## _wp15252

## Source details

**Canonical URL:** [_wp15252](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15252.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15252.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15252.pdf.json)

---

### II. THE REER ELASTICITY OF EXPORTS — Empirical Strategy
- Sample: 46 countries for the period 1996-2012.
- Objective: Investigate impact of real exchange rate changes on export volumes.
- Regression framework:
  - Panel specification based on cross-country literature (Freund and Pierola (2012); Eichengreen and Gupta (2013)).
  - Annual real export growth expressed as a function of annual real exchange rate growth.
  - Export data sources: OECD-WTO TiVA (total exports, goods plus services), UN-COMTRADE (merchandise and manufacturing exports).
  - Real Effective Exchange Rates data: IMF Information Notice System (INS); where INS not available, JP Morgan and national sources used.
  - Controls: country fixed effects (iδ), time fixed effects (tδ), lagged GDP, export-weighted GDP of the country’s main trading partners, and other commonly used controls.
  - Standard errors: clustered across cross-section (White cross-section covariance method).
- Regression specification:
  - itittiitit ControlREERExp
  - itExp denotes real export growth of country i at time t, itREER is growth in the real effective exchange rate, β captures effect of a depreciation on export growth, and itControl denotes controls.
- Data construction note:
  - Annual real exports growth time series obtained by transforming nominal USD exports to local currency using nominal exchange rates and deflating by Consumers Price Indices (CPI).
  - Rationale: CPI deflation commonly used in literature; export deflators may be noisy and subject to compositional effects due to commodity price volatility.

### II. THE REER ELASTICITY OF EXPORTS — REER Elasticity of Exports Over Time (Key findings)
- Dependent variables analyzed:
  - Panel A: growth of (real) manufacturing exports.
  - Panel B: growth of exports in goods.
  - Panel C: growth of total exports in goods and services.
- Elasticity estimates:
  - Panel A, manufacturing exports:
    - 1996-2003: -0.88
    - 2004-2012: -0.68
  - Panel B, goods exports:
    - 1996-2003: -0.94
    - 2004-2012: -0.88
  - Panel C, total exports:
    - 1996-2003: -0.91
    - 2004-2012: -0.78
- Overall magnitude:
  - Decline appears to be around 20 percent from 1996-2003 to 2004-2012, though only weak evidence that the estimated change is statistically significant.
- Robustness to outliers:
  - Omitting countries that experienced financial crises or hyperinflations in late 1990s (e.g., Bulgaria, Indonesia, Romania, Russia) does not change direction of results; increasing excluded countries narrows the decline.
- Sensitivity to deflators and data sources:
  - Using total export volumes from World Economic Outlook (IMF) database (not relying on CPI) yields smaller point estimates and a less steep decreasing pattern.
  - Deflating exports using Wholesale Price Indices (WPI) shows REER elasticity declines over time but less steeply than baseline (CPI).
  - Using US CPI to deflate nominal exports yields smaller manufacturing elasticities than Table 1 but still a declining pattern.
  - Deflating manufacturing export values with export deflators of goods (in USD) from WEO yields no statistically significant impact of REER changes on manufacturing export growth — possibly due to compositional effects and volatile commodity prices.

### II. THE REER ELASTICITY OF EXPORTS — Discussion (Timing and possible explanations)
- Rolling regressions (7-year windows) indicate:
  - Average REER elasticity gradually decreased since mid-1990s and remained almost flat since early 2000s.
  - Observed decline predates the global financial crisis; the crisis is one factor among others.
- Four classes of explanations for lower REER elasticity:
  1. Slower trade growth: trade growth expanded in the 1990s with production fragmentation into GVCs and decreased in the 2000s as fragmentation decelerated (Constantinescu et al., 2015).
  2. Trade policy: tariff barriers declined faster in the 1990s; temporary trade barriers (anti-dumping and countervailing duties) may be set in response to currency movements (Bown and Crowley (2012)).
  3. Declining exchange rate pass-through: high performing firms, large exporters, and exporters of high quality goods more likely to absorb exchange rate movements in mark-ups (Berman et al., 2012; Amiti et al., 2014; Chen and Juvenal, 2014); if such firms gain importance, export volumes respond less to depreciation.
  4. Other factors beyond the global financial crisis.
- Robustness notes:
  - Rolling regressions with 5, 6, and 8 year windows show similar results.
  - Contemporaneous IMF (2015a) work shows no statistically significant decline in REER elasticity of total export volumes in a sample of more than 50 economies over past three decades.

