeches.
Transparency and communication are not the same
Transparency is the principle; communication is its operational use to shape beliefs.
Definition
Central bank communication is the strategic provision of information by the central bank about its mandate, its assessment of the economy, its decisions, its instruments, the balance of risks, and its likely reaction function — addressed to different audiences through different channels for specific policy and institutional purposes.
The principle
Transparency
The institutional commitment to openness: what the central bank makes public, on what schedule, and with what completeness.
Typical outputs
decisions · rules · minutes · forecasts · data · voting records
Operational use
Communication
The active, time-specific practice of conveying policy content through deliberate language, timing, and interaction.
Typical channels
language · statements · speeches · press conferences · interviews
A bank can be highly transparent — many documents published on time — and still communicate poorly: the reaction function may be unclear and messages may be inconsistent. Transparency is necessary for modern communication; it is not sufficient.
Key forms of central bank communication
The channels through which the regime makes itself observable
📄
Policy announcements
Statements and press releases issued at the close of monetary policy meetings, recording the decision and its rationale.
📚
Regular reports
Periodic publications such as Inflation Reports, Monetary Policy Reports, and Financial Stability Reports.
🎤
Press conferences
Live briefings by the governor or chair, including a Q&A in which the reaction function is probed in real time.
📑
Meeting minutes
Written records of policy discussions, voting patterns, and the balance of arguments behind the decision.
📊
Forecasts & projections
Numerical projections for inflation, output, and the policy path; SEP and dot plot at the Fed; staff projections at the ECB.
📣
Speeches & testimony
Public addresses by Council and Board members, and parliamentary testimony delivered between meetings.
Six channels, one regime. Each is a way the audience can read the bank's stance, outlook, reaction function, or risk view.
Every word matters
Fed Face.
Markets read the Chair's expression, not only the words.
- Negative expressions move prices within three-minute windows.
- One standard deviation more negativity: SPY -0.53 bp, VIX +3.76 bp, EUR/USD -0.18 bp.
- The effect remains after controlling for verbal tone, hawkishness, the rate decision, and market conditions.
- Negative faces line up with later FOMC minutes, suggesting genuine information, not only noise.
Curti and Kazinnik (2023), "Let's face it: Quantifying the impact of nonverbal communication in FOMC press conferences."
Journal of Monetary Economics
Federal Reserve communication timeline
Layered scheduled, periodic, and archival communication across one year
Source: Federal Reserve FOMC calendars
Federal Reserve communication is a layered system: meeting-day signals (statement, press conference, dot plot), delayed scheduled releases (minutes), continuous discretionary streams (speeches, testimony, interviews, research), and long-lag archival material (transcripts, +5 years).
Evolution of central bank communication
From "do not give reasons" to "the explanation is the policy"
“Not only do expectations about policy matter, but, at least under current conditions, very little else matters.”
— Michael Woodford, Jackson Hole Conference, 2005
🔒
1930s – 1980s
Era of secrecy
“It is a dangerous thing to start to give reasons.”
— Sir Stephen Harvey, Bank of England
“Never explain, never excuse.”
— Montagu Norman, Bank of England
🔇
1990s
Transition period
“Since becoming a central banker I have learned to mumble with great incoherence.”
“If I've made myself too clear, you must have misunderstood me.”
— Alan Greenspan, Fed
💬
2000s
Modern era
“Sometimes the explanation is the policy.”
— Janet Yellen, Fed
📣
Post-2008
Forward-guidance era
“Monetary policy is 98 percent talk and 2 percent action.”
— Ben Bernanke, Fed
Objectives of central bank communication
What communication is meant to do, in the language of the regime
🎯
Enhance policy effectiveness
Strengthen transmission by making the stance, strategy, and outlook legible, so that the intended financial-conditions response is what actually occurs.
⚓
Anchor expectations
Hold $\pi^{e}$ at long horizons close to target, raising the credibility weight $\chi$ and reducing the persistence of supply shocks.
📈
Reduce uncertainty
Improve the predictability of policy decisions, lowering term premia and dampening volatility around announcements.
⚖
Ensure accountability
Discharge the democratic bargain that comes with independence: explain the mandate, the decisions, and the rationale to elected representatives and the public.
🛡
Build credibility
Foster public confidence through consistent and clear communication of policy commitments, so that words about the future are priced as binding rather than discounted as cheap talk.
Communication does not have one purpose. It is the joint instrument by which the regime delivers transmission, anchoring, predictability, accountability, and credibility.
Three axes
Same sentence, three independent effects — analyse along each axis separately
Axis 01
Audience
who hears it — and with what attention
Markets
high attention, fast reaction, narrow vocabulary
Price-setters
low attention; decisions feed inflation
Households
very low attention; salient prices dominate
Political bodies
arena of accountability and legitimacy
Peer central banks
spillovers, coordination, credibility
Axis 02
Channel
through what medium, format, and lag
Decisions
terse statements; high signal density per word
Explanatory
press conferences and reports; looser register
Deliberation
minutes and transcripts; lagged, dissent visible
Forecasts
projections and dot plots; quantitative outlook
Outreach
speeches, testimony, blogs, listening events
Axis 03
Objective
what the message is meant to achieve
Anticipation
make the reaction function readable from data
Path-shaping
move expected future short rates
Anchoring
tie long-run inflation to the target
Accountability
legitimacy and democratic standing
Crisis stabilisation
compress tail risk in transmission
A central bank does not address a representative listener. Identical language can be decisive for markets and invisible to households; one objective may demand a different audience-and-channel mix from the next. Consistency across channels is itself a problem to be solved.
Communication is also listening
The communication problem is solved by the round trip, not the broadcast
→
Outward
The bank conveys policy content to its audiences.
←
Inward
The same encounters return information about beliefs, expectations, and conditions on the ground.
A message succeeds only if the audience clears all three
Condition 01
Notice
Has the audience seen the message at all? For households the answer is usually no.
Condition 02
Understand
Has the content been parsed correctly? Technical vocabulary creates predictable misreadings.
Condition 03
Interpret
Does the audience map the message to the bank's intended meaning, or to a different one?
