## Appendix Figure A1: Import intensity and spending on freight; 2018

## Source details

**Canonical URL:** [Appendix Figure A1: Import intensity and spending on freight; 2018](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022061-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022061-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022061-print-pdf.pdf.json)

---

### Overview
- Since the second half of 2020, indicators of container shipping costs increased by over 500 percent from pre-pandemic levels, while the cost of shipping bulk commodities by sea had tripled.
- Goods imports amount to some 38 percent of GDP on average in 2018; associated freight costs amount to some 7.5 percent of the value of imported goods.
- Freight costs exceed 15 percent of the value of imported goods in much of Sub-Saharan Africa and among small island states.
- The Baltic Dry Index (BDI) is used to proxy global shipping costs; it starts in January 1985, covers 100 percent of the bulk dry cargo in transit on the world’s oceans, and (since 2016) correlates with a container shipping price index compiled by Freightos at 0.85 at monthly frequency.

### Transmission channels and empirical approach
- Direct channel: higher shipping costs increase the local price of imported goods proportionately to shipping cost shares in product costs.
- Indirect channel: higher shipping costs raise costs of imported intermediate inputs, putting upward pressure on domestically produced goods; potential second-round effects via wage bargaining indexed to past inflation.
- Headline CPI decomposition used: π_t = (1−δ) π_t^D + δ π_t^I, where δ is the share of imported goods in the CPI basket.
- Empirical strategy: Jordà (2005) local projections estimated on monthly data to obtain impulse responses for horizons k = {0, 1, ..., 18}.
- Baseline regression controls include: global output gap; country i’s output gap; month-over-month growth rate of global oil prices; month-over-month growth rate of global food prices.
- Lags: number of lags l = 12 to control for additive seasonal effects.
- Robustness: heteroskedasticity-robust standard errors clustered at the country level; observations with dependent variable π below 1st percentile or above 99th percentile (1985–2021 distribution) are excluded.
- Instrumental Variables (IV) approach to strengthen causal interpretation: instrument changes in shipping costs using Suez Canal closure events.

### Data sample and key data features
- Baseline sample: monthly data since February 1992 to December 2021 for 46 countries (30 advanced economies and 16 emerging economies).
- Price series available for the sample: headline inflation, core inflation (excluding food and energy), producer prices, and import prices.
- BDI features: daily frequency from January 1985; measures average price to transport dry bulk materials across more than 20 oceanic shipping routes.

### Main empirical findings (baseline)
- A one-standard-deviation (21.8 percentage points) increase in global shipping costs typically increases domestic headline inflation by 0.15 percentage point over 12 months; the effect peaks after 12 months and reverts six months later.
- The elasticity of domestic inflation to global shipping costs is estimated to be 0.0067 at a horizon of 12 months.
- Effects are positive, persistent, and statistically significant for headline inflation, core inflation, producer prices, and import prices.
- Responses for import and producer prices materialize faster and are larger in magnitude than for headline and core inflation.
- When standardized to one standard deviation, shipping cost shocks produce inflation effects similar in magnitude but more persistent than shocks from global oil and food prices.

### Effect of shipping cost shocks on headline inflation and household consumption
- A one-standard-deviation increase in global shipping costs raises headline inflation; the estimated contribution is 0.3 percent of GDP and thus approximately 0.45 percent of household consumption.
- Shipping costs are more volatile than oil and food: standard deviation of 21.8 percentage points for shipping costs versus 10.8 and 3.0 percentage points for oil and food, respectively.
- Time profile:
  - Inflation from the BDI rises gradually and peaks after 12 months.
  - About 90 percent of the impact from an oil price shock materializes within four months.
  - Impact from a food price shock peaks after seven months.

### Effects across inflation measures and prices
- Core inflation:
  - Response to a global shipping cost shock is statistically significant beyond 6 months but only one-third as large as the impact on headline inflation.
  - Persistence similar to headline inflation, building gradually and peaking at 14 months.
- Producer and import prices:
  - Highly statistically significant responses at all horizons up to 12 months.
  - A one-standard-deviation increase in shipping costs peaks at an impact of 0.3 to 0.4 percentage point after 12 months.
  - Over 90 percent of the impact on these prices is in place within four months of the shock.
- Inflation expectations:
  - A one-standard-deviation increase in shipping costs increases 12-month-ahead inflation expectations by about seven basis points.
  - Response is highly persistent, rising until 12 months and returning to zero after 16 months.

