## Global Commodity Inflation Pass-through: Vulnerability of Small Island Developing States (wpiea2025138-print-pdf) — Section V content

## Source details

**Canonical URL:** [Global Commodity Inflation Pass-through: Vulnerability of Small Island Developing States (wpiea2025138-print-pdf) — Section V content](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025138-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025138-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025138-print-pdf.pdf.json)

---

### Key findings
- Sample and method:
  - Dataset: 168 countries (including 31 SIDS).
  - Main methodological approach: local projections (LP) method of Jordà (2005).
- Pass-through magnitudes and asymmetry:
  - Pass-through from global food inflation shocks to domestic inflation in SIDS is considerably larger than for other countries.
  - Difference in pass-through from oil inflation shocks between SIDS and non-SIDS is less apparent.
  - Asymmetry in pass-through from positive vs. negative global food price shocks is evident for both country groups and is slightly larger for SIDS.
- Historical inflation context for SIDS:
  - Average inflation over the last two decades: 3.1 percent.
  - This is 0.9 percentage points above the average inflation experienced in AEs and similar to the average of EMDEs.
  - During the 2000-10 period, the gap between SIDS and AE average inflation was 1.6 percentage points.
  - The headline inflation gap of SIDS has closed in the last decade to about 0.5 percentage points.

### Data and sample
- Main dependent variable:
  - Point-to-point domestic inflation = year-on-year change of the quarterly headline consumer price index (Ha, Kose, and Ohnsorge, 2023).
- Sample construction:
  - Data beginning 1992Q1; initial sample of 202 countries.
  - Excluded 34 countries with hyperinflation (point-to-point inflation above 50 percent during 1992Q1-2022Q4).
  - Final dataset: 168 countries, of which 31 SIDS and 137 non-SIDS.
  - Hyperinflation country list (34 excluded): Albania, Angola, Armenia, Belarus, Brazil, Bulgaria, Croatia, Ecuador, Ghana, Guinea-Bissau, Indonesia, Kazakhstan, Lao PDR, Latvia, Liechtenstein, Lithuania, Malawi, Mongolia, Nigeria, North Macedonia, Romania, Russia, Serbia, South Sudan, Sri Lanka, Sudan, Suriname, Sao Tome and Principe, Turkey, Uganda, Ukraine, Venezuela, Zambia, Zimbabwe.
- International commodity price measures:
  - International point-to-point inflation rates for food and crude oil prices from IMF’s Primary Commodity Price Database; quarterly indices via three-month simple average.
- Controls and additional series:
  - G7 core inflation (OECD COICOP 1999 dataset), bilateral exchange rate (period average, national currency units per US dollar), Global Supply Chain Pressure Index (GSCPI) three-month simple average.
  - CPI basket weights for food & beverage and transport components from IMF’s CPI database (COICOP) used in pre-weighting exercises.
- Quarterly domestic activity proxy limitations:
  - Majority of SIDS lack quarterly GDP; baseline quarterly regressions omit direct domestic activity control.
  - Robustness check: annual data with HP-filtered output gap supports baseline quarterly findings.

### Empirical methodology
- Objective:
  - Examine persistence and differences between global food- and oil-inflation shocks on domestic headline inflation in SIDS vs. non-SIDS and test asymmetries for positive vs negative shocks.
- Baseline local-projection specification (quarterly):
  - 휋_{t+h,i} − 휋_{t−1,i} = 훼_i + 훾_{i,t} 휋_{t−1,i} + β^h_{SIDS,oil} D_i^{SIDS} X_t + β^h_{Non−SIDS,oil} D_i^{Non−SIDS} X_t + β^h_{SIDS,food} D_i^{SIDS} F_t + β^h_{Non−SIDS,food} D_i^{Non−SIDS} F_t + 휃 Y_{i,t} + u_{i,t}
  - Definitions:
    - 휋_{t+h} = year-on-year headline CPI inflation in country i h-quarters ahead.
    - X_t = year-on-year change in global oil prices.
    - F_t = year-on-year change in global food prices.
    - Dummies D_i^{SIDS}, D_i^{Non−SIDS} indicate group membership.
    - Y_{i,t} controls: bilateral nominal exchange rate, GSCPI, global core inflation.
    - 훾_{i,t} captures lagged inflation persistence; 훼_i country fixed effects; u_{i,t} iid error.
  - Baseline lags: one (robust to two lags).
  - Horizons: h = {0,...,8} quarters estimated by OLS.
  - Confidence bands: Driscoll and Kraay (1998) standard errors.
- Asymmetry specification:
  - Positive/negative shock dummies: D_{food+} = 1 if 휋_{t}^{food} > 0, D_{food−} = 1 if 휋_{t}^{food} < 0 (analogous for oil).
- CPI pre-weighting:
  - Pre-weight global shocks by country food & non-alcoholic beverage and transport CPI weights (sample reduced to 2000Q1-2022Q4; CPI-weighted sample: 135 countries, 23 SIDS).

