## _wp16240

## Source details

**Canonical URL:** [_wp16240](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16240.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16240.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16240.pdf.json)

---

### Correlations between bilateral exchange rates and NEER; importance of the United States
- The median correlation between the nominal effective exchange rate (NEER) and the bilateral exchange rate against the U.S. dollar has been smaller than 0.5 in many cases.
- The distinction between bilateral and effective rates has grown more relevant as the United States’ share of global imports has fallen substantially over the last five decades.
- Panel A reports rolling three-year correlations between NEER and the bilateral exchange rate against the U.S. dollar starting in 1993; box plots display interquartile ranges and medians for each country.
- Panel B shows the five-year moving average of the annual share of total imports from the United States over total world import (excluding the United States).

### Methodology: construction of NEER and empirical strategy
- NEERi,t constructed as a weighted average of bilateral exchange rates of trading partners vis-à-vis the U.S. dollar, weighted by their import shares (lagged one year) from the IMF’s Direction of Trade Statistics.
- Monthly percentage change specification:
  - ∆neeri,t = Σj=1^J ωi j,t (∆ei,t − ∆e j,t), i ≠ j
  - ei,t = natural logarithm of country i’s bilateral exchange rate (local currency per U.S. dollar)
  - ωi j,t = share of exports from country j to country i in country i’s total imports (lagged one year)
- Cost of production in import partners proxied by:
  - ∆mPPIi,t = Σj=1^J ωi j,t ∆PPIj,t, i ≠ j, where PPIj,t is the natural logarithm of country j’s producer price index.
- Bilateral exchange rates are weighted by lagged import flows that vary annually to better capture impact of currency depreciation on domestic prices.
- Empirical specification includes six lags and is estimated by ordinary least squares on monthly data.
- Local projections method (per Jordà (2005)) used instead of VARs to enhance robustness and uncover non-linearities and interactions.
  - Implementation: fit separate regressions for each horizon h = 1,2,...,h and compute cumulative impact as 1 + ρ̂1 + ρ̂2 + ... + ρ̂h.
  - Additional steps: use serial-correlation-robust estimator; include residual from horizon h−1 regression as additional regressor for horizon h to improve efficiency and address bias identified by Teulings and Zubanov (2014).
- Estimated response: cumulative impact of an innovation in the NEER on the consumer price index (CPI); no structural identification of exchange rate shocks is imposed.

### Data description and sample
- Dependent variable: seasonally adjusted headline consumer price index (IMF’s International Financial Statistics and Haver Analytics).
- International oil and food prices: composite indices in U.S. dollars from the IMF’s World Economic Outlook.
- Output gap proxied by the cyclical component of industrial production (HP filter with smoothing coefficient equal to 129,600 on monthly data); end-point bias addressed by linearly extrapolating HP trend from 2013 to the last two years in sample.
- Sample period for baseline estimates: January 2000 to December 2015.
- Summary statistics for 1995-2016:
  - Emerging economies:
    - ∆pi,t: N = 7,691; Mean = 0.7; Std. Dev. = 1.9; Quartiles = [0.2, 0.8]
    - ∆NEERi,t: N = 7,512; Mean = 0.3; Std. Dev. = 3.6; Quartiles = [-0.8, 0.9]
    - gapi,t: N = 7,191; Mean = 0.0; Std. Dev. = 4.6; Quartiles = [-1.9, 2.4]
    - ∆mPPIi,t: N = 7,512; Mean = 0.4; Std. Dev. = 0.8; Quartiles = [0.0, 0.7]
  - Advanced economies:
    - ∆pi,t: N = 8,292; Mean = 0.2; Std. Dev. = 0.4; Quartiles = [0.0, 0.3]
    - ∆NEERi,t: N = 7,992; Mean = -0.1; Std. Dev. = 1.4; Quartiles = [-0.7, 0.5]
    - gapi,t: N = 8,131; Mean = 0.0; Std. Dev. = 4.6; Quartiles = [-1.9, 2.3]
    - ∆mPPIi,t: N = 7,992; Mean = 0.2; Std. Dev. = 0.6; Quartiles = [-0.1, 0.6]

### Estimation results: exchange rate pass-through (cumulative CPI responses)
- Baseline: cumulative impulse responses of headline consumer prices to a 1-percent innovation in the NEER with 95-percent confidence bands.
- Panel pooled estimates (cumulative impact after two years):
  - Advanced economies: 0.13
  - Emerging markets: 0.39
- Regional cumulative pass-through after two years (emerging market economies):
  - Asia: 0.2
  - Latin America: 0.28
  - Europe (emerging Europe): 0.5
- Differences in speed of adjustment:
  - Advanced economies outside the euro area: bulk of price response occurs within 12 months.
  - Latin American emerging markets: effect peaks after 9 months.
  - Emerging Europe and Asia: price response is more gradual.
- Figures referenced:
  - Figure 2: cumulative impulse response of headline consumer prices to a 1 percent innovation in the NEER (A. Advanced Economies; B. Emerging Market Economies); shaded bands correspond to 95-percent confidence intervals.
  - Figure 3, Panel A: pass-through by panel group (ADV, ADV (Ex cl. Euro Area), EME, EME Asia, EME Europe, EME LATAM) showing pass-through at 12 months and 24 months, implied pass-through, and import content.
  - Figure 3, Panel B: pass-through over rolling time windows (series plotted for 2007, 2011, 2015 for each panel group).