### Cross-country heterogeneity and role of country characteristics
- Larger headline inflation effects in countries with a higher share of imported final consumption.
- Medium-term effects on headline and especially core inflation are more muted in countries with monetary policy frameworks that have track records of delivering low inflation.
- Inflationary impacts tend to be larger in small island and less developed economies.
- Freight costs are decreasing in GDP per capita and increasing in a country’s weighted distance from trading partners.
- By income:
  - Effect somewhat smaller in advanced economies than in emerging and developing countries; low-income countries see the largest effect.
  - Precision does not allow rejecting equality with baseline estimates.
- By region and geography:
  - Impact larger in Latin America and Asia than in Europe.
  - Larger in landlocked countries than in countries with direct ocean access.
  - Largest response in island countries (Caribbean and Pacific), where maximum impact is more than twice as large as the baseline.

### Robustness checks and IV evidence
- Expanded samples:
  - Full set of 143 countries (CPI data 1985–2021): one-standard-deviation shock leads to an increase in headline inflation of 0.2 percentage points after 12 months.
  - Balanced panel of 63 countries (January 1990–December 2021): results qualitatively and quantitatively consistent with baseline.
- Alternative controls and specifications:
  - Added country industrial production growth, China’s industrial production growth, VIX index, and nominal effective exchange rate (with 12 lags); baseline results are consistent across these specifications.
- IV strategy using Suez Canal closures:
  - Identified three blockage episodes: November 2004 (~3 days), February 2006 (1 day), March 2021 (6 days).
  - First-stage: regression of log changes in the BDI on Suez blockage yields a t-statistic of over 25.
  - Kleibergen‒Paap rk Wald F statistic is much higher than Stock-Yogo critical value.
  - IV estimates confirm baseline: significant impact on consumer prices increasing over the 6- to 18-month horizon; core inflation, PPI, and import prices also rise significantly.

### Non-linearity, timing, and dynamics
- Non-linearities significant only in first five months after an increase in global shipping costs: larger increases lead to faster pass-through to headline inflation.
- For horizons of six to 15 months, quadratic term is no longer significant — larger and smaller shocks have similar elasticity.
- At 18 months, quadratic term sign inverts and is highly significant: larger shocks have smaller impact on inflation.
- Stability over time:
  - Re-estimated for two periods (1990–2005 and 2006–2021) using 143 economies: impact on headline inflation remained unchanged across periods.
  - Responses peak between 10 and 13 months; recent sample shows a peak impact of 0.15.

### Factors affecting pass-through and second-round effects
- Import share of domestic final consumption (EORA input-output table; average 1990–2014):
  - Countries with high import share (over 24.6 percent) see larger impacts on headline inflation.
  - Core inflation differences between high and low import-share bins significant at horizons between 13 and 16 months.
- Degree of backward integration into global supply chains (share of foreign value added in exports):
  - Headline and core inflation responses significantly larger for countries with greater backward integration; low-integration countries show no statistically significant core inflation response.
- Monetary policy framework and anchoring:
  - Economies with high past inflation: core inflation increases about 0.08 after 14 months.
  - Economies with low past inflation: core inflation response is 0.025 after 14 months.
  - Countries without inflation targeting experience larger impacts on headline and core inflation; differences statistically significant.
  - Core inflation much stronger where expectations are poorly anchored; differences statistically significant.
  - Lower pass-through to headline and core inflation where disagreement among professional forecasters is lower; differences modest but statistically significant.
- Wages:
  - Wages rise for one year following an increase in shipping costs (annual sample of 18 countries, 1985–2021).
  - Wage increases larger in countries where inflation expectations are less anchored.

### Key timing and magnitudes (selected exact figures)
- Shipping costs standard deviation: 21.8 percentage points.
- Oil price standard deviation: 10.8 percentage points.
- Food price standard deviation: 3.0 percentage points.
- Peak of BDI effect on headline and core inflation: 12 months (headline) and 14 months (core).
- Impact on producer/import prices: peaks at 0.3 to 0.4 percentage point after 12 months.
- Increase in inflation expectations at 12 months: about seven basis points.
- Full-sample expanded estimate: 0.2 percentage points increase in headline inflation after 12 months (143-country sample).
- Recent-sample peak impact: 0.15.

*International Monetary Fund — IMF Working Papers: Shipping Costs and Inflation (content unit: Appendix Figure A1: Import intensity and spending on freight; 2018).*

### Appendix Figure A1: Import intensity and spending on freight; 2018......................................................

