## wpiea2025079-print-pdf ## Source details **Canonical URL:** [wpiea2025079-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025079-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025079-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025079-print-pdf.pdf.json) --- ### Model insights (Simple New‑Keynesian model with experience learning) - Under rational expectations and strict inflation targeting (휋t = 0 for all t): - Et[휋t+1] = 0 → yt = 0 for all t (divine coincidence). - Equilibrium policy rate: it = 0 for all t. - Introducing experience learning (past high inflation 휋H > 0 with memory decay parameter 휆 ∈ (0,1)): - Expectations: Et[휋t+1] = 휆t 휋H. - With strict inflation targeting (휋t = 0) the Phillips curve implies yt = −(훽 휆t 휋H) / 휅. - Output gap is negative while memory persists; inflation stabilization has real (output) costs. - Policy rate consistent with strict inflation targeting: it = [1 + (훽 휎 / 휅) (1 − 휆)] 휆t 휋H. - Nominal interest rate is above steady state during the transition; stabilization requires a more hawkish stance. - Determinants of larger stabilization costs and stronger required tightening: - Larger past inflation exposure (휋H). - More recent exposure (smaller t). - More persistent memory (larger 휆). - Stickier prices (smaller 휅). - Higher risk aversion/intertemporal substitution parameter (higher 휎). - Conceptual implications: - Backward-looking experienced learning combined with forward-looking expectations → path dependence and central bank credibility matter. - Imperfect anchoring raises stabilization costs and necessitates real adjustments even with rational expectations about the future. ### Empirical strategy and data (Taylor‑rule panel estimation) - Baseline panel Taylor rule specification: - ic,t = αc + ρ ic,t−1 + β inf.gapc,t + γ Ygapc,t + θ ΔNEERec,t + μ ΔNEERec,t−1 + ϖ i.US,t−1 + εc,t. - Inflation‑gap measures used: - Contemporaneous headline inflation gap. - Contemporaneous core inflation gap. - One‑year ahead inflation expectations gap. - Controls: - Lagged policy rate, changes in nominal effective exchange rate, US policy rate; some exercises replace US rate with time fixed effects. - Heterogeneity: - Allow β and γ to vary with country characteristics at IT adoption (financial development, trade openness, capital account openness, central bank independence, past inflation). - Key data sources and constructs: - Inflation expectations: Consensus Economics. - Inflation targets and policy rates: BIS and AREAER. - NEER and inflation: IMF IFS. - Output gaps: quarterly real GDP (Haver), HP filter applied. - Financial development: Svirydzenka (2016) / Sahay and others (2015). - Central bank independence: Romelli (2022, 2024) index (6 dimensions). - Capital account openness: Chinn‑Ito index; FX intervention data from Adler and others (forthcoming). - Past inflationary experience: average inflation from 1960 (or first WDI appearance) to ten years before IT adoption. - Dummy = 1 if historical average inflation above 75th percentile. - Continuous transformed measure 휋̅ = 휋 / (100 + 휋). ### Main empirical results - Expected inflation vs. observed inflation in Taylor rules: - Coefficients on the expected inflation gap more than double those on observed headline or core inflation gaps. - A one percent deviation of expected inflation from the target triggers on average a change in the policy interest rate of around 20-22 basis points. - Observed inflation metrics trigger less than 10 basis points on average. - Heterogeneity by income and timing of IT adoption: - Positive and significant response of policy rates to expected inflation gaps in both AEs and EMDEs. - Within EMDEs, early adopters (adopted prior to 2006) respond more forcefully: - Taylor rule coefficient close to 40 basis points for any one percentage point deviation of expected inflation from target (about two times late adopters and AEs). - Credibility puzzle (time under IT and anchoring): - Interaction of expected inflation gap with years under IT: coefficient is negative but small and not statistically significant. - Alternative anchoring measures show coefficients in expected directions but not statistically significant. - Role of past inflationary history: - Interaction of expected inflation gap with high‑past‑inflation dummy: countries with an inflationary past exhibit larger and statistically significant policy rate responses. - Countries with a high inflationary past change their interest rates by an additional 26 basis points—making the overall effect close to 50 basis points in such countries. - Difference in expected inflation gap coefficient for high‑past‑inflation countries is more than double that of other countries. - Results robust when using average past inflation in levels and when interacting with the output gap. - Inflation persistence channel: - Interaction of expected inflation gap with a persistence gauge (one‑year ahead expectation elasticity on a 20‑quarter rolling basis) is positive but not statistically significant. - Interaction of expected inflation gap with average past inflation remains positive and significant. - Conclusion: inflation persistence does not fully explain stronger policy responses in countries with a history of high inflation. ### Discussion and policy implications - Empirical findings mirror model predictions: inflationary memory leads central banks to react more strongly to deviations of expected inflation from target. - Possible contributing channels: - Experience‑based preferences of individuals and policymakers: stronger aversion to inflation among those exposed to high inflation. - Central banks internalize population and policymaker inflationary memory: aim to keep expectations anchored even at short‑run output cost. - Challenges to standard inflation‑targeting tenets: - Full credibility assumption questioned: "credibility puzzle" — time under IT does not reliably reduce policy responses to expected inflation gaps. - Path dependence assumption challenged: inflationary past materially affects contemporary monetary policy reaction functions. - Policy interpretation: - Where inflationary memory is strong, central banks optimally tolerate short‑run output costs (negative output gaps, higher policy rates) to anchor expectations. - Credibility-building may not be automatic with time under IT; history and experience matter for policy design. - Modelling implications: - Importance of incorporating path dependency and less-than-perfect credibility into theoretical models of inflation targeting. ### Robustness - Main finding (stronger policy response in countries with historical high inflation) remains after controlling for: - Financial development, trade openness, capital account openness, central bank independence. - Extended specifications interacting gaps with dummies for high financial development, high trade openness, high capital account openness, high central bank independence at IT adoption: - Difference in response to expected inflation often becomes larger and remains statistically significant. - Sensitivity checks: - Conclusions hold when controlling for different sets of fixed effects (columns 2-4). - Results unaffected when adjusting Chinn‑Ito KA openness by FX intervention magnitude (column 5) or replacing high trade openness dummy with commodity exporter dummy (column 6). - Pre-COVID dummy does not affect baseline results. - Excluding outliers Turkey and Russia does not affect findings. - Years as IT specified linearly rather than fixed effects does not affect estimated