## The Macroeconomic Consequences of Undermining Central Bank Independence: Evidence from Governor Transitions (wpiea2026040-source-pdf)

## Source details

**Canonical URL:** [The Macroeconomic Consequences of Undermining Central Bank Independence: Evidence from Governor Transitions (wpiea2026040-source-pdf)](https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026040-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2026/english/wpiea2026040-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2026/english/wpiea2026040-source-pdf.pdf.json)

---

### Abstract and primary conclusions
- Paper studies macroeconomic consequences of undermining central bank independence through politically motivated transitions of central bank governors.
- Dataset: 132 central bank governor transitions in 28 advanced and emerging market economies since 2000.
- Main empirical findings:
  - Tenures of governors with politically motivated appointments are associated with higher and more volatile inflation, realized and expected.
  - Professional forecasters expect politically appointed governors to be more dovish in response to shifts in inflation.
  - Following announcements of politically motivated governor transitions:
    - Nominal and real short rates decline.
    - Expected and realized inflation rise.
    - GDP growth increases, consistent with an expansionary short-run macroeconomic impulse.
  - Effects are more pronounced when the incoming governor professes unorthodox views on monetary policy, implying a temporary growth–inflation trade-off.
  - Long-term inflation expectations rise only in the case of unorthodox governors with politically motivated appointments; otherwise long-term expectations do not materially change.

### Dataset, scope, and classifications
- Coverage and scope:
  - Time span: 2000 to 2024.
  - Observations: 132 governor transitions across 28 countries.
  - Country composition: 11 advanced economies and 17 emerging markets.
  - Coverage in 2024: 70 percent of GDP and 47 percent of global population.
- Dataset components:
  - (i) classification of transitions into politically and non-politically motivated;
  - (ii) timing (announcement dates of exits and appointments);
  - (iii) governor profiles (biographies and press coverage);
  - (iv) classification of incoming governor as orthodox or unorthodox.
- Defining politically motivated transitions (PMTs):
  - Classification based on biographies and newspaper articles reporting reasons for exits and appointments to identify political pressure versus non-political reasons.

### New stylized facts from the dataset
- Concentration and frequency:
  - Politically motivated transitions tend to be concentrated in a handful of countries.
  - Most advanced economies in the sample do not experience politically motivated transitions.
  - Four emerging markets have at least three quarters of transitions classified as politically motivated.
- Profile of appointees:
  - Governors appointed in politically motivated transitions are more likely to be unorthodox and have credentials considered less technocratic than governors appointed in non-politically motivated transitions.
- Long-term inflation expectations and PMT prevalence:
  - Long-term inflation expectations exceed targets by about 1 percent where politically motivated transitions are the majority.
  - Long-term inflation expectations exceed targets by over 2 percent where politically motivated transitions are the norm.
  - Expectations remain close to target in countries without political transitions.
  - No analogous relationship is found with de jure measures of central bank independence.

### Definitions, taxonomy, and sample statistics
- PMT classification decision logic:
  - If transition followed clear rule-based procedures AND no mention of executive displeasure or political favor in news → NMT.
  - Else if removal before end of term OR term-end transition with news suggesting friction with executive OR political motivation for appointment:
    - If exit motivated by personal scandal → NMT.
    - Else → PMT.
- Governor taxonomy (orthodox vs unorthodox) dimensions:
  - Mandate interpretation, Toolkit, Independence, Communication & rules, Risk attitude.
- Sample-wide counts and shares:
  - Full sample: 508 transitions.
  - Politically motivated: 213.
  - Routine (non-politically motivated): 295.
  - Share politically motivated (Full sample): 213 of 508 = 38 percent.
  - Emerging markets: 454 transitions; 219 politically motivated; share politically motivated 48 percent.
  - Advanced economies: 54 transitions; 4 politically motivated; share politically motivated 13 percent.
  - Orthodox: 327 transitions.
  - Unorthodox: 181 transitions.
- Governor background and education (selected comparisons PMT vs NMT):
  - PMT incoming governors main background shares: Central banking 16, Policy 16, Private sector 12, Politics 6, Academia 0.
  - Routine incoming governors main background shares: Central banking 46, Policy 25, Private sector 4, Politics 1, Academia 0.
  - PhD incidence:
    - PMT: 34 percent had a PhD.
    - NMT: 60 percent had a PhD.
  - Promotion from within central bank:
    - PMT: 30 percent.
    - NMT: 52 percent.
  - Prior minister position:
    - PMT: 22 percent.
    - NMT: 18 percent.
  - Orthodox vs unorthodox outcomes:
    - 36 percent of PMTs resulted in an unorthodox governor.
    - 12 percent of NMTs resulted in an unorthodox governor.
- Correlations with politically motivated dummy (Table 4, selected):
  - Positive correlations larger than 0.2: Political party, before: 0,33; Political party, now: 0,35; Politics: 0,23; Private sector: 0,24; Law degree: 0,23; Unorthodox: 0,28.
  - Negative correlations: Central banking: -0,24; PhD: -0,23; Foreign Education: -0,19; Economics degree: -0,21; Promoted: -0,24.

### Tenure patterns and macro links
- Average tenures (Table 5):
  - All: 5.1
  - Politically motivated (PMT): 3.4
  - Routine (NMT): 6.1
  - Unorthodox: 4.3
  - Orthodox: 5.3
  - PMT×Unorthodox: 3.2
  - PMT×Orthodox: 3.6
  - NMT×Unorthodox: 6.6
  - NMT×Orthodox: 6.1
- Stylized facts:
  - Governors leaving during politically motivated transitions have shorter tenures than those leaving during non-politically motivated transitions.
  - After conditioning on type of transition, tenure differences by governor type disappear.
  - Countries with more frequent PMTs tend to have long-term inflation expectations that hover close to inflation targets; countries with a majority of PMTs (share above 50 percent) tend to have long-term inflation expectations exceed targets.
  - No relationship found between de jure measures of central bank independence and the share of PMTs.