Communication is part of the information-production process of monetary policy — not only of its transmission. Sending a message does not solve the problem. The round trip does.
cf. Haldane, A Little More Conversation, A Little Less Action (2017)
01/B
Section B
Trinity Term · 2026
Risk and
Uncertainty.
Distributions, fan charts, scenarios, and the balance of risks.
Risk, uncertainty, and the balance of risks
When probabilities are knowable — and when the distribution itself is in question
Risk
— known unknowns
Outcomes can be enumerated and assigned probabilities. The bank acts under a known distribution.
$$\{(y_i,p_i)\}_{i=1}^{N}, \qquad \sum_{i=1}^{N} p_i = 1$$
$$E_t[Y_{t+h}] \;=\; \sum_{i=1}^{N} p_i \, y_i$$
Communicated by
fan charts — tracing the width and tilt of the conditional distribution.
Uncertainty
— unknown unknowns
The probabilities, or the relevant model, are not known with confidence. The distribution itself is in dispute.
the $p_i$ are unknown — the bank chooses among a set of candidate models
$$Y_{t+h} \sim F_m(\cdot), \qquad m \in \mathcal{M}$$
Communicated by
scenarios, alternative-paths analysis, and qualitative risk language.
Within a known distribution, the width signals dispersion and the tilt signals the balance of risks — fan charts trace both. When the distribution itself is in dispute, only scenarios and explicit language can convey what the baseline may miss.
Forecasts are distributions
A baseline summarises a distribution — policy depends on more than the centre
A statement such as "inflation will be 2.3%" is shorthand for one moment of the conditional distribution
$\,Y_{t+h}\mid\mathcal{I}_t\,$, where $\mathcal{I}_t$ is the information set at date $t$.
Centre
$E_t[Y_{t+h}]$
The modal/mean path — the figure quoted in the press release.
Width
$\operatorname{Var}_t(Y_{t+h})$
Spread of the bands at any horizon — how confident the bank is.
Skew
$\operatorname{Skew}_t(Y_{t+h})$
Asymmetry above versus below the modal path — the balance of risks.
Tail mass
$P\!\left(Y_{t+h}<\bar{Y}^{\text{bad}}\right)$
Probability mass below a catastrophic threshold.
Draghi's intervention did not move $E_t[Y_{t+h}]$ much; it collapsed $P(Y_{t+h}<\bar{Y}^{\text{bad}})$ — the perceived probability of catastrophic euro-area outcomes. Communication can act on any moment of the forecast distribution, not only on the centre.
01/C
Section C
Trinity Term · 2026
Expectations.
Static, adaptive, and rational expectation formation inside the Phillips curve.
One Phillips curve, four expectation rules
The same equation, four theories of how $\pi_t^e$ is formed
Expectations-augmented Phillips curve
$$\pi_t \;=\; \pi_t^{e} \;+\; \alpha\,(y_t - y_e) \;+\; u_t$$
inflation = expected inflation + slack term + supply shock
Each paradigm closes the model with a different rule for $\pi_t^{e}$
Rule 01
Static
$$\pi_t^{e} = \bar{\pi}$$
A fixed nominal anchor; expectations do not respond to data.
Rule 02
Adaptive
$$\pi_t^{e} = \pi_{t-1}^{e} + \lambda(\pi_{t-1} - \pi_{t-1}^{e})$$
Beliefs revised mechanically toward most recent realisation.
Rule 03
Rational
$$\pi_t^{e} = E_{t-1}[\pi_t \mid \mathcal{I}_{t-1}]$$
Conditional expectation given the model and information set.
Rule 04
Behavioural
$$\pi_t^{e} = \mathbb{F}_{t-1}[\pi_t]$$
Subjective expectation $\mathbb{F}$ — bounded attention, noisy updating, biases.
The choice of expectation rule determines what the slack coefficient $\boldsymbol{\alpha}$ identifies, what the residual $\boldsymbol{u_t}$ contains, and whether announcing a change in policy can move inflation without changing $\boldsymbol{y_t}$.
Static expectations: a fixed nominal reference point
Wage- and price-setters enter every period with the same belief
Rule
$$\pi_t^{e} \;=\; \bar{\pi}$$
$$\pi_t - \pi_t^{e} \;=\; \pi_t - \bar{\pi}$$
The forecast error is then the deviation of realised inflation from a fixed reference point — not from a moving belief:
Three implications
A useful benchmark for a world in which inflation expectations are passive.
Higher inflation today does not mechanically change tomorrow's expected inflation.
The model generates level effects — not a theory of inflation persistence.
Visualisation
Realised inflation deviates from a fixed belief; the belief itself does not adjust.
Static expectations are not a claim that agents are sophisticated; the opposite. Expected inflation is treated as given, so policy works through current demand — not through changing beliefs about future inflation.
Static expectations and the Phillips curve
Substituting the static rule into the E-A Phillips curve yields a level relation
Derivation
01
Start from the expectations-augmented Phillips curve:
$$\pi_t = \pi_t^{e} + \alpha(y_t - y_e) + u_t$$
02
Impose the static rule $\pi_t^{e} = \bar{\pi}$:
$$\pi_t = \bar{\pi} + \alpha(y_t - y_e) + u_t$$
03
Equivalently, the reduced form:
$$\pi_t - \bar{\pi} \;=\; \alpha(y_t - y_e) + u_t$$
Reading
A positive output gap raises $\pi_t$ above $\bar{\pi}$, but realised inflation does not feed back into expected inflation.
There is no acceleration mechanism: a sustained positive gap shifts the level of inflation, not its rate of change.
The model implies a stable short-run trade-off between economic slack and the level of inflation.
Adaptive expectations
Beliefs revised toward the most recent realisation — persistence from the rule itself
Adaptive expectations rule
$$\pi_t^{e} = \pi_{t-1}^{e} + \lambda\,(\pi_{t-1} - \pi_{t-1}^{e}), \qquad 0 < \lambda \leq 1$$
$\lambda$ is the speed of updating: the weight placed on last period's forecast error.
