## REFERENCES

## Source details

**Canonical URL:** [REFERENCES](https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026103-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2026/english/wpiea2026103-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2026/english/wpiea2026103-source-pdf.pdf.json)

---

### I. Introduction and scope
- Focus:
  - Explains why the post-COVID inflation surge varied markedly across countries and how energy price changes propagated to headline inflation.
- Sample and periods:
  - Cross-country panel covering 2015–2024.
- Principal questions:
  - Which factors (supply shocks, macroeconomic conditions, institutional characteristics, policy responses) explain cross-country differences in post-COVID inflation?
  - How did changes in energy prices transmit to headline inflation?

### II. Key findings (headline)
- Dominant drivers of cross-country variation in post-COVID inflation:
  - Historical inflation (pre-COVID inflation).
  - Magnitude of the domestic energy price shock.
- Pass-through from energy prices to headline inflation:
  - Estimated using local projections for a broad sample of advanced and emerging economies.
  - The inflation pass-through has become more pronounced in the post-pandemic period.
  - Pass-through amplification particularly evident in:
    - non-inflation targeting regime countries (IT vs non-IT distinction),
    - emerging markets,
    - countries that relied less on fossil fuel subsidies to counter supply shocks.

### III. Methods and analysis
- Cross-sectional analysis:
  - Cross-country regressions relating π_Post−COVID to π_Pre−COVID and other covariates.
- Local projections:
  - Used to trace second-round effects of energy price shocks and explore heterogeneity in pass-through.
  - Coefficient of interest βh: impact of a 1 percent change in energy prices on cumulative CPI inflation over horizon h.
- Robustness and supplementary analyses:
  - Annex I: Robustness checks for cross-sectional regressions (robust estimators, quantile regressions).
  - Annex II: Pass-through to core inflation.
  - Annex III: Variables and their sources.

### IV. Empirical findings — Cross-country regressions (exact reported metrics)
- Baseline specification:
  - π_Post−COVID = α + β π_Pre−COVID + γ X + ε
  - π_Post−COVID: average annual post-COVID inflation over 2020Q1-2024Q4.
  - π_Pre−COVID: average annual pre-COVID inflation over 2015Q1-2019Q4.
- Sample sizes and model fit:
  - Cross-country sample: 130 countries (where reported).
  - Table 1: R-squared range reported: 0.542, 0.543, 0.506, 0.568, 0.571, 0.574, 0.585, 0.758.
  - Table 2: Column (1) R-squared = 0.846 when including Pre-COVID inflation and Post-COVID energy inflation jointly.
- Pre-COVID inflation (robust predictor):
  - Reported coefficients in Table 1 include: 1.198***, 1.205***, 1.016***, 0.897***, 0.890***, 0.893***, 0.895***, 0.880*** (standard errors shown in table).
  - Robust/quantile examples in Annex I: 0.665*** (0.054), 0.702*** (0.056), 0.885*** (0.052), 0.839*** (0.060), 0.827*** (0.060), 0.826*** (0.059), 0.831*** (0.057), 0.867*** (0.066), and others up to 0.997** (0.415) in alternate specs.
- Post-COVID energy inflation and fundamentals:
  - Post-COVID energy inflation coefficients reported: 0.470*** and 0.482*** (std. err. 0.080 and 0.074) in Table 2; other reported values include 0.259*** (0.033) and 0.153*** (0.033); quantile: 0.387*** (0.052), 0.420*** (0.057).
  - Post-COVID REER growth examples: 0.421*** (0.087) and 0.290*** (0.088); quantile: 0.446*** (0.094) and 0.322** (0.140).
  - Pre-COVID life expectancy: -0.149** (0.072) in one reported column.
  - Other fundamentals (post-COVID real GDP growth, change in FX reserves/GDP, trade openness, demographic shares, pre-COVID inflation volatility) generally not robustly significant once initial inflation and energy shock are included.
- Institutional factors (sensitivity and mixed evidence):
  - Most institutional indicators not significantly associated after conditioning on pre-COVID inflation and energy shocks.
  - Exceptions and sensitive estimates:
    - Electoral regime index: 1.701* (0.948) in one specification.
    - Central bank independence index: 5.911*** (2.082) in one specification; other reported CB independence values include 5.018*** (1.375) and 3.837* (1.926) across columns.
- Policy variables (reduced-form associations, endogeneity caveat):
  - No robust association between post-COVID inflation and changes in policy rates, public debt ratio, or REER appreciation in reduced-form results generally.
  - Frequent macroprudential tightening and FX interventions are positively and significantly associated with inflation in some specifications (likely reflecting endogeneity).
  - Relative size of fossil fuel subsidies (share of GDP): negative association with inflation in some specs — countries increasing subsidies above historical averages experienced more moderate inflation (interpreted as cushioning effect, causality not identified).

### V. Dynamic analysis of energy-price pass-through (local projections and event studies)
- Sample for pass-through analysis:
  - 70 advanced and emerging economies for 2015Q1-2024Q4.
  - Of these, 33 are IT countries and 37 are non-IT countries.
- Aggregate pass-through estimates (cumulative elasticities, exact reported):
  - Total-sample cumulative elasticity: 0.07 over 8 quarters (statistically significant).
  - Post-COVID cumulative elasticity: 0.1.
  - Pre-COVID cumulative elasticity: 0.02.
  - Interpretation reported: Pass-through roughly tripled in the post-pandemic period relative to pre-COVID.
- Heterogeneity across country groups (exact reported elasticities):
  - Emerging economies cumulative elasticity: 0.06 (significant).
  - Advanced economies: mostly insignificant pass-through.
  - IT countries cumulative elasticity: 0.04 (significant).
  - Non-IT countries cumulative elasticity: 0.14.
  - Countries with median fossil fuel subsidies above sample median: cumulative elasticity 0.05 (significant).
  - Countries with median subsidies below sample median: cumulative elasticity 0.09.
- Asymmetric effects (exact reported peaks):
  - Positive energy price changes: headline inflation reacts, peaking at 0.11.
  - Negative energy price changes: impact small and statistically insignificant.
- Time- and regime-specific combined results (selected reported estimates):
  - Post-COVID (Panel A): non-IT cumulative elasticity 0.15 (significant); IT countries insignificant.
  - Pre-COVID (Panel B): IT cumulative elasticity 0.01 (significant); non-IT insignificant.