### Empirical results — International food inflation shocks
- Baseline quarterly IRF (Figure 6, 68 percent confidence bands):
  - A 10-percentage point shock to international food inflation:
    - SIDS: domestic inflation increases by 0.9 percentage point at its peak after three quarters; dissipates after about two years.
    - Non-SIDS: domestic inflation reaches just below 0.5 percentage point a full year after the shock; effectively disappears after two years.
  - SIDS show larger and more persistent impact relative to non-SIDS for at least the first year.
- Statistical significance:
  - F-tests for parameter equivalence (Table 1) show the estimated pass-through difference is statistically significant at the 5 percent (or greater) level in each of the first 4 quarters following the global food-inflation shock; significance dissipates thereafter.
  - Table 1 p-values for International Food Inflation Shock (t=1,2,3,4,8): 0.0062***, 0***, 0***, 0.0392**, 0.0856*.

### Empirical results — International crude oil inflation shocks
- Baseline quarterly IRF (Figure 7):
  - A 10-percentage point shock in international crude-oil inflation:
    - SIDS: domestic headline inflation increases by 0.1 percentage point at peak after two quarters; impact returns to zero after about a year.
    - Impacts for both groups are notably smaller than for food shocks.
  - Disparity between SIDS and non-SIDS less apparent for oil shocks.
- Statistical significance:
  - F-tests indicate no statistically significant difference between SIDS and non-SIDS at most horizons (Table 1 p-values for oil shock t=1..8: 0.4283, 0.1988, 0.2127, 0.0545*, 0.4224). Only t=4 shows significance at 10 percent.
- Interpretation:
  - Smaller oil-pass-through may reflect domestic oil policies (subsidies, price ceilings, interventions) that affect pass-through; longer-horizon impacts less conclusive.

### Asymmetric pass-through: positive vs negative shocks
- Evidence:
  - Asymmetric pass-through for food exists for both SIDS and non-SIDS (Figure 8).
  - Positive food shocks have larger domestic inflation impacts than negative shocks (negative-shock IRFs inverted for comparability).
  - SIDS:
    - Positive shock cumulative impact peaks around 0.15 percentage points after 4–5 quarters.
    - Asymmetries more strongly statistically significant for SIDS and diminish beyond two years.
  - Non-SIDS:
    - Positive shock peak below 0.1 percentage point.
- Implication:
  - Even with zero net incidence of positive and negative shocks over time, asymmetry implies cumulative domestic inflation impact is likely positive for SIDS.
- Statistical evidence (Table 1 asymmetry p-values):
  - International Food Inflation Shock (SIDS) t=1..4: 0***, 0***, 0***, 0***; t=8: 0.0965*.
  - International Food Inflation Shock (Non-SIDS) t=1..4: 0.0011***, 0.0001***, 0.0004***, 0.0011***; t=8: 0.2351.

### Robustness checks
- Annual specification with output gap:
  - HP-filtered real GDP output-gap proxy with 휆 = 6.25 (annual frequency).
  - Annual local projections including output gap preserve key baseline findings:
    - Food-shock differences between SIDS and non-SIDS remain marked and large.
    - Test for parameter equivalence for food shocks rejected at 10 percent level 1 year after the shock.
    - Oil shock responses remain muted.
- Additional robustness:
  - Quarterly specification including a COVID dummy (value =1 for 2020Q2 through 2022Q4) and two lags yields baseline results robust.
- Endogeneity considerations:
  - Global commodity-price shocks plausibly supply-driven; endogeneity concerns limited, especially for SIDS (too small to drive global demand).
  - Prior work suggests pass-through estimates negligibly impacted by endogeneity controls; if non-SIDS estimates were upward biased, true SIDS vs non-SIDS gap would be larger.