coefficients. - No significant asymmetries by sign of inflation gap; no evidence of non-linearities (squared term small and insignificant). - Accounting for regional shocks yields broadly robust results; expected inflation coefficient remains positive and significant though lower than baseline. ### Key summary statistics (Annex I, Table 1) - Policy rate: Average 4.78, St. Dev. 3.65, Min -0.5, Max 26.5 - NEER (change): Average -0.53, St. Dev. 7.96, Min -56.59, Max 33.81 - Inflation gap (observed): Average 0.59, St. Dev. 2.87, Min -7.91, Max 28.86 - Inflation gap (expected): Average 0.26, St. Dev. 1.01, Min 3.79, Max 11.58 - Output gap: Average -0.003, St. Dev. 2.33, Min 32.35, Max 10.38 - Note: Summary statistics exclude data for Turkey ### Sample composition (Annex I, Table 2) — selected entries - Advanced Economies: AUS (1993), CAN (1992), CZE (1997), GBR (1992), ISR (1997), KOR (2001), NOR (2001), NZL (1989), SWE (1993) - Emerging Markets: ALB (2009), BRA (1999), CHL (1999), COL (1999), DOM (2012), GEO (2009), GTM (2005), HUN (2001), IDN (2005), IND (2015), KAZ (2015), MDA (2013), MEX (2001), PER (2003), PHL (2002), POL (1998), PRY (2013), ROU (2005), RUS (2015), SRB (2006), THA (2000), TUR (2006), ZAF (2000) ### Representative coefficient magnitudes (selected table highlights) - Lagged policy rate coefficients frequently around 0.86–0.91 (*** p<0.01 in many specs). - Expected inflation gap coefficients often range from 0.1845* to 0.5904*** in different specifications and subsamples. - EMDE early adopters: expected inflation coefficient examples 0.3993***, 0.4117***, 0.3246***. - Interaction: Inf. gap x Pre-IT high inflation dummy often around 0.2601** to 0.9993*** across tables and specifications. - Observations in many panel regressions: 2,617 (varying across specifications); Number of groups typically 32 or close. ### Country-by-country findings (Annex II) - Country‑level Taylor rule estimated: i_{c,t} = α_c + ρ_c i_{c,t−1} + β_c inf.gap_{c,t} + γ_c Ygap_{c,t} + θ_c ΔNEEER_{c,t} + μ_c ΔNEEER_{c,t−1} + φ_c i.^US_{t−1} + ε_{c,t}. - Distributional finding: - β_c (inflation gap coefficient) is significantly higher than γ_c (output gap coefficient). - Panel estimates lie on the lower end of the interquartile range for both β_c and γ_c. - Robust relation: - Country-level β_c strongly correlated with average past inflation prior to IT adoption. - γ_c not correlated with average past inflation. - Table A2.1 (selected coefficients): - Coefficients on Pre-IT inflation for expected inflation coefficient regressions: 0.995*** (0.317), 1.021*** (0.336), 1.002*** (0.329). - For output gap coefficient regressions: 0.150 (0.115), 0.175 (0.173), 0.144 (0.125). - Observations: 28; R-squared for β regressions around 0.415–0.441; for γ regressions 0.091–0.151. - Bootstrapped standard errors reported in parentheses. ### Institutional arrangements (Annex IV) — selected entries - Inflation targets (2025, target form): - Albania: H CPI, 3%, Point, Medium term. - Australia: H CPI, 2%-3%, Range, Medium term. - Brazil: H CPI, 3%, Point, ±1.5 pp tolerance band, Yearly target until 2024, continuous since 2025. - Canada: H CPI, 2%, Point, ±1 pp tolerance band, Six-eight quarters. - India: H CPI, 4%, Point, ±2 pp tolerance band, Five years. - Japan: H CPI, 2%, Point, Yearly. - Turkey: H CPI, 5%, Point, ±2 pp tolerance band, Two years. - United Kingdom: H CPI, 2%, Point, At all times. - Uganda: Core CPI, 5%, Point, ±3 pp tolerance band, One-three years. - Decision‑making committees (Table A4.2) — examples: - Albania: 9 members; Meetings per year: 8; Governor's term: 7 years. - Australia: 9 members; External members?: 6; Meetings per year: 8; Governor's term: 7 years. - Georgia: 14 members; Meetings per year: 8; Governor's term: 7 years. - Poland: 10 members; External members?: 9; Meetings per year: 12**; Governor's term: 6 years. - Russia: 15 members; Meetings per year: 12; Governor's term: 5 years. - Ukraine: 7 members; Meetings per year: 8; Governor's term: 8 years. - Accountability and transparency (Table A4.3) — examples: - Parliamentary hearings and publication practices vary: many countries have parliamentary hearings (annual or twice a year), minutes and votes publication differ by country. - Inflation Report frequency examples: many countries publish 4 times per year; some publish 2 or 6 times. *Source: wpiea2025079-print-pdf — https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025079-print-pdf.pdf* ### conclusions of the paper. ### conclusions of the paper. ### Model insights (Simple New‑Keynesian model with experience learning) - Under rational expectations and strict inflation targeting (휋t = 0 for all t), the model implies: - Et[휋t+1] = 0 → yt = 0 for all t (divine coincidence). - Equilibrium policy rate supporting this outcome: it = 0 for all t (no deviation from steady state). - Introducing experience learning (past high inflation 휋H > 0 with memory decay parameter 휆 ∈ (0,1)): - Expectations evolve as Et[휋t+1] = 휆t 휋H. - With strict inflation targeting (휋t = 0) the Phillips curve implies: - yt = −(훽 휆t 휋H) / 휅. - The output gap is negative while memory persists; inflation stabilization has real (output) costs. - The policy rate consistent with strict inflation targeting is: - it = [1 + (훽 휎 / 휅) (1 − 휆)] 휆t 휋H. - The nominal interest rate is above steady state during the transition: inflation stabilization requires a more hawkish stance. - Determinants of larger stabilization costs and stronger required policy tightening: - Larger past inflation exposure (휋H). - More recent exposure (smaller t). - More persistent memory (larger 휆). - Stickier prices (smaller 휅). - Higher risk aversion/intertemporal substitution parameter (higher 휎). - Conceptual implications: - Backward-looking experienced learning combines with forward-looking expectations: path dependence and central bank credibility matter for optimal policy. - Even with rational expectations about the future, imperfect anchoring (legacy doubts about the central bank’s ability) raises stabilization costs and necessitates real adjustments. ### Empirical strategy and data (Taylor‑rule panel estimation) - Baseline econometric specification (panel Taylor rule): - ic,t = αc + ρ ic,t−1 + β inf.gapc,t + γ Ygapc,t + θ ΔNEERec,t + μ ΔNEERec,t−1 + ϖ i.US,t−1 + εc,t. - Three alternative inflation‑gap measures used: contemporaneous headline inflation gap, contemporaneous core inflation gap, and one‑year ahead inflation expectations gap. - Controls include lagged policy rate, changes in nominal effective exchange rate, and US policy rate; some exercises replace US rate with time fixed effects. - Heterogeneity explored by allowing inflation and output gap coefficients to vary with country characteristics at IT adoption (financial development, trade openness, capital account openness, central bank independence, past inflation). - Key data sources and constructs: - Inflation expectations: Consensus Economics (professional forecasters). - Inflation targets and policy rates: BIS and AREAER. - Nominal effective exchange rates and inflation: IMF IFS. - Output gaps: quarterly real GDP from national sources via Haver, HP filter applied. - Financial development: Svirydzenka (2016) / Sahay and others (2015) index. - Central bank independence: Romelli (2022, 2024) index (6 dimensions). - Capital account openness: Chinn‑Ito index; FX intervention data from Adler and others (forthcoming) in some exercises. - Past inflationary experience: average inflation from 1960 (or first WDI appearance) to ten