### VI. Robustness checks and identification (Annex I)
- Estimators and methods:
  - Robust regression estimator (iterative weighted least squares; Andersen, 2008) and quantile regression for the median.
- Robustness results (consistent patterns, exact examples):
  - Pre-COVID inflation remains the most robust predictor; reported coefficients across robustness tables include many values in the 0.6–0.95 range (examples above).
  - Post-COVID energy inflation and REER growth maintain positive and often significant coefficients across specifications.
  - Exchange rate peg dummy examples: -1.156** (0.495), -1.392* (0.807); quantile: -1.286** (0.569), -2.281** (0.901).
  - Policy/institutional coefficient examples across robustness tables:
    - Debt rule dummy: 1.220** (0.607), 1.712** (0.723).
    - Expenditure rule dummy: 0.705 (0.476), 1.166* (0.643).
    - Revenue rule dummy: -0.268 (0.636), -2.249** (1.002).
    - Change in central bank policy rate: 0.558*** (0.133) in one robust regression; alternative: 0.182 (0.109).
    - Change in macroprudential instruments: 1.105*** (0.415) and 0.795 (0.566) in different columns.
    - Change in Debt/GDP ratio: -0.044*** (0.015) and -0.092*** (0.031).
    - Post-COVID FX interventions in % of GDP: 1.195* (0.601) and 0.315 (0.908).
    - Change in fossil fuel subsidies in % of GDP: -0.280 (0.185) and -0.084 (0.326).
  - Observations and R-squared examples across tables: observations reported include 129, 129, 84, 90, 90, 90, 90, 56; other counts 103, 126, 120, 120, 126, 129, 120, 127, 87; R-squared examples include 0.546, 0.567, 0.786, 0.696, 0.688, 0.693, 0.711, 0.811, 0.787, 0.523, 0.573, 0.697, 0.564, 0.603, 0.522, 0.483, 0.782.

### VII. Mechanism, interpretation, and limitations
- Core mechanism emphasized in source:
  - Pre-existing inflation history acts as a conditioning state variable that alters transmission (the slope) from energy shocks to broader inflation via expectations, wage-setting, and second-round effects, rather than mechanically adding to the level of inflation.
- Empirical implication:
  - Differences in post-COVID inflation persistence largely reflect heterogeneity in propagation and feedback channels (expectations drift, SREs), not primarily differences in initial shock magnitudes.
- Limitations and caution:
  - Reduced-form regressions document conditional correlations; coefficients are not presented as causal estimates.
  - Policy coefficient interpretation complicated by endogeneity (policy actions often respond to inflationary pressures).

### VIII. Policy-relevant implications and trade-offs (as drawn in source)
- Main policy-relevant points preserved verbatim from source interpretation:
  - Monitoring and managing domestic exposure to energy price shocks is critical for inflation stabilization.
  - Institutional and policy frameworks (including inflation-targeting regimes and fossil fuel subsidy policies) shaped the magnitude of pass-through from energy shocks to inflation.
  - Fossil fuel subsidies can reduce pass-through but impose fiscal costs and distributional considerations.
  - Monetary tightening more effective in influencing speed of disinflation than in preventing initial supply-driven rises (literature-framed point).
  - Interpreting policy coefficients in reduced-form requires caution because observed positive associations between tightening interventions and inflation may reflect responsive policy actions.

### IX. Annex II and Annex III (variables, figures, country lists)
- Annex II:
  - Shows energy price pass-through to core CPI inflation (Figure A_II_1). Figure presentation details: line = point estimates; shaded area = 95 percent confidence interval; filled circles indicate significance.
- Annex III:
  - Table A_III_1 lists IT and non-IT countries (IT: 37 listed; Non-IT: 33 listed as in source).
  - Table A_III_2 provides variable definitions and sources (examples preserved exactly: infl, infl_e, Real GDP growth, FX reserves/GDP ratio, Trade openness, Life expectancy, Share of >65 population, REER growth, Inflation volatility, Macroprudential instruments, FX interventions in % of GDP, Debt/GDP ratio, Fossil fuel subsidies in % of GDP, CB independence index, IT dummy, AE dummy, Exchange rate peg dummy, expenditure/revenue/balanced budget/debt rule dummies, Central bank policy rate, Trade union density rate, Political stability index, V-DEM indices, etc.).