### Country-specific heterogeneity: selected exact estimates (two-year or 12/24 months as reported)
- Selected two-year or 12- and 24-month pass-through estimates (point estimates, standard errors, p-values where provided):
  - Argentina: 12 months = 0.55 (s.e. 0.13, p-value 0.00); 24 months = 0.36 (s.e. 0.11, p-value 0.00)
  - Hong Kong SAR: 12 months = 0.52 (s.e. 0.17, p-value 0.00); 24 months = 1.28 (s.e. 0.29, p-value 0.00)
  - Pakistan: 12 months = 0.68 (s.e. 0.19, p-value 0.00); 24 months = 1.33 (s.e. 0.37, p-value 0.00)
  - Slovenia: 12 months = 0.58 (s.e. 0.09, p-value 0.00); 24 months = 1.15 (s.e. 0.17, p-value 0.00)
  - United States: 12 months = 0.05 (s.e. 0.09, p-value 0.60); 24 months = 0.11 (s.e. 0.14, p-value 0.41)
  - Japan: 12 months = 0.07 (s.e. 0.02, p-value 0.00); 24 months = 0.17 (s.e. 0.04, p-value 0.00)
  - Brazil: 12 months = 0.10 (s.e. 0.03, p-value 0.00); 24 months = 0.23 (s.e. 0.04, p-value 0.00)
  - Germany: 12 months = 0.08 (s.e. 0.06, p-value 0.20); 24 months = -0.01 (s.e. 0.06, p-value 0.87)
  - France: 12 months = -0.03 (s.e. 0.07, p-value 0.66); 24 months = -0.18 (s.e. 0.07, p-value 0.01)
  - Canada: 12 months = -0.02 (s.e. 0.03, p-value 0.41)
- Additional country-level entries from Annex Table A2 (selected):
  - Argentina: 12 months = 0.33, 0.13, 0.01; 24 months = 0.30, 0.12, 0.01; Import Content: 0.11
  - Australia: 12 months = 0.03, 0.03, 0.36; 24 months = 0.04, 0.04, 0.26; Import Content: 0.16
  - China: 12 months = -0.05, 0.24, 0.82; 24 months = -0.79, 0.69, 0.25; Import Content: 0.06
  - India: 12 months = 0.08, 0.06, 0.14; 24 months = 0.11, 0.14, 0.44; Import Content: 0.05
  - Indonesia: 12 months = 0.17, 0.03, 0.00; 24 months = 0.31, 0.06, 0.00; Import Content: 0.14
  - Turkey: 12 months = 0.42, 0.06, 0.00; 24 months = 0.61, 0.08, 0.00; Import Content: 0.15
  - Ukraine: 12 months = 0.78, 0.14, 0.00; 24 months = 0.21, 0.28, 0.47; Import Content: 0.23
  - Singapore: 12 months = -0.30, 0.10, 0.00; 24 months = -0.72, 0.19, 0.00; Import Content: 0.34

### Decline in exchange rate pass-through (aggregate findings)
- Exchange rate pass-through to consumer prices has systematically decreased in all country groups over the past two decades.
- Decline occurred in both advanced and emerging economies; more pronounced in emerging economies.
- In Latin America, average pass-through has fallen to only one-third of its 1995–2006 levels.
- Figure 3 reports two benchmarks based on import content (discussed in Section IV).

### Benchmarks for first-round pass-through and import content
- Decomposition into:
  - first-round effects: contribution of imported goods prices measured at the border (import prices)
  - second-round effects: response of all other prices in the consumption basket (including distribution and retail margins on imported goods)
- Under perfect competition and complete pass-through to import prices, expected first-round effect of a depreciation on consumer prices = import content of final household consumption.
- Import content computed using Eora multi-region input-output tables:
  - Import content = direct component (imports of final consumption goods) + indirect component (imported inputs embodied in domestically produced goods consumed by households).
- Pooled panel estimates of pass-through to import prices are close to complete in both advanced and emerging market economies.
- In individual-economy regressions, null hypothesis of complete pass-through to import prices cannot be rejected at a 95-percent confidence level in 29 of 40 countries.

### Import content trends and measurement issues
- Import content of consumption expenditure has steadily increased in many economies over the past 20 years.
- Direct imports dominate import content, but indirect component can be sizable.
- Import content is higher in advanced economies than in emerging markets; small highly open advanced economies (e.g., Luxembourg and Singapore) contribute to high import content.
- Measures like imports-to-GDP are imperfect proxies for import content of consumption:
  - Example: in 2010, economies with input-output based import content ≈ 20 percent had imports-to-GDP ranging from <20 percent to ≈70 percent; economies with import content ≈ 40 percent had imports-to-GDP as large as 170 percent.