### Appendix Figure A1: Import intensity and spending on freight; 2018

### Overview
- Since the second half of 2020, indicators of container shipping costs increased by over 500 percent from pre-pandemic levels, while the cost of shipping bulk commodities by sea had tripled.
- Goods imports amount to some 38 percent of GDP on average in 2018; associated freight costs amount to some 7.5 percent of the value of imported goods.
- Freight costs exceed 15 percent of the value of imported goods in much of Sub-Saharan Africa and among small island states.
- The Baltic Dry Index (BDI) is used to proxy global shipping costs; it starts in January 1985, covers 100 percent of the bulk dry cargo in transit on the world’s oceans, and (since 2016) correlates with a container shipping price index compiled by Freightos at 0.85 at monthly frequency.

### Transmission channels and empirical approach
- Direct channel: higher shipping costs increase the local price of imported goods proportionately to shipping cost shares in product costs.
- Indirect channel: higher shipping costs raise costs of imported intermediate inputs, putting upward pressure on domestically produced goods; potential second-round effects via wage bargaining indexed to past inflation.
- Headline CPI decomposition used: π_t = (1−δ) π_t^D + δ π_t^I, where δ is the share of imported goods in the CPI basket.
- Empirical strategy: Jordà (2005) local projections estimated on monthly data to obtain impulse responses for horizons k = {0, 1, ..., 18}.
- Baseline regression controls include: global output gap; country i’s output gap; month-over-month growth rate of global oil prices; month-over-month growth rate of global food prices.
- Lags: number of lags l = 12 to control for additive seasonal effects.
- Robustness: heteroskedasticity-robust standard errors clustered at the country level; observations with dependent variable π below 1st percentile or above 99th percentile (1985–2021 distribution) are excluded.
- Instrumental Variables (IV) approach to strengthen causal interpretation: instrument changes in shipping costs using Suez Canal closure events.

### Data sample and key data features
- Baseline sample: monthly data since February 1992 to December 2021 for 46 countries (30 advanced economies and 16 emerging economies).
- Price series available for the sample: headline inflation, core inflation (excluding food and energy), producer prices, and import prices.
- BDI features: daily frequency from January 1985; measures average price to transport dry bulk materials across more than 20 oceanic shipping routes.

### Main empirical findings (baseline)
- A one-standard-deviation (21.8 percentage points) increase in global shipping costs typically increases domestic headline inflation by 0.15 percentage point over 12 months; the effect peaks after 12 months and reverts six months later.
- The elasticity of domestic inflation to global shipping costs is estimated to be 0.0067 at a horizon of 12 months.
- Effects are positive, persistent, and statistically significant for headline inflation, core inflation, producer prices, and import prices.
- Responses for import and producer prices materialize faster and are larger in magnitude than for headline and core inflation.
- When standardized to one standard deviation, shipping cost shocks produce inflation effects similar in magnitude but more persistent than shocks from global oil and food prices.

### Cross-country heterogeneity and role of country characteristics
- Larger headline inflation effects in countries with a higher share of imported final consumption.
- Medium-term effects on headline and especially core inflation are more muted in countries with monetary policy frameworks that have track records of delivering low inflation.
- Inflationary impacts tend to be larger in small island and less developed economies.
- Freight costs are decreasing in GDP per capita and increasing in a country’s weighted distance from trading partners.

### Comparisons and literature context
- Prior findings: Herriford et al. (2016) find modest but significant effects of shipping costs on US core PCE inflation, peaking after 11 months.
- UNCTAD (2021) suggests container freight rates at 2021 levels would raise global consumer prices by 1.5 percent relative to a counterfactual without the freight rate surge.
- OECD (2021) finds a persistent 50 percent increase in shipping costs would raise CPI inflation by about 0.2 percentage point after four quarters.
- Contribution: larger country sample, monthly data, multiple inflation measures, IV strategy exploiting Suez Canal closures, and examination of structural and monetary policy interactions.

*International Monetary Fund — IMF Working Papers: Shipping Costs and Inflation (content unit: Appendix Figure A1: Import intensity and spending on freight; 2018).*

### 0.3 percent of GDP and thus approximately 0.45 percent of household consumption.

### wpiea2022061-print-pdf - 0.3 percent of GDP and thus approximately 0.45 percent of household consumption.

### Effect of shipping cost shocks on headline inflation
- A one-standard-deviation increase in global shipping costs raises headline inflation; the estimated contribution is 0.3 percent of GDP and thus approximately 0.45 percent of household consumption.  
- Shipping costs are more volatile than oil and food: standard deviation of 21.8 percentage points for shipping costs versus 10.8 and 3.0 percentage points for oil and food, respectively.  
- The inflationary effects from shipping costs are quantitatively similar to those from oil and food price shocks because of this higher volatility.  
- Time profile:
  - Inflation from the BDI rises gradually and peaks after 12 months.
  - About 90 percent of the impact from an oil price shock materializes within four months.
  - Impact from a food price shock peaks after seven months.