years before IT adoption. Two constructs: - Dummy = 1 if historical average inflation above 75th percentile in sample. - Continuous transformed measure 휋̅ = 휋 / (100 + 휋) to compress hyperinflation values. ### Main empirical results - Inflation expectations vs. observed inflation in Taylor rules: - Coefficients on the expected inflation gap more than double those on observed headline or core inflation gaps. - A one percent deviation of expected inflation from the target triggers on average a change in the policy interest rate of around 20-22 basis points. - Observed inflation metrics trigger less than 10 basis points on average. - Heterogeneity by income and timing of IT adoption: - Positive and significant response of policy rates to expected inflation gaps in both AEs and EMDEs. - Within EMDEs, early adopters (adopted prior to 2006) respond more forcefully: Taylor rule coefficient close to 40 basis points for any one percentage point deviation of expected inflation from target (about two times late adopters and AEs). - Credibility puzzle (time under IT and anchoring): - Interaction of expected inflation gap with years under IT: coefficient is negative (as expected) but small and not statistically significant → longer track record does not systematically translate into smaller interest‑rate movements for a given expected inflation gap. - Alternative credibility/anchoring measures (rolling variance of the inflation gap over past two years; Bems et al. anchoring subcomponent) show coefficients in expected directions but are not statistically significant given available samples. - Role of past inflationary history: - Interaction of expected inflation gap with high‑past‑inflation dummy: countries with an inflationary past exhibit larger and statistically significant policy rate responses. - Countries with a high inflationary past change their interest rates by an additional 26 basis points—making the overall effect close to 50 basis points in such countries (economically significant). - The difference in expected inflation gap coefficient for high‑past‑inflation countries is more than double that of other countries. - Results robust when using average past inflation in levels and when interacting with the output gap. - Inflation persistence channel: - Adding a gauge of inflation persistence (one‑year ahead expectation elasticity to contemporaneous inflation on a 20‑quarter rolling basis) and its interaction with the expected inflation gap: - The interaction with persistence is positive but not statistically significant. - The interaction of the expected inflation gap with average past inflation remains positive and significant. - Conclusion: inflation persistence does not fully explain the stronger policy response in countries with a history of high inflation. ### Discussion and policy implications - Empirical findings mirror model predictions: a legacy of high inflation (inflationary memory) leads central banks to react more strongly (more hawkishly) to deviations of expected inflation from target. - Possible contributing channels: - Experience‑based preferences of individuals and policymakers: those exposed to high inflation have stronger aversion to inflation (links to Magud and Pienknagura (2024); Melmendier (2021); Rogoff (1985) logic). - Central banks internalize population and policymaker inflationary memory into policy: aim to keep expectations anchored even at short‑run output cost. - Challenges to standard inflation‑targeting tenets: - Full credibility assumption is questioned by the "credibility puzzle"—time under IT does not yet reliably reduce policy responses to expected inflation gaps. - No path dependence assumption is challenged: inflationary past materially affects contemporary monetary policy reaction functions. - Policy interpretation: - Where inflationary memory is strong, central banks optimally tolerate short‑run output costs (negative output gaps, higher policy rates) to anchor expectations and stabilize inflation. - Credibility-building may not be automatic with time under IT; history and experience matter for policy stance and should be accounted for in models and policy design. ### Robustness - Main finding (stronger policy response in countries with historical high inflation) remains after controlling for country characteristics that could be shaped by inflationary past: - Financial development, trade openness, capital account openness, central bank independence. - Extended specifications (interacting gaps with dummies for high financial development, high trade openness, high capital account openness, high central bank independence at IT adoption) confirm relevance of inflationary past; the difference in response to expected inflation often becomes larger and remains statistically significant. *Source: wpiea2025079-print-pdf — https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025079-print-pdf.pdf* ### conclusions hold when we control for the different set of fixed effects described earlier (columns 2-4). ### V. Concluding remarks ### Summary of main findings - Inflation targeting was adopted for the first time about 35 years ago and has become widespread among central banks in AEs and EMs. - Despite important similarities among IT central banks, the paper documents important differences in the conduct of monetary policy. - Exposure to historical episodes of high inflation is a key determinant of monetary policy heterogeneity. - Empirical findings show that, on average, central banks that adopted an IT framework respond preponderantly to deviations of inflation expectations from the target rate. - For a given expected inflation gap, central banks in countries with a history of high inflation adjust their policy rate in a more aggressive way. - This stronger response potentially explains why the coefficient for the expected inflation in the Taylor rule does not appear to fall over time nor to be lower in central banks with greater inflation expectations’ anchoring (the “credibility puzzle”). - Central banks with past high inflation experiences continue to react as strongly as in the past to inflation expectations shocks, probably due to fears of the return of unanchored inflation. ### Robustness and sensitivity checks (high-level) - Conclusions hold when controlling for different sets of fixed effects (columns 2-4). - Results are unaffected when: - adjusting the de jure measure of openness (from Chinn-Ito, KA openness) by the magnitude of interventions in foreign exchange markets (column 5)14, or - replacing the high trade openness dummy with a dummy taking value one if the country is a commodity exporter (column 6)15. - Many variables affect the response of policy rates to expected inflation gaps: - Deeper financial markets or more independent central banks marginally reduce the reaction of policy interest rates to expected inflation gap shocks, as given by the negative interaction term of these variables and the expected inflation gap (Table 8). - Integration to international financial markets (openness of the financial account), once corrected by FX interventions, shows a similar marginal reduction. - Initial levels of trade openness do not significantly weigh on the parameters of the Taylor rule (interaction coefficient not statistically significant). - Interacting expected inflation and output gaps with country-specific variables at the time of IT adoption (Table 9) confirms prior findings: - Interaction between the expected inflation gap and the average past inflation level is