### Comparison of estimated pass-through vs. benchmarks (key conclusions)
- Two benchmarks reported alongside pass-through estimates:
  - import content of consumption (“import content”)
  - product of import content and estimated pass-through to import prices (“implied pass-through”)
- For advanced economies:
  - Estimated pass-through to consumer prices is below both benchmarks.
  - Import content is above the implied pass-through, suggesting incomplete exchange rate pass-through to import prices.
- For emerging market economies:
  - Estimated pass-through to consumer prices is generally above both benchmarks, suggesting sizable second-round effects.
- The difference between overall pass-through to consumer prices and both benchmarks has narrowed substantially in emerging market economies.
- Generalized decline in overall exchange rate pass-through to consumer prices cannot be accounted for by changes in the role of import prices alone.

### Decomposition, trends, and contribution of first- vs second-round effects
- Implied pass-through declined by only 0.01 between the first and last estimation windows in the referenced economies.
- Only about 5 percent of the documented decline in pass-through to consumer prices can be accounted for by changes in first-round effects; the remainder is associated with falling second-round effects.
- Baseline sample (24-month horizon): 62 estimates of second-round effects:
  - 33 economies display positive second-round effects.
  - 29 economies display negative second-round effects.
- Subsample with import price data (40 economies): distribution shifted to the right; only three negative values; larger concentration between 0 and 0.2.

### Two-stage testing strategy and determinants of pass-through
- Two-stage estimation:
  - First stage: time-varying estimates of exchange rate pass-through for each economy estimated by rolling 12-year windows starting in January of each year since 1995.
  - Second stage: regress β̂0,i,τ12 on explanatory variables X i,τ averaged over the corresponding window τ and country i with time fixed effects ψτ; weighted least squares used with weights equal to inverse of variance of estimated pass-through coefficients.
- Determinants (Table 3 key points):
  - Standard measures of price stability (mean and standard deviation of the inflation rate) are positively related to exchange rate pass-through.
    - A 1 percent increase in the mean rate of inflation is associated with a pass-through response that is 1.4 percent stronger.
  - Median depreciation and standard deviation of the multilateral exchange rate are associated with higher pass-through.
  - Variability of consensus inflation forecasts at a 12-month horizon: where forecasts are more volatile, exchange rate pass-through is higher.
  - Log disagreement among professional forecasters of inflation:
    - Strong and significant positive relationship with exchange rate pass-through.
    - In a full specification, lower forecast disagreement remains associated with smaller pass-through.
    - A decline in disagreement from the top to the bottom of the interquantile range is associated with a decline in pass-through of about 0.07 percentage points.
  - With country fixed effects included, variability of inflation and forecast disagreement remain positively related to pass-through; coefficient on forecast disagreement remains large.

### Forecast disagreement as a monetary credibility measure
- Forecast disagreement measured as standard deviation across fixed-event forecasts combined into a synthetic 12-month fixed horizon (monthly frequency for 41 countries).
- Strong correlation documented between forecast disagreement and the average (squared) deviation of inflation from central bank targets between 2000 and 2015.
- Interpretation: better-anchored inflation expectations (lower disagreement) make firms less likely to pass along exchange rate movements to consumer prices.

### Second-round effects regression and robustness
- Second-round effects constructed as:
  - Ω̂i,τ = β̂0,i,τ12 − δ i,τ
  - δ i,τ = benchmark for expected first-round effects (average for window τ and country i)
- Regression: Ω̂i,τ = θ X i,τ + ψτ + ε i,τ
- Using second-round effects with complete pass-through benchmark:
  - Strong cross-sectional correlation between forecast disagreement and second-round effects.
  - Decline in disagreement from top to bottom of interquartile range associated with decline in pass-through of about 0.10 percentage point.
- Allowing for incomplete pass-through to import prices:
  - Main findings broadly unchanged.
  - Coefficient on disagreement larger and remains strongly significant with all controls and country fixed effects.
- Conclusion: for given levels and volatility of inflation and the exchange rate, more credible monetary policy substantially reduces exchange rate pass-through to consumer prices beyond the border.

### Concluding findings and policy implications
- Empirical decomposition separates pass-through into first-round (import pricing at the dock) and second-round (domestically-produced goods and services) components.
- Despite increasing import content of domestic consumption, generalized fall in exchange rate pass-through occurred; modest declines in pass-through to import prices imply reductions in second-round effects largely explain the fall in overall pass-through.
- Monetary policy performance, including measures of price stability and forecast disagreement, is statistically significant and economically important in explaining cross-country and time variation in exchange-rate pass-through.
  - An increase in disagreement among professional forecasters of inflation from the 25th to the 75th percentiles is associated with an increase in the estimated pass-through coefficient by 0.1.
  - Average cumulative pass-through after two years is about 0.3 in the sample.
- Policy implication: where central bank credibility is established, it may be less costly for monetary policy to stabilize inflation and real activity while allowing the exchange rate to help the economy adjust to external shocks.

