## 1.5 percent of GDP increase in real net exports, with substantial cross-country variation

## Source details

**Canonical URL:** [1.5 percent of GDP increase in real net exports, with substantial cross-country variation](https://www.imf.org/-/media/files/publications/wp/2017/wp1758.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2017/wp1758.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2017/wp1758.pdf.json)

---

### Main findings
- A 10 percent real effective depreciation in an economy’s currency is associated with, on average, a 1.5 percent of GDP increase in real net exports, with substantial cross-country variation around this average.
- Most of the response of trade to exchange rate movements materializes within the first year.
- There is little evidence that exchange rates and trade have disconnected over time; rolling estimations, large-depreciation event analysis, and sector-level Chow tests all point to broad stability in trade elasticities.
- There is no consistent evidence that GVC expansion has substantially weakened trade elasticities at the macroeconomic or cross-sector average level.
- Tighter financial conditions and lower economic slack reduce the response of export volume to exchange rates.
- The choice of deflator for trade volume significantly affects results: the exchange rate pass-through into consumer prices is much weaker than into border prices, and CPI-deflated trade volume is a biased measure of true volume.

### Empirical estimates and key statistics
- Sample and coverage:
  - Sample includes 60 economies: 23 AEs and 37 EMDEs.
  - Depending on the economy, the sample period starts between 1980 and 1989 and ends in 2014.
- Exchange rate pass-through (macro, long-term averages):
  - A 10 percent exchange rate depreciation in real effective terms reduces export prices in local currency by 5.5 percent.
  - A 10 percent exchange rate depreciation in real effective terms increases import prices in local currency by 6.1 percent.
  - Most of these pass-through effects materialize within one year.
  - Corresponding response of export prices in domestic currency to a 10 percent REER depreciation would be a rise of 4.5 percent (–10 × (0.552 – 1)).
- Price elasticity of trade volumes (macro, long-term averages):
  - A 10 percent increase in export prices reduces export volumes by about 3 percent in the long term.
  - A 10 percent increase in import prices reduces import volumes by about 3 percent in the long term.
- Marshall-Lerner condition:
  - The average estimated elasticities satisfy the Marshall-Lerner condition under incomplete pass-through.
- Global value chain (GVC) context:
  - The foreign value content of exports currently averages only about 25 percent across economies.
  - The share of foreign value added in world exports rose from about 15 percent in the 1970s to about 25 percent in 2013.
- Rolling estimation windows:
  - Rolling 10-year intervals from 1990–1999 up to 2005–14 were used to test stability.

### Large currency depreciation episodes (event analysis)
- Identification and sample:
  - Large depreciation defined via two criteria: (i) nominal depreciation threshold at the 90th percentile of annual depreciations (separate thresholds for AEs and EMDEs), and (ii) unusually large change relative to previous year (90th percentile of changes).
  - Thresholds: for the first criterion the AE threshold is a depreciation of 13 percent vis-à-vis the dollar; for EMDEs the threshold is 20 percent. For the second criterion AE and EMDE thresholds are both about 13 percentage points.
  - Episodes excluded if annual inflation above 30 percent or associated with banking crises (banking-crisis association defined as occurring within three years of a crisis).
  - Applying the methodology to 1980–2014 yields 66 large depreciation episodes; 17 episodes occurred in AEs.
- Average dynamics following identified large depreciations (baseline, non-banking-crisis episodes):
  - The real effective exchange rate declines on average 25 percent over five years during identified depreciation episodes.
  - Export prices in foreign currency fall by about 10 percent, with much of the adjustment occurring in the first year.
  - Export volume increases gradually with a cumulated change of about 10 percent over five years.
  - Implied average price elasticity of exports from these episodes is about 0.7 (higher than macro estimate of about 0.3).
- Subsample comparisons over time:
  - Splitting the 66 episodes into earlier (33 episodes in 1997 or earlier) and later (33 episodes after 1997) shows that exchange rate and export volume responses are very similar across subsamples; export price responses somewhat weaker in the first two years after 1997 but similar in the long term.
- Role of initial conditions and financial health:
  - Episodes with initial economic slack (demeaned real GDP growth below the median) show larger export gains—about an additional 7 percentage points after five years compared with the baseline.
  - Depreciations associated with banking crises show much weaker export responses: estimated elasticity of export prices to REER of about 0.25 (about half the baseline), and the response of real exports is near zero.

### Sector-level analysis and GVCs
- Data and methods:
  - Sector-level dataset built from UN Comtrade and OECD TiVA covering 46 countries, period 1995–2011.
  - Sector-level prices constructed as GEKS indexes; sector-level exchange rates constructed with a Tornqvist method that uses export destination shares.
  - Sector-level regressions estimated in reduced-form (effects of REER on volumes) to mitigate measurement-noise bias in price indices.
- Main sector-level results:
  - For three out of four trade elasticities, there is no rejection of stability over time; some weakening observed only for import volumes.
  - No consistent or economically meaningful evidence that higher GVC participation (foreign value-added shares) systematically weakens exchange rate effects on trade at the country-sector average level.
  - Some specifications indicate that sectors with higher foreign value added in exports are associated with weaker export-volume responses, but effects are inconsistent across import/export side and model specifications.
  - Little evidence that industries with a larger share of a country’s exports have smaller trade elasticities; reverse causality appears to play an insignificant role in biasing macro-level elasticity estimates.