### Impact of consumer basket (CPI) weights
- Food CPI weight pre-weighting (sample 2000Q1-2022Q4; 135 countries, 23 SIDS):
  - Accounting for food & non-alcoholic beverage CPI weights does not materially alter baseline results.
  - A 10-percentage point shock in international food inflation (CPI-weighted):
    - SIDS: domestic inflation increases by about 1 percentage point at peak after three quarters; returns to zero after two years.
  - F-tests (Table 2) show disparity concentrated within 1 year of shock; significance dissipates after 3 quarters.
  - Table 2 p-values for CPI-weighted food shock t=1..8: 0.0139**, 0.0031***, 0.0069***, 0.0599*, 0.0797*.
  - Conclusion: Larger food-share in SIDS CPI does not meaningfully explain higher susceptibility; other factors (supply-chain connectivity, domestic market structures) account for significant unexplained vulnerability.
- CPI-weighted asymmetry:
  - Asymmetric pass-through persists after weighting; SIDS more responsive to positive food-price shocks than non-SIDS.
  - Food CPI weights do not explain asymmetry.
- Transport CPI weight results for oil shocks:
  - Accounting for transport weights has limited impact.
  - With transport weights, a 10-percentage point shock in international crude-oil inflation:
    - SIDS: domestic inflation increases by 0.04 percentage points at peak after two quarters; impact returns to zero within first year.
  - Conclusion: transport weights do not alter main conclusion that oil pass-through is modest and similar across groups.

### Conclusion — summary of main findings
- SIDS inflation is approximately twice as vulnerable to international food inflation shocks as in non-SIDS (local-projection estimates for 168 countries).
- Vulnerability and disparity for international crude oil inflation shocks is less apparent.
- Results robust across specifications and checks.
- Global-supply-chain connectivity and domestic market structures are potential major contributing factors to SIDS inflation vulnerabilities.
- Relative lack of pricing power of firms in SIDS may increase susceptibility to global food inflation shocks.

### Policy implications and options
- No easy or obvious policy solution; suggested directions include:
  - Improve efficiency and orientation of domestic production towards diversification and food self-sufficiency.
  - Regional and international collaboration to mitigate geographic constraints.
  - Pooling orders regionally (regional food hub) to exercise greater market power in purchases.
- Recognize SIDS’ limited short- to medium-term policy options due to import dependence and use of exchange-rate pegs.

### Research gaps and further work
- Priorities for further research:
  - Investigate second-round effects of global commodity-price shocks to assess medium-term persistence of inflationary effects.
  - Examine how inflation for different goods within the CPI basket is affected by global price fluctuations.
  - Identify drivers of pass-through and evaluate policy reform options to mitigate SIDS vulnerabilities.

*Source: IMF Working Paper (wpiea2025138-print-pdf), Section V content as supplied.*

### Introduction ...........................................................................................................

### Introduction

### Key findings
- Dataset: 168 countries (including 31 SIDS).
- Main methodological approach: local projections (LP) method of Jordà (2005).
- Pass-through results:
  - Pass-through from global food inflation shocks to domestic inflation in SIDS is considerably larger than for other countries.
  - Difference in pass-through from oil inflation shocks between SIDS and non-SIDS is less apparent.
  - Asymmetry in pass-through from positive vs. negative global food price shocks is evident for both country groups, and is slightly larger for SIDS.
- Historical inflation statistics for SIDS:
  - Average inflation over the last two decades: 3.1 percent.
  - This is 0.9 percentage points above the average inflation experienced in AEs and similar to the average of EMDEs.
  - During the 2000-10 period, the gap between SIDS and AE average inflation was 1.6 percentage points (Figure 3).
  - The headline inflation gap of SIDS has closed in the last decade to about 0.5 percentage points.