### X. Overall conclusions and suggested future research (exactly as reported)
- Data and estimates summary:
  - Cross-country regressions across 130 economies and dynamic analysis for 70 economies were used.
  - Pre-pandemic inflation levels and the scale of the energy price shock explain most of the cross-country variation in post-COVID inflation (85 percent reported when combined).
- Three headline conclusions (verbatim):
  1. Pre-pandemic inflation levels and domestic energy price shocks explain most cross-country variation in post-COVID inflation outcomes.
  2. Strength of supply-shock transmission to domestic prices intensified in the post-COVID period; pass-through roughly tripled.
  3. Policy and institutional variables play a nuanced role—some factors (e.g., fossil fuel subsidies) mechanically dampen pass-through but carry fiscal and distributional trade-offs; many governance and policy indicators show no direct statistical association once shocks and pre-pandemic conditions are accounted for.
- Policy-relevant implications (preserved):
  - Supply-driven inflation episodes are mediated by pre-existing inflation credibility and exposure to energy price swings; anchoring expectations and shock-absorption capacity matter.
  - Fossil fuel subsidies can reduce pass-through but impose fiscal costs and distributional considerations that must be weighed.
  - Interpreting policy coefficients in reduced-form regressions requires caution because of policy endogeneity.
- Suggested directions for future research:
  - Examine nonlinearities and the interplay between fiscal and monetary policy during large global shocks to better identify causal channels and policy trade-offs.

*Source: IMF Working Paper "One Global Shock, Many Inflation Paths: Explaining Post-COVID Inflation Divergence" (content unit: REFERENCES and introductory material).*

### REFERENCES .............................................................................................................

### REFERENCES

### I. Introduction and scope
- Focus: Explains why the post-COVID inflation surge varied markedly across countries and how energy price changes propagated to headline inflation.
- Sample and periods:
  - Cross-country panel covering 2015–2024.
- Principal questions:
  - Which factors (supply shocks, macroeconomic conditions, institutional characteristics, policy responses) explain cross-country differences in post-COVID inflation?
  - How did changes in energy prices transmit to headline inflation?

### II. Key findings
- Dominant drivers of cross-country variation in post-COVID inflation:
  - Historical inflation.
  - Magnitude of the domestic energy price shock.
- Less robust or subdominant factors:
  - Other structural factors are not robustly associated with inflation variation over this period.
- Pass-through from energy prices to headline inflation:
  - Estimated using local projections for a broad sample of advanced and emerging economies.
  - The inflation pass-through has become more pronounced in the post-pandemic period.
  - Pass-through amplification is particularly evident in:
    - non-inflation targeting regime countries (IT vs non-IT distinction emphasized in figures and analysis),
    - emerging markets,
    - countries that have relied less on fossil fuel subsidies to counter supply shocks.

### III. Methods and analysis
- Cross-sectional analysis:
  - Cross-country regressions to assess the impact of supply shocks, macroeconomic conditions, institutional characteristics, and policy responses on inflation outcomes.
- Local projections:
  - Used to trace second-round effects of energy price shocks and to explore heterogeneity in pass-through.
- Robustness and supplementary analyses:
  - Annex I: Robustness checks for cross-sectional regressions.
  - Annex II: Pass-through to core inflation.
  - Annex III: Variables and their sources.

### IV. Interpretation and implications
- Even with common global shocks (energy and food price increases, logistics bottlenecks, supply-chain disruptions), outcomes diverged substantially across countries with similar exposures and macro frameworks.
- Institutional anchors and traditional predictors of inflation (e.g., central bank independence, well-anchored expectations) may have lost some explanatory power amid the scale and simultaneity of global shocks; historical inflation experience can be a crucial channel.
- Policy-relevant implications (inferred from findings in the source text):
  - Monitoring and managing domestic exposure to energy price shocks is critical for inflation stabilization.
  - Institutional and policy frameworks (including inflation-targeting regimes and fossil fuel subsidy policies) shaped the magnitude of pass-through from energy shocks to inflation.

### V. Figures and tables (as listed in the source)
- Figures (selected titles from the source):
  - 1. Average annual CPI inflation: Pre- and post-COVID
  - 2. Post-COVID CPI inflation around the world
  - 3. Distribution of CPI inflation
  - 4. Dynamics of price levels
  - 5. Event study analysis: Evolution of energy price inflation in IT and non-IT countries
  - 6. Energy price pass through: Total sample
  - 7. Energy price pass-through: Pre- versus post-COVID period
  - 8. Energy price pass-through: Advanced versus emerging economies
  - 9. Energy price pass-through: IT versus non-IT countries
  - 10. Energy price pass-through: IT versus non-IT countries
  - 11. Energy price pass-through: Asymmetric effects of positive and negative energy price changes
  - 12. Energy price pass-through: IT versus non-IT countries during pre- and post-COVID period
- Tables (selected titles from the source):
  - 1. Estimation results: The impact of country-specific characteristics
  - 2. Estimation results: The impact of fundamental factors
  - 3. Estimation results: The impact of institutional factors
  - 4. Estimation results: The impact of policies

*Source: IMF Working Paper "One Global Shock, Many Inflation Paths: Explaining Post-COVID Inflation Divergence" (content unit: REFERENCES and introductory material).*

### Section 5 concludes and draws policy lessons. Given the cross-country scope and reduced-form

### Section 5 concludes and draws policy lessons. Given the cross-country scope and reduced-form

### Literature review
- Post-pandemic inflation surge framed as largely driven by common, global supply-side forces: pandemic-related supply disruptions, rapid demand rebound, sharp increase in energy and food prices.
- Early literature emphasizes global synchronization and supply-side dominance; later work focuses on cross-country divergence in persistence.
- Key empirical regularities:
  - “Lived experience” (historical inflation) strongly correlates with post-pandemic inflation persistence (Gagnon and Kamin, 2025; Gagnon and Rose, 2024).
  - Second-round effects (SREs) from commodity/energy shocks to wages and core inflation exist but are conditional and heterogeneous (Baba and Lee, 2022; De Jonghe and others, 2025).
  - Expectations formation (gradual updating and disagreement) can amplify persistence where historical inflation is higher (Coibion and Gorodnichenko, 2012, 2015).
- Policy literature highlights:
  - Monetary tightening more effective in influencing speed of disinflation than in preventing initial supply-driven rise (Dynan and Elmendorf, 2024; Reifschneider, 2024).
  - Fiscal measures (energy price subsidies, cost-of-living supports) dampened headline inflation and pass-through in the short run (IMF, 2022, 2024; Hsu, 2024; Fatás, 2024).
- Paper’s contribution: shift from outcomes to transmission — interpret inflation history as a conditioning state variable that governs how energy shocks feed into broader price dynamics. Differences in persistence largely reflect differences in transmission rather than differences in shock size.