### Tables and selected numeric estimates (from macro and sector tables)
- Table 4 (simple average of individual economy estimates for 60 economies, 1980–2014; based on Producer Price Index):
  - Export prices (Long-Term): 0.552
  - Import prices (Long-Term): –0.605
  - Exports price elasticity of volumes (Long-Term): –0.321
  - Imports price elasticity of volumes (Long-Term): –0.298
  - Export prices (One-Year Effect): 0.625
  - Import prices (One-Year Effect): –0.580
  - Exports price elasticity of volumes (One-Year Effect): –0.260
  - Imports price elasticity of volumes (One-Year Effect): –0.258
- Table 4 (based on Consumer Price Index):
  - Export prices (Long-Term): 0.457
  - Import prices (Long-Term): –0.608
  - Exports price elasticity of volumes (Long-Term): –0.328
  - Imports price elasticity of volumes (Long-Term): –0.333
  - Export prices (One-Year Effect): 0.599
  - Import prices (One-Year Effect): –0.546
  - Exports price elasticity of volumes (One-Year Effect): –0.200
  - Imports price elasticity of volumes (One-Year Effect): –0.200
- Memorandum — Non–Commodity Exporters:
  - Long-Term Elasticity (Export prices): 0.571
  - Long-Term Elasticity (Import prices): –0.582
  - Exports price elasticity of volumes (Long-Term): –0.461
  - Imports price elasticity of volumes (Long-Term): –0.272
- Table 5 excerpts (Pass-through into export prices, full sample and subperiods):
  - Pass-Through into Export Prices (All Countries): Full 0.569*** ; 1990–2001 0.557*** ; 2002–2014 0.457***
  - Pass-Through into Import Prices (All Countries): Full -0.612*** ; 1990–2001 -0.549*** ; 2002–2014 -0.632***
  - Price Elasticities of Exports (All Countries): Full -0.207*** ; 1990–2001 -0.147*** ; 2002–2014 -0.255***
- Sector-level regression highlights (Table 7 and Table 8):
  - Sector-level export-price pass-through: average ∆ln(Exchange rate) coefficients 0.360***, 0.297***, 0.402*** (std. errs: (0.040), (0.050), (0.066)).
  - Sector-level import-price pass-through: average ∆ln(Exchange rate) coefficients -0.717***, -0.633***, -0.685*** (std. errs: (0.026), (0.048), (0.062)).
  - Sector-level export-volume coefficients: average ∆ln(Exchange rate) coefficients -0.244***, 0.217, 0.221 (std. errs: (0.124), (0.152), (0.184)).
  - Sector-level import-volume coefficients: average ∆ln(Exchange rate) coefficients 1.147***, 1.423***, 1.829*** (std. errs: (0.107), (0.147), (0.178)).

### Caveats and robustness considerations
- Endogeneity and aggregation biases:
  - Reverse causality from trade volume to trade prices and heterogeneity of demand elasticities across goods can bias macro estimates, typically downward.
  - Such biases matter primarily if their underlying factors exhibit trends; robustness checks include large-depreciation episode analysis and sector-level estimations.
- Deflator choice:
  - CPI-deflated trade volume is biased because CPI pass-through is weaker than border prices and CPI includes many non-traded goods and services.
- Sensitivity and robustness tests:
  - Results robust to alternate estimation methods, alternative episode definitions, and exclusion of outliers during high inflation episodes where approximate growth rates are inaccurate.
  - Reduced-form replications (Ahmed, Appendino, and Ruta sample) show apparent declines in elasticities are sensitive to excluding high-inflation observations, robust regression, growth-rate definition, and choice of export deflators.
  - Table B1 robustness summary (Full sample 1996–2012, Elasticity of Total Exports to REER):
    - Baseline: -1.046***
    - CPI inflation < 30%: -0.776***
    - Robust regression: -0.714***
    - Log linear specification: -0.734***
    - Export prices (not CPI): -0.180***
    - No domestic real GDP control: -0.177***

### Policy implications
- Traditional policy considerations that rely on the exchange rate–trade relationship remain justifiable: exchange rate changes continue to have strong effects on export and import prices and on trade volumes.
- Flexible exchange rates still play a significant role in facilitating the resolution of trade imbalances.
- Exchange rate movements have macroeconomic significance not only through elasticities but also because shares of exports and imports in GDP have risen; even marginal declines in elasticities can coincide with equal or greater macroeconomic effects.
- When GVC participation has weakened exchange rate effects on trade for specific economies or sectors, larger exchange rate adjustments may be required to resolve trade imbalances in those cases.
- The transmission of exchange rate changes to inflation and monetary policy is relevant given continued pass-through into trade prices; the CPI understates border-price pass-through and thus can mislead econometric analyses if used as trade deflator.

*Source: IMF Working Paper (wp1758) — “1.5 percent of GDP increase in real net exports, with substantial cross-country variation”*

### 1.5 percent of GDP increase in real net exports, with substantial cross-country variation

### 1.5 percent of GDP increase in real net exports, with substantial cross-country variation

### Main findings
- A 10 percent real effective depreciation in an economy’s currency is associated with, on average, a 1.5 percent of GDP increase in real net exports, with substantial cross-country variation around this average.
- Most of the response of trade to exchange rate movements materializes within the first year.
- There is little evidence that exchange rates and trade have disconnected over time; rolling estimations, large-depreciation event analysis, and sector-level Chow tests all point to broad stability in trade elasticities.
- There is no consistent evidence that GVC expansion has substantially weakened trade elasticities at the macroeconomic or cross-sector average level.
- Tighter financial conditions and lower economic slack reduce the response of export volume to exchange rates.
- The choice of deflator for trade volume significantly affects results: the exchange rate pass-through into consumer prices is much weaker than into border prices, and CPI-deflated trade volume is a biased measure of true volume.