### Effects across inflation measures and prices
- Core inflation:
  - Response to a global shipping cost shock is statistically significant beyond 6 months but only one-third as large as the impact on headline inflation.
  - Persistence similar to headline inflation, building gradually and peaking at 14 months.
- Producer and import prices:
  - Highly statistically significant responses at all horizons up to 12 months.
  - A one-standard-deviation increase in shipping costs peaks at an impact of 0.3 to 0.4 percentage point after 12 months.
  - Over 90 percent of the impact on these prices is in place within four months of the shock.
- Inflation expectations:
  - A one-standard-deviation increase in shipping costs increases 12-month-ahead inflation expectations by about seven basis points.
  - Response is highly persistent, rising until 12 months and returning to zero after 16 months.

### Baseline sample and robustness checks
- Baseline: unbalanced panel of 46 countries with data on headline inflation, core inflation, producer prices and import prices.
- Expanded samples:
  - Full set of 143 countries (CPI data 1985–2021): one-standard-deviation shock leads to an increase in headline inflation of 0.2 percentage points after 12 months.
  - Balanced panel of 63 countries (January 1990–December 2021): results qualitatively and quantitatively consistent with baseline.
- Alternative controls and specifications:
  - Added country industrial production growth, China’s industrial production growth, VIX index, and nominal effective exchange rate (with 12 lags); baseline results are consistent across these specifications.
- IV strategy using Suez Canal closures:
  - Identified three blockage episodes: November 2004 (~3 days), February 2006 (1 day), March 2021 (6 days).
  - First-stage: regression of log changes in the BDI on Suez blockage yields a t-statistic of over 25.
  - Kleibergen‒Paap rk Wald F statistic is much higher than Stock-Yogo critical value.
  - IV estimates confirm baseline: significant impact on consumer prices increasing over the 6- to 18-month horizon; core inflation, PPI, and import prices also rise significantly.

### Heterogeneity across income groups and regions
- By income:
  - Effect somewhat smaller in advanced economies than in emerging and developing countries; low-income countries see the largest effect.
  - Precision does not allow rejecting equality with baseline estimates.
- By region and geography:
  - Impact larger in Latin America and Asia than in Europe.
  - Larger in landlocked countries than in countries with direct ocean access.
  - Largest response in island countries (Caribbean and Pacific), where maximum impact is more than twice as large as the baseline.

### Stability over time
- Re-estimated for two periods (1990–2005 and 2006–2021) using 143 economies:
  - Impact on headline inflation remained unchanged across periods.
  - Responses peak between 10 and 13 months; recent sample shows a peak impact of 0.15.
  - Cannot reject equality of responses across periods.

### Non-linearity in shock size
- Augmented specification with quadratic term finds:
  - Non-linearities significant only in first five months after an increase in global shipping costs: larger increases lead to faster pass-through to headline inflation.
  - For horizons of six to 15 months, quadratic term is no longer significant — larger and smaller shocks have similar elasticity.
  - At 18 months, quadratic term sign inverts and is highly significant: larger shocks have smaller impact on inflation.

### Factors affecting pass-through to inflation
- Import share of domestic final consumption (EORA input-output table; average 1990–2014):
  - Defined bins by quartiles; countries with high import share (over 24.6 percent) see larger impacts on headline inflation.
  - Core inflation differences between high and low import-share bins significant at horizons between 13 and 16 months.
- Degree of backward integration into global supply chains (share of foreign value added in exports):
  - Split at median into low/high integration bins.
  - Headline and core inflation responses significantly larger for countries with greater backward integration; low-integration countries show no statistically significant core inflation response.
- Monetary policy framework and anchoring:
  - Split by average inflation in the 1990s into “high” and “low past inflation” bins:
    - Economies with high past inflation: core inflation increases about 0.08 after 14 months.
    - Economies with low past inflation: core inflation response is 0.025 after 14 months (virtually no pass-through).
  - Alternative proxies for monetary policy strength:
    - Inflation targeting regime: countries without inflation targeting experience larger impacts on headline and core inflation; differences statistically significant.
    - Estimated anchoring of inflation expectations (Choi and others 2022): headline similar across groups; core inflation much stronger where expectations are poorly anchored; differences statistically significant.
    - Disagreement among professional forecasters: lower pass-through to headline and core inflation where disagreement is lower; differences modest but statistically significant.
- Wages (annual sample of 18 countries, 1985–2021):
  - Wages rise for one year following an increase in shipping costs (evidence of second-round effects).
  - Wage increases larger in countries where inflation expectations are less anchored.
- Joint interaction robustness:
  - Time fixed effects with individual interaction terms: all interactions have statistically significant positive coefficients at some horizon, usually between 9 and 18 months.
  - Full model with time fixed effects and all interaction terms: import share of consumption and average past inflation are highly significant determinants of headline and core inflation responses; alternative monetary-policy proxies not significant due to collinearity.