positive, statistically significant, and larger in magnitude than in Table 7. - Interaction of past inflation and the output gap is non-significant. - Results robust to inclusion of time and year of IT adoption fixed effects and to interaction with a commodity exporter dummy. - Using contemporaneous country-specific variables (Table 10) confirms Table 9 results: the interaction of the expected inflation gap and past inflation remains positive and statistically significant. - Robustness to periods, samples, and specifications (Table 11): - Adding a dummy for pre-COVID years (to test COVID shock effects) does not affect baseline results (columns 1 and 2). - Excluding outliers Turkey and Russia does not affect findings (columns 3 and 4). - Including years as IT linearly rather than fixed effects does not affect estimated coefficients for expected inflation and output gaps (column 5). - Allowing the coefficient for the expected inflation gap to vary by sign of the gap (interaction with a positive-gap dummy) yields a non-significant and small coefficient for the asymmetric term (column 6), suggesting non-significant asymmetries for the average country. - Inclusion of changes in commodity terms-of-trade (column 7) or the ratio of the output gap and inflation volatilities (column 8) does not affect estimated Taylor rule coefficients. - Testing for non-linearities by including the expected inflation gap squared (column 9) shows no evidence of non-linearities (squared term small and statistically insignificant). - Role of regional shocks (Table 12): - Accounting for regional shocks (e.g., terms-of-trade shocks or natural disasters) while presenting estimates without years as IT, with years as IT linearly, and as fixed effects yields broadly robust results. - Coefficients for expected inflation and output gaps remain positive and significant; the expected inflation coefficient is substantially larger in magnitude, albeit lower than in the baseline estimation. ### Interpretations and potential mechanisms - Two potential explanations (not formally tested in the paper): - Policy makers (central bank board members) may be scarred by their own experiences with high inflation and have a strong preference for price stability. - Policy makers in countries with an inflationary memory may internalize the impact that such history has on price and wage dynamics and thus need to respond more aggressively to deviations from target to break a potential inflationary spiral. - These explanations align with evidence that individuals who experienced past high inflation express a stronger preference for price stability (Magud and Pienknagura, 2024). ### Policy and modelling implications - Policy implications: - Inflationary history may affect not only the conduct of monetary policy but also the transmission of monetary policy. This channel is not explored in this paper and is an important avenue for future research. - Modelling implications (main contribution): - Importance of incorporating path dependency and less-than-perfect credibility in theoretical models of inflation targeting. - Standard inflation targeting models have omitted path dependency and imperfect credibility, and including these elements has non-trivial implications. *Source: wpiea2025079-print-pdf - conclusions hold when we control for the different set of fixed effects described earlier (columns 2-4).* ### Annex I. Tables ### Annex I. Tables ### Summary statistics (Table 1) - Policy rate: Average 4.78, St. Dev. 3.65, Min -0.5, Max 26.5 - NEER (change): Average -0.53, St. Dev. 7.96, Min -56.59, Max 33.81 - Inflation gap (observed): Average 0.59, St. Dev. 2.87, Min -7.91, Max 28.86 - Inflation gap (expected): Average 0.26, St. Dev. 1.01, Min 3.79, Max 11.58 - Output gap: Average -0.003, St. Dev. 2.33, Min 32.35, Max 10.38 - Note: Summary statistics exclude data for Turkey ### Sample composition (Table 2) - Advanced Economies (with year of IT adoption in parenthesis): AUS (1993), CAN (1992), CZE (1997), GBR (1992), ISR (1997), KOR (2001), NOR (2001), NZL (1989), SWE (1993) - Emerging Markets (with year of IT adoption in parenthesis): ALB (2009), BRA (1999), CHL (1999), COL (1999), DOM (2012), GEO (2009), GTM (2005), HUN (2001), IDN (2005), IND (2015), KAZ (2015), MDA (2013), MEX (2001), PER (2003), PHL (2002), POL (1998), PRY (2013), ROU (2005), RUS (2015), SRB (2006), THA (2000), TUR (2006), ZAF (2000) ### Expected inflation vs. observed inflation (Table 3) — key coefficient patterns - Lagged policy rate (Policy rate (t-1)): coefficients range around 0.8692***, 0.8960***, 0.8723***, 0.8518***, 0.8866***, 0.8627***, 0.8678***, 0.9084***, 0.8719*** (Driscoll-Kraay SEs shown) - XR depreciation: negative coefficients, examples include -0.0220*, -0.0250*, -0.0196*, -0.0225*, -0.0262*, -0.0202*, -0.0323***, -0.0391***, -0.0294*** - XR depreciation (t-1): positive coefficients, examples include 0.0243**, 0.0219**, 0.0203**, 0.0251**, 0.0228**, 0.0210**, 0.0251**, 0.0241**, 0.0205* - US Policy rate (t-1): positive and significant in some specifications, e.g., 0.0735***, 0.0630***, 0.0748***, 0.0604***, 0.0524***, 0.0642*** - Output gap: positive and significant across specifications, examples 0.0651***, 0.0777***, 0.0687***, 0.0693***, 0.0826***, 0.0732***, 0.0839***, 0.0907***, 0.0841*** - Inflation gap (observed/core/expected): expected inflation gap notably larger where included, e.g., Inflation gap (expected inf) 0.2207***, 0.2253***, 0.1845* in some columns - Implied b/(1-r) reported in columns with values such as 0.7085**, 0.8689**, 1.7281***, 0.6732***, 0.8333**, 1.6406***, 0.5849*, 0.7708, 1.4405** (SEs shown) - Observations: 2,612; 2,366; 2,617 (varies by column). Number of groups: 32 or 28. Adjusted R-squared ranges 0.896–0.923 - Driscoll-Kraay standard errors; significance codes: *** p<0.01, ** p<0.05, * p<0.1 ### Effects by income levels (Table 4) - Sample splits: AEs, EMDE early adopters, EMDE late adopters (observations: 1,010; 927; 680) - Lagged policy rate: coefficients around 0.8888***, 0.8476***, 0.8960***, 0.8885***, 0.8367***, 0.8957***, 0.9109***, 0.8711***, 0.8891*** - XR depreciation: mixed signs; for EMDE late adopters strongly negative and significant (e.g., -0.0734***, -0.0764***, -0.0779***) - XR depreciation (t-1): positive and significant for EMDE late adopters (e.g., 0.0653***, 0.0651**, 0.0620**) - Output gap: positive and significant across groups (examples 0.0375**, 0.0851***, 0.0602***) - Inflation gap (expected inf): larger coefficients for EMDE early adopters (e.g., 0.3993***, 0.4117***, 0.3246*** in some specs) and smaller or insignificant for AEs in some columns (e.g., 0.1823*** for AEs) - Adjusted R-squared: high across splits, e.g., 0.960, 0.916, 0.849; time FE and Years as IT FE included in some specifications ### The credibility puzzle (Table 5) - Core patterns: - Lagged policy rate stable around 0.8670*** to 0.8948*** across specifications - XR depreciation negative (examples -0.0198*, -0.0293***, -0.0303***, -0.0321***) - XR depreciation (t-1) positive and significant in many columns (e.g., 0.0208**, 0.0215**) - Output gap positive and significant (examples 0.0686***, 0.0941***, 0.0798***) - Inflation gap (expected inf) large and significant in many specifications (e.g., 0.3294***, 0.3228***, 0.3210***, 0.3164***, 0.3253***) - Interactions and additional terms tested: - Inflation gap (expected inf) x years as IT: negative but not significant in shown coefficients (e.g., -0.0066, -0.0062, -0.0084) - Output gap x years as IT: small negative coefficients, sometimes