### Descriptive context and documented dynamics
- SIDS face extreme vulnerability to external economic and inflation shocks due to small size and isolated geographic location; vulnerabilities highlighted during the 2008-2009 global financial crisis and the COVID-19 pandemic.
- Inflation among SIDS is higher than in AEs and exhibits greater variation over time; variability corresponds partly to variability in global commodity prices, notably during:
  - post-GFC commodity-price declines in 2009, and
  - post-COVID commodity-price increases in 2021-22.
- Regional heterogeneity:
  - Pacific Island and Caribbean SIDS were disproportionately impacted by the 2008 global food crisis.
  - Geographic and supply-chain connectivity differences noted: Pacific and West African SIDS have more limited integration within global supply chains; Caribbean SIDS benefit from relative geographic advantage with respect to East-West container shipping routes (United Nations Conference on Trade and Development, 2014).

### Data references and visual documentation
- Figures referenced in the text:
  - Figures (1 & 2): evolution of global food and oil inflation alongside headline inflation in SIDS and other groupings (AE, EMDEs).
  - Figure 3: gap dynamics between SIDS and AE average inflation over time.
- Source explicitly mentioned for a supply-chain measure: Federal Reserve Bank of New York, Global Supply Chain Pressure Index, https://www.newyorkfed.org/research/policy/gscpi and authors' calculations.

### Contribution to existing literature
- Directly related studies and comparative findings:
  - Gelos and Ustyugova (2017): commodity-price shock pass-through varies with food share in CPI, fuel intensity, pre-existing high inflation, governance, and central bank independence.
  - Furceri and others (2016): global food price shocks impacts similar between AEs and EMs using annual data (2000-2013).
  - Choi and others (2018): oil-inflation pass-through similar for AEs and EMs, with some asymmetric pass-through (higher for positive shocks); drivers include transportation share in CPI and energy-price subsidies.
  - Tillmann (2024): asymmetric pass-through across supply-chain conditions for European countries.
  - The present study documents higher inflation pass-through for SIDS relative to non-SIDS, primarily for food, and shows pass-through asymmetry for positive vs. negative shocks.
- Broader “global inflation” literature:
  - Papers suggesting increasing global co-movement of inflation: Mumtaz and Surico (2008), Ciccarelli and Mojon (2010), Neely and Rapach (2011).
  - Countervailing evidence on domestic drivers: Parker (2018) (limits of global inflation to middle and low-income countries), Bems and others (2022) (importance of domestic factors in 19 EMs).
  - Mechanisms highlighted in other work: input-output linkages (Auer, Levchenko and Saure, 2019) and shipping costs (Carriere-Swallow and others, 2023).
- Prior SIDS-focused work:
  - Downes and others (2020): dominance of import-price and supply-side factors in SIDS inflation, with a role for monetary policy stance.
  - Downes, Holder and Leon (1987); Cumberbatch (1995): significance of external prices and domestic influences (consumer credit, wages, labour productivity, unemployment).
  - Greenidge and DaCosta (2009): evidence of cost-push and demand-pull inflation in Barbados, Jamaica, Guyana, Trinidad and Tobago (1970–2006).
  - Bai, Tumbarello, and Wu (2016): in twelve Pacific Island countries, international oil price shocks had smaller but more persistent effects than food price shocks, driven mainly from the demand-side.
  - Impavido (2018): Barbados’ inflation tends to converge to that of the United States, with import prices such as oil influencing short-term dynamics.

### Paper structure (as presented)
- Section II: Data used for analysis.
- Section III: Empirical methodology.
- Section IV: Baseline results on commodity-inflation spillovers to SIDS vs. non-SIDS, tests for asymmetric effects across positive/negative shocks, and robustness tests supporting baseline conclusions.