### Empirical estimates and key statistics
- Sample and coverage:
  - Sample includes 60 economies: 23 AEs and 37 EMDEs.
  - Depending on the economy, the sample period starts between 1980 and 1989 and ends in 2014.
- Exchange rate pass-through (macro, long-term averages):
  - A 10 percent exchange rate depreciation in real effective terms reduces export prices in local currency by 5.5 percent.
  - A 10 percent exchange rate depreciation in real effective terms increases import prices in local currency by 6.1 percent.
  - Most of these pass-through effects materialize within one year.
  - Corresponding response of export prices in domestic currency to a 10 percent REER depreciation would be a rise of 4.5 percent (–10 × (0.552 – 1)).
- Price elasticity of trade volumes (macro, long-term averages):
  - A 10 percent increase in export prices reduces export volumes by about 3 percent in the long term.
  - A 10 percent increase in import prices reduces import volumes by about 3 percent in the long term.
- Marshall-Lerner condition:
  - The average estimated elasticities satisfy the Marshall-Lerner condition under incomplete pass-through.
- Global value chain (GVC) context:
  - The foreign value content of exports currently averages only about 25 percent across economies.
  - The share of foreign value added in world exports rose from about 15 percent in the 1970s to about 25 percent in 2013.
- Rolling estimation windows:
  - Rolling 10-year intervals from 1990–1999 up to 2005–14 were used to test stability.

### Large currency depreciation episodes (event analysis)
- Identification and sample:
  - Large depreciation defined via two criteria: (i) nominal depreciation threshold at the 90th percentile of annual depreciations (separate thresholds for AEs and EMDEs), and (ii) unusually large change relative to previous year (90th percentile of changes).
  - Thresholds: for the first criterion the AE threshold is a depreciation of 13 percent vis-à-vis the dollar; for EMDEs the threshold is 20 percent. For the second criterion AE and EMDE thresholds are both about 13 percentage points.
  - Episodes excluded if annual inflation above 30 percent or associated with banking crises (banking-crisis association defined as occurring within three years of a crisis).
  - Applying the methodology to 1980–2014 yields 66 large depreciation episodes; 17 episodes occurred in AEs.
- Average dynamics following identified large depreciations (baseline, non-banking-crisis episodes):
  - The real effective exchange rate declines on average 25 percent over five years during identified depreciation episodes.
  - Export prices in foreign currency fall by about 10 percent, with much of the adjustment occurring in the first year.
  - Export volume increases gradually with a cumulated change of about 10 percent over five years.
  - Implied average price elasticity of exports from these episodes is about 0.7 (higher than macro estimate of about 0.3).
- Subsample comparisons over time:
  - Splitting the 66 episodes into earlier (33 episodes in 1997 or earlier) and later (33 episodes after 1997) shows that exchange rate and export volume responses are very similar across subsamples; export price responses somewhat weaker in the first two years after 1997 but similar in the long term.
- Role of initial conditions and financial health:
  - Episodes with initial economic slack (demeaned real GDP growth below the median) show larger export gains—about an additional 7 percentage points after five years compared with the baseline.
  - Depreciations associated with banking crises show much weaker export responses: estimated elasticity of export prices to REER of about 0.25 (about half the baseline), and the response of real exports is near zero.

### Sector-level analysis and GVCs
- Data and methods:
  - Sector-level dataset built from UN Comtrade and OECD TiVA covering 46 countries, period 1995–2011.
  - Sector-level prices constructed as GEKS indexes; sector-level exchange rates constructed with a Tornqvist method that uses export destination shares.
  - Sector-level regressions estimated in reduced-form (effects of REER on volumes) to mitigate measurement-noise bias in price indices.
- Main sector-level results:
  - For three out of four trade elasticities, there is no rejection of stability over time; some weakening observed only for import volumes.
  - No consistent or economically meaningful evidence that higher GVC participation (foreign value-added shares) systematically weakens exchange rate effects on trade at the country-sector average level.
  - Some specifications indicate that sectors with higher foreign value added in exports are associated with weaker export-volume responses, but effects are inconsistent across import/export side and model specifications.
  - Little evidence that industries with a larger share of a country’s exports have smaller trade elasticities; reverse causality appears to play an insignificant role in biasing macro-level elasticity estimates.

### Caveats and robustness considerations
- Endogeneity and aggregation biases:
  - Reverse causality from trade volume to trade prices and heterogeneity of demand elasticities across goods can bias macro estimates, typically downward.
  - Such biases matter primarily if their underlying factors exhibit trends; robustness checks include large-depreciation episode analysis and sector-level estimations.
- Deflator choice:
  - CPI-deflated trade volume is biased because CPI pass-through is weaker than border prices and CPI includes many non-traded goods and services.
- Sensitivity:
  - Results robust to alternate estimation methods, alternative episode definitions, and exclusion of outliers during high inflation episodes where approximate growth rates are inaccurate.