### Associations across tenures and mechanisms (performance gaps)
- Tenure associations (controlling for country and time fixed effects):
  - Realized inflation, short- and long-term inflation expectations are higher under politically appointed governors.
  - Inflation volatility, both realized and expected, is larger under politically appointed governors.
  - Transmission of inflation surprises to long-term inflation expectations is more than twice as large during tenures of politically appointed governors compared to other governors; difference driven by unorthodox governors.
  - Perceived Taylor rules: professional forecasters expect politically appointed governors to have lower sensitivity to inflation when setting the policy rate, particularly for unorthodox governors.

### Anchoring of inflation expectations (estimates)
- Model: Et πig,t+h = ρ Et−1 πig,t−1+h + (1−ρ) π* ig + δ(πigt − π* ig), where δ measures anchoring.
- Short-term transmission:
  - A one percentage point higher annualized inflation surprise translates into a 0.46 percentage points higher 1-year inflation expectation for the full sample.
- Estimated anchoring coefficients for longer horizons: 0.14, 0.9, 0.07 and 0.04 for 2-, 3-, 4- and 5-year horizons respectively.
- Differences by governor type (1-year and 3-year horizons):
  - 1-year horizon:
    - PMT governors: anchoring coefficient 0.79.
    - NMT governors: anchoring coefficient 0.36.
    - PMT–unorthodox: 0.94.
    - PMT–orthodox: 0.45.
  - 3-year horizon:
    - PMT governors: 0.16.
    - NMT governors: 0.06.
    - PMT–unorthodox: 0.20.
    - PMT–orthodox: 0.03.
- Growth expectations:
  - No evidence that real GDP growth surprises transmit to long-term growth expectations for any governor type.

### Perceived Taylor rules and forecaster expectations
- Policy-rule setup: ιigt = r* ig + π* ig + φ(πigt − π* ig) + uigt; focus on φ.
- Forecasters’ perceived rule (estimates reported in Table 6):
  - For NMT governors, a one percentage point increase in expected inflation is associated with a 0.73 percentage point increase in the expected policy rate.
  - PMT × E(inflation) interaction coefficient: −0.239*** (indicating 0.24 percentage points lower sensitivity for PMT governors).
  - PMT-u × E(inflation): −0.241***.
  - PMT-o × E(inflation): −0.141**.
- Robustness:
  - Results hold using expected real policy rate, with robust standard errors and R2 values ranging from 0.845 to 0.970; observations count: 5357 across specifications.

### Forecaster-level perceived reaction functions (Table 7 highlights)
- Key coefficients (selected):
  - Expected inflation coefficients around 0.361***, 0.219***, -0.781***, 0.181*** across specifications (standard errors reported in parentheses).
  - PMT x E(inflation): approximately -0.168*** to -0.212*** across specifications.
  - PMT-u x E(inflation): approximately -0.199*** to -0.216***.
  - PMT-o x E(inflation): smaller, sometimes insignificant (e.g., -0.0649** to -0.0779).
- Interpretation:
  - PMT governors are perceived to be more dovish (lower φ), driven primarily by unorthodox PMT governors.
  - Results robust to controls including policy-rate inertia, expected GDP growth, and expected depreciation.

### Causal identification and dynamic impacts (LP-DiD)
- Identification strategy:
  - Local projection difference-in-differences (LP-DiD) controlling for macroeconomic pre-trends and selection on observables.
  - Dataset for LP-DiD covers 26 countries from 2000 to 2024 (Venezuela and Ukraine excluded in this sample).
  - Two treatment indicators: binary PMT and binary NMT; treatment is non-absorbing and staggered.
  - Clean controls approach used and controls include up to 3 lags of first differences of outcome variables.
- LP-DiD specification:
  - ∆h y c,t+h = βP h PMT c,t + βN h NMT c,t + αh c + αh t + sum_{l=1}^3 γh l X c,t−l + v h c,t for h = 1,...,24.
  - Study horizons up to 24 months after the transition; average effects reported for 0-12 months, 13-24 months, and 0-24 months.
  - Inference based on joint wild bootstrap clustered at country level.

### Dynamic causal effects following announcements
- Aggregate PMT effects (0–24 months):
  - Short nominal and real rates decline after a politically motivated transition; no notable change after NMT.
  - Expected and realized inflation rise following PMTs; average effect becomes significantly positive during the first year after transition (estimates subject to substantial uncertainty).
  - Real GDP growth accelerates in the short run after PMTs.
  - Fiscal balance expectations remain broadly stable and statistically indistinguishable from zero following PMTs.
  - Exchange-rate responses are imprecisely estimated and exhibit pre-trends.
- Heterogeneity by governor type (equation 12 results):
  - Short rate declines begin around six months after transition for both orthodox and unorthodox subsamples; magnitude larger for unorthodox governors.
  - Real rate declines more pronounced for unorthodox governors, especially beyond twelve months.
  - Realized and expected inflation responses stronger for politically motivated appointments of unorthodox governors; increases tend to emerge after approximately nine months for unorthodox appointments.
  - GDP growth (actual and expected) rises following PMTs; rises slightly more robust statistically for orthodox appointments.
  - Fiscal balance expectations remain near zero for both governor types.
  - Overall: effects substantially larger and more persistent when incoming politically appointed governor is classified as unorthodox.

### Effects on long-term expectations (synthetic control)
- Methodology:
  - Synthetic control method (SCM) applied for long-term (5-year ahead) inflation expectations due to bi-annual frequency of series.
  - Horizon: 9 quarters with middle quarter as treatment; donor pool excludes countries with their own transition during the 9-quarter window.
- Long-term inflation expectations (SCM results):
  - PMT group: long-term inflation expectations diverge from synthetic control in two years following transition; after two years PMT expectations are 0.2 percentage points higher than control group median.
  - NMT group: track control closely pre- and post-treatment.
  - Unorthodox PMT: long-term inflation expectations rise by about 1 percentage point relative to control over two years post-appointment.
  - Orthodox PMT: long-term inflation expectations remain broadly unchanged relative to control.
- Long-term growth expectations:
  - No evidence that long-term growth expectations change materially following any transition type.
  - Appointments of unorthodox governors: long-term growth expectations decline somewhat relative to control by about 0.2 percentage points (small).