marginally significant (e.g., -0.0017*) - Variance of inflation gap over past 2 years (t-1) coefficient example -0.0105 (not significant) - Alternative specifications include sensitivity measures; many interaction terms reported with SEs - Observations: mostly 2,617; some smaller samples 2,374 or 1,246 when limited to subsamples; Number of groups 32 (or 22 for subsample) - Adjusted R-squared up to 0.935 in some columns ### History matters (Table 6) - Regressions include lagged policy rate, changes in NEER and its lag (coefficients not shown) - Key coefficients: - Lagged policy rate: 0.8595***, 0.8471***, 0.8479***, 0.8604*** - XR depreciation: negative and often significant (e.g., -0.0203*, -0.0208*, -0.0207*, -0.0302***) - XR depreciation (t-1): positive and significant (e.g., 0.0202**, 0.0209**, 0.0208**, 0.0213**) - US Pol. rate (t-1): positive and significant in several columns (e.g., 0.0893***, 0.0801***, 0.0799***), but -0.1045 (not significant pattern) in one specification - Output gap: positive and significant (e.g., 0.0676***, 0.0722***, 0.0638***, 0.0739***) - Inflation gap (expected inf): 0.1951**, 0.2006**, 0.2007**, 0.1677* across columns - Inf. gap x Pre-IT high inflation dummy: positive and significant (e.g., 0.2601**, 0.2694**, 0.2672**, 0.2412**) - Observations: 2,617; Number of groups 32; Adjusted R-squared up to 0.916 ### Pre-IT inflation levels and persistence (Table 7) - Lagged policy rate: coefficients 0.8641***, 0.8499***, 0.8507***, 0.8620***, 0.9027***, 0.9108*** - XR depreciation: negative and significant in some columns (e.g., -0.0200*, -0.0205*, -0.0290**, -0.0321**, -0.0325**) - US Pol. rate (t-1): positive and significant in several columns (e.g., 0.0912***, 0.0813***, 0.0817***); some large coefficients in later columns (0.1554***, 0.1619***) - Output gap: positive and significant (e.g., 0.0698***, 0.0747***, 0.0621***, 0.0693***, 0.0672***, 0.0651***) - Inflation gap interactions: - Inf. gap x Pre-IT inflation: strongly positive and significant in multiple columns (e.g., 0.5890**, 0.6544**, 0.6497**, 0.5887**, and in later specs 0.3296***, 0.3105**) - Inf. gap x Persistence measure and Persistence measure reported with coefficients (e.g., Inf. gap x Persistence measure 0.1980; Persistence measure -0.2036) but not significant here - Observations vary: 2,513 and subset 1,872; Number of groups 31 or 30. Adjusted R-squared around 0.899–0.914/0.904 ### Financial depth, global integration, CBI, and net commodity exporter (Table 8) - Output gap: positive and significant across columns (e.g., 0.0979***, 0.0964***, 0.1104***, 0.1112***, 0.0739***, 0.0956***) - Inflation gap (expected inf): positive and significant (e.g., 0.3498***, 0.3443***, 0.2823**, 0.2833**, 0.2875***, 0.3689***) - Inf. gap x Pre-IT high inflation dummy: positive and significant in many specs (e.g., 0.3189***, 0.3517***, 0.3192**, 0.3433***, 0.2838**, 0.2039**) - Inf. gap x high initial FD dummy: negative and significant (e.g., -0.2598***, -0.2791***, -0.2353***, -0.2343***, -0.1677***, -0.2029***) - Output gap x high initial KA dummy: negative and significant across many specs (e.g., -0.0726***, -0.0622***, -0.0769***, -0.0699***, -0.0755***) - Inf. gap x high initial KA dummy (modified): -0.3823** in one column - Inf. gap x high initial CBI dummy: negative and sometimes significant (e.g., -0.1605*, -0.1509*, -0.1629**) - Inf. gap x commodity exporter dummy: 0.1190 (not significant) in one column; Output gap x commodity exporter dummy 0.0819** in one column - Observations 2,375 (or 2,171 in one column); Number of groups 29; Adjusted R-squared up to 0.917 ### Actual inflation and net commodity exporters (Table 9) - Output gap: positive and significant in all columns shown (e.g., 0.0503***, 0.0694***, 0.0365***, 0.0581***) - Inflation gap (expected inf): coefficients negative but not significant in shown columns (e.g., -0.1698, -0.1796, -0.0912, -0.1105) - Inf. gap x Pre-IT inflation: very large positive and significant (e.g., 0.9993***, 0.9199***, 0.7753***, 0.7281***) - Inf. gap x FD index (initial): negative and in one column significant (e.g., -0.9502*; other specs -0.8942*, -0.7494, -0.7242) - Output gap x FD index (initial): positive and in some columns significant (e.g., 0.3642***, 0.3226**, 0.2455**, 0.1909*) - Inf. gap x KA index (initial): positive and sometimes significant (e.g., 0.5651**, 0.5267**, 0.5535, 0.5253) - Output gap x KA index (initial): negative and significant across columns (e.g., -0.1618***, -0.1512***, -0.1846***, -0.1682***) - Inf. gap x CBIE index (initial): positive and significant in shown columns (e.g., 1.0326***, 1.0590***, 1.0217**, 1.0405***) - Output gap x CBIE index (initial): positive and in some columns significant (e.g., 0.2700**, 0.2371**) - Observations: 2,513 across columns; Number of groups 31; Adjusted R-squared 0.906–0.916 ### Contemporaneous state variables (Table 10) - Regressions include contemporaneous values of interacted country-specific variables (FD index, trade openness, CBI index, KA index) — coefficients for these contemporaneous terms not shown for simplicity - Output gap: coefficients reported include 0.0667, 0.0751, 0.0402, 0.0522 (not significant in these columns) - Inflation gap (expected inf): positive and significant in some columns, e.g., 0.4499**, 0.5904***, 0.4451**, 0.5593*** - Inf. gap x Pre-IT inflation: positive and significant across columns (e.g., 0.9726***, 0.8819**, 0.8766***, 0.7889**) - Inf. gap x FD index: negative and sometimes significant (e.g., -0.4366*, -0.6144**, -0.3999, -0.5695**) - Output gap x KA index: negative and significant in several columns (e.g., -0.0654**, -0.0707**, -0.0896***, -0.0935***) - Output gap x commodity exporter dummy: 0.0494* and 0.0466* in two columns - Observations: 2,059; Number of groups 28; Adjusted R-squared 0.920–0.931 ### The effects of the COVID shock, outliers, inflation gap direction, terms-of-trade and non-linearities (Table 11) - Lagged policy rate: examples 0.8139***, 0.8310***, 0.8793***, 0.8556***, 0.8701***, 0.8719***, 0.8964***, 0.9034***, 0.8720*** - XR depreciation: generally negative with varying significance (examples -0.0128, -0.0209**, -0.0217**, -0.0201**, -0.0282**, -0.0294***, -0.0197**, -0.0322***, -0.0294***) - XR depreciation (t-1): coefficients vary (examples 0.0100, 0.0108, 0.0149, 0.0100, 0.0203*, 0.0206*, 0.0031, 0.0179*, 0.0205*) - Output gap: positive and significant across many specs (e.g., 0.1187***, 0.1049***, 0.0766***, 0.0697***, 0.0796***, 0.0841***, 0.0912***, 0.0808***, 0.0841***) - Inflation gap (expected inf): positive and often significant (e.g., 0.3314***, 0.3010***, 0.1857*, 0.2675***, 0.1882*, 0.1932**, 0.2417***, 0.1047, 0.1832**) - Change in commodity terms-of-trade (t-1): 0.0640*** in one column - Ratio of inflation to output gap volatilities: 0.0572* in one specification - Inflation gap squared (expected inf): 0.0000 (not significant) - Observations and samples vary: examples Observations 2,117; 2,117; 2,583; 2,546; 2,617; 2,617; 1,944; 2,286; 2,617 across columns; Number of groups 32, 32, 31, 32, 32, 31, 32, 32, 32 - Samples include Pre-2020, Exc. Russia, Exc. Turkey, All; Time FE and Years as IT FE included in many specifications - Adjusted R-squared reported examples: 0.920, 0.929, 0.923, 0.930, 0.912, 0.914, 0.946, 0.921, 0.914 ### Region-specific time-varying shocks (Table 12) - Lagged policy rate: 0.8739***, 0.8739***, 0.8771*** - XR depreciation: -0.0254**, -0.0254**, -0.0268** - XR depreciation (t-1): 0.0169 (not significant in shown columns) - Output gap: 0.0566***, 0.0566***, 0.0637*** - Inflation gap (expected inf): 0.1686*, 0.1686*, 0.1633* - Years as IT: 0.0031 (not significant) - Regression setup: Region-time fixed effects included in these specifications; Observations 2,617, Number of groups 32, Adjusted R-squared 0.932–0.933 *Source: Annex I. Tables (content unit: wpiea2025079-print-pdf - Annex I. Tables).