### Key timing and magnitudes (selected)
- Shipping costs standard deviation: 21.8 percentage points.
- Oil price standard deviation: 10.8 percentage points.
- Food price standard deviation: 3.0 percentage points.
- Peak of BDI effect on headline and core inflation: 12 months (headline) and 14 months (core).
- Impact on producer/import prices: peaks at 0.3 to 0.4 percentage point after 12 months.
- Increase in inflation expectations at 12 months: about seven basis points.
- Full-sample expanded estimate: 0.2 percentage points increase in headline inflation after 12 months (143-country sample).
- Recent-sample peak impact: 0.15.

*Source: wpiea2022061-print-pdf - 0.3 percent of GDP and thus approximately 0.45 percent of household consumption.*

### 1992. We have described the strength and sequence of the transmission of these shocks through import

### wpiea2022061-print-pdf - 1992. We have described the strength and sequence of the transmission of these shocks through import

### Main finding
- A one-standard-deviation increase in global shipping costs increases domestic headline inflation by about 0.15 percentage point, with the effect building up over the course of 12 months.
- This pass-through has "remained strong over time," possibly reflecting increased openness of countries to international trade.

### Transmission sequence and channels
- Shocks transmit through import prices, producer prices, and into core and headline inflation.
- Figures indicate impulse response functions (IRFs) where t=0 denotes the year of the shock and confidence bands are reported (90 percent and 95 percent).

### Heterogeneity across countries and factors influencing pass-through
- Strength of pass-through depends crucially on:
  - Import share of domestic consumption.
  - Degree of integration into global supply chains (backward GVC linkages).
  - Strength of the monetary framework (including inflation targeting regimes and anchoring of inflation expectations).
- Observed heterogeneity:
  - Larger impacts in emerging and low-income countries that tend to have weaker monetary frameworks.
  - Highest impacts among small island countries that rely heavily on imported goods.
- Interaction results shown in figures:
  - Response of domestic prices interacts with import content (fourth quartile vs first quartile).
  - Response interacts with backward GVC linkages (above-median vs below-median).
  - Response interacts with average inflation in the 1990s (below-median vs above-median).
  - Response interacts with inflation-targeting regime dummy (inflation targeting vs other frameworks).
  - Response interacts with estimated anchoring of inflation expectations (anchoring below median vs above median).
  - Response interacts with disagreement among professional forecasters (disagreement above median vs below median).

### Magnitudes and statistics (summary from tables)
- Baseline sample summary (Table 1):
  - Full sample means: Headline 2.45, Core 2.17, Import prices 1.90, Producer prices 2.29.
  - Full sample standard deviations: Headline 2.91, Core 2.60, Import prices 9.53, Producer prices 5.90.
  - N (Full sample): Headline 10,349; Core 10,349; Import prices 10,348; Producer prices 10,349.
  - Advanced economies mean: Headline 1.78, Core 1.59, Import prices 1.13, Producer prices 1.37; Std. dev. Headline 1.58, Core 1.22, Import prices 8.82, Producer prices 4.70; N 7,277 for each series.
  - Emerging economies mean: Headline 4.04, Core 3.55, Import prices 3.70, Producer prices 4.45; Std. dev. Headline 4.36, Core 4.07, Import prices 10.84, Producer prices 7.63; N 3,072 for most series.
- Table 2 sample moments (1985–2021 and 2006–2021) (selected):
  - Baltic Dry Index (mom % chg) mean 0.36, Std. Dev. 21.81, N 10,486 (1985–2021); mean 0.24, Std. Dev. 24.10, N 7,891 (2006–2021).
  - Global oil price (mom % chg) mean 0.34, Std. Dev. 10.71, N 10,486; mean 0.09, Std. Dev. 11.48, N 7,891.
  - Inflation expectations (12m ahead) mean 0.03, Std. Dev. 0.02, N 9,808 (1985–2021); mean 0.02, Std. Dev. 0.02, N 7,351 (2006–2021).
  - Import share of domestic consumption mean 0.21, Std. Dev. 0.10, N 5,673 (1985–2021); mean 0.24, Std. Dev. 0.10, N 3,125 (2006–2021).