### Stylized facts: Post-COVID inflation surge
- Two successive global shocks: pandemic disruptions and commodity-price spike after Russia’s invasion of Ukraine.
- Pre-COVID baseline: decade of historically low and stable global inflation; many emerging markets experienced declining, better-anchored inflation expectations.
- Post-COVID dynamics:
  - Headline inflation rose sharply and synchronously across advanced and emerging economies, reaching multidecade highs by late 2022.
  - Initial drivers:
    - Supply: factory closures, labor shortages, lockdown restrictions → production bottlenecks, shipping delays, surging freight costs, shortages of intermediate inputs.
    - Demand: fiscal transfers, accommodative monetary policy, rapid reopening → strong rebound in consumption, surge in demand for durable goods colliding with constrained supply.
  - Geopolitical shock (Russia–Ukraine) triggered energy market volatility (oil, natural gas, refined products) and global food price spikes (grain, fertilizer), with disproportionate effects on food-import-dependent economies.
- Cross-country heterogeneity:
  - Energy-importing economies, economies with limited fiscal space, and countries with weaker exchange rates experienced larger and more persistent price surges.
  - Regional patterns: Asia-Pacific experienced relatively modest inflation; Eastern European countries hit harder due to energy links with Russia and higher energy intensity.
- Empirical illustrations (data periods and notes preserved):
  - Average annual CPI inflation compared for 2015Q1-2019Q4 (pre-COVID) and 2020Q1-2024Q4 (post-COVID).
  - Panel and map evidence indicate a rightward shift and flattening of global inflation distribution after COVID, with increased cross-country dispersion.
  - Dynamics of CPI index normalized at 100 in 2020Q1 for 130 countries show wide divergence: in a small set price levels rose by 10-15 percent over the 5-year period post-pandemic (corresponds to 2-3 percent average annual inflation); in most other countries increase exceeded 15 percent and surpassed 60 percent in some countries.

### Empirical analysis: Strategy and interpretation
- Two-step empirical strategy:
  1. Cross-country regressions on a sample of 130 economies to relate average annual post-COVID inflation to pre-COVID inflation and determinants.
  2. Local projections framework to measure pass-through from energy price shocks to headline inflation as a proxy for SREs; compare pass-through before and after the pandemic and across country groups.
- Emphasis: reduced-form regressions document robust empirical regularities; coefficients are conditional correlations, not causal estimates. Many variables, including institutional features and policy measures, may be jointly determined with inflation outcomes.

### Empirical findings — Cross-country regressions (key numeric results preserved)
- Baseline regression form:
  - π_Post−COVID = α + β π_Pre−COVID + γ X + ε
  - π_Post−COVID: average annual post-COVID inflation over 2020Q1-2024Q4
  - π_Pre−COVID: average annual pre-COVID inflation over 2015Q1-2019Q4
- Sample sizes and model fit:
  - Cross-country sample: 130 countries (where reported).
  - Table 1: R-squared range in reported specifications: 0.542, 0.543, 0.506, 0.568, 0.571, 0.574, 0.585, 0.758.
  - Table 2: Column (1) reports R-squared = 0.846 when including Pre-COVID inflation and Post-COVID energy inflation jointly.
- Pre-COVID inflation:
  - Strong, positive, and statistically significant association with post-COVID inflation across specifications.
  - Coefficient examples from Table 1: Pre-COVID inflation coefficients reported include 1.198***, 1.205***, 1.016***, 0.897***, 0.890***, 0.893***, 0.895***, 0.880*** (standard errors shown in table).
- Fundamental factors (Table 2):
  - Post-COVID energy inflation: positive and significant association with headline post-COVID inflation.
    - Coefficients: 0.470*** and 0.482*** in reported specifications (standard errors 0.080 and 0.074).
  - Joint explanatory power: Pre-COVID inflation and Post-COVID energy inflation explain about 85 percent of variation (R-squared = 0.846).
  - Pre-COVID life expectancy: negative association in some specifications (coefficient -0.149** in one reported column; standard error 0.072); interpreted as proxying deeper structural characteristics, not causal.
  - Other fundamentals (post-COVID real GDP growth, change in FX reserves/GDP, trade openness, demographic shares, REER growth, pre-COVID inflation volatility) generally not statistically significant once initial inflation and energy shock are accounted for.
- Institutional factors (Table 3):
  - Most institutional indicators (rule of law, political stability, governance, corruption, public administration, democracy) not significantly associated with post-COVID inflation after conditioning on pre-COVID inflation and common shocks.
  - Exception: electoral regime index shows a positive association in one specification (coefficient 1.701* with standard error 0.948).
  - Central bank independence index: not consistently significant; one reported coefficient 5.911*** in a specification with large standard errors (2.082).
  - Interpretation: institutions may matter more for inflation persistence and disinflation dynamics than for the initial synchronized shock.
- Policy variables (Table 4 summary in text):
  - No significant association between post-COVID inflation and changes in policy rates, public debt ratio, or real effective exchange rate appreciation in reduced-form results.
  - Frequent macroprudential tightening and FX interventions are positively and significantly associated with inflation; likely reflects endogeneity (policy responses to high inflation drivers rather than causes).
  - Relative size of fossil fuel subsidies (as share of GDP): significant and negative association with inflation — countries increasing such subsidies above historical averages experienced more moderate inflation (interpretation: cushioning effect), though causality not identified.