*IMF Working Papers — Introduction (excerpt).*

### Section V concludes with some discussion of the policy implications of our findings.

### Global Commodity Inflation Pass-through: Vulnerability of Small Island Developing States (wpiea2025138-print-pdf) — Section V content

### Data
- Main dependent variable: point-to-point domestic inflation, measured as the year-on-year change of the quarterly headline consumer price index, sourced from the global inflation database of Ha, Kose, and Ohnsorge, (2023).
- Sample:
  - Data beginning from 1992Q1; initial sample of 202 countries.
  - Excluded 34 countries with episodes of hyperinflation (defined as point-to-point inflation above 50 percent) observed during 1992Q1-2022Q4.
  - Final dataset: 168 countries, of which 31 SIDS and 137 non-SIDS.
- International commodity price measures:
  - International point-to-point inflation rates for food and crude oil prices calculated from year-on-year change of food and crude oil price indices in IMF’s Primary Commodity Price Database; quarterly indices obtained by three-month simple average for each quarter.
- Controls and additional series:
  - Proxy of global non-commodity inflationary pressure: contemporaneous changes in point-to-point core inflation rates of G7 countries (G7 core price index from COICOP 1999 dataset via OECD Data Explorer).
  - Bilateral exchange rate (period average, national currency units per US dollar) from IMF’s International Financial Statistics Database.
  - Global Supply Chain Pressure Index (GSCPI) three-month simple average provided by Benigno et al. (2022).
- CPI basket weights:
  - Consumer basket weights for food & beverage and transport components from IMF’s CPI database (COICOP classification) used to explore influence of relative CPI basket weights.
- Quarterly domestic activity proxy limitations:
  - Majority of SIDS lack quarterly GDP or reliable proxies (e.g., electricity consumption); thus baseline quarterly regressions omit a direct domestic activity control.
  - Robustness check using annual data constructs an output-gap series; results support baseline quarterly findings.
- Hyperinflation country list (34 countries excluded): Albania, Angola, Armenia, Belarus, Brazil, Bulgaria, Croatia, Ecuador, Ghana, Guinea-Bissau, Indonesia, Kazakhstan, Lao PDR, Latvia, Liechtenstein, Lithuania, Malawi, Mongolia, Nigeria, North Macedonia, Romania, Russia, Serbia, South Sudan, Sri Lanka, Sudan, Suriname, Sao Tome and Principe, Turkey, Uganda, Ukraine, Venezuela, Zambia, Zimbabwe.
- Note on alternative proxy: Tiedemann and others (2024) conclude cyclical unemployment is a poor measure of labor market slack, especially in developing economies.

### Empirical Methodology
- Objectives:
  - Examine persistence and differences between global food- and oil-inflation shocks on domestic headline inflation in SIDS vs. non-SIDS.
  - Investigate asymmetries in commodity-price pass-through between positive and negative global commodity-price shock episodes.
- Method:
  - Local projections method (Jordà, 2005) applied to an adapted Phillips curve; impulse response functions (IRFs) presented.
  - Panel: 168 countries (31 SIDS) for 1992Q1 to 2022Q4.
- Baseline local-projection specification (quarterly):
  - Equation (1) (as reported): 
    - 휋_{t+h,i} − 휋_{t−1,i} = 훼_i + 훾_{i,t} 휋_{t−1,i} + β^h_{SIDS,oil} D_i^{SIDS} X_t + β^h_{Non−SIDS,oil} D_i^{Non−SIDS} X_t + β^h_{SIDS,food} D_i^{SIDS} F_t + β^h_{Non−SIDS,food} D_i^{Non−SIDS} F_t + 휃 Y_{i,t} + u_{i,t}
    - 휋_{t+h} = year-on-year headline CPI inflation in country i h-quarters ahead.
    - X_t = year-on-year change in global oil prices.
    - F_t = year-on-year change in global food prices.
    - Dummies D_i^{SIDS} and D_i^{Non−SIDS} indicate country grouping.
    - Y_{i,t} controls: bilateral nominal exchange rate, GSCPI, global core inflation.
    - 훾_{i,t} captures lagged inflation persistence; 훼_i country fixed effects; u_{i,t} iid error.
  - Baseline number of lags: one (results robust to two lags).
  - Horizons estimated: h = {0,...,8} quarters using OLS.
  - Confidence bands: Driscoll and Kraay (1998) standard errors (robust to spatial and temporal dependence as T becomes large).
- Asymmetry specification:
  - Differentiate positive vs negative commodity shocks by introducing dummies:
    - For food: D_{food+} = 1 if 휋_{t}^{food} > 0, else 0; D_{food−} = 1 if 휋_{t}^{food} < 0, else 0.
    - For oil: D_{oil+} and D_{oil−} defined analogously.
  - Compare impact of positive vs negative shocks on domestic inflation for SIDS and non-SIDS.
- CPI basket pre-weighting:
  - Following Choi and others (2018), pre-weight global commodity inflation shocks by each country’s food & non-alcoholic beverage and transport CPI weights.
  - Due to limited basket-weight data, sample reduced to 2000Q1-2022Q4 for this exercise.
  - Pre-weighted baseline specification given as Equation (4) (reporting same structure with δ_{i,t} representing CPI weights).
  - CPI weights imputation: nearest observation or interpolation with simple average when gaps exist.
  - Pre-weighted analysis sample: 135 countries, of which 23 SIDS.