### III. GVCs AND THE REER ELASTICITY OF EXPORTS — Measures of GVC Integration
- Gross trade decomposition:
  - Gross exports = Domestic Value Added (DV) + Foreign Value Added (FV).
  - DV components: (1) exported in final goods, (2) exported in intermediates, (3) exported in intermediates re-exported to third countries, (4) exported in intermediates that return home.
- GVC measures at country-sector level (TiVA data):
  - GVC_Forward Linkages_i,k = IV_i,k / EXP_i,k
    - IV_i,k = domestic content of a country-sector in third-country exports.
    - EXP_i,k = country-sector gross exports.
  - GVC_Backward Linkages_i,k = FV_i,k / EXP_i,k
    - FV_i,k = foreign content (import content) in a country-sector’s gross exports.
  - GVC_Participation_i,k = IV_i,k / EXP_i,k + FV_i,k / EXP_i,k
    - Participation index: sum of foreign value added embodied in a country’s exports and indirect value-added exports expressed as a percentage of gross exports.
  - GVC_Position_i,k = ln (1+IV_i,k / EXP_i,k) – ln (1+FV_i,k / EXP_i,k)
    - Position index: positive for upstream specialization, negative for downstream specialization.
- Intuition for effects on REER elasticity:
  - In traditional fully domestic production, depreciation increases exports via expenditure switching.
  - With GVCs, backward linkages raise cost of imported intermediates when currency depreciates, lowering competitiveness gains.
  - Forward linkages: depreciation increases competitiveness of downstream producers, stimulating demand for their goods and potentially counterbalancing expenditure-switching.
  - Conjectures:
    - Backward and forward linkages are associated with lower exchange rate elasticity of exports.
    - Higher GVC participation lowers REER elasticity of exports.
    - GVC position sign is ambiguous.

### III. GVCs AND THE REER ELASTICITY OF EXPORTS — Empirical approach
- Data and construction:
  - Value-added exports data from OECD-WTO TiVA used to construct measures of GVC participation, backward and forward linkages, and GVC position for countries and manufacturing subsectors, from 1996 to 2012.
  - Where TiVA is published only for selected years, time-series interpolations from Duval et al. (2014) are used in some regressions.
- Regression strategies:
  1. Add interaction of REER with GVC variables to manufacturing exports growth regressions; use five-year average time series due to TiVA publication frequency.
  2. Increase cross-sectional variation by using manufacturing exports growth at the industry level and interacting REER with industry-level GVC participation.
  3. Consider value-added exports and alternative REER measures that correct for GVCs (REER in Goods Value Added, REER in Tasks Value Added) and compare elasticity estimates for value-added exports with original estimates.

### III. GVCs AND THE REER ELASTICITY OF EXPORTS — Manufacturing Exports (Key empirical finding)
- Main finding:
  - Higher participation in GVCs is related to lower elasticity of manufacturing exports to REER.
- Evidence:
  - Table 2 Columns (2) and (3) show countries with higher GVC participation present lower exchange rate elasticity of manufacturing exports for both annual data and five-year averages.
- Robustness:
  - Main conclusions are robust when using WPI instead of CPI as deflator to compute real exports.

### GVC participation and REER elasticity — core findings
- GVC participation lowers the elasticity of manufacturing exports to the REER.
- Five-year average sample of the second panel of Table 2:
  - Average participation in GVCs = 36.05.
  - Slope on the REER in column (2) of the second panel = -1.106.
  - Positive slope on the interaction term for participation in GVCs = 0.006.
  - Corrected elasticity = -1.106+0.006x36.05 = -0.87.
- Interpretation and comparative magnitudes:
  - A country with no participation in GVCs would have a REER elasticity of -1.1 (second panel, Table 2).
  - On average, GVC participation reduces REER elasticity by approximately 22 percent (-1.11+0.87/-1.11).
  - Using the 20th and 80th percentiles of participation in GVCs, the implied reduction in REER elasticity is between 13 percent and 28 percent (Figure 4).
  - Backward linkages (import-content of exports) explain the variation of the elasticity with GVC participation more than forward linkages (second panel, column (4), Table 2).