### Mechanism and interpretation
- Core mechanism: Pre-existing inflation history operates as a conditioning state variable that alters transmission (the slope) from energy shocks to broader inflation via expectations, wage-setting, and SREs, rather than mechanically adding to the level of inflation.
- Empirical implication: Differences in post-COVID inflation persistence reflect heterogeneity in propagation and feedback channels (expectations drift, second-round effects), not primarily differences in initial shock magnitudes.
- Caution: Reduced-form framework limits causal claims; observed associations may capture unobserved country characteristics or endogenous policy responses.

*Source: https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026103-source-pdf.pdf*

### Annex I presents a range of robustness checks using robust estimator that downweighs

### Annex I presents a range of robustness checks using robust estimator that downweighs observations with larger absolute residuals using iterative weighted least squares (Andersen, 2008) and quantile regressions. The results remain qualitatively similar.

### Robustness checks and identification
- Robust estimators (iterative weighted least squares) and quantile regressions were applied; results remain qualitatively similar.
- Cross-sectional regressions using pre-COVID variables as determinants of post-COVID inflation were also run to reduce endogeneity risks; results remain qualitatively unchanged and “can be shared upon request.”
- The interpretation of reduced-form policy coefficients is complicated by endogeneity: policy actions often respond to inflationary pressures rather than driving them.

### Key empirical findings on policy determinants
- Short-run policy adjustments played a secondary role relative to the interaction between large supply shocks and pre-existing domestic conditions.
- Lack of robust association with monetary and fiscal policy changes reinforces the view that post-COVID inflation was largely driven by factors outside the immediate control of macroeconomic policy.
- Positive association observed between macroprudential tightening, foreign exchange intervention, and inflation likely reflects policy responses to inflationary pressures (endogeneity), making causal interpretation difficult.
- Negative association between fossil fuel subsidies and inflation points to a direct channel: subsidies dampened pass-through of energy price shocks to consumer prices, but involved fiscal costs and distributional implications.
- Cross-country differences in inflation outcomes were associated less with discretionary policy moves and more with how supply shocks propagated through domestic prices, conditional on initial inflation environments and available policy buffers.

### Dynamic analysis of energy-price pass-through: sample and methods
- Sample for pass-through analysis: 70 advanced and emerging economies for the period 2015Q1-2024Q4.
  - Of these, 33 are IT countries and 37 are non-IT countries.
- Event-study specification and local-projection specifications were used to estimate cumulative pass-through elasticities, controlling for country fixed effects and time fixed effects.
- The coefficient of interest, βh, denotes the impact of a 1 percent change in energy prices on cumulative CPI inflation over horizon h.

### Aggregate pass-through estimates and temporal variation
- Total-sample cumulative elasticity: 0.07 over 8 quarters and statistically significant.
- Pre- versus post-COVID:
  - Post-COVID cumulative elasticity: 0.1.
  - Pre-COVID cumulative elasticity: 0.02.
- Interpretation: Pass-through roughly tripled in the post-pandemic period relative to pre-COVID years, reflecting larger energy supply shocks and erosion of some inflation-dampening forces from the 2010s.

### Heterogeneity across country groups
- Advanced versus emerging economies:
  - Emerging economies cumulative elasticity: 0.06 and significant.
  - Advanced economies: mostly insignificant pass-through.
  - Interpretation: energy price shocks propagated faster to headline inflation in emerging economies.
- IT versus non-IT countries:
  - IT countries cumulative elasticity: 0.04 and significant.
  - Non-IT countries cumulative elasticity: 0.14.
  - Interpretation: inflation-targeting frameworks moderate but do not eliminate pass-through; IT countries show stronger expectations anchoring.
- Role of fossil fuel subsidies:
  - Countries with median fossil fuel subsidies above the overall sample median: cumulative elasticity 0.05 and statistically significant.
  - Countries with median subsidies below the overall sample median: cumulative elasticity 0.09.
  - Interpretation: relatively higher subsidies helped attenuate transmission of energy price shocks to inflation, at the expense of fiscal costs and possible long-term inefficiencies.

### Asymmetric effects of positive versus negative energy shocks
- Pass-through asymmetry:
  - Positive energy price changes: headline inflation reacts significantly, peaking at 0.11.
  - Negative energy price changes: impact is small and statistically insignificant.
- Interpretation: price rigidity and incomplete reversal contributed to persistent cross-country differences in inflation dynamics after energy price hikes.

### Time- and regime-specific combined results
- IT versus non-IT in pre- versus post-COVID periods (selected reported estimates):
  - Post-COVID period (Panel A): non-IT countries cumulative elasticity 0.15 and significant; IT countries insignificant.
  - Pre-COVID period (Panel B): IT countries cumulative elasticity 0.01 and significant; non-IT countries insignificant.
- Interpretation: largest pass-through observed in non-IT countries during the post-COVID period; monetary regime and period interact in shaping pass-through.