### Interpretation, mechanisms, and model implications
- Short-term trade-off:
  - Political interference in central bank leadership induces a short-term inflation–growth trade-off: higher growth in the short run comes at the expense of higher inflation, even when incoming governors are orthodox.
- Credibility costs:
  - Costs to central bank credibility are more pronounced when politically motivated transitions install unorthodox governors: higher long-term inflation expectations and reduced perceived responsiveness of policy rates to inflation.
- Heterogeneity by governor type:
  - Incoming orthodox governors face better inflation–growth trade-offs; unorthodox appointees are perceived as a negative credibility shock that raises expected long-term inflation without improving expected long-term growth.
- Simple model motivation:
  - New Keynesian setup with central bank loss L_t = 1/2 [ π_t^2 + α (x_t − θ)^2 ], where θ captures preference for a positive output gap (political pressure) and α captures dovishness.
  - Steady-state implications:
    - Steady-state inflation: π∗ = κθ / (1−β+ κ/2 α)
    - Steady-state output gap: x∗ = θ / (1 + κ/2 α(1−β))
    - Real interest rate gap: r∗ − rn = − θ/σ / (1 + κ/2 α(1−β))
  - Implication: increases in θ and/or α raise steady-state inflation and weaken long-term anchoring.

### Key empirical conclusions (paper-level summary)
- Dataset: 132 governor transitions across 28 advanced and emerging economies since 2000.
- Politically motivated changes in central bank leadership are associated with:
  - Higher and more volatile inflation.
  - More poorly anchored inflation expectations.
  - More dovish expected policy responses.
  - Declines in nominal and real short-term interest rates following announcement.
  - Short-run rise in GDP growth, indicating an expansionary macroeconomic impulse.
- Effects are strongest when incoming governors hold unorthodox monetary policy views.
- Long-term inflation expectations rise only for politically motivated appointments of unorthodox governors; long-term growth expectations do not move materially.
- Overall interpretation: political interference generates a temporary trade-off between higher growth and higher inflation, but political appointment of unorthodox governors weakens medium-to-long-term central bank credibility.

*Prepared by Marijn A. Bolhuis, Rui C. Mano and Hedda Thorell; IMF Working Paper WP/26/40; March 2026. Source: wpiea2026040-source-pdf*

### Section 1

### The Macroeconomic Consequences of Undermining Central Bank Independence: Evidence from Governor Transitions

### Abstract and primary conclusions
- Paper studies macroeconomic consequences of undermining central bank independence through politically motivated transitions of central bank governors.
- Dataset: 132 central bank governor transitions in 28 advanced and emerging market economies since 2000.
- Main empirical findings:
  - Tenures of governors with politically motivated appointments are associated with higher and more volatile inflation, realized and expected.
  - Professional forecasters expect politically appointed governors to be more dovish in response to shifts in inflation.
  - Following announcements of politically motivated governor transitions:
    - Nominal and real short rates decline.
    - Expected and realized inflation rise.
    - GDP growth increases, consistent with an expansionary short-run macroeconomic impulse.
  - Effects are more pronounced when the incoming governor professes unorthodox views on monetary policy, implying a temporary growth–inflation trade-off.
  - Long-term inflation expectations rise only in the case of unorthodox governors with politically motivated appointments; otherwise long-term expectations do not materially change.

### Dataset, scope, and classifications
- Coverage and scope:
  - Time span: 2000 to 2024.
  - Observations: 132 governor transitions across 28 countries.
  - Country composition: 11 advanced economies and 17 emerging markets.
  - Coverage in 2024: 70 percent of GDP and 47 percent of global population.
- Dataset components:
  - (i) classification of transitions into politically and non-politically motivated;
  - (ii) timing (announcement dates of exits and appointments);
  - (iii) governor profiles (biographies and press coverage);
  - (iv) classification of incoming governor as orthodox or unorthodox.
- Defining politically motivated transitions (PMTs):
  - Classification based on biographies and newspaper articles reporting reasons for exits and appointments to identify political pressure versus non-political reasons.

### New stylized facts from the dataset
- Concentration and frequency:
  - Politically motivated transitions tend to be concentrated in a handful of countries.
  - Most advanced economies in the sample do not experience politically motivated transitions.
  - Four emerging markets have at least three quarters of transitions classified as politically motivated.
- Profile of appointees:
  - Governors appointed in politically motivated transitions are more likely to be unorthodox and have credentials considered less technocratic than governors appointed in non-politically motivated transitions.
- Long-term inflation expectations and PMT prevalence:
  - Long-term inflation expectations exceed targets by about 1 percent where politically motivated transitions are the majority.
  - Long-term inflation expectations exceed targets by over 2 percent where politically motivated transitions are the norm.
  - Expectations remain close to target in countries without political transitions.
  - No analogous relationship is found with de jure measures of central bank independence.

### Associations across tenures and mechanisms
- Tenure associations (controlling for country and time fixed effects):
  - Realized inflation, short- and long-term inflation expectations are higher under politically appointed governors.
  - Inflation volatility, both realized and expected, is larger under politically appointed governors.
  - The transmission of inflation surprises to long-term inflation expectations is more than twice as large during tenures of politically appointed governors compared to other governors; difference driven by unorthodox governors.
  - Perceived Taylor rules: professional forecasters expect politically appointed governors to have lower sensitivity to inflation when setting the policy rate, particularly for unorthodox governors.

### Causal identification and dynamic impacts
- Identification strategy:
  - Use local projection based difference-in-differences (LP-DiD) controlling for macroeconomic pre-trends and selection on observables.
  - Verify effects are not driven by concurrent expectations of looser fiscal policy.
- Dynamic causal effects following announcement of a politically motivated transition:
  - Nominal and real short rates decline.
  - Expected and realized inflation rise.
  - Real GDP growth accelerates.
  - Effects are sharper and more statistically robust when the incoming governor is unorthodox, but are present even for incoming orthodox governors.
- Long-term expectations (limited data; synthetic control method used):
  - Following a politically motivated transition with an unorthodox incoming governor, long-term inflation expectations rise by 1 percentage point more than the control group in the two years after the appointment.
  - Long-term inflation expectations remain unchanged following non-politically motivated transitions and politically motivated transitions with an orthodox incoming governor.
  - No evidence that long-term growth expectations change following any of the transitions.