### Policy implications
- Traditional policy considerations that rely on the exchange rate–trade relationship remain justifiable: exchange rate changes continue to have strong effects on export and import prices and on trade volumes.
- Flexible exchange rates still play a significant role in facilitating the resolution of trade imbalances.
- Exchange rate movements have macroeconomic significance not only through elasticities but also because shares of exports and imports in GDP have risen; even marginal declines in elasticities can coincide with equal or greater macroeconomic effects.
- When GVC participation has weakened exchange rate effects on trade for specific economies or sectors, larger exchange rate adjustments may be required to resolve trade imbalances in those cases.
- The transmission of exchange rate changes to inflation and monetary policy is relevant given continued pass-through into trade prices; the CPI understates border-price pass-through and thus can mislead econometric analyses if used as trade deflator.

*Source: IMF Working Paper (wp1758) — “1.5 percent of GDP increase in real net exports, with substantial cross-country variation”*

### REFERENCES

### REFERENCES

### Bibliographic compilation
- Comprehensive list of academic and policy works cited, including articles, working papers, book chapters, and IMF staff reports covering topics such as:
  - Exchange rate pass-through, trade elasticities, and pricing-to-market (examples: Atkeson and Burstein 2008; Gopinath, Itskhoki, and Rigobon 2010; Campa and Goldberg 2005, 2008).
  - Global value chains and their role in trade responsiveness to exchange rates (examples: Ahmed, Appendino, and Ruta 2015, 2016; Johnson and Noguera 2012; Constantinescu, Mattoo, and Ruta 2015).
  - Trade collapse, trade finance, and banking/crisis effects on trade (examples: Ahn, Amiti, and Weinstein 2011; Ronci 2004; Iacovone and Zavacka 2009).
  - Empirical methods and data issues relevant to price indices and panel estimation (examples: Feenstra 1994; Ivancic, Diewert, and Fox 2011; Pedroni 2004; Pesaran and Smith 1995).
- Includes IMF institutional publications and country reports addressing trade elasticities, global value chains, and exchange-rate–trade interactions (examples: IMF 2015a–2015e; IMF WEO chapters referenced).

### Notable methodological and data sources cited
- Methods: Modern robust regression, local projections for impulse responses, panel cointegration, heterogeneous panel long-run estimation (references include Andersen 2008; Jordà 2005; Pedroni 2004; Pesaran and Smith 1995).
- Data sources used in the analysis (as reported in Table 3):
  - Export Prices: IMF staff calculations using export value divided by export volume
  - Export Volume: IMF, World Economic Outlook database
  - Export Value: IMF, World Economic Outlook database
  - Import Prices: IMF staff calculations using import value divided by import volume
  - Import Volume: IMF, World Economic Outlook database
  - Import Value: IMF, World Economic Outlook database
  - International Commodity Price Index: IMF, Global Assumptions database
  - International Energy Price Index: IMF, Global Assumptions database
  - Nominal Effective Exchange Rate: IMF, Information Notice System
  - Nominal GDP: IMF, World Economic Outlook database
  - Real Effective Exchange Rate: IMF, Information Notice System
  - Real GDP: IMF, World Economic Outlook database
  - Trade-Weighted Foreign CPI: IMF staff calculations
  - Trade-Weighted Foreign Demand: IMF, Global Economic Environment database
  - Trade-Weighted Foreign PPI: IMF staff calculations
  - Unit Labor Cost: OECD Economic Outlook; U.S. Bureau of Labor Statistics; and IMF staff calculations (IMF staff calculations use data from Haver Analytics; International Labor Organization; IMF, World Economic Outlook database; and IMF, International Financial Statistics)
  - Backward Participation and Forward Participation: OECD–WTO, Trade in Value Added database
- Note: CPI = consumer price index; PPI = producer price index.

### Tables and key quantitative results (selected)
- Table 4. Exchange Rate Pass-Through and Price Elasticities (simple average of individual economy estimates for 60 economies during 1980–2014):
  - Based on Producer Price Index:
    - Export prices (Long-Term): 0.552
    - Import prices (Long-Term): –0.605
    - Exports price elasticity of volumes (Long-Term): –0.321
    - Imports price elasticity of volumes (Long-Term): –0.298
    - Marshall-Lerner Condition Satisfied? Yes
    - Export prices (One-Year Effect): 0.625
    - Import prices (One-Year Effect): –0.580
    - Exports price elasticity of volumes (One-Year Effect): –0.260
    - Imports price elasticity of volumes (One-Year Effect): –0.258
    - Marshall-Lerner Condition Satisfied? Yes
  - Based on Consumer Price Index:
    - Export prices (Long-Term): 0.457
    - Import prices (Long-Term): –0.608
    - Exports price elasticity of volumes (Long-Term): –0.328
    - Imports price elasticity of volumes (Long-Term): –0.333
    - Marshall-Lerner Condition Satisfied? Yes
    - Export prices (One-Year Effect): 0.599
    - Import prices (One-Year Effect): –0.546
    - Exports price elasticity of volumes (One-Year Effect): –0.200
    - Imports price elasticity of volumes (One-Year Effect): –0.200
    - Marshall-Lerner Condition Satisfied? Yes
  - Memorandum — Non–Commodity Exporters (excludes economies where primary products > 50 percent of exports, on average, 2009–2013):
    - Long-Term Elasticity (Export prices): 0.571
    - Long-Term Elasticity (Import prices): –0.582
    - Exports price elasticity of volumes (Long-Term): –0.461
    - Imports price elasticity of volumes (Long-Term): –0.272
    - Marshall-Lerner Condition Satisfied? Yes
  - Source: IMF staff estimates.