*Source: Authors’ calculations and analysis as reported in the document.*

### 1993. These correlations have frequently been far from unity in several economies, with the

### _wp16240 - 1993. These correlations have frequently been far from unity in several economies, with the

### Correlations between bilateral exchange rates and NEER; importance of the United States
- The median correlation between the nominal effective exchange rate (NEER) and the bilateral exchange rate against the U.S. dollar has been smaller than 0.5 in many cases.
- The distinction between bilateral and effective rates is becoming more relevant over time as the United States’ share of global imports has fallen substantially over the last five decades.
- Panel A (distributions) reports rolling three-year correlations between NEER and the bilateral exchange rate against the U.S. dollar starting in 1993; box plots display interquartile ranges and medians for each country.
- Panel B shows the five-year moving average of the annual share of total imports from the United States over total world import (excluding the United States).

### Methodology: construction of NEER and empirical strategy
- Nominal effective exchange rate (NEERi,t) constructed as a weighted average of bilateral exchange rates of trading partners vis-à-vis the U.S. dollar, weighted by their import shares (lagged one year) from the IMF’s Direction of Trade Statistics.
- Monthly percentage change formula: ∆neeri,t = Σj=1^J ωi j,t (∆ei,t − ∆e j,t), i ≠ j, where ei,t is the natural logarithm of country i’s bilateral exchange rate (local currency per U.S. dollar); ωi j,t is the share of exports from country j to country i in country i’s total imports (lagged one year).
- Monthly change in cost of production in country i’s import partners proxied by: ∆mPPIi,t = Σj=1^J ωi j,t ∆PPIj,t, i ≠ j, where PPIj,t is the natural logarithm of country j’s producer price index.
- Departing from trade-weighted NEER, bilateral exchange rates are weighted by lagged import flows that vary annually to better capture impact of currency depreciation on domestic prices.
- Empirical specification includes six lags and is estimated by ordinary least squares on monthly data.
- Local projections method used (per Jordà (2005)) instead of VARs to enhance robustness to misspecification and allow uncovering non-linearities and interactions.
  - Implementation: fit separate regressions for each horizon h = 1,2,...,h and compute cumulative impact as 1 + ρ̂1 + ρ̂2 + ... + ρ̂h.
  - Additional efficiency and bias-reduction steps: use serial-correlation-robust estimator; include residual from horizon h−1 regression as additional regressor for horizon h to improve efficiency and address bias identified by Teulings and Zubanov (2014).
- The estimated response is the cumulative impact of an innovation in the NEER on the consumer price index (CPI); no structural identification of exchange rate shocks is imposed.

### Data description
- Dependent variable: seasonally adjusted headline consumer price index (IMF’s International Financial Statistics and Haver Analytics).
- International oil and food prices: composite indices in U.S. dollars from the IMF’s World Economic Outlook.
- Output gap proxied by the cyclical component of industrial production (HP filter with smoothing coefficient equal to 129,600 on monthly data); end-point bias addressed by linearly extrapolating HP trend from 2013 to the last two years in sample.
- Sample period for baseline estimates: January 2000 to December 2015.
- Sample coverage listed by economy; summary statistics reported for 1995-2016:
  - Emerging economies (N reported per variable)
    - ∆pi,t: N = 7,691; Mean = 0.7; Std. Dev. = 1.9; Quartiles = [0.2, 0.8]
    - ∆NEERi,t: N = 7,512; Mean = 0.3; Std. Dev. = 3.6; Quartiles = [-0.8, 0.9]
    - gapi,t: N = 7,191; Mean = 0.0; Std. Dev. = 4.6; Quartiles = [-1.9, 2.4]
    - ∆mPPIi,t: N = 7,512; Mean = 0.4; Std. Dev. = 0.8; Quartiles = [0.0, 0.7]
  - Advanced economies (N reported per variable)
    - ∆pi,t: N = 8,292; Mean = 0.2; Std. Dev. = 0.4; Quartiles = [0.0, 0.3]
    - ∆NEERi,t: N = 7,992; Mean = -0.1; Std. Dev. = 1.4; Quartiles = [-0.7, 0.5]
    - gapi,t: N = 8,131; Mean = 0.0; Std. Dev. = 4.6; Quartiles = [-1.9, 2.3]
    - ∆mPPIi,t: N = 7,992; Mean = 0.2; Std. Dev. = 0.6; Quartiles = [-0.1, 0.6]