### Empirical Results — Impact of International Food Inflation Shocks
- Baseline quarterly IRF findings (Figure 6, 68 percent confidence bands, country-clustered standard errors):
  - SIDS show larger and more persistent impact relative to non-SIDS for at least the first year after a shock.
  - A 10-percentage point shock to international food inflation:
    - Domestic inflation in SIDS increases by 0.9 percentage point at its peak after three quarters, before dissipating after about two years.
    - Domestic inflation in non-SIDS reaches just below 0.5 percentage point a full year after the shock, before effectively disappearing after two years.
  - Conclusion: SIDS are more vulnerable to international food price shocks, especially short-term, with longer persistence.
- Parameter equivalence tests (F-tests):
  - Estimated pass-through difference between SIDS and non-SIDS is statistically significant at the 5 percent (or greater) level in each of the first 4 quarters following the global food-inflation shock, dissipating thereafter (Table 1).

### Empirical Results — Impact of International Crude Oil Inflation Shocks
- Baseline quarterly IRF findings (Figure 7):
  - Disparity between SIDS and non-SIDS is less apparent for oil shocks.
  - F-tests indicate no statistically significant difference between SIDS and non-SIDS at any period following the shock (Table 1 shows p-values).
  - Magnitude:
    - A 10-percentage point shock in international crude-oil inflation increases domestic headline inflation in SIDS by 0.1 percentage point at peak after two quarters; impact returns to zero after about a year.
    - Impacts for both groups are notably smaller than for food shocks.
  - Interpretation: Smaller influence of global oil-price shocks may reflect domestic oil policies (subsidies, price ceilings, interventions) that affect pass-through; impacts at longer horizons less conclusive.

### Table 1 — F-test p-values for parameter equivalence (Baseline SIDS vs. Non-SIDS)
- International Food Inflation Shock (p-values shown for t=1,2,3,4,8):
  - t=1: 0.0062***
  - t=2: 0***
  - t=3: 0***
  - t=4: 0.0392**
  - t=8: 0.0856*
- International Oil Inflation Shock:
  - t=1: 0.4283
  - t=2: 0.1988
  - t=3: 0.2127
  - t=4: 0.0545*
  - t=8: 0.4224
- Food Commodity Price Shock Asymmetry (p-values):
  - International Food Inflation Shock (SIDS): t=1: 0***, t=2: 0***, t=3: 0***, t=4: 0***, t=8: 0.0965*
  - International Food Inflation Shock (Non-SIDS): t=1: 0.0011***, t=2: 0.0001***, t=3: 0.0004***, t=4: 0.0011***, t=8: 0.2351
- Notes: p-values; bolded and starred values denote rejection of parameter equivalence across the two regimes at 10 (*), 5 (**), and 1 (***) percent level. Values shown for t = 1 to 4 and 8 quarters after shock.

### Asymmetric Inflationary Response of SIDS vs non-SIDS to Global Shocks
- Motivation: SIDS characteristics (imperfect market structures, high import dependencies) can generate asymmetric, non-linear pass-throughs with more persistent and volatile inflation responses.
- Asymmetric international food inflation shocks:
  - Evidence of asymmetric pass-through for both SIDS and non-SIDS (Figure 8 Panels A and B).
  - Positive shocks have larger domestic inflation impacts than negative shocks (IRFs for negative shocks inverted for comparability).
  - SIDS-specific findings:
    - Positive shock cumulative impact peaks at around 0.15 percentage points after 4-5 quarters.
    - Non-SIDS peak for positive shock below 0.1 percentage point.
    - Asymmetries are more strongly statistically significant for SIDS; asymmetries diminish beyond two years.
  - Implication: Even if net incidence of positive and negative shocks is zero over time, cumulative domestic inflation impact is likely to be positive due to asymmetry.