* ### Annex II. Results from Country-by-Country ### Annex II. Results from Country-by-Country ### Taylor Rules: Specification and Estimation - Estimated Taylor rule for each IT country: i_{c,t} = α_c + ρ_c i_{c,t−1} + β_c inf.gap_{c,t} + γ_c Ygap_{c,t} + θ_c ΔNEEER_{c,t} + μ_c ΔNEEER_{c,t−1} + φ_c i.^US_{t−1} + ε_{c,t} (A1) - Definitions used (as in the main text): - i_{c,t} is the policy rate in country c at time t. - Ygap_{c,t} is the output gap in country c at time t (output detrended through HP filter). - ΔNEEER_{c,t} is the change in the nominal effective exchange rate in period t. - i.^US_{t−1} is the monetary policy rate in the US at time t−1. - inf.gap_{c,t} is the inflation gap in country c at time t. - For simplicity, results focus on the inflation gap constructed using inflation expectations. ### Distribution of Coefficients: Inflation vs Output Gaps - Key qualitative finding: - The coefficient for the inflation gap (β_c) is significantly higher compared to the coefficient for the output gap (γ_c). - The estimated coefficients from the panel regressions lie on the lower end of the interquartile range for both β_c and γ_c. - Figure A2.1 (referenced): plots distribution of all coefficients β_c and γ_c. ### Robustness: Past Inflationary History and Country-specific Estimates - Country-specific estimates are used to assess whether past inflationary history shapes central bank monetary policy. - Regression results (referenced as Table A3.1, columns 1 and 4) show: - The inflation gap coefficient (β_c) is strongly correlated with average past inflation experienced by a country prior to IT adoption. - The output gap coefficient (γ_c) is not correlated with average past inflation. - Additional robustness checks (referenced as columns 2 and 3): - Regression of the inflation gap coefficient on past inflation plus additional controls (both values at time of IT adoption and average from adoption to latest year). - Inclusion of these controls does not affect the magnitude nor the statistical significance of the relationship between β_c and past inflation. - Similar exercises (referenced as columns 5 and 6) confirm that the output gap coefficient remains unrelated to past inflation. ### Table A2.1: Country-specific Estimates and Past Inflationary History — Key Statistics - Dependent variables summarized as two groups: Expected inflation coefficient (columns 1–3) and Output gap coefficient (columns 4–6). - Coefficients on Pre-IT inflation: - Column (1): 0.995*** (0.317) - Column (2): 1.021*** (0.336) - Column (3): 1.002*** (0.329) - Column (4): 0.150 (0.115) - Column (5): 0.175 (0.173) - Column (6): 0.144 (0.125) - Other covariates (coefficient estimates and bootstrapped standard errors in parentheses): - Initial FD index: - Column (1): -0.0746 (0.502) - Column (4): 0.200 (0.166) - Initial trade over GDP: - Column (1): -0.00173 (0.00193) - Column (4): -9.23e-06 (0.000707) - Initial KA index: - Column (1): 0.0181 (0.170) - Column (4): -0.0197 (0.0644) - Initial CBIE index: - Column (1): -0.223 (0.319) - Column (4): 0.0105 (0.188) - Average FD index: - Column (2): -0.00109 (0.0160) - Column (5): 0.00192 (0.00403) - Average trade over GDP: - Column (2): -5.68e-05 (0.000121) - Column (5): 1.46e-05 (4.93e-05) - Average KA index: - Column (2): 0.00791 (0.0127) - Column (5): -0.00256 (0.00362) - Average CBIE index: - Column (2): -0.00423 (0.0168) - Column (5): -0.00387 (0.0104) - Constant terms: - Column (1): 0.237*** (0.0542) - Column (2): 0.486 (0.348) - Column (3): 0.265 (0.197) - Column (4): 0.0427*** (0.0145) - Column (5): -0.0332 (0.151) - Column (6): 0.0841 (0.0905) - Sample and fit statistics: - Observations: 28 (for each column) - R-squared: - Column (1): 0.415 - Column (2): 0.441 - Column (3): 0.436 - Column (4): 0.091 - Column (5): 0.151 - Column (6): 0.135 - Notes: - Bootstrapped standard errors in parentheses. - Significance indicators: *** p<0.01, ** p<0.05, * p<0.1. *Source: Annex II. Results from Country-by-Country, wpiea2025079-print-pdf* ### Annex IV: Institutional Arrangements of Inflation ### Annex IV: Institutional Arrangements of Inflation ### Targeting Countries (Table A4.1: Individual Countries’ Inflation Target) - Albania — Target set by: CB; Target measure: H CPI; Target in 2025: 3%; Target type: Point; Target horizon: Medium term - Armenia — Target set by: G and CB; Target measure: H CPI; Target in 2025: 4%; Target type: Point, ±1.5 pp tolerance band; Target horizon: One-three years - Australia — Target set by: G and CB; Target measure: H CPI; Target in 2025: 2%-3%; Target type: Range; Target horizon: Medium term - Brazil — Target set by: G and CB (a); Target measure: H CPI; Target in 2025: 3%; Target type: Point, ±1.5 pp tolerance band; Target horizon: Yearly target until 2024, on a continuous basis since 2025 - Canada — Target set by: G and CB; Target measure: H CPI; Target in 2025: 2%; Target type: Point, ±1 pp tolerance band; Target horizon: Six-eight quarters - Chile — Target set by: CB; Target measure: H CPI; Target in 2025: 3%; Target type: Point, ±1 pp tolerance band; Target horizon: Two years - Colombia — Target set by: G and CB (b); Target measure: H CPI; Target in 2025: 3%; Target type: Point, ±1 pp tolerance band; Target horizon: Not disclosed - Costa Rica — Target set by: CB; Target measure: H CPI and Core CPI; Target in 2025: 3%; Target type: Point, ±1 pp tolerance band; Target horizon: Not disclosed - Czech Republic — Target set by: CB; Target measure: H CPI; Target in 2025: 2%; Target type: Point, ±1 pp tolerance band; Target horizon: 12-18 months - Dominican Republic — Target set by: G and CB (b); Target measure: H CPI; Target in 2025: 4%; Target type: Point, ±1 pp tolerance band; Target horizon: Two years - Georgia — Target set by: CB; Target measure: H CPI; Target in 2025: 3%; Target type: Point; Target horizon: Three years - Ghana — Target set by: G and CB; Target measure: H CPI; Target in 2025: 8%; Target type: Point, ±2 pp tolerance band; Target horizon: Four quarters - Guatemala — Target set by: G and CB (b); Target measure: H CPI; Target in 2025: 4%; Target type: Point, ±1 pp tolerance band; Target horizon: Medium term - Hungary — Target set by: CB; Target measure: H CPI; Target in 2025: 3%; Target type: Point, ±1 pp tolerance band; Target horizon: Five-eight quarters - Iceland — Target set by: G and CB; Target measure: H CPI; Target in 2025: 2.5%; Target type: Point; Target horizon: On average - India — Target set by: G and CB; Target measure: H CPI; Target in 2025: 4%; Target type: Point, ±2 pp tolerance band; Target horizon: Five years - Indonesia — Target set by: G and CB; Target measure: H CPI; Target in 2025: 2.5%; Target type: Point, ±1 pp tolerance band; Target horizon: Three years - Israel — Target set by: G and CB; Target measure: H CPI; Target in 2025: 1%-3%; Target type: Range; Target horizon: Within two years - Jamaica — Target set by: G; Target measure: H CPI; Target in 2025: 4%-6%; Target type: Range; Target horizon: Three fiscal years - Japan — Target set by: CB; Target measure: H CPI; Target in 2025: 2%; Target type: Point; Target horizon: Yearly - Kazakhstan — Target set by: CB; Target measure: H CPI; Target in 