### Quantitative magnitudes and illustrative calculations
- Five-year average sample example:
  - REER slope (raw) = -1.106.
  - Interaction slope for participation = 0.006.
  - Average participation = 36.05.
  - Corrected elasticity = -1.106+0.006x36.05 = -0.87.
- Cross-panel summary:
  - Country with no GVC participation: REER elasticity = -1.1.
  - Average reduction due to GVCs = approximately 22 percent.
  - Reduction range using 20th and 80th percentiles = between 13 percent and 28 percent.
- Sector-average example (Table 3):
  - For average country-sector GVC participation of 4.5, predicted elasticity = -0.99.

### Sectoral analysis (industry-level results)
- Industry-level exports (1996-2011) confirm cross-country findings and show a stronger decreasing evolution of elasticity over time than total manufacturing exports.
- Elasticity observations:
  - First half of the sample (1996-2003) elasticity is higher than most recent years (2004-2011), though without statistical significance in some specifications.
- Robustness:
  - Sectoral results reinforce main findings and are considered evidence against endogeneity bias.

### Value-added exports and REER (value-added REER measures)
- Comparative regressions (Table 4) — selected coefficients (independent variable / dependent variable):
  - Total Exports/REER: -0.88***, -0.94***, -0.68***, 0.25
  - Value Added Exports/REER: -0.75***, -0.81***, -0.58***, 0.22
  - Value Added Exports / REER in Value Added Terms (tasks value added): -0.20***, -0.17*, -0.25***, -0.08
  - Value Added Exports / REER in Value Added Terms (goods value added): -0.27**, -0.23*, -0.32***, -0.09
- Two main results from value-added analysis:
  1. The value-added REER elasticity does not present the same decreasing pattern over time as the standard REER elasticity. This holds across alternative value-added REER measures.
  2. Elasticities measured in value-added terms are lower than corresponding elasticities measured using standard REER and gross exports, suggesting currency depreciations have substantially less impact when exports are measured in value-added terms.
- Caveat:
  - Value-added REER uses GDP deflator whereas gross REER uses CPI deflator. Comparisons using the same gross REER show value-added exports are less elastic than gross exports, indicating deflator differences are not the main driver.

### Conclusion and policy-relevant implications
- Main conclusions:
  - GVC participation lowers the responsiveness of manufacturing exports to REER changes by 22 percent on average and by close to 30 percent for countries with the highest GVC participation.
  - The REER elasticity of manufacturing exports has slightly declined over time in the study period, consistent with increasing GVC integration.
  - When REERs and exports are measured in value-added terms, estimated REER elasticities are substantially smaller and do not show a declining time pattern.
- Policy implications and suggested research directions:
  - Models and policy prescriptions that rely on exchange rate depreciations to stimulate exports should account for a country’s GVC integration.
  - Future research should explore alternative channels and dynamics, including:
    - Lags of depreciations due to GVC structure adjustment and industry-specific lag structures.
    - Pricing-to-market behavior conditional on market power and GVC roles.
    - The impact of exchange rate changes on countries’ ability to anchor to GVCs and capture cross-border production benefits.
- Practical implication:
  - Currency depreciations primarily impact the domestic value added embodied in gross exports; countries with large import-content of exports will see a weaker export response to depreciations.

*Source: _wp15252 - Section III focuses on the role of integration in GVCs and contains the main finding of the paper (excerpts from the provided PDF content).*