### Overall conclusions and implications
- Data and estimates summary:
  - Cross-country regressions across 130 economies and dynamic analysis for 70 economies were used.
  - Pre-pandemic inflation levels and the scale of the energy price shock explain most of the cross-country variation in post-COVID inflation (85 percent).
- Three headline conclusions:
  1. Pre-pandemic inflation levels and domestic energy price shocks explain most cross-country variation in post-COVID inflation outcomes.
  2. Strength of supply-shock transmission to domestic prices intensified in the post-COVID period; pass-through roughly tripled.
  3. Policy and institutional variables play a nuanced role—some factors (e.g., fossil fuel subsidies) mechanically dampen pass-through but carry fiscal and distributional trade-offs; many governance and policy indicators show no direct statistical association once shocks and pre-pandemic conditions are accounted for.
- Policy-relevant implications:
  - Supply-driven inflation episodes are mediated by pre-existing inflation credibility and exposure to energy price swings; anchoring expectations and shock-absorption capacity matter.
  - Fossil fuel subsidies can reduce pass-through but impose fiscal costs and distributional considerations that must be weighed.
  - Interpreting policy coefficients in reduced-form regressions requires caution because of policy endogeneity; observed positive associations between tightening interventions and inflation may reflect responsive policy actions.
- Suggested directions for future research:
  - Examine nonlinearities and the interplay between fiscal and monetary policy during large global shocks to better identify causal channels and policy trade-offs.

*IMF WORKING PAPER — One Global Shock, Many Inflation Paths: Explaining Post-COVID Inflation Divergence*

### References

### References and Annex I: Robustness Checks for Cross-Sectional Regressions

### Key referenced literature (selection from the source)
- Adler, Gustavo; Kyun Suk Chang; Rui Mano; Yuting Shao. 2025. “Foreign Exchange Intervention: A Data Set of Official Data Estimates.” Journal of Money, Credit and Banking 57 (5): 1241-1273.
- Amaglobeli, David; Mengfei Gu; Emine Hanedar; Gee Hee Hong; Celine Thevenot. 2023. “Policy Responses to High Energy and Food Prices.” IMF Working Paper 23/74.
- Andersen, Robert. 2008. Modern Methods for Robust Regression.
- Blanchard, Olivier and Ben S. Bernanke. 2023. NBER Working Paper No. 31417; 2024. NBER Working Paper No. 32532.
- Ha, Jongrim; Ayhan M. Kose; Franziska Ohnsorge. 2023. “One-Stop Source: A Global Database of Inflation.” Journal of International Money and Finance 137: 102896.
- Additional IMF, BIS, PIIE, ECB, World Bank, Brookings, and other working papers and policy contributions cited (full list in source).

### Annex I — Notes on methodology used across robustness tables
- Dependent variable in these regressions: the post-COVID inflation (except Table A_I_9 where dependent variable is 2015Q1-2019Q4 inflation).
- Estimation methods reported:
  - Robust regression estimator that downweighs observations with larger absolute residuals using iterative weighted least squares (Andersen, 2008).
  - Quantile regression for the median (for Tables A_I_5 to A_I_8 and A_I_9).
- Significance notation used throughout: *** p<0.01, ** p<0.05, * p<0.1.
- Data sources cited for regressions: World Bank’s global inflation dataset (Ha and others, 2023), IMF fiscal rules database, AREAER, WEO, WDI, V-DEM, IMF iMapp, Adler and others (2024), Romelli (2023), and authors’ calculations.

### Robustness results — consistent and notable coefficient estimates (exact values as reported)
- Pre-COVID inflation (frequently the most robust predictor of post-COVID inflation):
  - Examples of reported coefficients: 0.665*** (std. err. (0.054)), 0.702*** (0.056), 0.885*** (0.052), 0.839*** (0.060), 0.827*** (0.060), 0.826*** (0.059), 0.831*** (0.057), 0.867*** (0.066).
  - Other reported estimates in alternate specifications/quantile regressions: 0.909*** (0.059), 0.917*** (0.059), 0.945*** (0.059), 0.872*** (0.063), 0.946*** (0.073).
  - In Table A_I_9 (2010Q1-2014Q4 inflation as control predicting 2015Q1-2019Q4 inflation): 0.435** (0.198), 0.414** (0.199), 0.361* (0.201), 0.814*** (0.238), 0.821*** (0.245), 0.824*** (0.246), 0.828*** (0.246), 0.997** (0.415).
- Post-COVID energy inflation:
  - 0.259*** (0.033) in one specification.
  - 0.153*** (0.033) in another reported column.
  - In quantile regressions: 0.387*** (0.052), 0.420*** (0.057).
- Post-COVID REER growth:
  - 0.421*** (0.087) and 0.290*** (0.088) in different columns.
  - Quantile estimates: 0.446*** (0.094) and 0.322** (0.140).
- Exchange rate peg dummy:
  - -1.156** (0.495) in one robust regression column.
  - -1.392* (0.807) in another.
  - Quantile: -1.286** (0.569) and -2.281** (0.901) in some specifications.
- Fiscal/institutional policy variables with notable coefficients:
  - Debt rule dummy: 1.220** (0.607) and 1.712** (0.723) in reported columns.
  - Expenditure rule dummy: 0.705 (0.476) and 1.166* (0.643) in different specs.
  - Revenue rule dummy: -0.268 (0.636) and -2.249** (1.002) in different specs.
  - Balanced budget rule dummy: 0.707 (0.742) and 1.072 (0.927).
- Central bank independence and political/institutional indices (selected columns):
  - Post-COVID CB independence index: 5.018*** (1.375) in one column and 3.837* (1.926) in another.
  - Post-COVID electoral regime index: 1.160*** (0.412) in one column and 0.833 (0.596) in another.
  - Post-COVID freedom of expression index: 1.601** (0.765) in one column and 2.737 (3.133) in another.
  - Post-COVID range of consultation index: -0.033 (0.174) in one column and -1.207** (0.532) in another.
  - In quantile regressions, Post-COVID CB independence index reported as 4.561* (2.473) and 4.724* (2.720).
- Policy response / instrument change coefficients (selected):
  - Change in central bank policy rate: 0.558*** (0.133) in one robust regression; 0.182 (0.109) in another column; quantile regression examples: 0.398*** (0.139) and 0.471** (0.220).
  - Change in macroprudential instruments: 1.105*** (0.415) and 0.795 (0.566) in some robust-regression columns; quantile example: 1.135* (0.672) and 0.735 (1.139).
  - Change in Debt/GDP ratio: -0.044*** (0.015) and -0.092*** (0.031) in robust-regression columns; quantile examples: -0.034 (0.023) and -0.066 (0.062).
  - Change in average REER appreciation: 0.296*** (0.058) and 0.094 (0.080) in robust-regression columns; quantile: 0.344*** (0.069) and -0.066 (0.157).
  - Post-COVID FX interventions in % of GDP: 1.195* (0.601) in one column and 0.315 (0.908) in another; quantile examples: 1.192 (0.755) and -0.206 (1.807).
  - Change in fossil fuel subsidies in % of GDP: -0.280 (0.185) and -0.084 (0.326) in reported columns; quantile examples: -0.237 (0.301) and 0.137 (0.658).
- Constants, sample sizes, and R-squared examples (exactly as reported):
  - Example constant values: 3.300*** (0.268), 3.194*** (0.291), 2.943*** (0.413), 2.678*** (0.417), 3.179*** (0.315), 2.518*** (0.712), 2.156*** (0.575), 1.014 (1.202).
  - Observations reported across tables: 129, 129, 84, 90, 90, 90, 90, 56; other reported observation counts: 103, 126, 120, 120, 126, 129, 120, 127, 87; and many others across specifications (e.g., 84, 110, 126, 119, 77).
  - R-squared examples: 0.546, 0.567, 0.786, 0.696, 0.688, 0.693, 0.711, 0.811; other R-squared examples: 0.787, 0.523, 0.573, 0.697, 0.564, 0.603, 0.522, 0.483, 0.782.
- Quantile-regression median specifications generally reinforce the dominant role of pre-COVID inflation and highlight similar signs for REER growth and energy inflation, while some institutional/policy coefficients show sensitivity across specifications.