### Baseline Robustness Checks
- Annual specification with output gap:
  - Constructed output-gap proxy using HP-filtered real GDP series; smoothing parameter 휆 = 6.25 for annual frequency.
  - Annual local projections include output gap alongside other controls in Y_{i,t}.
  - Annual results (Figure 9) preserve key baseline findings:
    - Differences in inflation dynamics between SIDS and non-SIDS in response to food shocks remain marked and large in magnitude.
    - Test for parameter equivalence for food shocks rejected at 10 percent level 1 year after the shock.
    - Responses to oil inflation shocks remain muted with less differentiation.
  - Limitation: Annual frequency reduces temporal detail on persistence but mitigates omitted-variable concerns.
- Additional robustness:
  - Introducing a COVID dummy (value =1 for 2020Q2 through 2022Q4) and running quarterly specification with two lags yields baseline results robust (results not reported).
- Endogeneity concerns:
  - Potential endogeneity of domestic and global commodity-price inflation would matter only if global commodity price shocks were demand-driven; this plausibly applies only to non-SIDS (SIDS too small to drive global demand).
  - Evidence cited that supply-side factors and supply-chain disruptions are key drivers of global food and oil price shocks, reducing endogeneity concerns.
  - Studies (Adjemian and others (2023); Furceri and others (2016); Choi and others (2018)) indicate pass-through estimates are negligibly impacted by approaches to control endogeneity; if non-SIDS pass-through were upward biased by endogeneity, the true SIDS vs non-SIDS difference would be larger than reported.

### Impact of Consumer Basket Weights
- Rationale: Larger CPI basket shares of food/transport could mechanically increase pass-through for SIDS.
- Food & non-alcoholic beverage CPI weight results:
  - Pre-weighting global food shock by country food & non-alcoholic beverage weight (Equation (4)); sample restricted to 2000Q1-2022Q4; baseline sample restricted to match.
  - Findings (Figure 10, Panel B vs Panel A):
    - Accounting for CPI basket weights does not materially alter baseline results; SIDS greater share of food in CPI basket does not meaningfully explain higher susceptibility.
    - With CPI weights, a 10-percentage point shock in international food inflation:
      - SIDS domestic inflation increases by about 1 percentage point at peak after three quarters; impact returns to zero after two years.
    - F-tests: disparity concentrated within 1 year of shock; statistically significant difference dissipating after 3 quarters (Table 2).
    - CPI weights explain some modest aspects of vulnerabilities (especially in second year) but a significant proportion remains unexplained — pointing to factors such as global supply chain connectivity and domestic market structures.
- CPI-weighted asymmetry:
  - Asymmetric pass-through persists after weighting; domestic inflation in SIDS more responsive to positive food-price shocks than in non-SIDS (Figure 11).
  - Sizes of food CPI weights have no bearing on asymmetry.
- Table 2 — F-test p-values for CPI-weighted shocks (SIDS vs Global Counterparts):
  - International Food Inflation Shock:
    - t=1: 0.0139**
    - t=2: 0.0031***
    - t=3: 0.0069***
    - t=4: 0.0599*
    - t=8: 0.0797*
  - International Oil Inflation Shock:
    - t=1: 0.4236
    - t=2: 0.2943
    - t=3: 0.4247
    - t=4: 0.2835
    - t=8: 0.0881*
  - Food Commodity Price Shock Asymmetry (p-values):
    - International Food Inflation Shock (SIDS): t=1: 0.0001***, t=2: 0***, t=3: 0***, t=4: 0***, t=8: 0.1767
    - International Food Inflation Shock (Non-SIDS): t=1: 0.002***, t=2: 0.0001***, t=3: 0.0001***, t=4: 0.0005***, t=8: 0.6946
  - Notes: p-values; bolded and starred values denote rejection of parameter equivalence across the two regimes at the 10 (*), 5 (**), and 1 (***) percent level. Values shown for t = 1 to 4 and 8 quarters after shock.
- Transport CPI weight results:
  - Accounting for transport weights has limited impact on baseline results for oil-price pass-through.
  - With transport weights, SIDS cumulative impact from a 10-percentage point shock in international crude-oil inflation:
    - Domestic inflation increases by 0.04 percentage points at peak after two quarters; impact returns to zero within the first year (Figure 12).
  - Consistent with Choi and others (2018): similar share of oil in consumption baskets across economies is an important determinant of inflation response; nonetheless, transport weight adjustments do not alter the main conclusion that oil pass-through is modest and similar across SIDS and non-SIDS.