### Baseline regression results (Table 3) — coefficients scaled to one-standard-deviation shocks
- Headline inflation (Panel A): Shipping costs coefficients by horizon
  - k=1: 0.01705* (std. error 0.00943)
  - k=3: 0.06409*** (0.01646)
  - k=6: 0.08632*** (0.01935)
  - k=12: 0.14667*** (0.02234)
  - k=18: 0.08627*** (0.02218)
  - N: 10,337 (k=1) down to 9,460 (k=18). R2: 0.88 (k=1), 0.75 (k=3), 0.58 (k=6), 0.24 (k=12), 0.17 (k=18).
- Core inflation (Panel B): Shipping costs coefficients by horizon
  - k=1: 0.00571 (0.00668)
  - k=3: 0.02198** (0.01103)
  - k=6: 0.02463* (0.01366)
  - k=12: 0.04807*** (0.01837)
  - k=18: 0.05047*** (0.01642)
  - N: 10,217 down to 9,209. R2: 0.90, 0.79, 0.62, 0.31, 0.21.
- Producer price inflation (Panel C): Shipping costs coefficients by horizon
  - k=1: 0.14722*** (0.03683)
  - k=3: 0.24206*** (0.04229)
  - k=6: 0.27033*** (0.04436)
  - k=12: 0.29089*** (0.04246)
  - k=18: 0.16058*** (0.03563)
  - N: 10,325 down to 9,432. R2: 0.87, 0.71, 0.49, 0.22, 0.16.
- Import price inflation (Panel D): Shipping costs coefficients by horizon
  - k=1: 0.10454 (0.07454)
  - k=3: 0.23434** (0.09583)
  - k=6: 0.24332** (0.11516)
  - k=12: 0.37763*** (0.10038)
  - k=18: 0.00945 (0.06202)
  - N: 10,246 down to 8,956. R2: 0.55, 0.42, 0.29, 0.13, 0.10.
- Notes: Heteroskedasticity-robust standard errors clustered at the country level. ***, **, * denote 99, 95, 90 percent significance.

### Instrumental variable estimation (Table 4)
- Fitted shipping costs coefficients (response to one-standard-deviation shock):
  - k=1: -0.0038 (0.0105)
  - k=3: 0.0249 (0.0173)
  - k=6: 0.1072*** (0.0314)
  - k=12: 0.0991** (0.0457)
  - k=18: 0.1315*** (0.0366)
- 1st stage F-statistics: 60.4 (k=1), 63.5 (k=3), 60.5 (k=6), 60.4 (k=12), 32.5 (k=18).
- N: 10,409 down to 9,714. R2 reported: 0.99, 0.96, 0.85, 0.51, 0.09 by horizon.

### Non-linear effects (Table 5)
- Shipping costs (linear) coefficients:
  - k=1: -0.0271* (0.0140)
  - k=3: -0.0198 (0.0218)
  - k=6: 0.0619** (0.0272)
  - k=12: 0.1301*** (0.0287)
  - k=18: 0.1681*** (0.0362)
- Shipping costs (quadratic) coefficients:
  - k=1: 0.0540*** (0.0191)
  - k=3: 0.1028*** (0.0255)
  - k=6: 0.0300 (0.0325)
  - k=12: 0.0203 (0.0333)
  - k=18: -0.0993*** (0.0355)
- N: 10,337 down to 9,460. R2: 0.88, 0.75, 0.58, 0.24, 0.17.

### Additional empirical evidence and figures
- Indices of shipping costs during the COVID-19 pandemic displayed with January 2019=100 (Freightos, New ConTex).
- Comparative IRFs show impacts of shipping costs relative to global oil and food price shocks.
- Appendix figures show:
  - Import intensity and spending on freight (2018) for the 46-country baseline sample.
  - Correlates of freight costs (2005-18 average) with log GDP per capita (slope -0.14, R2 0.31) and log distance from trading partners (slope 0.035, R2 0.11); p-values < 0.001 for slope tests.