### Interpretation points directly supported by the reported numbers
- Pre-COVID inflation is the most consistently positive and statistically significant predictor of post-COVID inflation across robust and quantile specifications (many coefficients in the 0.6–0.95 range and frequently significant at p<0.01).
- Post-COVID energy inflation and post-COVID REER growth register positive and often statistically significant coefficients in multiple specifications.
- Exchange rate peg shows a negative and in some specs statistically significant association with post-COVID inflation.
- Some fiscal rule dummies (debt rule, expenditure rule, revenue rule) and institutional indices (CB independence, electoral regime, freedom of expression, range of consultation) show large and sometimes significant coefficients, but their significance and sign vary across specifications, indicating sensitivity to model choice and sample.
- Policy-instrument changes (central bank policy rate changes, macroprudential changes, changes in Debt/GDP ratio, REER appreciation) appear in multiple specifications with notable coefficients, highlighting the empirical relevance of both monetary and macroprudential/fiscal dimensions in cross-country post-COVID inflation differences.

*Source: References list and Annex I (robustness tables) of the IMF working paper “One Global Shock, Many Inflation Paths: Explaining Post-COVID Inflation Divergence.”*

### Annex II: Pass-through to core inflation

### Annex II: Pass-through to core inflation

### Energy price pass-through (Figure A_II_1)
- The figure shows the cumulative impact of a 1 percent change of energy prices on core CPI inflation.
- Presentation details preserved from the source:
  - The line denotes point estimates.
  - The shaded area denotes the 95 percent confidence interval based on robust standard errors.
  - Filled circles indicate significance.

### Context and placement
- Figure caption: "Energy price pass-through to core inflation: Total sample"
- Note: The figure is part of the IMF Working Paper "One Global Shock, Many Inflation Paths: Explaining Post-COVID Inflation Divergence".

---

### Annex III: Variables and their sources

### Table A_III_1 — List of IT and non-IT countries
- Source: IMF AREAER, 2023 vintage, which is based on the de jure classification of monetary policy regimes.
- Sampling notes from the source:
  - From the initial sample of 40 IT countries in the AREAER we have excluded Finland and Spain given that they have adopted Euro since then.
  - In addition, we have excluded Ghana, Guatemala, Moldova, Turkiye and Uganda, since they have experienced very high annual inflation rates (more than 10 percent) in the pre-COVID sample which casts doubt on their de facto IT implementation.
  - Ukraine is removed from the IT sample given the exceptional wartime circumstances and Euro Area is removed from the sample since it is a special arrangement consisting of a group of countries making it incompatible with the rest of the sample.
- IT countries (numbered as in the source):
  1. Albania
  2. Armenia (Strong)
  3. Australia
  4. Brazil
  5. Canada (Strong)
  6. Chile
  7. Colombia
  8. Czechia (Strong)
  9. DominicanRepublic
  10. Georgia (Strong)
  11. Hungary
  12. Iceland (Strong)
  13. India
  14. Indonesia (Strong)
  15. Israel (Strong)
  16. Japan (Strong)
  17. Kazakhstan
  18. Korea (Strong)
  19. Mexico (Strong)
  20. NewZealand (Strong)
  21. Norway
  22. Paraguay (Strong)
  23. Peru (Strong)
  24. Philippines
  25. Poland
  26. Romania (Strong)
  27. Russia (Strong)
  28. Serbia (Strong)
  29. SouthAfrica
  30. Sweden
  31. Thailand
  32. UnitedKingdom
  33. Uruguay (Strong)
  34. Tunisia
  35. UnitedArabEmirates
  36. UnitedStates
  37. Vietnam
- Non-IT countries (numbered as in the source):
  1. Algeria
  2. Azerbaijan
  3. Bahamas
  4. Bahrain
  5. Belarus
  6. BosniaandHerzegovina
  7. Botswana
  8. BruneiDarussalam
  9. Bulgaria
  10. CapeVerde
  11. China
  12. CostaRica
  13. Denmark
  14. Ecuador
  15. ElSalvador
  16. Fiji
  17. HongKongSAR
  18. Jamaica
  19. Jordan
  20. Kuwait
  21. Macedonia
  22. Malaysia
  23. Mauritius
  24. Mongolia
  25. Montenegro
  26. Morocco
  27. Oman
  28. SaintKittsAndNevis
  29. SaudiArabia
  30. Singapore
  31. Switzerland
  32. Tonga
  33. TrinidadandTobago