### Interpretation and implications
- Short-term trade-off:
  - Political interference in central bank leadership induces a short-term inflation–growth trade-off: higher growth in the short run comes at the expense of higher inflation, even when incoming governors are orthodox.
- Credibility costs:
  - Costs to central bank credibility are more pronounced when politically motivated transitions install unorthodox governors: higher long-term inflation expectations and reduced perceived responsiveness of policy rates to inflation.
- Heterogeneity by governor type:
  - Incoming orthodox governors face better inflation–growth trade-offs; unorthodox appointees are perceived as a negative credibility shock that raises expected long-term inflation without improving expected long-term growth.

### Relation to existing literature and theoretical framework
- Builds on literature linking de jure central bank independence to lower and less volatile inflation, while addressing empirical identification limitations of de jure indices.
- Complements studies of governor appointments and political motivations (Romer and Romer [2004]; Göh lmann and Vaubel [2007]; Ehrmann and Fratzscher [2011]; Vuletin and Zhu [2011]; Ennser-Jedenastik [2014]; Mishra and Reshef [2019]; Malmendier et al. [2021]; Ioannidou et al. [2022]).
- Offers cross-country, high-frequency LP-DiD evidence and synthetic control analysis on short- to long-term transmission of political shocks to policy, inflation, and credibility.
- Connects to theoretical mechanisms (Barro and Gordon [1983a]; Rogoff [1985]; Afrouzi et al. [2024]; Clayton and Schaab [2025]; Svec and Tortorice [2025]) on inflation bias, delegation, and political economy of credibility.

### JEL classification and keywords
- JEL Classification Numbers: D72, E31, E52, E58, F41
- Keywords: Central bank independence, political interference, central bank governors, monetary policy credibility, inflation dynamics

*Prepared by Marijn A. Bolhuis, Rui C. Mano and Hedda Thorell; IMF Working Paper WP/26/40; March 2026.*

### Section 2

### Section 2

### Definition and classification of politically motivated transitions (PMTs)
- Governors transitions are classified using two sets of criteria:
  - First, exclude transitions that follow clear, rule-based procedures AND where there is no mention of displeasure with the incumbent or particular political favor with the incoming governor in contemporaneous news sources leading up to the removal and/or appointment.
  - Second, for remaining transitions (removals before end of term, or term-end transitions with news suggesting friction with the executive, or political motivation for appointment) all three authors independently read news for the transitions to ascertain the motive for exit or appointment.
- Transitions motivated by personal scandals were removed from the politically-motivated category; only transitions motivated by short-term political considerations were classified as politically-motivated transitions (PMTs). All other transitions were classified as non-politically motivated (NMTs).
- Primary news sources: centralbanking.com, supplemented when needed with Bloomberg and the Financial Times.
- Decision logic (summarized):
  - If transition followed clear rule-based procedures AND no mention of executive displeasure or political favor in news → NMT.
  - Else if transition was removal before end of term OR term-end transition with news suggesting friction with executive OR political motivation for appointment:
    - If exit motivated by personal scandal → NMT.
    - Else → PMT.

### Timing, data collection, and governor profiles
- Announcement dates:
  - Hand-collected earliest announcement dates among international sources (Bloomberg, Financial Times, Centralbanking.com) and local newspapers.
- Exact start and end dates in office:
  - Hand-collected primary from central banks’ own websites, complemented by news when missing.
- Governor profile variables hand-collected include:
  - name, date of birth, gender, citizenship(s), education (major and level of degree), foreign education, main work background before appointment (academia, central banking, policy, politics, private sector), promoted from within central bank, connection to executive through family or political party, party membership, prior minister position, prior work at other central bank or international financial institutions.

### Taxonomy and classification of orthodox vs unorthodox governors
- Broad definitions:
  - Orthodox: sticks to conventional instruments, rule-like policy frameworks (e.g., inflation targeting), tight separation between monetary and fiscal policy.
  - Unorthodox: favors unconventional tools, looser adherence to traditional rules, willingness to experiment or coordinate with fiscal authorities.
- Dimensions in taxonomy (as used to classify):
  - Mandate interpretation (narrow price stability vs flexible broader goals).
  - Toolkit (conventional short-term policy rate, reserve requirements vs QE, negative rates, yield-curve control, direct credit programs, temporary direct financing).
  - Independence (maintains institutional independence vs more willing to coordinate/tolerate fiscal financing).
  - Communication & rules (transparent, rule-like vs experimental rhetoric).
  - Risk attitude (cautious about balance-sheet/moral-hazard risks vs willing to accept larger balance sheets/higher risk).
- Classification procedure:
  - Taxonomy fed into a prompt to an LLM that collects public newspaper articles mentioning governor profiles fitting taxonomy.
  - Classification uses newspaper articles spanning the full term of the governor (not limited to announcement period).