*Source: wpiea2022061-print-pdf - Shipping Costs and Inflation (International Monetary Fund).*

### Appendix Figure A3:  Baltic Dry  Index, January  1985  to February 2022

### Appendix Figure A3: Baltic Dry Index, January 1985 to February 2022

### Overview
- Coverage: Baltic Dry Index monthly averages from January 1985 to February 2022.
- Source: Bloomberg.

### Related empirical figures (descriptions)
- Appendix Figure A4: Elasticity of headline inflation to global shocks (percentage points)
  - Panels: A. Global shipping costs; B. Global oil prices; C. Global food prices.
  - Note: Impact of a one percentage point increase in each global shock on headline price inflation in the baseline sample of 46 economies. Solid line is the impulse response function (IRF); dark shaded region indicates the 90 percent confidence band; light shaded region indicates the 95 percent confidence band. t=0 denotes the year of the shock.
- Appendix Figure A5: The impact of global shipping cost shocks on headline inflation in alternative samples
  - Panel A: All available CPI data; 1985–2021 (143 economies).
  - Panel B: Balanced CPI panel; 1990–2021 (62 economies).
  - Note: Impact of a one standard deviation increase in global shipping costs on domestic headline inflation. Solid line is the IRF; shaded regions indicate 90 percent and 95 percent confidence bands. t=0 denotes the year of the shock.
- Appendix Figure A6: Impact of shipping cost shocks on headline inflation; robustness models
  - Note: Impact of one standard deviation increase in shipping costs on domestic headline inflation in baseline sample of 46 economies. Solid line is baseline IRF; dark shaded region indicates 90 percent confidence band; light shaded region indicates 95 percent confidence band. Dotted lines are IRFs for three robustness models that include: (i) contemporaneous and 12 lags of the change in the VIX index (black); (ii) contemporaneous and 12 lags of the change in the nominal effective exchange rate (red); and (iii) contemporaneous and 12 lags of the change in domestic and Chinese industrial production. t=0 denotes the year of the shock (purple).
- Appendix Figure A7: Impact of shipping cost shocks on headline inflation; baseline sample
  - Panel A: By income group.
  - Panel B: By geographic region.
  - Note: Solid line is IRF for full sample; dotted lines are IRFs for sub-samples. t=0 denotes the year of the shock.
- Appendix Figure A8: Average import share of domestic consumption (percentage of domestic consumption)
  - Source: EORA import-output table.
  - Note: Shows the average import share of domestic consumption for the 46 economies in the baseline sample.
- Appendix Figure A9: Impact of shipping cost shocks on wages (percentage points)
  - Panel A: Linear model.
  - Panel B: Interaction with estimated anchoring of inflation expectations.
  - Note: Impact of a one standard deviation increase in shipping costs (21.8 percentage points based on monthly frequency to ensure comparability to the baseline results) on wages in a sample of 18 economies. t=0 denotes the year of the shock. Panel B: shipping costs variable interacted with a dummy indicating bins of data over an estimate for the degree of inflation anchoring from Choi and others (2022). Purple line: IRF for economies with below median anchoring; blue line: IRF for economies with above median anchoring.

### Key statistics and sample information
- Baltic Dry Index monthly frequency: 1985m1-2022m1 (monthly average).
- Freightos global container index frequency: Weekly frequency; monthly average; sample 2016m10-2022m1.
- CTS global container index frequency: Monthly frequency; sample 2011m2-2021m11.
- New ConTex (Container Ship Time Charter Assessment Index) frequency: Monthly frequency; sample 2007m10-2022m1.
- Appendix Table A1: Pairwise correlation coefficients (overlapping samples at monthly frequency)
  - Baltic Dry — Baltic Dry: 1.00
  - Baltic Dry — CTS: 0.59
  - Baltic Dry — Freightos: 0.85
  - Baltic Dry — New ConTex: 0.38
  - CTS — CTS: 1.00
  - CTS — Freightos: 0.99
  - CTS — New ConTex: 0.75
  - Freightos — Freightos: 1.00
  - Freightos — New ConTex: 0.99
  - New ConTex — New ConTex: 1.00
- Baseline sample (Appendix Table A2)
  - TOTAL (46) 11,530 Jan-92 Dec-21
  - Example country entries (Economy, N, Start, End): Australia 279 Jul-98 Sep-21; Austria 262 Jan-00 Oct-21; Belgium 250 Jan-01 Oct-21; Brazil 359 Jan-92 Nov-21; China 201 Jan-05 Nov-21; United States 360 Jan-92 Dec-21; Korea 360 Jan-92 Dec-21; Taiwan Province of China 360 Jan-92 Dec-21.
- Wages (total labor compensation) frequency: Annual data (IMF World Economic Outlook).
- Import share of domestic consumption: Annual frequency 1990-2014 (EORA Global Input-Output table).
- Inflation expectations: Synthetic 12-months-ahead using weighted average of current and next year fixed-event forecasts (Consensus Economics).