### Estimation results: exchange rate pass-through (cumulative CPI responses)
- Baseline: cumulative impulse responses of headline consumer prices to a 1-percent innovation in the NEER with 95-percent confidence bands.
- Panel estimates pooled by income designation (advanced versus emerging market economies):
  - Cumulative impact after two years:
    - Advanced economies: 0.13
    - Emerging markets: 0.39
- Regional results for emerging market economies (cumulative pass-through after two years reported as percentages):
  - Asia: 0.2
  - Latin America: 0.28
  - Europe (emerging Europe): 0.5
- Differences in speed of adjustment:
  - Advanced economies outside the euro area: bulk of price response occurs within 12 months; process relatively fast and smooth.
  - Latin American emerging markets: effect peaks after 9 months (faster pass-through).
  - Emerging Europe and Asia: response of prices is more gradual (slower pass-through).
- Figures and panels:
  - Figure 2: cumulative impulse response of headline consumer prices to a 1 percent innovation in the NEER (A. Advanced Economies; B. Emerging Market Economies); shaded bands correspond to 95-percent confidence intervals.
  - Figure 3, Panel A: pass-through by panel group (ADV, ADV (Ex cl. Euro Area), EME, EME Asia, EME Europe, EME LATAM) showing pass-through at 12 months and 24 months, implied pass-through, and import content.
  - Figure 3, Panel B: pass-through over rolling time windows (series plotted for 2007, 2011, 2015 for each panel group).

### Interpretation and analytical notes
- Reported pass-through estimates reflect cumulative percentage increases in headline CPI one and two years after each percentage point increase in NEER.
- Estimates are reduced-form and conditional on the average constellation of shocks that moved exchange rates during the sample period; no structural identification of exchange rate shocks is imposed.
- Local projections allow uncovering potential non-linearities in impulse responses and accommodate alternative lag structures and interactions across countries and sectors.
- Robustness considerations:
  - Serial correlation in residuals addressed with robust estimators.
  - Efficiency improved by augmenting horizon-h regressions with residuals from horizon-(h−1) regressions to mitigate loss of efficiency with increasing horizon h and to address a potential bias pointed out by Teulings and Zubanov (2014).

*Source: Authors’ calculations and figures as reported in the document.*

### 2003. In Figure 3, Panel B reports the results for the two-year cumulative exchange rate pass-

### _wp16240 - 2003. In Figure 3, Panel B reports the results for the two-year cumulative exchange rate pass-through to consumer prices of a one-percent innovation in the nominal effective exchange rate.

### Decline in exchange rate pass-through (aggregate findings)
- The exchange rate pass-through to consumer prices has systematically decreased in all country groups over the past two decades.
- The decline occurred in both advanced and emerging economies, though the decline has been more pronounced in emerging economies.
- Important reductions in pass-through are estimated for emerging Asia, Europe, and Latin America.
- In Latin America, the average pass-through has fallen to only one-third of its 1995–2006 levels.
- Figure 3 reports two benchmarks based on import content; these benchmarks are described and discussed in Section IV.

### Country-specific estimation and heterogeneity (method and sample)
- Country-specific models estimate pass-through to consumer prices over the baseline sample period 2000-15 using equation (2) specified in the source.
- Table 2 reports pass-through estimates for each economy in the sample.
- There is substantial heterogeneity in estimated exchange rate pass-through across countries: point estimates at a two-year horizon range from negative (although not statistically significant) to larger than unity for a handful of emerging economies.
- Note on samples and variants:
  - The sample for Argentina uses data from January 2000 to December 2010; CPI data after December 2006 corresponds to private analysts’ estimates.
  - Pass-through estimates using the bilateral exchange rate against the U.S. dollar are reported in Table A1 (not reproduced here).

### Selected country pass-through estimates from Table 2 (exact reported values)
- Argentina: Pass-through after 12 months = 0.55 (s.e. 0.13, p-value 0.00); Pass-through after 24 months = 0.36 (s.e. 0.11, p-value 0.00).
- Hong Kong SAR: Pass-through after 12 months = 0.52 (s.e. 0.17, p-value 0.00); Pass-through after 24 months = 1.28 (s.e. 0.29, p-value 0.00).
- Pakistan: Pass-through after 12 months = 0.68 (s.e. 0.19, p-value 0.00); Pass-through after 24 months = 1.33 (s.e. 0.37, p-value 0.00).
- Slovenia: Pass-through after 12 months = 0.58 (s.e. 0.09, p-value 0.00); Pass-through after 24 months = 1.15 (s.e. 0.17, p-value 0.00).
- United States: Pass-through after 12 months = 0.05 (s.e. 0.09, p-value 0.60); Pass-through after 24 months = 0.11 (s.e. 0.14, p-value 0.41).
- Japan: Pass-through after 12 months = 0.07 (s.e. 0.02, p-value 0.00); Pass-through after 24 months = 0.17 (s.e. 0.04, p-value 0.00).
- Brazil: Pass-through after 12 months = 0.10 (s.e. 0.03, p-value 0.00); Pass-through after 24 months = 0.23 (s.e. 0.04, p-value 0.00).
- Germany: Pass-through after 12 months = 0.08 (s.e. 0.06, p-value 0.20); Pass-through after 24 months = -0.01 (s.e. 0.06, p-value 0.87).
- France: Pass-through after 12 months = -0.03 (s.e. 0.07, p-value 0.66); Pass-through after 24 months = -0.18 (s.e. 0.07, p-value 0.01).
- Examples of near-zero or negative short-run estimates: Canada 12-month = -0.02 (s.e. 0.03, p-value 0.41); New Zealand 12-month = 0.03 (s.e. 0.03, p-value 0.28).