Special case: $\lambda = 1$
$$\pi_t^{e} = \pi_{t-1}$$
last period's inflation becomes today's expectation
Mechanism
Expectations are backward-looking: $\lambda$ governs the speed at which beliefs absorb the most recent forecast error. Higher realised inflation revises expected inflation upward in the next period, regardless of cause.
Implication
Inflation inherits persistence from its own past. A sequence of shocks — even temporary ones — lifts inflation for prolonged periods if beliefs keep adjusting upward.
Disinflation becomes harder the longer inflation has been elevated: yesterday's inflation feeds directly into today's beliefs, and policy must work against that inheritance — not just against current demand.
Adaptive expectations and the Phillips curve
The output gap drives the change in inflation — the accelerationist Phillips curve
Derivation
01
E-A Phillips curve:
$$\pi_t = \pi_t^{e} + \alpha(y_t - y_e) + u_t$$
02
Impose adaptive expectations with $\lambda = 1$:
$$\pi_t^{e} = \pi_{t-1}$$
03
Substitute:
$$\pi_t = \pi_{t-1} + \alpha(y_t - y_e) + u_t$$
04
Take first differences — the accelerationist Phillips curve:
$$\Delta \pi_t \;=\; \alpha(y_t - y_e) + u_t$$
Reading
The output gap now drives the change in inflation, not its level. This is the Phillips curve associated with Friedman (1968) and Phelps (1967).
A stable long-run trade-off disappears: holding $y_t > y_e$ does not permanently lower unemployment — it generates continuing inflation increases.
A central bank that persistently targets output above equilibrium is choosing an inflation path with no natural stopping point.
The persistence comes from the expectation-formation rule itself, not from sticky prices, contracts, or supply shocks. Change the rule and the dynamics change.
Adaptive expectations in the post-war record
The Great Inflation as a slow-moving inheritance — not a sequence of independent shocks
U.S. CPI inflation, 1965–1981
The Great Inflation
The 1970s pattern is consistent with persistence and backward-looking belief formation: each new outcome fed into the next round of expectations.
Negative supply shocks and accommodative policy lifted realised inflation, and wage indexation made the propagation mechanism especially powerful.
Disinflation, when it came, required a regime change deliberate enough to be perceived as one — not just tighter policy.
Limitations of the adaptive rule
Mechanically backward-looking. No role for forward information or expectation about the future.
Insensitive to announcements. The rule does not allow expectations to respond directly to credible policy regime changes.
Not model-consistent. The expectation-formation rule itself is fixed and exogenous to the model agents inhabit.
Rational expectations
Beliefs are the model-consistent conditional expectation, given the information set
Definition
$$\pi_t^{e} \;=\; E_{t-1}[\pi_t \mid \mathcal{I}_{t-1}]$$
$\mathcal{I}_{t-1}$: information set at the time the expectation is formed
Orthogonality (the testable property)
$$E_{t-1}\!\left[\pi_t - \pi_t^{e}\right] \;=\; 0$$
forecast errors are mean-zero given $\mathcal{I}_{t-1}$
What the assumption means
- Agents use all information in $\mathcal{I}_{t-1}$ when forming $\pi_t^e$.
- Systematic forecast-error patterns can be learned away.
- Only genuinely new information generates errors: $E_{t-1}[\pi_t-\pi_t^e]=0$.
Why it matters for policy
- A systematic policy rule is already embedded in $E_{t-1}[\pi_t\mid\mathcal{I}_{t-1}]$.
- Only unanticipated policy moves generate forecast errors.
- Credibility and communication become more central because they shape $\mathcal{I}_{t-1}$.
Rational expectations do not say that agents are always right — they say that agents are not predictably wrong. The discipline is on the structure of errors, not on the contents of beliefs.
What rational expectations does — and does not — claim
Three misconceptions and the actual content of the assumption
The misconception
What is actually claimed
01
Agents are always correct about the future.
→
Forecast errors occur, but they have zero mean conditional on the information set: $\boldsymbol{E_{t-1}[\pi_t - \pi_t^{e}] = 0}$.
02
Monetary policy is irrelevant — only surprises matter, and surprises are by construction unpredictable.
→
With sticky prices and wages, policy still has real effects. Unanticipated policy moves outcomes; the systematic part is anticipated and built into private decisions.
03
Communication is decorative — what matters is what the bank does.
→
Beliefs $\boldsymbol{\to}$ expectations $\boldsymbol{\to}$ current decisions. A regime that is understood and believed alters dynamics immediately; one that is announced but not believed does not.
Under rational expectations, credibility and communication are more central, not less. The bank does not have to fool the public — but it does have to be understood and believed.
Rational expectations: empirical illustration
New Zealand's inflation-targeting regime as a credible nominal anchor
N.Z. CPI inflation, annual average, 1990–1998
Annual average CPI inflation. Reserve Bank of New Zealand became inflation-targeting in 1990.
Reading the evidence
Inflation fell from roughly 7% in 1990 to near 1% by 1992 — faster than any backward-looking rule could generate from a sequence of demand shocks alone.
The shift to inflation targeting is a candidate for a credible nominal anchor: a regime change visible enough that private agents updated $\pi_t^{e}$ directly, without repeated policy surprises.
Temporary deviations from target after 1995 did not destroy credibility once the regime itself was understood.
A credible regime change, understood and believed, can shift current inflation dynamics immediately. Under rational expectations, credible communication is part of the transmission mechanism — not a sideshow.
Lawrence Schembri, Deputy Governor of the Bank of Canada
🔗
Source: Bank of Canada, 2019. ·
Link to speech
“It's not often that a policy performs better than expected. Our inflation-control target did just that, and continues to do so. Over the past 27 years, we have reduced inflation as measured by the CPI and maintained it at a level close to our 2 per cent target, with no persistent episodes outside our control range. It has served as an anchoring and coordinating mechanism, allowing Canadians to make better economic decisions and achieve better economic outcomes. In addition, there has been much less volatility in interest rates and output growth.”
◎ First lesson
The clarity and simplicity of the 2 per cent target has facilitated its communication and broad acceptance, and that has helped enhance the target's credibility.
⚓ Second lesson
As inflation expectations have become firmly anchored at the 2 per cent target, the effectiveness of the policy has increased.