### Robustness and regression specification highlights
- Appendix Table A4: Robustness specifications for interactions with time fixed effects (baseline sample of 46 economies over 1992-2021).
  - Panel A: Single interaction models; Panel B: Joint model with all interactions.
  - Dependent variables: Headline inflation and Core inflation.
  - Number of observations (Headline inflation, Panel A): N 10,473; 10,411; 10,253; 10,088; 9,923; 9,869; 9,759; 9,596 for k=1, k=3, k=6, k=9, k=12, k=13, k=15, k=18 respectively.
  - R2 (Headline inflation, Panel A): 0.89; 0.79; 0.65; 0.55; 0.46; 0.46; 0.46; 0.45 for k=1, k=3, k=6, k=9, k=12, k=13, k=15, k=18 respectively.
  - Number of observations (Core inflation, Panel A): N 10,353; 10,270; 10,083; 9,896; 9,710; 9,649; 9,525; 9,345 for k=1, k=3, k=6, k=9, k=12, k=13, k=15, k=18 respectively.
  - R2 (Core inflation, Panel A): 0.91; 0.81; 0.66; 0.52; 0.41; 0.39; 0.36; 0.33 for k=1, k=3, k=6, k=9, k=12, k=13, k=15, k=18 respectively.
  - Selected interaction coefficients and significance (examples, Headline inflation, Panel A, k=12):
    - Import share (high): 0.0498 (0.0313)
    - Past inflation (high): 0.0231 (0.0268)
    - No IT regime: 0.0807* (0.0433)
    - Anchoring (low): 0.0005 (0.0307)
    - Disagreement (high): 0.0503** (0.0238)
  - Note: Responses of headline and core inflation following a one standard deviation shock to the Baltic Dry Index; differential impact for countries in the indicated bin versus others. Table reports estimates for specifications that include time fixed effects instead of global control variables, and are estimated on the baseline sample of 46 economies over 1992-2021. Heteroskedasticity-robust standard errors clustered at the country level are reported in parentheses. ***, **, and * denote statistical significance at 99, 95, and 90 percent confidence levels.

### Variable sources and measurement notes (Appendix Table A3)
- Consumer Price Index: Haver Analytics.
- Core CPI: Haver Analytics (Chile spliced using historical variation in IPCX1 for 2003-2011).
- Import price index: Haver Analytics (China and Philippines: quarterly frequency; India: annual frequency; Brazil, Malaysia, Mexico, Peru, Switzerland, and Turkey: original series denominated in US dollars multiplied by the nominal exchange rate to express in local currency units).
- Industrial production index: Haver Analytics (China: spliced backwards using variation in quarterly real GDP for 1991-1997).
- Baltic Dry Index: Bloomberg (Daily frequency data; monthly average).
- Freightos global container index: Bloomberg (Weekly frequency; monthly average).
- CTS global container index: Bloomberg (Monthly frequency).
- Container Ship Time Charter Assessment Index: Bloomberg and Hamburg Shipbrokers’ Association (Monthly frequency).
- World oil price: Bloomberg West Texas Intermediate Crude Oil Prices.
- World food price index: IMF Primary Commodity Prices (Monthly since January 1992).
- Output gap and World output gap: IMF World Economic Outlook (Annual data).
- Nominal effective exchange rate: IMF Information Notice System (Local currency units/USD).
- VIX index: Bloomberg (Equity price volatility index from Chicago Board Options Exchange).
- Inflation expectations and disagreement about future inflation: Consensus Economics.
- Distance from trade partners: CEPII GeoDist Database and UN COMTRADE (Weights constructed using total trade in 2019).
- Landlocked country dummy: CEPII GeoDist Database.
- Inflation targeting dummy: 1 if inflation targeting, 0 otherwise.
- Advanced/developing dummy: IMF World Economic Outlook (1 if advanced, 0 if developing).

*Source: IMF Working Paper — Shipping Costs and Inflation (figures and tables from the Appendix).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022061-print-pdf.pdf_