### Section III focuses on the role of integration in GVCs and contains the main finding of the paper and

### _wp15252 - Section III focuses on the role of integration in GVCs and contains the main finding of the paper and

### II. THE REER ELASTICITY OF EXPORTS — Empirical Strategy
- Sample: 46 countries for the period 1996-2012.
- Objective: Investigate impact of real exchange rate changes on export volumes.
- Regression framework:
  - Panel specification based on cross-country literature (Freund and Pierola (2012); Eichengreen and Gupta (2013)).
  - Annual real export growth expressed as a function of annual real exchange rate growth.
  - Export data sources: OECD-WTO TiVA (total exports, goods plus services), UN-COMTRADE (merchandise and manufacturing exports).
  - Real Effective Exchange Rates data: IMF Information Notice System (INS); where INS not available, JP Morgan and national sources used.
  - Controls: country fixed effects (iδ), time fixed effects (tδ), lagged GDP, export-weighted GDP of the country’s main trading partners, and other commonly used controls.
  - Standard errors: clustered across cross-section (White cross-section covariance method).
- Regression specification (as presented):
  - itittiitit ControlREERExp
  - where itExp denotes real export growth of country i at time t, itREER is growth in the real effective exchange rate, β captures effect of a depreciation on export growth, and itControl denotes controls.

- Data construction note:
  - Annual real exports growth time series obtained by transforming nominal USD exports to local currency using nominal exchange rates and deflating by Consumers Price Indices (CPI).
  - Rationale: CPI deflation commonly used in literature; export deflators may be noisy and subject to compositional effects due to commodity price volatility.

### II. THE REER ELASTICITY OF EXPORTS — REER Elasticity of Exports Over Time (Key findings)
- Dependent variables analyzed:
  - Panel A: growth of (real) manufacturing exports.
  - Panel B: growth of exports in goods.
  - Panel C: growth of total exports in goods and services.
- Focus rationale for manufacturing:
  - Manufacturing associated with cross-border supply chains driving the explanation for decline in REER elasticity.
  - Availability of sectoral and sub-sectoral GVC integration information from TiVA.
  - Narrower sectors reduce endogeneity concerns.
- Elasticity estimates (Panel A, manufacturing exports):
  - 1996-2003: -0.88
  - 2004-2012: -0.68
- Elasticity estimates (Panel B, goods exports):
  - 1996-2003: -0.94
  - 2004-2012: -0.88
- Elasticity estimates (Panel C, total exports):
  - 1996-2003: -0.91
  - 2004-2012: -0.78
- Overall magnitude:
  - Decline appears to be around 20 percent from 1996-2003 to 2004-2012, though only weak evidence that the estimated change is statistically significant.
- Robustness to outliers:
  - Omitting countries that experienced financial crises or hyperinflations in late 1990s (e.g., Bulgaria, Indonesia, Romania, Russia) does not change direction of results; increasing excluded countries narrows the decline.
- Sensitivity to deflators and data sources:
  - Using total export volumes from World Economic Outlook (IMF) database (not relying on CPI) yields smaller point estimates and a less steep decreasing pattern.
  - Deflating exports using Wholesale Price Indices (WPI) shows REER elasticity declines over time but less steeply than baseline (CPI).
  - Using US CPI to deflate nominal exports yields smaller manufacturing elasticities than Table 1 but still a declining pattern.
  - Deflating manufacturing export values with export deflators of goods (in USD) from WEO yields no statistically significant impact of REER changes on manufacturing export growth — possibly due to compositional effects and volatile commodity prices.

### II. THE REER ELASTICITY OF EXPORTS — Discussion (Timing and possible explanations)
- Rolling regressions (7-year windows) indicate:
  - Average REER elasticity gradually decreased since mid-1990s and remained almost flat since early 2000s.
  - Observed decline predates the global financial crisis; the crisis is one factor among others.
- Four classes of explanations for lower REER elasticity:
  1. Slower trade growth: trade growth expanded in the 1990s with production fragmentation into GVCs and decreased in the 2000s as fragmentation decelerated (Constantinescu et al., 2015).
  2. Trade policy: tariff barriers declined faster in the 1990s; temporary trade barriers (anti-dumping and countervailing duties) may be set in response to currency movements (Bown and Crowley (2012)).
  3. Declining exchange rate pass-through: high performing firms, large exporters, and exporters of high quality goods more likely to absorb exchange rate movements in mark-ups (Berman et al., 2012; Amiti et al., 2014; Chen and Juvenal, 2014); if such firms gain importance, export volumes respond less to depreciation.
  4. Other factors beyond the global financial crisis.
- Robustness notes:
  - Rolling regressions with 5, 6, and 8 year windows show similar results.
  - Contemporaneous IMF (2015a) work shows no statistically significant decline in REER elasticity of total export volumes in a sample of more than 50 economies over past three decades.