↻ Third lesson
The more successful we are at hitting the target, the more credible the policy is and the more confident Canadians are that inflation will remain on target.
01/D
Section D
Trinity Term · 2026
Anchoring
expectations.
Credibility, persistence, and the output cost of returning inflation to target.
Anchored expectations
One forecasting rule, one parameter $\chi$, three nested regimes
expected inflation = weight on the target + weight on yesterday's inflation
Anchored expectation rule
$$ \pi_t^{\,e} \;=\; \chi\,\pi^{T} \;+\; (1-\chi)\,\pi_{t-1}, \qquad \chi \in [0,\,1] $$
$$ \boldsymbol{\chi = 1} $$
Fully anchored
$$ \boldsymbol{\pi_t^{\,e} \;=\; \pi^{T}} $$
Expectations are locked to the target: $\boldsymbol{\pi^T}$ is the forecast every period. Past inflation has zero weight.
$$ \boldsymbol{0 < \chi < 1} $$
Partially anchored
$$ \boldsymbol{\pi_t^{\,e} \;=\; \chi\,\pi^{T} + (1-\chi)\,\pi_{t-1}} $$
Expectations are a weighted average. The target pulls beliefs toward $\boldsymbol{\pi^T}$; experience pulls them toward $\boldsymbol{\pi_{t-1}}$.
$$ \boldsymbol{\chi = 0} $$
Unanchored
$$ \boldsymbol{\pi_t^{\,e} \;=\; \pi_{t-1}} $$
Expectations are purely adaptive. Yesterday's inflation, $\boldsymbol{\pi_{t-1}}$, becomes the forecast; the target has no role.
$\boldsymbol{\chi}$ is the credibility weight. It is not a parameter the modeller chooses; it is an institutional outcome of the regime, and it is the object that central bank communication is meant to raise.
Credibility as a measurable object
$\chi$ is inferred from whether long-horizon beliefs stay near target when current inflation moves
Where credibility is observed
Professional forecasters
SPF and market economists give explicit forecasts. Useful for asking whether experts still put weight on $\boldsymbol{\pi^T}$.
Household surveys
Noisier and more dispersed. They reveal whether the target is understood outside financial markets.
Breakeven inflation
Difference between nominal and inflation-indexed bond yields. It embeds expectations plus risk and liquidity premia.
5y5y inflation swap
A derivatives-market price for average inflation over years 5 to 10. Standard long-horizon anchoring gauge.
The latent object
$$ \pi_t^{\,e} = \chi\,\pi^{T} + (1-\chi)\pi_{t-1} $$
$\boldsymbol{\chi}$ is not directly observed. We infer it from how expectations respond to news.
High anchoring: one-year measures move with oil, gas, food, or tax shocks, but five-to-ten-year measures stay close to target.
Low anchoring: long-horizon measures drift when current inflation rises. Markets and households begin to doubt the regime.
How asset prices measure anchoring
Breakeven inflation
$$ i_t^{N} - i_t^{R} $$
$\boldsymbol{i_t^N}$ is the yield on a nominal government bond. $\boldsymbol{i_t^R}$ is the real yield on an inflation-indexed bond, such as TIPS or linkers. The spread prices expected inflation plus inflation-risk and liquidity premia.
5y5y inflation swap
A swap is a derivative contract: one side pays a fixed inflation rate, the other pays realised inflation. The 5y5y rate is the fixed rate for average inflation over years 5 to 10, so it looks past near-term shocks.
How to read the signal
Level. Is the long-horizon rate near the inflation target?
Sensitivity. Does it jump when current CPI or energy prices jump?
Caveat. It is not a pure expectation: premia and market liquidity must be stripped out.
Analytical setup, and a propagation law
One supply shock, $T$ periods, one credibility weight $\chi$, output gap held at zero
1. Phillips curve
$$ \pi_t = \pi_t^{\,e} + \alpha\,x_t + \varepsilon_t $$
Cost-push shock $\varepsilon_t$, output gap $x_t$, slope $\alpha > 0$.
2. Initial conditions
$$ \pi_{-1} = \pi^{T}, \quad x_{t} = 0 $$
Economy starts at target. Output gap held at zero to isolate propagation through expectations alone.
3. The shock
$$ \varepsilon_0 = \bar{\varepsilon} > 0,\;\; \varepsilon_t = 0\;\forall\,t \geq 1 $$
A single adverse supply shock at $t = 0$, nothing thereafter.
4. Anchored rule
$$ \pi_t^{\,e} = \chi\pi^{T} + (1-\chi)\pi_{t-1} $$
Hybrid rule from the previous slide. Single weight $\chi$ on the target.
Substitute, iterate · impact and propagation
$$ \pi_0 - \pi^{T} \;=\; \bar{\varepsilon}, \qquad
\pi_t - \pi^{T} \;=\; (1-\chi)\,(\pi_{t-1} - \pi^{T}) \;=\; (1-\chi)^{t}\,\bar{\varepsilon} $$
1. Define the state variable. Let $g_t \equiv \pi_t-\pi^T$. This is the inflation gap: how far inflation is from target after the initial shock.
2. Remove new disturbances. For $t\geq1$, $\varepsilon_t=0$ and $x_t=0$. Any movement in inflation now comes only from expectations carrying forward the old gap.
3. AR(1) law of motion. The gap follows $g_t=(1-\chi)g_{t-1}$. The eigenvalue is $\boldsymbol{1-\chi}$: the multiplier on yesterday's state.
4. Persistence mechanism. Iterating gives $g_t=(1-\chi)^t\bar{\varepsilon}$. Higher credibility means higher $\chi$, a smaller eigenvalue, and faster decay back to target.
Three cases nested by $\chi$, and why $\chi = 0$ is the costly corner
Reading $\pi_t - \pi^{T} = (1-\chi)^{t}\,\bar{\varepsilon}$ at the three benchmark values
$\chi = 1$ · fully anchored
$$ \pi_0 - \pi^{T} = \bar{\varepsilon}, \;\; \pi_t = \pi^{T}\;\forall\,t \geq 1 $$
One-period level effect. Inflation is back at target the period after the shock with no internal propagation.