### Stylized facts on governor transitions (2000–2024)
- Sample and overall shares:
  - Full sample: 508 transitions.
  - Politically motivated: 213.
  - Routine (non-politically motivated): 295.
  - Share politically motivated (Full sample): 213 of 508 = 38 percent.
- By economy type:
  - Emerging markets: 454 transitions; 219 politically motivated; share politically motivated 48 percent.
  - Advanced economies: 54 transitions; 4 politically motivated; share politically motivated 13 percent.
- By governor type:
  - Orthodox: 327 transitions; 210 politically motivated? (Table shows Orthodox 327, 210, 431 % — note: table layout indicates Orthodox total 327, politically motivated 210, share 43 percent).
  - Unorthodox: 181 transitions; 28 politically motivated; share 64 percent (table shows Unorthodox 181, 28, 64 %).
- Stylized fact 1 — concentration:
  - Politically motivated transitions are more prevalent in emerging markets and highly concentrated in some countries.
  - 17 out of 28 countries experience at least one PMT during 2000 to 2024:
    - 13 out of 17 in emerging markets.
    - 4 out of 17 in advanced economies.
  - Eleven countries only experience non-politically motivated transitions during 2000 to 2024:
    - 7 in advanced economies and 4 in emerging markets.
  - Four countries have at least 75 percent of all transitions politically motivated (top left of distribution).
- Stylized fact 2 — profiles of incoming governors (comparison PMT vs routine):
  - Main background before appointment (shares):
    - PMT incoming governors: Academia 0, Central banking 16, Policy 16, Politics 6, Private sector 12.
    - Routine incoming governors: Academia 0, Central banking 46, Policy 25, Politics 1, Private sector 4.
  - Level of education:
    - PhD: PMT 17 percent; Routine 34 percent; All 60 percent (table shows PhD 17 50 34 60 — interpret as PMT 17, routine 50? but text clarifies: "34 percent of all incoming governors in politically motivated transitions had a PhD, while 60 percent had it in the case of non-politically motivated transitions." Use text values:)
      - PMT: 34 percent had a PhD.
      - NMT: 60 percent had a PhD.
  - Foreign education:
    - Lower for politically motivated transitions (text states this qualitatively).
  - Degree major (most common overall: economics).
  - Political party membership:
    - For PMTs: 50 percent had been a party member in the past; 30 percent were a party member when starting their term.
  - Promotion from within central bank:
    - PMT: 30 percent promoted within central bank.
    - NMT: 52 percent promoted within central bank.
  - Prior minister position before appointment:
    - PMT: 22 percent.
    - NMT: 18 percent.
  - Orthodox vs unorthodox outcomes:
    - 36 percent of PMTs resulted in an unorthodox governor.
    - 12 percent of NMTs resulted in an unorthodox governor.
- Table 4 correlations (correlation between politically motivated dummy and characteristics):
  - Positive correlations larger than 0.2:
    - Political party, before: 0,33
    - Political party, now: 0,35
    - Politics: 0,23
    - Private sector: 0,24
    - Law degree: 0,23
    - Unorthodox: 0,28
  - Negative correlations:
    - Central banking: -0,24
    - PhD: -0,23
    - Foreign Education: -0,19
    - Economics degree: -0,21
    - Promoted: -0,24
  - Other correlations:
    - Academia: -0,18
    - Policy: 0,07
    - Business: 0,01
    - Minister: 0,03
    - IFI: -0,13
- Stylized fact 3 — tenure and macro performance links:
  - Average tenures:
    - Governors leaving during politically motivated transitions: 3.4 years on average.
    - Governors stepping down during non-politically motivated transitions: 6.1 years on average.
    - Unorthodox governors: 4.3 years average tenure.
    - Orthodox governors: 5.3 years average tenure.
    - After conditioning on type of transition, tenure differences by governor type disappear.
  - Relationship with macro indicators:
    - Countries with more frequent PMTs tend to have long-term inflation expectations that hover close to inflation targets; countries with a majority of PMTs (share above 50 percent) tend to have long-term inflation expectations exceed targets.
    - No relationship found between de jure measures of central bank independence (Romelli [2024]) and the share of politically motivated transitions.
- Literature links and interpretations:
  - High governor turnover correlates with political and regime instability and high inflation (prior literature: Dreher et al. [2008, 2010]; Bodea and Garriga [2023]).
  - Governor characteristics affect policy preferences and outcomes (examples: Chappell Jr et al. [1995]; Malmendier et al. [2021]; Göh lmann and Vaubel [2007]; Mishra and Reshef [2019]).

*Source: wpiea2026040-source-pdf - Section 2*

### Section 3

### 3. Macroeconomic Effects

### 3.1. Monthly data
- Matched governor transition announcement dates with monthly macroeconomic data: realized inflation, short rates (3 month maturity), long rates (10 year maturity), GDP growth rates, and exchange rates vis-à-vis the dollar.
- Forward-looking expectations from Consensus Economics (collected once a month):
  - Short rates: 3 and 12 months ahead.
  - Exchange rate: 3, 12 and 24 months ahead.
  - Current and next year expectations for inflation, real GDP growth, and overall fiscal balance.
  - Long-term inflation expectations for 2 to 10 years ahead available at lower frequency (bi-annually until 2014, and quarterly from 2014) and therefore unsuitable for LP-DiD analysis.
- Transformation of fixed-event forecasts (current-year and next-year) into 12-month-ahead forecasts via a geometric transformation:
  - π12m t = 100 × [ (1 + πcurrent year t /1200)^(12−month+1) × (1 + πnext year t /1200)^(month−1) − 1 ]
- Fiscal balance expectations expressed as percent of nominal GDP; when only nominal balance forecast is available, expected nominal GDP for current and next year is inferred by combining forecasts for real GDP and inflation; fixed-event forecasts for fiscal balance transformed into 12-month ahead forecasts by taking a simple weighted mean.
- CPI-release expectations from Bloomberg’s survey of professional forecasters: median forecast (50th percentile) just prior to each announcement used as market consensus.
- Consensus Economics reports published on the second or third Monday of the month.

### 3.2. Performance gaps: simple associations between outcomes and governor tenures
- Empirical framework:
  - Panel specification: Yigt = β TenurePMT igt + αt + αi + εigt, where TenurePMT is a dummy equal to one if the governor was appointed during a politically motivated transition (PMT).
  - Extended specification distinguishing PMT-unorthodox and PMT-orthodox: Yigt = βu TenurePMTu igt + βo TenurePMTo igt + αt + αi + εigt.
- Inflation performance (Figure 6):
  - Under PMT governors, countries experience higher actual inflation and higher short- and long-term inflation expectations compared to NMT tenures.
  - This holds for unconditional gaps, gaps relative to time averages, and gaps relative to time and country averages.
  - Differences particularly pronounced for PMT governors classified as unorthodox.
- Volatility of inflation outcomes (Figure 7):
  - Inflation and inflation expectations are more volatile during tenures of governors appointed in PMT compared to NMT tenures.
  - Volatility gaps are particularly pronounced for PMT–unorthodox governors.
- Performance relative to predecessor (Equations 4–5; Figure 8):
  - Outcomes are generally worse during the tenure of a PMT governor relative to her predecessor.
  - Some effects arise from PMT governors following other PMT governors.
  - Unorthodox PMT governors drive these relative-worsening results.