### Table A_III_2 — Variable definitions and sources
- Variables (as labeled in the source) with definitions and sources preserved exactly:

- infl
  - Definition: Growth in CPI index (headline or core), y/y, percent (log diff)
  - Sources: Ha and others (2023), World Bank

- infl_e
  - Definition: Growth in energy price index, y/y, percent (log diff)
  - Sources: Ha and others (2023), World Bank

- Real GDP growth
  - Definition: Growth in real GDP, y/y, percent (log diff)
  - Sources: IMF WEO

- FX reserves/GDP ratio
  - Definition: Ratio of foreign exchange reserves over GDP (percent)
  - Sources: World Bank

- Trade openness
  - Definition: Sum of trade export and import over GDP (percent)
  - Sources: World Bank

- Life expectancy
  - Definition: Life expectancy at birth (years)
  - Sources: UNDP

- Share of >65 population
  - Definition: Share of >65 population in total population (percent)
  - Sources: World Bank

- REER growth
  - Definition: Growth in REER index, y/y, percent (log diff)
  - Sources: IMF WEO

- Inflation volatility
  - Definition: Standard deviation of CPI inflation (percent)
  - Sources: Ha and others (2023), Worldbank

- Macroprudential instruments
  - Definition: Summary of changes in 17 macroprudential indices
  - Sources: IMF IMaPP dataset

- FX interventions in % of GDP
  - Definition: Purchase of sale of FX (percent of GDP)
  - Sources: Adler and others (2025)

- Debt/GDP ratio
  - Definition: Public debt as a share of GDP
  - Sources: IMF WEO

- Fossil fuel subsidies in % of GDP
  - Definition: Total explicit fossil fuel subsidies as a share of GDP
  - Sources: IMF, Amaglobeli and others (2023)

- Rule of law index
  - Definition: Index measuring the perceptions of the rule of law
  - Sources: World Bank's Worldwide Governance Indicators (WGI)

- CB independence index
  - Definition: Central Bank Independence score (de jure)
  - Sources: Romelli (2022, 2024)

- IT dummy
  - Definition: Dummy: =1 for IT countries
  - Sources: IMF AREAER

- AE dummy
  - Definition: Dummy: =1 for advanced economies
  - Sources: IMF WEO

- Exchange rate peg dummy
  - Definition: Dummy: =1 for countries pegging their exchange rate
  - Sources: Adler and others (2025)

- Expenditure rule dummy
  - Definition: Dummy: =1 for countries with expenditure rule
  - Sources: IMF fiscal rules dataset

- Revenue rule dummy
  - Definition: Dummy: =1 for countries with revenue rule
  - Sources: IMF fiscal rules dataset

- Balanced budget rule dummy
  - Definition: Dummy: =1 for countries with balanced budget rule
  - Sources: IMF fiscal rules dataset

- Debt rule dummy
  - Definition: Dummy: =1 for countries with debt rule
  - Sources: IMF fiscal rules dataset

- Central bank policy rate
  - Definition: Central bank policy rate
  - Sources: LSEG

- Trade union density rate
  - Definition: The share of workers who are union members
  - Sources: OECD

- Political stability index
  - Definition: Index measuring the perceptions of political stability
  - Sources: World Bank's Worldwide Governance Indicators (WGI)

- Liberal democracy index
  - Definition: Extent of liberal democracy
  - Sources: V-DEM

- Freedom of expression index
  - Definition: Extent to which governments respect press and media freedom
  - Sources: V-DEM

- Clean election index
  - Definition: Extent to which election violence, government intimidation, fraud, large irregularities, and vote buying are absent
  - Sources: V-DEM

- Legislative constraints index
  - Definition: Measure of how effectively legislatures and oversight agencies investigate and constrain government actions
  - Sources: V-DEM

- Range of consultation index
  - Definition: Extent of consultations across political elites
  - Sources: V-DEM

- Engaged society index
  - Definition: Extent to which ordinary people independently discuss important policy changes among themselves
  - Sources: V-DEM

- Public goods index
  - Definition: Extent to which public goods are distributed equally across a society
  - Sources: V-DEM

- Public administration index
  - Definition: Extent to which public officials follow the law and show no favoritism or discrimination
  - Sources: V-DEM

- Public corruption index
  - Definition: Measure of corruption
  - Sources: V-DEM

- Electoral regime index
  - Definition: Extent to which political systems adhere to democratic principles
  - Sources: V-DEM

- Fossil fuel subsidy and fiscal variables are listed with their sources as above.

*One Global Shock, Many Inflation Paths: Explaining Post-COVID Inflation Divergence — Working Paper No WP/2026/103*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026103-source-pdf.pdf_