$\chi = 0.5$ · partially anchored
$$ \pi_t - \pi^{T} = (0.5)^{t}\,\bar{\varepsilon} $$
Geometric decay at rate $0.5$. Half-life of one period. About $1/16$ of the original deviation remains by $t = 4$.
$\chi = 0$ · unanchored
$$ \pi_t - \pi^{T} = \bar{\varepsilon}\;\forall\,t \geq 0 $$
Permanent level shift. Without policy action, the temporary shock is built into every subsequent period.
Why $\chi = 0$ is the costly corner
$$ \pi_t^{\,e} = \pi_{t-1} \;\;\Longrightarrow\;\; \pi_t = \pi_{t-1} + \alpha\,x_t + \varepsilon_t
\;\;\Longrightarrow\;\;
\pi_0 = \pi^{T}+\bar{\varepsilon},\;\pi_1 = \pi_0,\;\pi_2 = \pi_1,\;\ldots $$
Under $\chi = 0$ the previous period's inflation is the next period's expectation. With $x_t = 0$, the original deviation $\bar{\varepsilon}$ is written into the next round of wage- and price-setting, and into the round after that. Returning inflation to target requires the central bank to engineer a negative output gap. The sacrifice ratio that follows is the subject of the closing slide.
Fully anchored expectations: $\chi = 1$
The shock raises inflation once. The Phillips curve returns by itself.
Algebra at $\chi = 1$
$$ \pi_0 - \pi^{T} = \bar{\varepsilon}, \qquad \pi_t - \pi^{T} = 0 \;\;\forall\,t \geq 1 $$
Mechanism · period by period
- Economy starts at the medium-run equilibrium $A = Z$, with $\pi = \pi^{T}$ and $y = y_e$.
- Shock at $t = 0$ shifts the Phillips curve up to $PC(\pi^{e}=\pi^{T},\,\bar{\varepsilon})$. Inflation rises to $\pi_0$ at point $B$.
- Because $\chi = 1$, the next period's expectation is still $\pi^{T}$. The Phillips curve drops back to its original position.
- The central bank holds the policy rate at the stabilising rate $r_s$. Output never leaves $y_e$.
The credibility of the target absorbs the shock. The episode is a price-level event, not an inflation regime.
Carlin & Soskice (2024), Figure 4.7, panel (a)
Partially anchored expectations: $\chi = 0.5$
Geometric path back to target; policy traces a sequence of contractions along the MR line
Closed form at $\chi = 0.5$
$$ \pi_t - \pi^{T} \;=\; (0.5)^{t}\,\bar{\varepsilon} $$
Mechanism · period by period
- $t = 0$: shock takes inflation to $\pi_0$ at point $B$ on $PC(\pi^{e}=\pi^{T},\,\bar{\varepsilon})$.
- $t = 1$: expectation $\pi_1^{e} = 0.5\,\pi^{T} + 0.5\,\pi_0 \equiv \pi'$. Phillips curve shifts to $PC(\pi^{e}=\pi')$. The bank raises the rate to $r_0$, output falls to $y_1$, inflation lands at point $C$ on $\pi_1$.
- $t = 2$: expectation updates to $\pi_2^{e} = \pi''$. The curve shifts down. The rate eases to $r_1$, output recovers toward $y_2$, inflation lands at point $D$.
- The economy converges back to $A = Z$ along the MR line, with the deviation falling by half each period.
Half-credibility halves the deviation each period. The output cost is real but bounded; the path is short.
Carlin & Soskice (2024), Figure 4.7, panel (b)
Unanchored expectations: $\chi = 0$
A temporary shock becomes a permanent regime shift unless policy intervenes
Algebra absent policy
$$ \pi_0 = \pi^{T} + \bar{\varepsilon}, \;\; \pi_1 = \pi_0, \;\; \pi_2 = \pi_1, \;\ldots $$
Mechanism
- Expectations follow last period's inflation: $\pi_t^{e} = \pi_{t-1}$. The target plays no role.
- $t = 0$: shock places inflation at $\pi_0$ at point $B$.
- $t = 1$: $\pi_1^{e} = \pi_0$. The Phillips curve shifts up by the full amount of the shock and stays there absent policy. The bank raises the rate to $r_0$, drives output to $y_1$, and lands at point $C$.
- $t = 2$: a further sequence of negative output gaps via $r_1, y_2$ and point $D$ is required. Convergence is slow and costly.
The 1970s episode in stylised form. The Phillips curve does not return on its own; it is dragged back by a sustained contraction.
Carlin & Soskice (2024), Figure 4.7, panel (c)
What policy must do
The output cost of disinflation is the unanchored part of the shock divided by the Phillips-curve slope
Policy problem in period 1
$$ \begin{aligned}
\pi_0 &= \pi^{T}+\bar{\varepsilon}, \qquad \varepsilon_1 = 0, \qquad
\pi_1^{e} = \chi\pi^{T} + (1-\chi)\pi_0 \\[-0.2em]
\pi_1=\pi^{T} &= \pi_1^{e}+\alpha x_1+\varepsilon_1
= \chi\pi^{T} + (1-\chi)(\pi^{T}+\bar{\varepsilon}) + \alpha\,x_1
\;\Longrightarrow\;
x_1 = -\,\frac{(1-\chi)\,\bar{\varepsilon}}{\alpha}
\end{aligned} $$
The new shock is gone; policy offsets the inherited expectation gap $\boldsymbol{(1-\chi)\bar{\varepsilon}}$.
What monetary policy offsets
$$ \text{Inherited expectation gap} = (1-\chi)\bar{\varepsilon} $$
The central bank is not offsetting the original shock itself. It is offsetting the portion that wage- and price-setters carry into the next period. Credibility reduces that inherited component before interest rates do anything.
WHY $\alpha$ MATTERS
$$ \alpha x_1 = - (1-\chi)\bar{\varepsilon} $$
$\alpha$ is the slope of the Phillips curve. If inflation is insensitive to slack, a larger negative output gap is required. If expectations are well anchored, less slack is needed for the same return to target.