### III. GVCS AND THE REER ELASTICITY OF EXPORTS — Measures of GVC Integration
- Gross trade decomposition (Koopman et al. (2010) schematic reproduced in Figure 3):
  - Gross exports = Domestic Value Added (DV) + Foreign Value Added (FV).
  - DV components: (1) exported in final goods, (2) exported in intermediates, (3) exported in intermediates re-exported to third countries, (4) exported in intermediates that return home.
- GVC measures at country-sector level (using TiVA data):
  - GVC_Forward Linkages_i,k = IV_i,k / EXP_i,k
    - IV_i,k = domestic content of a country-sector in third-country exports (inputs produced domestically used in other countries’ exports).
    - EXP_i,k = country-sector gross exports.
  - GVC_Backward Linkages_i,k = FV_i,k / EXP_i,k
    - FV_i,k = foreign content (import content) in a country-sector’s gross exports.
  - GVC_Participation_i,k = IV_i,k / EXP_i,k + FV_i,k / EXP_i,k
    - Participation index: sum of foreign value added embodied in a country’s exports and indirect value-added exports expressed as a percentage of gross exports.
  - GVC_Position_i,k = ln (1+IV_i,k / EXP_i,k) – ln (1+FV_i,k / EXP_i,k)
    - Position index: positive for upstream specialization (high forward relative to backward linkages), negative for downstream specialization.
- Intuition for effects on REER elasticity:
  - In traditional fully domestic production, depreciation increases exports via expenditure switching.
  - With GVCs, backward linkages raise cost of imported intermediates when currency depreciates, lowering competitiveness gains.
  - Forward linkages: depreciation increases competitiveness of downstream producers, stimulating demand for their goods and potentially counterbalancing expenditure-switching.
  - Conjectures:
    - Backward and forward linkages are associated with lower exchange rate elasticity of exports.
    - Higher GVC participation lowers REER elasticity of exports.
    - GVC position sign is ambiguous (depends on whether upstream or downstream).

### III. GVCS AND THE REER ELASTICITY OF EXPORTS — GVCs and REER Elasticity (Empirical approach)
- Data and construction:
  - Value-added exports data from OECD-WTO TiVA used to construct measures of GVC participation, backward and forward linkages, and GVC position for countries and, when available, manufacturing subsectors, from 1996 to 2012.
  - Where TiVA is published only for selected years, time-series interpolations from Duval et al. (2014) are used in some regressions.
- Regression strategies (building on Equation (1)):
  1. Add interaction of REER with GVC variables to manufacturing exports growth regressions; use five-year average time series due to TiVA publication frequency.
  2. Increase cross-sectional variation by using manufacturing exports growth at the industry level and interacting REER with industry-level GVC participation.
  3. Consider value-added exports and alternative REER measures that correct for GVCs (REER in Goods Value Added, REER in Tasks Value Added (Bayoumi et al., 2013)) and compare elasticity estimates for value-added exports with original estimates.

### III. GVCS AND THE REER ELASTICITY OF EXPORTS — Manufacturing Exports (Key empirical finding)
- Main finding:
  - Higher participation in GVCs is related to lower elasticity of manufacturing exports to REER.
- Evidence:
  - Table 2 Columns (2) and (3) show countries with higher GVC participation present lower exchange rate elasticity of manufacturing exports for both annual data and five-year averages.
- Robustness:
  - Main conclusions are robust when using WPI instead of CPI as deflator to compute real exports.

*Source: _wp15252 - Section III focuses on the role of integration in GVCs and contains the main finding of the paper and (excerpts from the provided PDF content).*

### 36.05 in the five-year average sample of the second panel of Table 2 predicts an elasticity of -0.87

### _wp15252 - 36.05 in the five-year average sample of the second panel of Table 2 predicts an elasticity of -0.87

### GVC participation and REER elasticity — core findings
- GVC participation lowers the elasticity of manufacturing exports to the REER.
- In the five-year average sample of the second panel of Table 2:
  - Average participation in GVCs = 36.05.
  - Using the slope on the REER in column (2) of the second panel, -1.106, and the positive slope on the interaction term for participation in GVCs, 0.006, the corrected elasticity is:
    - -1.106+0.006x36.05 = -0.87.
- A country with no participation in GVCs would have a REER elasticity of -1.1 (second panel, Table 2).
- On average, GVC participation reduces REER elasticity by approximately 22 percent (-1.11+0.87/-1.11).
- Using the 20th and 80th percentiles of participation in GVCs, the implied reduction in REER elasticity is between 13 percent and 28 percent (Figure 4).
- Backward linkages (import-content of exports) explain the variation of the elasticity with GVC participation more than forward linkages (second panel, column (4), Table 2).