*Source: IMF Working Paper (wpiea2025138-print-pdf), Section V content as supplied.*

### Conclusion

### Conclusion

### Main findings
- Estimates using local projections for a sample of 168 countries suggest inflation in SIDS is approximately twice as vulnerable to international food inflation shocks as in non-SIDS.
- The level of vulnerability and disparity between SIDS and non-SIDS is less apparent with respect to international crude oil inflation shocks.
- Results hold under a range of alternate specifications and robustness tests.
- Evidence indicates global-supply-chain connectivity issues and domestic market structures are potential major contributing factors to the inflation vulnerabilities of SIDS.
- The relative lack of pricing power of firms in SIDS may cause them to be more prone to global food inflation shocks.

### Evidence on asymmetry of pass-through
- There is strong asymmetry in the food inflation pass-through, particularly for SIDS: international food inflation increases have larger pass-throughs than equivalently sized food inflation decreases.
- This asymmetry indicates a net inflationary impact of global food inflation developments over time for SIDS.
- Note from the empirical figure: x-axis in quarters; t=0 is the quarter of the shock (1% to global oil inflation). The plotted local projection impulse response functions are reported in percentage points, with SIDS shown in red and non-SIDS in blue; corresponding shaded areas represent their 68 percent confidence bands.

### Potential channels and supporting evidence
- Market structure differences likely contribute:
  - If food imports into SIDS are supplied by a small number of firms, or relatively small firms, there may be greater degree of market power (and hence pricing power).
  - Smaller firms may tend to have larger markups than larger firms under some circumstances (De Loecker and Eeckhout (2017)); competition policy for small economies differs from larger economies (Gal (2003)).
  - Operating in small markets with small order sizes and heightened uncertainty (variable shipping costs; risks of product spoilage) can necessitate larger markups and greater profits for suppliers to remain viable.
  - SIDS consumers may have fewer substitutes available, forcing acceptance of higher prices when some global food prices increase.
- The role of market power aligns with the “sellers’ inflation” hypothesis:
  - During the pandemic, industrial firms with larger market shares not only passed through the entirety of cost shocks, but increased prices more than warranted by capitalizing on temporary monopoly power created by bottlenecks (Weber and Wasner (2023)).
  - More concentrated US industries displayed a 25 percent higher pass-through of costs to consumer prices than in other industries (Brauning and others (2022)).

### Policy implications and options
- There may not be an easy or obvious policy solution to the heightened global food-inflation pass-through facing SIDS.
- Possible policy directions include collaborative efforts to enhance connectivity and accessibility, and renewed focus on food security:
  - Improve efficiency and orientation of domestic production towards diversification and food self-sufficiency.
  - Regional and international collaboration is important because geographic limitations will remain a binding constraint.
  - Pooling orders at the regional level—such as by creating a regional food hub that can exercise greater market power in purchases—could help reduce vulnerability.
- Recognize SIDS’ limited range of short- to medium-term policy options due to dependence on imports and use of exchange-rate pegs.

### Research gaps and further work
- Further work is needed to identify drivers of pass-through and the best options for policy reforms.
- Specific research priorities:
  - Investigate second round effects of global commodity-price shocks to uncover the persistence of inflationary effects over the medium term.
  - Examine how inflation for different goods within the CPI basket is impacted by global price fluctuations.

*Source: Conclusion from "Global Commodity Inflation Pass-through: Vulnerability of Small Island Developing States", Working Paper No. WP/2025/138*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025138-print-pdf.pdf_