A one-number example
Suppose the shock raises inflation by $2$ percentage points and $\alpha = 0.5$. Then $x_1 = -4(1-\chi)$.
Fully anchored
$$ \chi=1 \Rightarrow x_1=0 $$
No demand contraction is needed.
Half anchored
$$ \chi=0.5 \Rightarrow x_1=-2 $$
Only half the shock is inherited.
Unanchored
$$ \chi=0 \Rightarrow x_1=-4 $$
The whole shock must be offset by slack.
The new lesson is a policy decomposition: rate policy moves $\boldsymbol{x_1}$, but communication and credibility move $\boldsymbol{\chi}$. A higher $\chi$ lowers the sacrifice ratio before the central bank creates unemployment.
Forward guidance: words about the future move yields today
Communication aimed at the path component of the longer-maturity yield
longer-maturity yield = average expected short rate + term premium
Reduced-form yield decomposition
$$ y_t^{(n)} \;\approx\; \tfrac{1}{n}\sum_{j=0}^{n-1} E_t\!\left[i_{t+j}\right] \;+\; TP_t^{(n)} $$
1. The path component
$$ \tfrac{1}{n}\sum_{j=0}^{n-1} E_t\!\left[i_{t+j}\right] $$
The average expected short rate over the bond's life. Communication that shifts $E_t[i_{t+j}]$ moves $y_t^{(n)}$ today, before any rate decision.
2. The lower bound
$$ i_t = i^{ELB} $$
When the current short rate is at its effective lower bound, the only remaining rate channel runs through $E_t[i_{t+j}]$ for $j > 0$.
3. The commitment case
$$ E_t\!\left[i_{t+j}\right] \;<\; i^{*}_{t+j} $$
Eggertsson and Woodford (2003): optimal policy commits to keep rates lower for longer than discretion would later choose, raising current output and inflation.
Forward guidance is the explicit application of the expectations channel built up in the previous sections. The Draghi episode worked partly through the path component, alongside the sovereign-specific premium $SP_t$ that decomposed the same equation in Lecture 1.
Forward guidance by form
How the future policy path is written into the message
Qualitative
General language signals the likely direction of future policy without a date or numerical threshold.
“Rates will remain low for an extended period.”
Direction, not a precise trigger
Calendar-based
The policy path is linked to specific dates, horizons, or minimum time periods.
“At least through mid-2013.”
Time-contingent guidance
State-dependent
Policy changes are conditioned on observable economic outcomes.
“Until unemployment falls below 6.5%.”
Outcome-contingent guidance
Forward guidance by meaning
The economic content is either information about the outlook or a commitment about future policy
Delphic
Forecast-based signal
The central bank reveals its economic outlook and likely policy response.
- Based on forecasts and staff assessment.
- Reveals the bank's view of future conditions.
- Primarily informational: markets learn what the bank knows or believes.
Impact: helps markets infer the reaction function and the bank's assessment of the economy.
Odyssean
Commitment-based signal
The central bank tries to bind future policy actions to dates, states, or thresholds.
- Contains an explicit policy commitment.
- Can be time-contingent or state-contingent.
- Stronger signal: markets price the promised future stance.
Impact: moves expectations more forcefully when the commitment is credible.
Form and meaning are separate. A date or threshold can be read as a forecast, a commitment, or both; the market response depends on which interpretation is credible.
01/E
Section E
Trinity Term · 2026
Central bank
independence
and reputation.
Institutional credibility, political pressure, and the reputation mechanism.
Independence in practice
The bargain is delegated instruments, not delegated democracy
Distinction 01
Goal vs instrument independence
Modern central banks usually do not choose the objective. They receive a mandate and choose the instrument path needed to deliver it.
Distinction 02
De jure vs de facto independence
Legal protections matter, but practical independence can still be weakened by governor turnover, fiscal pressure, or quasi-fiscal tasks.
Why delegate monetary policy?
Inflation bias
Insulation helps the bank choose the ex ante plan rather than a politically tempting surprise inflation.
Electoral horizons
Price stability operates over horizons longer than the electoral cycle, especially when tightening is unpopular.
Fiscal dominance
Credibility falls if markets expect rates to be subordinated to debt-service or government financing needs.
Independence is not freedom from democracy. It is operational insulation inside a democratically assigned mandate, paired with explanation, transparency, and evaluation.
The empirical record, and three institutional designs
Evidence disciplines the case for independence; accountability disciplines the institution
Lower inflation
Alesina & Summers, 1993
More independent central banks are associated with lower average inflation and lower inflation volatility, without clear long-run output costs.
Political turnover matters
Governor-transition evidence
Politically motivated leadership changes are followed by higher realised and expected inflation, lower short rates, and a temporary growth impulse.
Credibility is observable
Expectations at long horizons
If realised inflation moves but long-horizon expectations remain close to target, the regime is still credible. Repeated inconsistency erodes that anchor.
Same delegation logic, different constitutional designs
US
Federal Reserve
Goal
Congress: dual mandate
Tool
FOMC
Scrutiny
Testimony, Monetary Policy Report, minutes, SEP, press conferences, transcripts.
EU
ECB
Goal
Treaty: price stability
Tool
Governing Council
Scrutiny
Annual Report, ECON hearings, written questions, monetary policy accounts.
UK
Bank of England
Goal
Government remit, renewed annually
Tool
MPC
Scrutiny
Monetary Policy Report, Treasury Committee, open letters after large target misses.
The empirical claim is not that central banks should be unaccountable. It is that credible instrument independence works when it is paired with visible, enforceable accountability.
Bank of England: the open letter rule
An accountability technology built into the inflation-targeting regime
- Trigger: CPI inflation more than 1 percentage point above or below the 2% target.
- Purpose: explain the miss, identify the shocks, and set out the path back to target.
- Tone: procedural and analytical, not apologetic; accountability through explanation.
The governing framework, now populated
Each link carries content from the lecture; outcomes feed back into credibility
01
Mandate and institution
A target, instrument independence, and public accountability define the regime.
→
02
Communication
Statements, minutes, forecasts, speeches, testimony, outreach, and even nonverbal signals.
→
03
Expectations
Policy path, reaction function, macro outlook, inflation anchor, and tail-risk beliefs.