### Benchmarks for first-round pass-through (concept and construction)
- Decomposition: pass-through to consumer prices is decomposed into
  - first-round effects — contribution of imported goods prices measured at the border (import prices), and
  - second-round effects — response of all other prices in the consumption basket (including distribution and retail margins on imported goods).
- Under perfectly competitive markets and complete pass-through to import prices, the expected first-round effect of a depreciation on consumer prices corresponds directly to the import content of final household consumption.
- Empirical considerations:
  - Some studies document incomplete pass-through to import prices; mechanisms include pricing-to-market under imperfect competition, currency-of-invoicing choices, and consumer substitution toward domestic varieties.
  - The authors start with a benchmark assuming complete pass-through to import prices and also construct a second benchmark allowing for incomplete pass-through using their own estimated pass-through to import prices.
- Data and methods for benchmarks:
  - Import content of final consumption is computed using Eora multi-region input-output tables.
  - Import content = direct component (imports of final consumption goods) + indirect component (imported inputs embodied in domestically produced goods consumed by households).
  - Figure 4 reports pooled panel estimates of pass-through to import prices (local currency) using a local projection specification similar to (1); pooled estimates are close to complete in both advanced and emerging market economies.
  - In individual-economy regressions, the null hypothesis of complete exchange rate pass-through to import prices cannot be rejected at a 95-percent confidence level in 29 of 40 countries in the sample.

### Import content trends and measurement issues
- Figure 5: The import content of consumption expenditure has steadily increased in many economies over the past 20 years.
- Direct imports dominate the import content, but the indirect component can be sizable.
- Import content is higher in advanced economies than in emerging markets, partly due to small highly open advanced economies (e.g., Luxembourg and Singapore).
- Measures of trade openness (e.g., imports-to-GDP) are imperfect proxies for import content of consumption:
  - Imports-to-GDP includes imports of non-consumption goods, government-absorbed consumption goods, and re-exports — these are not part of household consumption.
  - Figure 6 shows that in 2010, economies with an input-output based import content of about 20 percent had imports-to-GDP ratios ranging from less than 20 percent to close to 70 percent; economies with import content close to 40 percent had import-to-GDP ratios as large as 170 percent.
  - The dissociation between trade-based openness metrics and input-output based import content may have grown over time as participation in global value chains increased.

### Comparison of estimated pass-through vs. benchmarks (key conclusions)
- The authors report two benchmarks alongside pass-through estimates: (1) import content of consumption (“import content”) and (2) product of import content and estimated pass-through to import prices (“implied pass-through”).
- For advanced economies:
  - Estimated pass-through to consumer prices is below both benchmarks.
  - Import content is above the implied pass-through, suggesting incomplete exchange rate pass-through to import prices.
- For emerging market economies:
  - Estimated pass-through to consumer prices is generally above both benchmarks, suggesting sizable second-round effects on consumer prices.
- Two additional observations:
  - The difference between overall pass-through to consumer prices and both benchmarks has narrowed substantially in emerging market economies, suggesting second-round effects are less pervasive than in the past.
  - The generalized decline in overall exchange rate pass-through to consumer prices cannot be accounted for by changes in the role of import prices alone.

*Italic: Source — _wp16240 - 2003. In Figure 3, Panel B reports the results for the two-year cumulative exchange rate pass-through to consumer prices of a one-percent innovation in the nominal effective exchange rate (PDF content).*

### 0.33 between the first and last estimation windows. During that period, our measure of

### _wp16240 - 0.33 between the first and last estimation windows. During that period, our measure of

### Decline in implied pass-through and decomposition
- Implied pass-through declined by only 0.01 in the referenced economies between the first and last estimation windows.
- Only about 5 percent of the documented decline in pass-through to consumer prices can be accounted for by changes in first-round effects; the remainder is associated with falling second-round effects.
- Average import content for each economy is reported for the estimation period of 2000 to 2015, with considerable heterogeneity across economies.

### Heterogeneity of second-round effects
- Baseline sample: 62 estimates of second-round effects at a horizon of 24 months.
  - 33 economies display positive second-round effects.
  - 29 economies display negative second-round effects.
- Subsample with import price data: 40 economies.
  - Distribution shifted to the right because pass-through to import prices is generally somewhat smaller than one.
  - Only three negative values; larger concentration of economies between 0 and 0.2.
- Note: Some point estimates for pass-through to import prices are slightly above one, but complete pass-through cannot be rejected in those cases.

### Two-stage testing strategy for monetary policy effects
- First stage: Time-varying estimates of exchange rate pass-through for each economy obtained by estimating equation (2) in rolling 12-year windows starting in January of each year since 1995.
- Second stage: Regress β̂0,i,τ12 on explanatory variables X i,τ measured as the average for the corresponding window τ and country i, with time fixed effects (ψτ) included:
  - β̂0,i,τ12 = θ X i,τ + ψτ + ε i,τ
- Weighted least squares used with weights equal to the inverse of the variance of the estimated pass-through coefficients from the first stage.
- Sample restricted to countries with data for all variables in X i,τ.

