## Global Value Chains and External Adjustment: Do Exchange Rates Still Matter? — Section I–IV (WP/19/300)

## Source details

**Canonical URL:** [Global Value Chains and External Adjustment: Do Exchange Rates Still Matter? — Section I–IV (WP/19/300)](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019300-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019300-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019300-print-pdf.pdf.json)

---

### Abstract — Core result
- Greater integration into international value chains reduces the exchange rate elasticity of gross trade volumes.
- This result holds both in the short and medium term, pointing to the rigidity of value chains.
- Greater value chain integration is associated with larger gross trade flows, relative to GDP, which tends to amplify the effect of exchange rate movements.
- Combining the two results suggests that, for most countries, integration into global value chains does not materially alter the working of exchange rates and the benefits of exchange rate flexibility in facilitating external adjustment remain.
- Paper identifiers: WP/19/300; Research Department; December 2019.

### Motivation and key research questions
- Conventional Mundell-Fleming: exchange rate movements change relative prices and through expenditure-switching affect export and import volumes.
- Question: Do global value chains (GVCs) — backward and forward linkages — alter how exchange rates affect trade prices, quantities, and the trade balance?
- Empirical focus: responses of trade prices and quantities to exchange rate movements in a panel of bilateral manufacturing trade among 37 advanced and emerging market economies (2000–14).

### Mechanisms: How GVCs alter exchange-rate effects
- Backward linkages (BWD):
  - Exporter marginal costs MC_a depend on exchange rates vis-à-vis upstream suppliers; depreciation of a raises marginal costs and dampens export quantity response.
  - Operational interpretation: BWD acts like a supply shifter.
- Forward linkages (FWD):
  - If exports from a to b are reexported to downstream buyers, downstream exchange rates affect demand for b’s exports and thus a’s intermediate inputs.
  - Depreciation can increase demand for imported inputs used for re-exports, dampening the import-volume response.
  - Operational interpretation: FWD acts like a demand shifter.
- Generic characterization: trade prices and volumes include bilateral exchange rate effects plus BWD supply shifter (MC_a→b) and FWD demand shifter (D_a→b).

### Thought experiment: unilateral depreciation (de_aall)
- Exports:
  - Stand-alone: quantities +
  - With BWD: prices + (BWD); quantities – (BWD)
- Imports:
  - Stand-alone: prices +; quantities –
  - With FWD: prices + (FWD); quantities + (FWD)
- Table 1 summary (directional effects preserved as in source).

### Construction of GVC shifters — methodology summary
- Backward GVC shifters:
  - Weighted sum of bilateral exchange rate movements relative to upstream suppliers, weights equal import-content shares from each upstream partner.
  - Computed at country-sector level using WIOD and mapped to bilateral level using sectoral composition of bilateral trade flows.
  - Includes direct and indirect imported intermediate inputs via (I_S − A_a,s)^(−1) M_a,s representation.
  - Change in marginal costs: ∆lnMC_G a,s = GVC_a,s abcd · ∆ln e_a,s (∆ln e_a,s is C×1 vector of bilateral exchange rate movements; positive = depreciation of a).
  - Bilateral ∆lnMC_a→b is sector-weighted average using X_a,s→b,s.
- Forward GVC shifters:
  - Forward linkage captures exports from a to b that are reexported to downstream country d:
    - GGVF_aas_fbb = x_ab,t^I × (I_S − Ã_b,t)^−1 R_b,t (GGVF is S×C; aggregation sums rows to 1×C).
    - Forward demand shifter: ∆lnD_aas_abba = GGVF_aas_fbb · ∆ln e_a,s
  - Section I ends before full formal forward-shifter formulas; methodology continues in later sections.