↓
04
Financial conditions
Long rates, term premia, sovereign spreads, exchange rates, equity prices, and credit conditions.
←
05
Output and inflation
The realised macro outcome: activity, labour market pressure, inflation, and inflation persistence.
←
06
Credibility feedback
Success makes later words more powerful. Repeated misses make the same words cheaper.
Forward path: communication works through expectations and financial prices before the macro data arrive.
↺
Feedback path: outcomes validate or erode the institution that made the message credible.
Communication is not adjacent to monetary policy. It is part of the transmission mechanism, and the loop closes through credibility.
When communication works, and when it fails
The message is only as strong as the regime behind it
Works when
Four regime conditions
Clear objective
A public nominal anchor: inflation target, mandate, or equivalent destination.
Readable reaction function
Markets can infer how the bank responds to inflation, activity, and risk.
Instruments behind the words
The audience sees the policy capacity that would make the statement actionable.
Institutional credibility
Independence, accountability, and a record consistent with the stated objective.
Fails when
Five recurring breakdowns
Cheap talk under weak credibility
Words are discounted before they are tested.
Framework complexity
A stance that is hard to read loses precision through expectations.
Policy news vs information news
Markets react to the bank's economic read, not only to intended policy.
Audience heterogeneity
Markets, firms, households, and politicians parse different messages.
False precision under uncertainty
Overconfident paths damage credibility when reality moves.
Diagnosis
When communication fails, first inspect the regime: objective, reaction function, instruments, and credibility. Speaking more rarely fixes a broken framework.
Independence can be costly
Independence: in practice
Two episodes that test the institutional framework against real political pressure
Clip 1 · Jerome Powell said he would not step down from his position as Fed chief even if President-elect Donald Trump asked him to.
Clip 2 · Federal Reserve chairman Jerome Powell said he would not step down if Trump asked and that it is "not permitted under law" for the White House to force him out.
Two short clips, two ways of seeing the same point. Independence is a practical condition, observable in how a central bank conducts itself under pressure, and reflected in the credibility premium it earns or loses.
Reputation under stress: Venezuela, 2005
FT, 7 July 2005 · "Venezuela poised to seize bank's reserves"
Financial Times · 7 July 2005
Venezuela poised to seize bank's reserves
Andy Webb-Vidal in Caracas
Venezuela's central bank is bracing itself for a hostile takeover bid by an unlikely suitor: the government of President Hugo Chávez.
Legislators loyal to Mr Chávez are close to approving a law that will allow the government to withdraw and spend at least $5bn (€4.2bn, £2.9bn) of the bank's international reserves, which currently stand at $29bn.
For more than a year, Mr Chávez has insisted that the level of reserves accumulated by the world's fifth-largest oil exporter is too high, and that the money would be better used for social programmes. Among Latin American economies, Venezuela has the highest level of reserves as measured by equivalent weeks' worth of imports.
The central bank, says Mr Chávez, should belong to "the people", and it must come under full control of his radical nationalistic "Bolivarian revolution".
Government-aligned deputies, who maintain a narrow but effective majority in the National Assembly, began the final debate on the law on Thursday and they predict its passage next week. But the move is leaving some economists aghast at what they see as the demise of the bank's role as guardian of the bolívar, Venezuela's national currency. The bank has tried to resist the law.
Jose Guerra, economic research chief at the bank until earlier this year, says the measure will undermine the value of the currency, as some of the dollars will be converted twice into bolívars. The move also in effect opens the door to enabling Mr Chávez to finance Venezuela's chronic fiscal deficit with part of the reserves, he added.
"The big loser in all of this will be the credibility and the reputation of the central bank as an institution," said Mr Guerra. "Who's to say that after the first $5bn is withdrawn there won't be another $5bn that's taken out?"
Venezuela's international reserves are invested in a mixture of US Treasuries, Euro-denominated bonds, cash and gold. Critics say other state entities, such as state oil company Petroleos de Venezuela and Bandes, a state development bank, have about $10bn in overseas accounts, and the government should use some of that money instead of the bank's international reserves.
Gaston Parra, the central bank's president, may resign if the law is passed, sources at the bank say, because of the perceived "illegality" of the government-proposed legislation. The central bank could challenge the constitutionality of the law in Venezuela's supreme court. But analysts see the court as controlled by the government.
Economists predict that the expenditure of part of the reserves will stoke inflationary pressures, although the impact may be limited in the medium-term because of the existence of price and exchange controls.
"Investors are more concerned with the signal that is sent by the measure, especially given what they see as potential for the seized funds to be used in a non-transparent fashion," said Vitali Meschoulam, emerging markets strategist at HSBC Securities in New York. "There is a concern that these funds will not be used for productive investments but rather to finance current spending, increasing the risk that inflation may get out of hand."
Fewer international reserves may harm Venezuela's ability to service its foreign debt if oil prices decline. Under the draft law, some of the reserves are earmarked for vaguely-worded "strategic situations".
Mr Chávez, who has been in power for more than six years, faces presidential elections at the end of next year.
Source: Andy Webb-Vidal, "Venezuela poised to seize bank's reserves", Financial Times, 7 July 2005. Scroll within the panel to read the full article.
Turkey: when the Fisher equation is read backwards
"Recep Tayyip Erdoğan believes high inflation is the outcome of high interest rates. "
Fisher equation
$$ i \;=\; r \;+\; \pi^{e} $$
Nominal rate $=$ real rate $+$ expected inflation.
The doctrine
- The textbook reading: a credible cut in $\pi^{e}$ permits a lower $i$. Causation runs from inflation to nominal rates.
- The Erdoğan doctrine inverts the arrow: cut $i$ first, and $\pi$ will follow down. The Fisher identity is read as a causal claim from $i$ to $\pi$.
The institutional question is not whether the doctrine is correct — it is whether a central bank can be made to act on it.
Once you open the dam …
"Something wrong with every one, not me"
Is it too late?
The 2023 reversal: orthodox policy resumed; the credibility stock has to be rebuilt
Why disinflation is slow
- $\chi$ near zero: the propagation law $(1-\chi)^{t}\bar{\varepsilon}$ implies long persistence. Each percentage point of disinflation costs $1/\alpha$ output.