Key reported average tenure statistics (Table 5):
- Average tenure (years)
  - All: 5.1
  - Politically motivated (PMT): 3.4
  - Routine (NMT): 6.1
  - Unorthodox: 4.3
  - Orthodox: 5.3
  - PMT×Unorthodox: 3.2
  - PMT×Orthodox: 3.6
  - NMT×Unorthodox: 6.6
  - NMT×Orthodox: 6.1

### 3.3. Anchoring of inflation expectations across governor tenures
- Model for long-term inflation expectations (Equation 6):
  - Et πig,t+h = ρ Et−1 πig,t−1+h + (1−ρ) π* ig + δ(πigt − π* ig)
  - δ measures degree of anchoring: δ = 0 implies long-term expectations equal central bank target; δ > 0 implies long-term expectations drift with realized inflation above target.
- Estimation approach:
  - Use monthly inflation surprises constructed from Bloomberg survey of professional forecasters for each CPI release; regress change in long-term inflation expectations on inflation surprise and controls.
  - Sample restricted to changes in long-term inflation expectations within the same governor tenure; exclude observations covering multiple tenures.
  - Restrict sample to countries with at least 10 years of inflation surprises.
- Estimated transmission (reported in text and Figure 9):
  - Short-term: a one percentage point higher annualized inflation surprise translates into a 0.46 percentage points higher 1-year inflation expectation for the full sample.
  - Estimated anchoring coefficients for longer horizons: 0.14, 0.9, 0.07 and 0.04 for 2-, 3-, 4- and 5-year horizons respectively.
- Differences by governor type:
  - Transmission of inflation surprises to expectations considerably stronger during PMT tenures than NMT tenures.
  - At 1-year horizon: anchoring coefficient for PMT governors 0.79 versus 0.36 for NMT governors.
  - For 3-year ahead expectations: transmission is 0.16 for PMT governors and 0.06 for NMT governors.
  - Differences driven entirely by PMT–unorthodox governors:
    - PMT–unorthodox: anchoring coefficient 0.94 at 1-year horizon, 0.20 at 3-year horizon.
    - PMT–orthodox: anchoring coefficients 0.45 at 1-year horizon and 0.03 at 3-year horizon.
  - Differences between unorthodox PMT and NMT governors significant at every horizon; no significant differences between orthodox PMT and NMT governors.
- Growth expectations:
  - No evidence that real GDP growth surprises transmit to long-term growth expectations for any governor type (Figure 10).

### 3.4. Perceived Taylor rules across governor tenures
- Monetary policy rule setup (Equation 8):
  - ιigt = r* ig + π* ig + φ(πigt − π* ig) + uigt
  - Focus on φ, the sensitivity of policy rate to the inflation gap.
- Perceived monetary policy rule estimated from forecasters’ expectations (Equation 9):
  - Et−1 ιigt = αig + ˜φ Et−1 πigt + ˜φ PMT Et−1 πigt + ueigt
  - Governor fixed effects αig capture governor-specific implicit inflation target and natural rate.
- Main estimates (Table 6):
  - For NMT governors, a one percentage point increase in expected inflation is associated with a 0.73 percentage point increase in the expected policy rate.
  - Forecasters expect PMT governors to have lower sensitivity to inflation:
    - The PMT × E(inflation) interaction coefficient is reported as −0.239*** (column (1) and repeated across specifications), indicating sensitivity is 0.24 percentage points lower for PMT governors relative to NMT governors.
  - When splitting PMT types:
    - PMT-u × E(inflation): −0.241*** (similar lowering effect).
    - PMT-o × E(inflation): −0.141** (smaller lowering effect).
- Additional notes on Table 6:
  - Dependent variables reported include expected policy rate and expected real policy rate across columns.
  - Robust standard errors reported in parentheses; ***, **, * denote significance at 1, 5, and 10 percent levels.
  - Observations across columns: 5357 (repeated).
  - Reported R2 values range from 0.845 to 0.970 across specifications.

*Source: Section 3 of the provided content unit.*

### Section 4

### Section 4

### Perceived monetary policy reaction function (forecaster-level)
- Forecaster-level specification (equation 10) estimates whether interest-rate forecasts become more compressed relative to dispersion in inflation forecasts under PMT governors versus NMT governors, introducing forecaster-country fixed effects and country-governor-time fixed effects to control for time-varying neutral rates and implicit inflation targets.
- Key coefficient estimates from Table 7 (selected):
  - Expected inflation: 0.361*** (0.0424); 0.219*** (0.0523); -0.781*** (0.0523); 0.181*** (0.0595); 0.387*** (0.0396); 0.222*** (0.0530); -0.778*** (0.0530); 0.183*** (0.0598).
  - PMT x E(inflation): -0.168*** (0.0207); -0.209*** (0.0568); -0.209*** (0.0568); -0.212*** (0.0669).
  - PMT-u x E(inflation): -0.199*** (0.0276); -0.213*** (0.0575); -0.213*** (0.0575); -0.216*** (0.0673).
  - PMT-o x E(inflation): -0.0649** (0.0284); -0.0251 (0.132); -0.0251 (0.132); -0.0779 (0.141).
- Fixed effects and controls:
  - Forecaster-country FE: Yes in all columns.
  - Country-time FE: No in column 1; Yes in columns 2-4 and 6-8.
  - Initial interest rate control and E[GDP growth] control included in columns 4 and 8.
- Summary statistics from Table 7:
  - Observations: 6844768182681826818268447681826818268182
  - R2: 0.8030.9650.9890.9680.8070.9660.9890.968
- Interpretation:
  - PMT governors are perceived to be more dovish, driven primarily by unorthodox PMT governors (strong negative PMT-u x E(inflation) coefficients).
  - Results are robust to using the real interest rate, controlling for policy-rate inertia, expected GDP growth, and expected depreciation of nominal and real exchange rates.
  - Forecaster-level specification addresses concerns about correlation between expected monetary policy shocks and expected inflation by absorbing forecaster disagreement and country-governor-time common beliefs.

### Outcomes following governor transitions using LP-DiD
- Identification strategy:
  - Use local projection difference-in-difference (LP-DiD) methods of Dube et al. [2025] to estimate dynamic effects of governor transitions.
  - Dataset covers 26 countries from 2000 to 2024 (Venezuela and Ukraine excluded in this sample; politically motivated transitions during pegged-currency periods, wartime participation, or IMF SBA/EFF program periods are excluded).
  - Two treatment indicators: binary politically motivated transition (PMT) and binary non-politically motivated transition (NMT). Treatment is non-absorbing and staggered.
  - Clean controls approach (Dube et al. [2025]) used to restrict controls to countries in which dynamic effects of prior treatment have stabilized.
  - Controls include up to 3 lags of lagged first differences of outcome variable, inflation, GDP growth, log exchange rate, and short rate.
- LP-DiD specification (equation 11):
  - ∆h y c,t+h = βP h PMT c,t + βN h NMT c,t + αh c + αh t + sum_{l=1}^3 γh l X c,t−l + v h c,t for h = 1,...,24.
  - Study horizons up to 24 months after the transition, with stabilization assumed after 12 months.
  - Average effects reported for 0-12 months, 13-24 months, and 0-24 months; inference based on joint wild bootstrap clustered at country level.