- Table 5. Trade Elasticities over Time: Stability Tests (Full sample and subperiods 1990–2001 and 2002–2014; stars denote statistical significance):
  1. Pass-Through into Export Prices (All Countries)
    - Full 0.569*** ; 1990–2001 0.557*** ; 2002–2014 0.457***
  - By Region — Asia:
    - Full 0.429*** ; 1990–2001 0.419*** ; 2002–2014 0.346***
  - By Region — Europe:
    - Full 0.658*** ; 1990–2001 0.647*** ; 2002–2014 0.687***
  - By Integration into Global Value Chains — Countries with Larger Increase:
    - Full 0.572*** ; 1990–2001 0.560*** ; 2002–2014 0.548***
  - By Integration into Global Value Chains — Countries with Smaller Increase:
    - Full 0.684*** ; 1990–2001 0.608*** ; 2002–2014 0.609***
  2. Pass-Through into Import Prices (All Countries)
    - Full -0.612*** ; 1990–2001 -0.549*** ; 2002–2014 -0.632***
  - By Region — Asia:
    - Full -0.671*** ; 1990–2001 -0.684*** ; 2002–2014 -0.668***
  - By Region — Europe:
    - Full -0.553*** ; 1990–2001 -0.528*** ; 2002–2014 -0.587***
  - By Integration into Global Value Chains — Countries with Larger Increase:
    - Full -0.621*** ; 1990–2001 -0.545*** ; 2002–2014 -0.618***
  - By Integration into Global Value Chains — Countries with Smaller Increase:
    - Full -0.650*** ; 1990–2001 -0.511*** ; 2002–2014 -0.720*** **
  3. Price Elasticities of Exports (All Countries)
    - Full -0.207*** ; 1990–2001 -0.147*** ; 2002–2014 -0.255*** *
  - By Region — Asia:
    - Full -0.329*** ; 1990–2001 -0.265*** ; 2002–2014 -0.489*** **
  - By Region — Europe:
    - Full -0.281*** ; 1990–2001 -0.303** ; 2002–2014 -0.375***
  - By Integration into Global Value Chains — Countries with Larger Increase:
    - Full -0.305*** ; 1990–2001 -0.343** ; 2002–2014 -0.373***
  - By Integration into Global Value Chains — Countries with Smaller Increase:
    - Full -0.402*** ; 1990–2001 -0.225 ; 2002–2014 -0.566*** *
- Notes associated with tables:
  - Table 4 sample: 60 economies, 1980–2014.
  - The formula for the Marshall-Lerner condition adjusted for imperfect pass-through is (–ERPT of P_X) (1 + price elasticity of X) + (ERPT of P_M) (1 + price elasticity of M) + 1 > 0, with X = exports, M = imports (Annex 3.3).
  - Table 3 provides detailed data-source mapping used in the empirical analysis.

### Coverage of economies (tables listing)
- Table 1: Economies Included in the Panel Estimation of Trade Elasticities — full list includes 90+ economies such as Albania, Algeria, Argentina, Australia, Austria, Bangladesh, Belgium, Bolivia, Brazil, Bulgaria, Cambodia, Canada, Chile, China, Colombia, Republic of Congo, Costa Rica, Côte d’Ivoire, Croatia, Czech Republic, Denmark, Dominican Republic, Ecuador, Egypt, El Salvador, Estonia, Finland, France, Germany, Ghana, Greece, Guatemala, Honduras, Hungary, India, Indonesia, Iran, Israel, Italy, Japan, Jordan, Kenya, Korea, Kuwait, Kyrgyz Republic, Latvia, Lebanon, Libya, FYR Macedonia, Malaysia, Mexico, Morocco, Netherlands, New Zealand, Nicaragua, Nigeria, Norway, Oman, Pakistan, Paraguay, Peru, Philippines, Poland, Portugal, Romania, Russia, Saudi Arabia, Slovak Republic, Slovenia, South Africa, Spain, Sri Lanka, Sweden, Switzerland, Syria, Taiwan Province of China, Thailand, Trinidad and Tobago, Tunisia, Turkey, Ukraine, United Arab Emirates, United Kingdom, United States, Uruguay, Venezuela, Yemen, Zambia.
- Table 2: Economies Included in the Individual Estimation of Trade Elasticities — split between Advanced Economies (Australia; Austria; Belgium; Canada; Denmark; Finland; France; Germany; Greece; Israel; Italy; Japan; Korea; Netherlands; New Zealand; Norway; Portugal; Spain; Sweden; Switzerland; Taiwan Province of China; United Kingdom; United States) and Emerging Market Economies (Algeria*; Argentina; Bangladesh; Bolivia*; Bulgaria; Chile*; China; Colombia*; Republic of Congo*; Costa Rica; Côte d'Ivoire*; Egypt; El Salvador; Guatemala; Honduras; Hungary; India; Indonesia; Islamic Republic of Iran*; Jordan; Kenya; Kuwait*; Malaysia; Mexico; Morocco; Nigeria*; Pakistan; Paraguay*; Philippines; Saudi Arabia*; South Africa*; Sri Lanka; Thailand; Trinidad and Tobago*; Tunisia; United Arab Emirates; Venezuela*).  
  - * Denotes commodity exporters (primary products > 50 percent of total exports, on average, 2009–2013).