- Markets price reversal risk into long-horizon $\pi^{e}$.
- The interest-rate channel transmits weakly when the rule itself is in doubt.
- The new and strong governor is doing monetary policy. But
- The institutional repair — statutory protection, dismissal rules, fiscal-monetary separation — has not been done.
Reputation is a stock, depleted in months and rebuilt in years. The constraint that produces credibility has not yet been put back.
Cat = inflation
01/F
Section F
Trinity Term · 2026
Hawks and
Doves.
Inflation aversion, MR slopes, and how markets infer preferences from communication.
Hawks and doves: the everyday distinction made formal
A position on inflation aversion, recovered by markets from communication
Hawk
Dove
Main concern
Inflation credibility, anchoring of expectations, second-round wage effects.
Activity, employment, financial conditions, avoiding excessive slack.
Shock response
Tighten earlier and more decisively after an inflation shock.
Tighten gradually; tolerate a temporary inflation overshoot.
Trade-off
Accepts more short-run output loss to restore target faster.
Accepts more temporary inflation to limit the output cost.
Market signal
Higher expected policy rates and real yields; demand compression.
Lower expected policy path; more support to demand and credit.
Hawk and dove are not adjectives. They describe a position on the inflation-aversion weight in the policymaker's loss function, recovered by markets from speeches, statements, and votes. The next two slides give that weight a name.
Two equations behind the labels
The reaction function the bank follows; the loss function it minimises
Reaction function · Taylor form
$$ i_t \;=\; r^{*} \;+\; \pi_t \;+\; \phi_\pi(\pi_t - \pi^{T}) \;+\; \phi_x\,x_t $$
$\phi_\pi$ is how much the policy rate responds to a one-point inflation gap. $\phi_x$ is how much it responds to a one-point output gap. A hawkish stance shows up as larger $\phi_\pi$ relative to $\phi_x$.
Loss function · the deeper object
$$ L_t \;=\; (y_t - y_e)^{2} \;+\; \beta\,(\pi_t - \pi^{T})^{2}, \qquad \beta > 0 $$
$\beta$ is the relative weight on inflation stabilisation. A hawk has high $\beta$; a dove has low $\beta$. The Taylor coefficients $\phi_\pi$ and $\phi_x$ are themselves derived from $\beta$ and from the slope of the Phillips curve.
Markets do not observe $\beta$ directly. Practitioner hawk/dove indices and academic text-as-data work on speeches and votes are formally exercises in recovering the cross-section of $\beta$ inside a committee. The next slide turns $\beta$ into a slope.
The MR curve and its slope: $-1/(\alpha\beta)$
Loss minimisation subject to the Phillips curve gives the analytical referent for hawk and dove
IS-PC-MR · after Carlin & Soskice. All three curves cross at $A,Z = (y_e, \pi^{T})$.
Loss + Phillips curve $\Rightarrow$ MR
$$ \min_{y_1}\;(y_1-y_e)^{2} + \beta(\pi_1-\pi^{T})^{2} $$
$$ \text{s.t.}\;\; \pi_1 = \pi^{e} + \alpha(y_1-y_e) $$
$$ \Longrightarrow\;\; y_1-y_e \;=\; -\,\alpha\beta\,(\pi_1-\pi^{T}) $$
Slope in $(y,\pi)$ space
$$ \left.\dfrac{d\pi}{dy}\right|_{\text{MR}} \;=\; -\,\dfrac{1}{\alpha\beta} $$
- Higher $\beta$ (preferences) flattens MR ⇒ more hawkish.
- Higher $\alpha$ (structure) flattens MR at fixed $\beta$: a steeper Phillips curve makes policy more responsive per unit of output cost.
- Symmetrically, lower $\beta$ or lower $\alpha$ steepens MR ⇒ more dovish.
In everyday usage, hawk and dove refer to differences in $\beta$ across policymakers, holding $\alpha$ approximately constant. The slope of MR is what markets are trying to recover from communication.
One shock, two regimes
A positive cost-push shock $s_t$ shifts PC up; the bank moves along its MR
Same shock, same Phillips-curve shift, two MR slopes, two outcomes. The dovish bank ends up at $B$; the hawkish bank at $C$. The closed-form ratios are on the next slide.
Inflation vs unemployment
The trade-off as a ratio in $\beta$
Closed-form response of inflation, output, and unemployment to a one-off shock $s_t$
Inflation gap
$$ \pi_t - \pi^{T} \;=\; \dfrac{s_t}{1 + \alpha^{2}\beta} $$
Higher $\beta$ shrinks the inflation gap.
Output gap
$$ y_t - y_e \;=\; -\,\dfrac{\alpha\beta}{1 + \alpha^{2}\beta}\,s_t $$
Higher $\beta$ enlarges the negative output gap.
Unemployment gap · Okun's law
$$ u_t - u^{n} \;\approx\; -\,\kappa\,(y_t - y_e) $$
The output sacrifice translates into higher unemployment.
The hawk-vs-dove choice is a position on the ratio $\alpha\beta/(1+\alpha^{2}\beta)$; which value of $\beta$ is welfare-maximising depends on the state of expectations and on how transitory the shock is.
The ECB Council, ranked dove $\rightarrow$ hawk
Practitioner hawk/dove tracking is empirically a recovery exercise on $\beta$ across members
The model's reading
- High-unemployment members (Greece, Italy, Spain, Portugal) cluster at the dovish end. Implied $\beta$ lower; MR steeper.
- Low-unemployment, inflation-averse members (Germany, Netherlands, Austria) cluster at the hawkish end. Implied $\beta$ higher; MR flatter.
- Executive Board members carry the centrist line.
The qualification
- Council members are not national delegates. The mandate is euro-area price stability.
- National experience shapes priors and rhetoric; decisions are taken in the euro-area interest.
Hawk and dove are conditional on the regime: within a credible regime, the welfare gap between moderate hawks and moderate doves is small.
End · Lecture 1
Words, anchors,
credibility.
Fatih Kansoy
Central Bank Communication ·
01/