### Determinants of cumulative 12-month pass-through (Table 3 results, key points)
- Standard measures of price stability (mean and standard deviation of the inflation rate) are positively related to exchange rate pass-through.
  - A 1 percent increase in the mean rate of inflation is associated with a pass-through response that is 1.4 percent stronger.
- Median depreciation and standard deviation of the multilateral exchange rate are associated with higher exchange rate pass-through.
- Variability of consensus inflation forecasts at a 12-month horizon:
  - Where forecasts of inflation are more volatile, exchange rate pass-through is higher.
- Log disagreement among professional forecasters of inflation:
  - Strong and significant positive relationship with exchange rate pass-through (column 6).
  - In a specification including all explanatory variables, disagreement acts as a good summary measure for price stability’s role: conditional on first and second moments of nominal variables, lower forecast disagreement remains associated with smaller pass-through.
  - Economically: a decline in disagreement from the top to the bottom of the interquantile range is associated with a decline in pass-through of about 0.07 percentage points.
- Inclusion of country fixed effects (column 8) is conservative due to only nine overlapping estimation windows per country, but:
  - Variability of inflation and degree of forecast disagreement remain positively related to pass-through.
  - Coefficient on forecast disagreement remains large.

### Forecast disagreement as a monetary credibility measure
- Disagreement among professional forecasters is measured as the standard deviation across fixed-event forecasts combined into a synthetic 12-month fixed horizon (monthly frequency for 41 countries).
- Figure 8 documents a strong correlation between forecast disagreement and the average (squared) deviation of inflation from central bank targets between 2000 and 2015.
- Interpretation: better-anchored inflation expectations (lower disagreement) make firms less likely to pass along exchange rate movements to consumer prices.

### Second-round effects and their determinants (Ω̂i,τ)
- Second-round effects constructed as:
  - Ω̂i,τ = β̂0,i,τ12 − δ i,τ
  - where δ i,τ is the benchmark for expected first-round effects (computed as the average for the corresponding window τ and country i).
  - Benchmarks: import content of consumption and its product with country-specific estimates of pass-through to import prices.
- Regression specification for second-round effects:
  - Ω̂i,τ = θ X i,τ + ψτ + ε i,τ
- Using second-round effects constructed assuming complete pass-through to import prices:
  - Strong cross-sectional correlation between forecast disagreement and second-round effects (Figure 9).
  - Economically: a decline in disagreement from the top to bottom of the interquartile range is associated with a decline in pass-through of about 0.10 percentage point (about one-third of the second-round effects estimated at the largest decile).
- Using alternative benchmark with estimated pass-through to import prices (Table 5):
  - Main findings broadly unchanged when allowing for incomplete pass-through to import prices.
  - Coefficient on disagreement larger and remains strongly significant when all explanatory variables included and with country fixed effects.
  - Conclusion: for a given level and volatility of inflation and the exchange rate, more credible monetary policy substantially reduces exchange rate pass-through to consumer prices beyond the border.

### Concluding findings and policy implications
- Empirical decomposition separates pass-through into first-round (import pricing at the dock) and second-round (domestically-produced goods and services) components.
- Despite an increase in the import content of domestic consumption, a generalized fall in exchange rate pass-through occurred; modest declines in pass-through to import prices imply reductions in second-round effects largely explain the fall in overall pass-through.
- Monetary policy performance, including measures of price stability and forecast disagreement, is statistically significant and economically important in explaining cross-country and time variation in exchange-rate pass-through.
  - An increase in disagreement among professional forecasters of inflation from the 25th to the 75th percentiles is associated with an increase in the estimated pass-through coefficient by 0.1.
  - Average cumulative pass-through after two years is about 0.3 in the sample.
  - Monetary credibility exerts considerable influence over second-round effects.
- Policy implication: where central bank credibility has been established, it may be less costly for monetary policy to stabilize inflation and real activity while allowing the exchange rate to help the economy adjust to external shocks.