*Source: wp1758 - REFERENCES (IMF staff compilation and tables).*

### 4. Price Elasticities of Imports

### 4. Price Elasticities of Imports

### Cross-country and regional price elasticities
- All Countries: -0.433***, -0.452***, -0.335***  
- Asia: -0.436***, -0.566***, -0.233  
- Europe: -0.470***, -0.484***, -0.446***  
- By Integration into Global Value Chains:
  - Countries with Larger Increase: -0.521***, -0.658***, -0.271** **  
  - Countries with Smaller Increase: -0.467***, -0.455***, -0.420***  
- Note: Blank space in the table column indicates no statistically significant difference. *p < 0.1; **p < 0.05; ***p < 0.01.  
- Source: IMF staff estimates.

### Large exchange rate depreciations not associated with banking crises
- Table lists numerous episodes across Advanced Economies and Emerging Market and Developing Economies (countries and years are reported in the source table).

### Sector-level exchange rate pass-through to prices (Table 7)
- Average ∆ln(Exchange rate) coefficients (export price change columns): 0.360***, 0.297***, 0.402*** (standard errors: (0.040), (0.050), (0.066))  
- Average ∆ln(Exchange rate) coefficients (import price change columns): -0.717***, -0.633***, -0.685*** (standard errors: (0.026), (0.048), (0.062))  
- Interaction with period dummy (period dummy = 0 if year < 2002, 1 otherwise):
  - Export price change: 0.079, 0.279***, 0.027 (std. errs: (0.042), (0.075), (0.104))
  - Import price change: -0.109***, -0.018, -0.060 (std. errs: (0.03), (0.071), (0.098))
- Interaction with inflation rate:
  - Export price change: -0.435*** (std. err: (0.079))
  - Import price change: 0.0232* (std. err: (0.013))
- Interaction with country-industry FVA share in gross exports:
  - Export price change: -0.020, -0.027
  - Import price change: 0.012***, 0.013*** (std. errs: (0.005), (0.006))
- Interaction with world industry FVA share in gross exports:
  - Export price change: -0.274***, 0.043
  - Import price change: 0.030***, 0.035*** (std. errs: (0.009), (0.010))
- Interaction with industry share in exports:
  - Export price change: 0.038, 0.038
  - Import price change: 0.009**, 0.006 (std. errs: (0.004), (0.005))
- Controls and model fit:
  - Country and industry fixed effects: Yes
  - Time fixed effects: Yes
  - Adjusted R-squared (export columns): 0.22, 0.26, 0.29
  - Adjusted R-squared (import columns): 0.33, 0.42, 0.46
  - Number of observations (export columns): 15,128; 7,261; 5,706
  - Number of observations (import columns): 15,634; 7,811; 6,143
- Notes: Significance at *** p < 0.01, ** p < 0.05, * p < 0.1. Standard errors are in the parentheses. Table reports estimated coefficients of equations (10) and (12). Additional variable definitions and data sources are reported in the table notes (period dummy, money growth, real GDP, inflation, country- and world-level FVA shares, industry share in exports, GFC dummy).

### Sector-level exchange rate effects on trade volumes (Table 8)
- Average ∆ln(Exchange rate) coefficients (export volume change columns): -0.244***, 0.217, 0.221 (std. errs: (0.124), (0.152), (0.184))  
- Average ∆ln(Exchange rate) coefficients (import volume change columns): 1.147***, 1.423***, 1.829*** (std. errs: (0.107), (0.147), (0.178))  
- Interaction with period dummy:
  - Export volume change: -0.552***, -0.949***, -0.786*** (std. errs: (0.133), (0.227), (0.277))
  - Import volume change: -0.826***, -1.634***, -1.604*** (std. errs: (0.111), (0.226), (0.276))
- Interaction with money growth: -1.246*** (export volume, std. err: (0.253)); 0.289 (import volume, std. err: (0.245))
- Interaction with inflation rate:
  - Export volume change: 1.552***, 1.777*** (std. errs: (0.389), (0.305))
- Interaction with country-industry FVA share in gross exports:
  - Export volume change: 0.320***, 0.197** , 0.026 (std. errs: (0.094), (0.104), (0.081))
  - Import volume change: 0.143 (std. err: (0.091))
- Interaction with world industry FVA share in gross exports:
  - Export volume change: 0.379, 0.368
  - Import volume change: 1.427***, 1.286*** (std. errs: (0.281), (0.328))
- Interaction with industry share in exports:
  - Export volume change: 0.051, -0.013, 0.090
  - Import volume change: 0.191** (std. err: (0.104))
- Controls and model fit:
  - Country and industry fixed effects: Yes
  - Time fixed effects: Yes
  - Adjusted R-squared (export columns): 0.12, 0.22, 0.23
  - Adjusted R-squared (import columns): 0.17, 0.25, 0.27
  - Number of observations (export columns): 10,631; 7,108; 5,647
  - Number of observations (import columns): 12,465; 7,376; 5,776
- Notes: Significance at *** p < 0.01, ** p < 0.05, * p < 0.1. Standard errors are in the parentheses. Table reports estimated coefficients of equations (10) and (12). Additional variable definitions and data sources are reported in the table notes.