### Effect of politically motivated transitions (aggregate)
- Interest rates and inflation:
  - Short rate and real rate decline after a politically motivated transition; no notable change after non-politically motivated transitions.
  - Realized inflation and inflation expectations increase following politically motivated transitions; average effect becomes significantly positive during the first year after transition (estimates subject to substantial uncertainty).
  - No evidence of differential pre-trends across PMT and NMT in the four primary charts.
- Output and fiscal expectations:
  - Real GDP growth accelerates in the short run after politically motivated transitions.
  - Fiscal balance expectations remain broadly stable and statistically indistinguishable from zero following politically motivated transitions.
- Exchange rates:
  - Responses of realized and expected exchange rates are not distinguishable across PMT and NMT; presence of pre-trends in exchange-rate responses suggests potential endogeneity bias.
- Inference and robustness:
  - Confidence intervals shown are country-clustered wild bootstrap 90-percent confidence intervals.
  - No significant pre-trends detected in main outcomes (except noted pre-trends for exchange rates).

### Heterogeneity: unorthodox versus orthodox governors
- Estimation allowing treatment effects to differ (equation 12):
  - ∆h y c,t+h = βPO h PMT orth c,t + βPU h PMT unorth c,t + βN h NMT c,t + αh c + αh t + sum_{l=1}^3 γh l X c,t−l + v h c,t.
- Short rate:
  - Decline in short rates begins around six months after transition for both subsamples.
  - Magnitude substantially larger for unorthodox governors; average effect significantly negative for the whole 2 years after the transition for unorthodox governors.
  - For orthodox governors the effect is insignificant in the first year and turns negative in the second year.
  - Slight pre-trend of increasing nominal short rates for unorthodox governors (not significantly different from NMT); no such pre-trend for orthodox governors.
- Real rate:
  - Difference across subsamples more pronounced for the real interest rate, particularly beyond twelve months.
  - Pattern consistent with weaker perceived commitment to price stability following unorthodox appointments, operating through inflation expectations.
  - For orthodox governors there are small declines in short-term real rates after the first year; the average effect is significantly negative over the two years following transition.
- Inflation (realized and expected):
  - Both realized and expected inflation respond more strongly to politically motivated appointments of unorthodox governors.
  - Inflation increases tend to emerge after approximately nine months for unorthodox appointments.
  - Response for orthodox governors is much smaller and statistically indistinguishable from zero; pre-trends are insignificant.
- GDP and expectations:
  - GDP growth (actual and expected) rises following politically motivated transitions; rises are slightly more robust statistically for orthodox appointments.
  - The weaker and less reliable GDP response for unorthodox appointments may reflect lower commitment and higher uncertainty about inflation targets perceived by agents.
- Fiscal balance and exchange rates:
  - Fiscal balance expectations remain close to zero and statistically insignificant for both orthodox and unorthodox appointments.
  - Realized and expected exchange-rate responses are imprecisely estimated and exhibit pre-trends (results potentially subject to endogeneity).
- Overall interpretation:
  - Politically motivated transitions produce dynamic effects on rates, inflation, and activity; these effects are substantially larger and more persistent when the incoming politically appointed governor is classified as unorthodox.
  - Evidence points to a mechanism operating through changes in inflation expectations and perceived commitment to price stability after unorthodox PMT appointments.

*Source: wpiea2026040-source-pdf - Section 4*

### Section 5

### wpiea2026040-source-pdf - Section 5

### Effects on long-term expectations (methodology)
- Long-term expectations data are available bi-annually; applying LP-DiD is infeasible because other variables are monthly.
- The synthetic control method (SCM) of Abadie et al. [2010] is applied to create control observations for each governor transition.
- For each transition:
  - Horizon: 9 quarters, with the middle quarter defined as the period in which treatment occurs.
  - Pre- and post-treatment periods: the four quarters before and after the middle quarter.
  - Donor pool: untreated countries with complete data during the same 9-quarter period.
  - Clean control requirement: donors are dropped if they undergo their own transition during the 9-quarter period.
  - Donor weights are selected to minimize the average absolute deviation between the treated unit and its synthetic counterpart in the change in inflation during the two-year pre-treatment window, to closely replicate the pre-transition trajectory.

### Effects on long-term inflation expectations (results)
- Outcomes plotted for long-term (5-year ahead) inflation expectations using SCM; summaries use medians to limit outlier influence.
- Politically motivated transitions (PMT):
  - PMT long-term inflation expectations track the synthetic control closely in the pre-treatment period.
  - PMT long-term inflation expectations diverge somewhat in the two years following the transition.
  - After two years, long-term inflation expectations of PMTs are 0.2 percentage points higher than the control group.
- Non-politically motivated transitions (NMT):
  - NMT track the control group closely in both pre- and post-treatment periods (no material divergence).
- Heterogeneity within PMTs:
  - Unorthodox PMT (politically motivated transition with an unorthodox governor):
    - Long-term inflation expectations rise by about 1 percentage point relative to the control group over the two years post-appointment.
  - Orthodox PMT (politically motivated transition with an orthodox governor):
    - Long-term inflation expectations remain broadly unchanged relative to the control group.
- Figure annotation details:
  - Red lines: summary statistics for countries undergoing a governor transition at time 0.
  - Blue lines: summary statistics for corresponding synthetic control group.
  - Dashed lines: medians; dotted lines: 25th and 75th percentiles.
  - Horizontal axis: half-year (six months) periods before and after transition announcement.