*Source: Authors’ calculations and analysis from the provided content.*

### REFERENCES

### _wp16240 - REFERENCES

### References cited
- Contains bibliographic entries on inflation dynamics, exchange rate pass-through, monetary policy, and related empirical and theoretical literature. Representative entries include:
  - Albagli, E., A. Naudon, and R. Vergara, 2015. “Inflation Dynamics in Latin America: Comparison with Global Trends and Implications for Monetary Policy,” Documentos de Política Económica 58, Banco Central de Chile: Santiago.
  - Bacchetta, P. and E. van Wincoop, 2002. “A Theory of the Currency Denomination of International Trade,” Journal of International Economics 67(2): 295-319.
  - Burstein, A., M. Eichenbaum, and S. Rebelo, 2005. “Large Devaluations and the Real Exchange Rate,” Journal of Political Economy 113(4): 742-784.
  - Cagan, P., 1956. “The Monetary Dynamics of Hyperinflation,” in Studies in the Quantity Theory of Money, ed. M. Friedman, University of Chicago Press: Chicago, IL.
  - Gopinath, G., 2015. “The International Price System,” Working paper 21646, National Bureau of Economic Research: Cambridge, MA.
  - Jasová, M., R. Moessner, and E. Takáts, 2016. “Exchange Rate Pass-Through: What Has Changed since the Crisis?” Working Paper 583, Bank for International Settlements: Basel.
- Topics span: exchange rate pass-through estimation and theory, inflation expectations, central bank governance and independence, methods for impulse response estimation, multi-region input-output databases, and empirical studies across industrialized and emerging markets.

### Annex — Table A1. Import Price Data Sources
- Purpose: Lists data sources used for import price series by economy and indicates whether local currency reporting is available.
- Format: Columns for Economy, Source, Local Currency.
- Sample entries (economy — source — local currency):
  - Austria — Haver Analytics / Eurostat — Yes
  - Brazil — Haver Analytics / FUNCEX — No
  - Canada — Haver Analytics / StatCan — Yes
  - Mexico — Haver Analytics / BMEX — No
  - Switzerland — Haver Analytics / SFS — No
  - United States — Haver Analytics / BLS — Yes

### Annex — Table A2. Estimates of Pass-through from Bilateral Exchange Rate
- Note: Beta represents the cumulative response of headline consumer prices to a 1-percent innovation in the nominal bilateral exchange rate vis-à-vis the U.S. dollar after 12 months and 24 months respectively, estimated with monthly data over 2000-2015. The import content corresponds to the share of household final consumption that is imported (including direct imports and the import content of domestically produced goods consumed locally) averaged over the same sample (2000-2015).
- Table structure: for each economy, reports:
  - Pass-through after 12 months: beta, s.e., p-value
  - Pass-through after 24 months: beta, s.e., p-value
  - Benchmark: Import Content
- Selected exact country estimates from the table (beta, s.e., p-value at 12 months; beta, s.e., p-value at 24 months; followed by Import Content where shown):
  - Argentina: 0.33, 0.13, 0.01; 0.30, 0.12, 0.01; Import Content: 0.11
  - Australia: 0.03, 0.03, 0.36; 0.04, 0.04, 0.26; Import Content: 0.16
  - Brazil: 0.10, 0.03, 0.00; 0.23, 0.05, 0.00; Import Content: 0.08
  - Canada: -0.03, 0.02, 0.20; 0.00, 0.04, 0.96; Import Content: 0.17
  - China: -0.05, 0.24, 0.82; -0.79, 0.69, 0.25; Import Content: 0.06
  - India: 0.08, 0.06, 0.14; 0.11, 0.14, 0.44; Import Content: 0.05
  - Indonesia: 0.17, 0.03, 0.00; 0.31, 0.06, 0.00; Import Content: 0.14
  - Japan: 0.05, 0.02, 0.04; 0.14, 0.05, 0.01; Import Content: 0.10
  - Malaysia: 0.01, 0.07, 0.91; -0.12, 0.11, 0.31; Import Content: 0.39
  - Mexico: 0.05, 0.03, 0.04; 0.07, 0.03, 0.04; Import Content: 0.12
  - Pakistan: 0.42, 0.27, 0.13; 0.76, 0.43, 0.08; Import Content: 0.04
  - Paraguay: 0.26, 0.05, 0.00; 0.46, 0.10, 0.00; Import Content: 0.28
  - Russia: 0.20, 0.08, 0.01; 0.54, 0.14, 0.00; Import Content: 0.14
  - Singapore: -0.30, 0.10, 0.00; -0.72, 0.19, 0.00; Import Content: 0.34
  - South Africa: 0.21, 0.04, 0.00; 0.29, 0.08, 0.00; Import Content: 0.15
  - Spain: 0.11, 0.03, 0.00; 0.07, 0.05, 0.18; Import Content: 0.16
  - Thailand: -0.17, 0.08, 0.05; -0.28, 0.09, 0.00; Import Content: 0.18
  - Turkey: 0.42, 0.06, 0.00; 0.61, 0.08, 0.00; Import Content: 0.15
  - Ukraine: 0.78, 0.14, 0.00; 0.21, 0.28, 0.47; Import Content: 0.23
  - United Kingdom: 0.07, 0.03, 0.02; 0.08, 0.08, 0.32; Import Content: 0.22
  - United States: Benchmark: (listed in table as the reference; Import Content column present)
- Notes on interpretation:
  - Beta is the cumulative effect on headline consumer prices of a 1-percent nominal bilateral exchange rate innovation vis-à-vis the U.S. dollar after 12 and 24 months.
  - Standard errors and p-values are reported alongside beta estimates.
  - Import Content is the average share of household final consumption that is imported (direct imports plus import content of domestically produced goods) over 2000-2015.

*Source: _wp16240 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16240.pdf_