### Key graphical findings and patterns from figures
- Figure 3.1 (Recent Exchange Rate Movements in Historical Perspective):
  - Major currencies have seen large movements in real effective terms recent years that are unusual compared with historical experience.
  - Panel date coverage: Last observation reported is June 2015; confidence band constructed from all 36-month evolutions since January 1980. Blue lines indicate most recent continuous appreciation/depreciation episodes (no interruptions > three months).
- Figure 3.2 (Long-Term Exchange Rate Pass-Through and Price Elasticities):
  - The estimated effects of exchange rate movements on trade prices and volumes have the expected sign for most economies considered.
  - Estimates based on annual data for 60 advanced and emerging market and developing economies from 1980 to 2014.
- Figure 3.8 (Trade Elasticities over Time in Different Regions):
  - Ten-year rolling windows ending in year t show little evidence of a general trend toward disconnect between exchange rates, trade prices, and total trade volumes across the global sample, Asia, and Europe.
  - Figure based on panel estimates using producer price index–based REER and export/import prices relative to foreign/domestic producer prices. Full sample spans 88 economies from 1990 to 2014.
- Figure 3.7 (Evolution of Global Value Chains):
  - Participation in global value chains has generally risen gradually, with substantial changes in some countries.
  - Backward and forward participation metrics (share of foreign value added in gross exports; intermediates used by trading partners as share of gross exports) show rising averages and dispersion over time.
- Figure 3.9 (Ratios of Exports and Imports to GDP, 1990–2014):
  - Exports and imports have been rising as a share of GDP (simple averages of economies in the sample).
- Figures on export dynamics following large depreciations:
  - Export prices and volumes display similar dynamics during the period through 1997 and in the period thereafter (Figure 3.10). Dashed lines denote 90 percent confidence intervals.

### Robustness tests on exchange rate–export volume elasticities (Appendix B)
- Reduced-form specification (equation (14)) replicates literature findings when using the Ahmed, Appendino, and Ruta (2016) sample (46 economies, 1996–2012):
  - For total exports, elasticity declines between subsamples from –1.2 to –0.7 (baseline replication).
  - For manufacturing exports, elasticity declines from –1.3 to –0.6.
- Robustness checks and their effects:
  - Excluding observations with CPI inflation in excess of 30 percent reduces the estimated decline:
    - Total exports: decline now from –0.8 to –0.7 (instead of –1.2 to –0.7).
    - Manufacturing exports: decline now from –0.7 to –0.6 (instead of –1.3 to –0.6).
    - The estimated decline is inside one standard error (0.1).
  - Robust regression (downweighting large-residual observations) results:
    - Total exports elasticity remains around –0.7 in both subsamples.
    - Manufacturing exports point estimate rises in absolute value from –0.6 to –0.7 (change within one standard error).
  - Using log first-differences to compute growth rates (instead of percentage changes) reduces the estimated decline:
    - Total exports: decline from –0.8 to –0.6 (instead of –1.2 to –0.7).
    - Manufacturing exports: decline from –0.8 to –0.7 (within one standard error).
    - Log differences dampen extreme values and are consistent with the log-linear model specification.
  - Deflating exports with appropriate export deflators (rather than CPI) reduces spurious correlation and lowers absolute sizes of elasticity estimates:
    - When total exports are deflated with total exports deflator and goods/manufacturing with goods deflator, absolute size of elasticity estimates decline and the drop across subsamples is less apparent (total exports elasticity unchanged at –0.2 across subsamples in this check).
  - Removing initial domestic GDP control from (14) has little effect on the conclusion that the drop in elasticity over time is fragile.
- Interpretation:
  - The apparent decline in exchange rate–export elasticities over time is sensitive to sample composition, outliers, growth-rate definitions, and choice of price deflators.

*Italic: Source: IMF staff estimates and figures/tables as reported in the source chapter.*

### conclusion is also not affected by expanding the sample from the 46 economies to the 88

### wp1758 - conclusion is also not affected by expanding the sample from the 46 economies to the 88

### Conclusion robustness: trade elasticities over time
- Expanding the sample from the 46 economies to the 88 economies included in Section II.D (see Table A3) does not affect the conclusion.
- Expanding the time sample to span 1990-2014, instead of 1996-2012, does not affect the conclusion.
- Overall finding: the analysis confirms that there is no disconnect over time for trade elasticities at the macroeconomic level, even at its reduced form.

### Guidance on deflator choice for total exports of goods and services
- When choosing the deflator for total exports of goods and services, one should be guided by economic theory and the effect of such choice on consistency of trade elasticity estimates.
- The authors state: "We believe that consistency of estimates takes precedence over accuracy."
- CPI is only a proxy for the correct export price index and its use is likely to result in biased estimates.

### Table B1 — Elasticity estimates: summary of robustness analysis
- All variables deflated using CPI inflation unless stated otherwise.
- Significance notation: * p < 0.1; ** p < 0.05; *** p < 0.01.
- Notes:
  - 1/ Robust regression downweights observations with larger absolute residuals using iterative weighted least squares (Andersen, 2008).
  - 2/ Growth rates computed as log first differences.