2025: 4%-6%; Target type: Range; Target horizon: Medium term - Kenya — Target set by: G and CB; Target measure: H CPI; Target in 2025: 5%; Target type: Point, ±2.5 pp tolerance band; Target horizon: Not discloased - Korea — Target set by: G and CB; Target measure: H CPI; Target in 2025: 2%; Target type: Point; Target horizon: Medium term - Mexico — Target set by: CB; Target measure: H CPI; Target in 2025: 3%; Target type: Point, ±1 pp tolerance band; Target horizon: Not disclosed - Moldova — Target set by: CB; Target measure: H CPI; Target in 2025: 5%; Target type: Point, ±1.5 pp tolerance band; Target horizon: On a continuous basis - New Zealand — Target set by: G and CB; Target measure: H CPI; Target in 2025: 2%; Target type: Point, ±1 pp tolerance band; Target horizon: Medium term - Norway — Target set by: G; Target measure: H CPI; Target in 2025: 2%; Target type: Point; Target horizon: Medium term - Paraguay — Target set by: CB; Target measure: H CPI; Target in 2025: 4%; Target type: Point, ±2 pp tolerance band; Target horizon: Not discloased - Peru — Target set by: CB; Target measure: H CPI; Target in 2025: 1%-3%; Target type: Range; Target horizon: On a continuous basis - Philippines — Target set by: G and CB; Target measure: H CPI; Target in 2025: 3%; Target type: Point, ±1 pp tolerance band; Target horizon: Two years - Poland — Target set by: CB; Target measure: H CPI; Target in 2025: 2.5%; Target type: Point, ±1 pp tolerance band; Target horizon: Medium term - Romania — Target set by: G and CB; Target measure: H CPI; Target in 2025: 2.5%; Target type: Point, ±1 pp tolerance band; Target horizon: Medium term - Russia — Target set by: G and CB; Target measure: H CPI; Target in 2025: 4%; Target type: Point; Target horizon: On a continuous basis - Serbia — Target set by: G and CB; Target measure: H CPI; Target in 2025: 3%; Target type: Point, ±1.5 pp tolerance band; Target horizon: On a continuous basis - South Africa — Target set by: G; Target measure: H CPI; Target in 2025: 3%-6%; Target type: Range; Target horizon: On a continuous basis - Sri Lanka — Target set by: G and CB; Target measure: H CPI; Target in 2025: 5%; Target type: Point, ±2 pp tolerance band; Target horizon: Two years - Sweden — Target set by: CB; Target measure: H CPI; Target in 2025: 2%; Target type: Point; Target horizon: Not discloased - Thailand — Target set by: G and CB; Target measure: H CPI; Target in 2025: 1%-3%; Target type: Range; Target horizon: Medium term - Turkey — Target set by: G and CB; Target measure: H CPI; Target in 2025: 5%; Target type: Point, ±2 pp tolerance band; Target horizon: Two years - Uganda — Target set by: CB; Target measure: Core CPI; Target in 2025: 5%; Target type: Point, ±3 pp tolerance band; Target horizon: One-three years - Ukraine — Target set by: CB; Target measure: H CPI; Target in 2025: 5%; Target type: Point; Target horizon: Medium term - United Kingdom — Target set by: G; Target measure: H CPI; Target in 2025: 2%; Target type: Point; Target horizon: At all times - Uruguay — Target set by: G and CB; Target measure: H CPI; Target in 2025: 3%-6%; Target type: Range; Target horizon: Two years - Uzbequistan — Target set by: CB; Target measure: H CPI; Target in 2025: 5%; Target type: Point; Target horizon: Not discloased *Note: (a) The inflation target is set by the National Monetary Council, composed of the Minister of Finance, the Minister of Planning and Budget, and the Central Bank Governor. *Note: (b) The inflation target is set by the central bank board, which includes the finance minister as a voting member. *Abbreviations: CB - Central Bank. G - Government. H CPI - Headline CPI. CPI - Consumer price index. pp - percentage point(s). ### Decision Making in Inflation Targeting Central Banks (Table A4.2) - Albania — Number on policy making committee: 9; External members?: No; Meetings per year: 8; Governor's term (years): 7 - Armenia — Number on policy making committee: 8; External members?: No; Meetings per year: 8; Governor's term (years): 6 - Australia — Number on policy making committee: 9; External members?: 6; Meetings per year: 8; Governor's term (years): 7 - Brazil — Number on policy making committee: 9; External members?: No; Meetings per year: 8; Governor's term (years): 4 - Canada — Number on policy making committee: 7; External members?: 2; Meetings per year: 8; Governor's term (years): 7 - Chile — Number on policy making committee: 5; External members?: No; Meetings per year: 8; Governor's term (years): 5 - Colombia — Number on policy making committee: 7; External members?: No; Meetings per year: 12; Governor's term (years): 4 - Costa Rica — Number on policy making committee: 7; External members?: No; Meetings per year: 8; Governor's term (years): 4 - Czech Republic — Number on policy making committee: 7; External members?: No; Meetings per year: 8; Governor's term (years): 6 - Dominican Republic — Number on policy making committee: 9; External members?: 6; Meetings per year: 12; Governor's term (years): 2 - Georgia — Number on policy making committee: 14; External members?: No; Meetings per year: 8; Governor's term (years): 7 - Ghana — Number on policy making committee: 7; External members?: 2; Meetings per year: 6; Governor's term (years): 4 - Guatemala — Number on policy making committee: 9; External members?: 4; Meetings per year: 8; Governor's term (years): 4 - Hungary — Number on policy making committee: 9; External members?: 4; Meetings per year: 12; Governor's term (years): 6 - Iceland — Number on policy making committee: 5; External members?: 2; Meetings per year: 6; Governor's term (years): 5 - India — Number on policy making committee: 6; External members?: 3; Meetings per year: 8; Governor's term (years): 5 - Indonesia — Number on policy making committee: 6; External members?: No; Meetings per year: 12; Governor's term (years): 5 - Israel — Number on policy making committee: 6; External members?: 3; Meetings per year: 12; Governor's term (years): 5 - Jamaica — Number on policy making committee: 5; External members?: 2; Meetings per year: 8; Governor's term (years): 5 - Japan — Number on policy making committee: 9; External members?: No; Meetings per year: 8; Governor's term (years): 5 - Kazakhstan — Number on policy making committee: 10; External members?: No; Meetings per year: 8; Governor's term (years): 5 - Kenya — Number on policy making committee: 9; External members?: 8; Meetings per year: 6; Governor's term (years): 4 - Korea — Number on policy making committee: 7; External members?: No; Meetings per year: 8; Governor's term (years): 4 - Mexico — Number on policy making committee: 5; External members?: No; Meetings per year: 8; Governor's term (years): 6 - Moldova — Number on policy making committee: 5; External members?: No; Meetings per year: 8; Governor's term (years): 7 - New Zealand — Number on policy making committee: 5 to 7; External members?: 1 to 2; Meetings per year: 7; Governor's term (years): 5 - Norway — Number on policy making committee: 5; External members?: 2; Meetings per year: 8; Governor's term (years): 6 - Paraguay — Number on policy making committee: 5; External members?: No; Meetings per year: 12; Governor's term (years): 5 - Peru — Number on policy making committee: 7; External members?: No; Meetings per year: 12; Governor's term (years): 5 - Philippines — Number on policy making committee: 7; External members?: No; Meetings per year: 8; Governor's term (years): 6 - Poland — Number on policy making committee: 10; External members?: 9; Meetings per year: 