### Effects on long-term growth expectations (results)
- No evidence that long-term growth expectations change materially following any transition type (Figure 19).
- Monetary accommodation by governors appointed during politically motivated transitions does not raise estimates of long-term growth.
- Appointments of unorthodox governors:
  - If anything, long-term growth expectations decline somewhat following unorthodox appointments, although the difference with the control group is small: 0.2 percentage points.
- Interpretation:
  - Observed responses are consistent with a vertical long-run Phillips curve and a negative impact on central bank credibility.
  - The absence of a meaningful revision to growth expectations suggests private agents do not perceive monetary accommodation to impact potential output.
  - The permanent effect appears to be an upward de-anchoring of inflation expectations after politically motivated appointments of unorthodox governors, consistent with forecasters updating their model of the central bank’s reaction function.

### Dynamic impacts and confidence intervals (figure notes)
- Figures plotting dynamic impacts on fiscal balance and expectations use country-clustered wild bootstrap 90-percent confidence intervals (shaded areas in red and blue).
- Blue solid line shows β_N_h and red shows β_PU_h (left) and β_PO_h (right) from equation (12).
- Horizontal axis in these figures shows months after transition announcement.
- Whiskers for average effect show country-clustered wild bootstrap for average effect.
- A large dip in the confidence interval for GDP growth in horizon 16 reflects a drop in 2020 for India due to COVID-19.

### Key empirical conclusions (paper-level summary from section)
- Dataset: 132 governor transitions across 28 advanced and emerging economies since 2000.
- Politically motivated changes in central bank leadership are associated with:
  - Higher and more volatile inflation.
  - More poorly anchored inflation expectations.
  - More dovish expected policy responses.
  - Declines in nominal and real short-term interest rates following announcement.
  - Short-run rise in GDP growth, indicating an expansionary macroeconomic impulse.
- Effects are strongest when incoming governors hold unorthodox monetary policy views.
- Long-term inflation expectations rise only for politically motivated appointments of unorthodox governors; long-term growth expectations do not move materially.
- Overall interpretation: political interference generates a temporary trade-off between higher growth and higher inflation, but political appointment of unorthodox governors weakens medium-to-long-term central bank credibility.

### Data exclusions and sources (selected details)
- Table A1 (country × periods excluded) lists exclusions for war, exchange rate pegs, SBA & EFF programs (examples include Russia February 2014 to February 2015; Argentina January 2000 to February 2002; Korea January 2000 to June 20010).
- Table A2 (data sources) summarizes frequencies, variables, units, and sources:
  - Daily variables include Government bond yields, Exchange rate against dollar, Stock market index, Policy rate, Interest swap rates, VIX index, Inflation expectations, GDP expectations, Policy rate expectations, Unemployment expectations, Exchange rate expectations, and release data—with sources Haver, GFD, Bloomberg, Refinitiv, FRED, OECD, Consensus, World Bank, ECB as indicated in the table.
  - Monthly variables include Inflation, Short rate (3m), Long rate (10y), GDP, Exchange rate per USD/EUR, inflation and GDP expectations across horizons, etc., with sources World Bank, WEO, ECB, Haver, FRED, OECD, Consensus.
  - Notes: World Bank is main source for monthly inflation; ECB data portal used for Euro area; when World Bank inflation not available, quarterly WEO data used and quarterly series interpolated; OECD used for Poland and Russia before February 2022; FRED used for Poland before January 2007; expectations for non-euro European countries originally formulated against EUR and converted to LCU/USD with realized series of USD/EUR.

### Simple model (motivation for empirical strategy on perceived policy rules)
- New Keynesian Phillips curve (NKPC) used:
  - π_t = β E_t[π_{t+1}] + κ x_t + u_t, (13)
  - where π_t is inflation at time t, x_t = y_t − y^n is the output gap, β ∈ (0,1) is the discount factor, κ > 0 is the slope of the Phillips curve, and u_t is a mean-zero cost–push shock.
- IS curve:
  - x_t = E_t[x_{t+1}] − σ(i_t − E_t[π_{t+1}] − r^n) + η_t, (14)
  - where i_t is the policy rate, r^n_t is the natural interest rate, and η_t is a mean-zero demand shock.
- Central bank loss function under discretion:
  - L_t = 1/2 [ π_t^2 + α (x_t − θ)^2 ], α > 0, θ > 0, (15)
  - θ captures preference for a positive output gap (which could reflect political pressure); α reflects ’dovishness’ of the central bank and captures preference for a stable output gap relative to stable inflation.
- Optimal policy after observing u_t and η_t:
  - x_t = θ − (κ/α) π_t, (16)
  - i_t = r^n + E_t[π_{t+1}] + (κ/ασ)(π_t − E_t[π_{t+1}]) − (1/σ) η_t, (17)
- Implications for long-term expectations under rational expectations and stationarity:
  - π_t = E_t[π_{t+1}] = π^*.

*Source: wpiea2026040-source-pdf - Section 5*

### Section 6

### Section 6

### Inflation bias and steady-state implications
- There is inflation bias such that steady-state inflation is strictly positive for θ >0.
- Steady-state inflation:
  - π∗ = κθ / (1−β+ κ/2 α)
- Long-term inflation expectations increase when θ increases.
- Steady-state output gap:
  - x∗ = θ / (1 + κ/2 α(1−β))
- Real interest rate gap is negative:
  - r∗ − rn = − θ/σ / (1 + κ/2 α(1−β))

### Exchange rate response — overview
- Figures present dynamic impacts on exchange rates and exchange rate expectations following transition announcements.
- Confidence intervals:
  - Shaded area in red and blue shows country-clustered wild bootstrap 90-percent confidence intervals.
- Estimators shown:
  - Blue solid line shows βNh and red shows βPh from equation (11).
- Horizontal axis shows months after transition announcement.
- Whiskers for average effect show country-clustered wild bootstrap for average effect.

### Exchange rate response — orthodox vs unorthodox transitions
- Separate dynamic impact figures for "unorthodox" and "orthodox" transitions are provided.
- Confidence intervals:
  - Shaded area in red and blue shows country-clustered wild bootstrap 90-percent confidence intervals.
- Estimators shown:
  - Blue solid line shows βNh.
  - Red lines show βPUh (left) and βPOh (right) from equation (12), respectively.
- Horizontal axis shows months after transition announcement.
- Whiskers for average effect show country-clustered wild bootstrap for average effect.

*Source: The Macroeconomic Consequences of Undermining Central Bank Independence: Evidence from Governor Transitions — Working Paper No. WP/2026/040*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026040-source-pdf.pdf_