- Elasticity of Total Exports to the Real Effective Exchange Rate
  - Full sample (1996–2012)
    - (1) Baseline: -1.046***
    - (2) As (1) but CPI inflation < 30%: -0.776***
    - (3) As (1) but without outliers (robust regression) 1/: -0.714***
    - (4) Log linear specification 2/: -0.734***
    - (5) As (4) but real exports defined using export prices (not CPI): -0.180***
    - (6) As (5) but no domestic real GDP control: -0.177***
  - Subsample 1996–2003
    - (1) Baseline: -1.245***
    - (2) CPI inflation < 30%: -0.822***
    - (3) Robust regression: -0.749***
    - (4) Log linear specification: -0.780***
    - (5) Export prices (not CPI): -0.196**
    - (6) No domestic real GDP control: -0.186**
  - Subsample 2004–2012
    - (1) Baseline: -0.664***
    - (2) CPI inflation < 30%: -0.664***
    - (3) Robust regression: -0.653***
    - (4) Log linear specification: -0.633***
    - (5) Export prices (not CPI): -0.174***
    - (6) No domestic real GDP control: -0.176***

- Elasticity of Goods Exports to the Real Effective Exchange Rate
  - Full sample (1996–2012)
    - (1) Baseline: -1.182***
    - (2) CPI inflation < 30%: -0.765***
    - (3) Robust regression: -0.688***
    - (4) Log linear specification: -0.814***
    - (5) Export prices (not CPI): -0.157*
    - (6) No domestic real GDP control: -0.156**
  - Subsample 1996–2003
    - (1) Baseline: -1.440***
    - (2) CPI inflation < 30%: -0.804***
    - (3) Robust regression: -0.856***
    - (4) Log linear specification: -0.861***
    - (5) Export prices (not CPI): -0.146
    - (6) No domestic real GDP control: -0.130
  - Subsample 2004–2012
    - (1) Baseline: -0.622***
    - (2) CPI inflation < 30%: -0.622***
    - (3) Robust regression: -0.626***
    - (4) Log linear specification: -0.674***
    - (5) Export prices (not CPI): -0.169***
    - (6) No domestic real GDP control: -0.192***

- Elasticity of Manufacturing Exports to the Real Effective Exchange Rate
  - Full sample (1996–2012)
    - (1) Baseline: -1.081***
    - (2) CPI inflation < 30%: -0.671***
    - (3) Robust regression: -0.691***
    - (4) Log linear specification: -0.746***
    - (5) Export prices (not CPI): -0.0899
    - (6) No domestic real GDP control: -0.0870
  - Subsample 1996–2003
    - (1) Baseline: -1.294***
    - (2) CPI inflation < 30%: -0.651***
    - (3) Robust regression: -0.592***
    - (4) Log linear specification: -0.765***
    - (5) Export prices (not CPI): -0.0377
    - (6) No domestic real GDP control: -0.0160
  - Subsample 2004–2012
    - (1) Baseline: -0.592***
    - (2) CPI inflation < 30%: -0.592***
    - (3) Robust regression: -0.653***
    - (4) Log linear specification: -0.663***
    - (5) Export prices (not CPI): -0.166*
    - (6) No domestic real GDP control: -0.191**

### Construction of the GEKS indices (country-sector export and import price indexes)
- Data and sectors
  - Country-sector level export and import price indexes constructed for 18 tradable sectors using UN Comtrade data at the annual frequency.
  - The 18 sectors are defined in the text.
  - Products are defined at the HS 2002 6-digit level.

- Method and properties
  - Adopted method: Ivancic et al. (2011), an adapted version of the multilateral GEKS approach.
  - GEKS index properties:
    - (1) makes maximum use of all possible matches between any two periods, and
    - (2) has no chain drift.

- Two-step construction (sector i, periods 0 to M, omit i for simplicity)
  - Step 1: obtain sector i GEKS price indexes for any two adjacent periods τ−1 and τ, denoted as GEKS_{τ−1,τ}.
    - GEKS_{τ−1,τ} is the geometric average of the ratios of bilateral Fisher indexes:
      - GEKS_{τ−1,τ} = ∏(P^F_{τ−1,l} / P^F_{τ,l})^{1/M}_{l=0} = ∏(P^F_{τ−1,l} ∙ P^F_{l,τ})^{1/M}_{l=0}
      - where M + 1 is the number of years, and P^F_{τ,l} denotes the Fisher index between period τ and l.
  - Step 2: calculate GEKS_{0,t} using all GEKS_{τ−1,τ} for τ = 1, ..., t:
    - GEKS_{0,t} = ∏_{τ=1}^{t} (GEKS_{τ−1,τ})^{1/t}

- Bilateral Fisher index construction
  - P^F_{j,k} = √(P^L_{j,k} ∗ P^P_{j,k})
  - Laspeyres index:
    - P^L_{j,k} = ∑(P_{n,k} ∙ Q_{n,j})_n / ∑(P_{n,j} ∙ Q_{n,j})_n
  - Paasche index:
    - P^P_{j,k} = ∑(P_{n,k} ∙ Q_{n,k})_n / ∑(P_{n,j} ∙ Q_{n,k})_n
  - In these indexes, summation is carried over products in sector i, with P_{n,t} indicating the price, and Q_{n,t} the quantity of product n in year t.

*Source: wp1758 - conclusion is also not affected by expanding the sample from the 46 economies to the 88*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2017/wp1758.pdf_