12**; Governor's term (years): 6 - Romania — Number on policy making committee: 9; External members?: 5; Meetings per year: 8; Governor's term (years): 5 - Russia — Number on policy making committee: 15; External members?: No; Meetings per year: 12; Governor's term (years): 5 - Serbia — Number on policy making committee: 5; External members?: No; Meetings per year: 12; Governor's term (years): 6 - Seychelles — Number on policy making committee: 7; External members?: No; Meetings per year: 12; Governor's term (years): 6 - South Africa — Number on policy making committee: 6; External members?: No; Meetings per year: 6; Governor's term (years): 5 - Sri Lanka — Number on policy making committee: 11; External members?: 2; Meetings per year: 6; Governor's term (years): 6 - Sweden — Number on policy making committee: 5; External members?: No; Meetings per year: 5; Governor's term (years): 6 - Thailand — Number on policy making committee: 7; External members?: 4; Meetings per year: 6; Governor's term (years): 5 - Turkey — Number on policy making committee: 7; External members?: 1; Meetings per year: 8; Governor's term (years): 5 - Uganda — Number on policy making committee: 12; External members?: No; Meetings per year: 6; Governor's term (years): 5 - Ukraine — Number on policy making committee: 7; External members?: No; Meetings per year: 8; Governor's term (years): 8 - United Kingdom — Number on policy making committee: 9; External members?: 4; Meetings per year: 8; Governor's term (years): 5 - Uruguay — Number on policy making committee: 6***; External members?: No; Meetings per year: 8; Governor's term (years): 5 - Uzbequistan — Number on policy making committee: 10; External members?: 2; Meetings per year: 8; Governor's term (years): 8 *Notes: */ Does not include government representatives. **/ Only 11 of the 12 meetings are monetary policy decision making meetings. ***/ Article 31 of the Central Bank law establishes that the MPC consists of the 3 Board member and 3 central bank senior staff, but only Board members have voting rights. ### Accountability and Transparency (Table A4.3) - Albania — Parliamentary hearings?: Yes, annual; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 4 - Armenia — Parliamentary hearings?: Yes, annual; Minutes published: Yes; Votes published: No; Inflation Report: Yes; Frequency: 4 - Australia — Parliamentary hearings?: Yes, twice a year; Minutes published: Yes; Votes published: No; Inflation Report: Yes; Frequency: 4 - Brazil — Parliamentary hearings?: Yes, twice a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Canada — Parliamentary hearings?: Yes, twice a year; Minutes published: Yes; Votes published: No; Inflation Report: Yes; Frequency: 4 - Chile — Parliamentary hearings?: Yes, twice a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Colombia — Parliamentary hearings?: Yes, twice a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Costa Rica — Parliamentary hearings?: Yes, at least once a year; Minutes published: Yes; Votes published: No; Inflation Report: Yes; Frequency: 4 - Czech Republic — Parliamentary hearings?: Yes, at least twice a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Dominican Republic — Parliamentary hearings?: No; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 2 - Georgia — Parliamentary hearings?: Yes, once a year; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 4 - Ghana — Parliamentary hearings?: No; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 6 - Guatemala — Parliamentary hearings?: Yes, twice a year; Minutes published: No; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Hungary — Parliamentary hearings?: Yes, twice a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Iceland — Parliamentary hearings?: Yes, twice a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - India — Parliamentary hearings?: No; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 2 - Indonesia — Parliamentary hearings?: Yes, every quarter; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 4 - Israel — Parliamentary hearings?: Yes, twice a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 2 - Jamaica — Parliamentary hearings?: Yes, at least twice a year; Minutes published: Yes; Votes published: No; Inflation Report: Yes; Frequency: 4 - Japan — Parliamentary hearings?: Yes, twice a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Kazakhstan — Parliamentary hearings?: No; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 4 - Kenya — Parliamentary hearings?: No; Minutes published: No; Votes published: No; Inflation Report: No; Frequency: N/A - Korea — Parliamentary hearings?: Yes, if requested from national assembly; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Mexico — Parliamentary hearings?: No; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Moldova — Parliamentary hearings?: Yes, once a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - New Zealand — Parliamentary hearings?: Yes, every quarter; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Norway — Parliamentary hearings?: Yes, annual; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Paraguay — Parliamentary hearings?: No; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Peru — Parliamentary hearings?: Yes, once a year; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 4 - Philippines — Parliamentary hearings?: Yes, every quarter and annual; Minutes published: Yes; Votes published: No; Inflation Report: Yes; Frequency: 4 - Poland — Parliamentary hearings?: Yes, once a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 3 - Romania — Parliamentary hearings?: No; Minutes published: Yes; Votes published: No; Inflation Report: Yes; Frequency: 4 - Russia — Parliamentary hearings?: Yes, annual; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 4 - Serbia — Parliamentary hearings?: No; Minutes published: No; Votes published: No (a); Inflation Report: Yes; Frequency: 4 - South Africa — Parliamentary hearings?: Yes, at least once a year; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 2 - Sri Lanka — Parliamentary hearings?: Yes, but unclear frequency; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 2 - Sweden — Parliamentary hearings?: Yes, twice a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 (c) - Thailand — Parliamentary hearings?: No; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 (b) - Turkey — Parliamentary hearings?: Yes, twice a year (two presentations); Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Uganda — Parliamentary hearings?: Yes, if Audit General raises issues; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 6 - United Kingdom — Parliamentary hearings?: Yes, four times a year; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Uruguay — Parliamentary hearings?: Possibly, but not required; Minutes published: Yes; Votes published: Yes; Inflation Report: Yes; Frequency: 4 - Uzbequistan — Parliamentary hearings?: Yes, once a year; Minutes published: No; Votes published: No; Inflation Report: Yes; Frequency: 4 *Notes: (a) Although there is a voting, agreement on inflation target is reached by consensus in practice. *Notes: (b) The Inflation Report is called Monetary Policy Report. *Notes: (c) A full monetary policy report is prepared four times a year (with every second monetary policy meeting), while a shorter update is prepared in the remainder of the meetings. *Source: 2022 AREAER Yearly Report and individual central bank's websites.* --- _Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025079-print-pdf.pdf_