### Quantitative magnitudes and illustrative calculations
- Five-year average sample example:
  - REER slope (raw) = -1.106.
  - Interaction slope for participation = 0.006.
  - Average participation = 36.05.
  - Corrected elasticity = -1.106+0.006x36.05 = -0.87.
- Cross-panel summary:
  - Country with no GVC participation: REER elasticity = -1.1.
  - Average reduction due to GVCs = approximately 22 percent.
  - Reduction range using 20th and 80th percentiles = between 13 percent and 28 percent.
- Sector-average example (Table 3):
  - For average country-sector GVC participation of 4.5, predicted elasticity = -0.99 (close to -1.0 from column (1)).

### Sectoral analysis (industry-level results)
- Industry-level exports (1996-2011) confirm cross-country findings and show a stronger decreasing evolution of elasticity over time than total manufacturing exports.
- Elasticity observations:
  - First half of the sample (1996-2003) elasticity is higher than most recent years (2004-2011), though without statistical significance in some specifications.
- Robustness:
  - Sectoral results reinforce main findings and are considered evidence against endogeneity bias (unlikely that developments in a particular industry would shift currency movements).

### Value-added exports and REER (value-added REER measures)
- Comparative regressions (Table 4) — key coefficients (independent variable / dependent variable):
  - Total Exports/REER: -0.88***, -0.94***, -0.68***, 0.25
  - Value Added Exports/REER: -0.75***, -0.81***, -0.58***, 0.22
  - Value Added Exports / REER in Value Added Terms (tasks value added): -0.20***, -0.17*, -0.25***, -0.08
  - Value Added Exports / REER in Value Added Terms (goods value added): -0.27**, -0.23*, -0.32***, -0.09
- Two main results from value-added analysis:
  - The value-added REER elasticity does not present the same decreasing pattern over time as the standard REER elasticity. This holds across alternative value-added REER measures.
  - Elasticities measured in value-added terms are lower than corresponding elasticities measured using standard REER and gross exports, suggesting currency depreciations have substantially less impact when exports are measured in value-added terms.
- Caveat:
  - Value-added REER uses GDP deflator whereas gross REER uses CPI deflator. However, comparisons using the same gross REER (first and second rows of Table 4) show value-added exports are less elastic than gross exports, indicating deflator differences are not the main driver of the finding.

### Conclusion and policy-relevant implications
- Main conclusions:
  - GVC participation lowers the responsiveness of manufacturing exports to REER changes by 22 percent on average and by close to 30 percent for countries with the highest GVC participation.
  - The REER elasticity of manufacturing exports has slightly declined over time in the study period, consistent with increasing GVC integration.
  - When REERs and exports are measured in value-added terms, estimated REER elasticities are substantially smaller and do not show a declining time pattern.
- Policy implications and suggested research directions:
  - Models and policy prescriptions that rely on exchange rate depreciations to stimulate exports should account for a country’s GVC integration; otherwise, policy predictions may be inaccurate.
  - Future research should explore alternative channels and dynamics, including:
    - Lags of depreciations due to GVC structure adjustment and industry-specific lag structures.
    - Pricing-to-market behavior conditional on market power and GVC roles.
    - The impact of exchange rate changes on countries’ ability to anchor to GVCs and capture cross-border production benefits (e.g., knowledge spillovers), noting GVC integration may be endogenous to macroeconomic conditions.
- Practical implication:
  - Currency depreciations primarily impact the domestic value added embodied in gross exports; countries with large import-content of exports will see a weaker export response to depreciations.

*Source: Excerpt from the IMF working paper content provided in the supplied PDF section.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15252.pdf_