### Empirical approach, data and identification
- Panel estimation at the country-pair level for bilateral and dollar exchange rates, following Gopinath et al. approach.
- Use newly constructed bilateral prices and quantities (COMTRADE-based constructions) and WIOD 2016 for GVC measures.
- Sample: 37 economies, period 2000–14; Observations: 18,708 (varies by specification).
- Controls: country-pair fixed effects; time fixed effects; exporter’s PPI growth; importer’s CPI and GDP growth (alternatives tested).
- Lags: 3; Standard errors clustered at the dyadic level.
- Trade balance generic expression:
  - dT/GDP_bs = X/Y_bs [de × (β_PPP + β_QPP) + de × ∆lnMC_bs (γ_P + γ_Q)] − M/Y_bs [de × (β_PPP + β_QPP) + de × ∆lnD_bs (δ_P + δ_Q)]
  - Stand-alone short-run response: contemporaneous β-coefficients; medium-run: contemporaneous + three lags.

### Key empirical findings — prices, quantities, and trade balance
- Prices:
  - A depreciation of the exporter’s currency leads to increased export prices.
  - Additional increase in export price elasticity due to backward integration: 0.277 in the short run and 0.255 in the medium run (compared with stand-alone effects 0.442 and 0.400 respectively in baseline blocks).
  - Amplification effect on import prices is also observed.
- Quantities:
  - GVC integration amplifies price responses but dampens quantity responses.
  - Stand-alone short-run export (import) volume elasticity: about 0.211 (−0.211) in a reported baseline block.
  - For an economy with average GVC integration, short-run export (import) volume elasticity: about 0.04 (−0.12) in reported estimates.
  - Dampening persists into the medium run, indicating rigidities in value chains over 0–3 years.
- Trade balance sensitivity:
  - For an economy with average degree of GVC participation and trade openness, the trade balance response (percent of GDP) drops:
    - by about half in the short term, relative to a country with no GVC integration;
    - by about one third in the medium term, relative to a country with no GVC integration.
  - Net implication: increased GVC integration has not materially altered the functioning of exchange rates because greater GVC integration is associated with larger trade openness, which amplifies trade-balance effects.

### Representative coefficient blocks (selected reported values preserved exactly)
- From baseline weighted regressions (short-run coefficients, standard errors in parentheses):
  - Stand-alone: 0.4416*** (0.0640); 0.5584*** (0.0640); 0.2112*** (0.0658); -0.2112*** (0.0658); 0.458*** (0.135)
  - GVC (average BWD and FWD): 0.277*** (0.0658); 0.149*** (0.0476); -0.176** (0.0723); 0.09531* (0.0535); -0.215** (0.0999)
  - Stand-alone + GVC: 0.719*** (0.0337); 0.708*** (0.0396); 0.0350 (0.0399); -0.116** (0.0491); 0.243** (0.100)
- Alternative reported blocks (examples from Tables):
  - Stand-alone: 0.571***, 0.690***, 0.130, -0.323***, 0.500*** (Table 4)
  - GVC: 0.192***, 0.0643, -0.116, 0.170***, -0.238** (Table 4)
  - Stand-alone + GVC: 0.763***, 0.754***, 0.0138, -0.153***, 0.263*** (Table 4)
- Robustness and alternative specifications report consistent qualitative patterns; reported observation counts and R-squared vary:
  - Observations: 18,708; R-squared values reported across specifications include 0.308, 0.315, 0.341, 0.349, 0.385, 0.389, 0.462, 0.464, 0.479, 0.482 depending on table and dependent variable.
  - Appendix tables include examples: Table A.6 (excluding intra-EA trade), Table A.7 (global controls), Table A.8 (exporter-time FE as supply control).

### GVC integration and dominant currency pricing (US dollar)
- Extended specification includes bilateral and USD exchange rates and USD-based GVC shifters.
- USD-based backward-supply and forward-demand shifters:
  - ∆lnMC_a,s^US$ = GGV_a,t_bbdd × ∆ln e_{a$,s}
  - ∆lnMC_{a→a,s}^{US$} = Σ [X_{a,s→a,s} / Σ X_{a,s→a,s s}] × ∆ln m c_{a s, s}^{US$}
  - ∆lnD_{a→a,s}^{US$} = GGVF_{a a,s_fbb} × ∆ln e_{a$,s}
- Functional form:
  - T_{a→a} = f_{a→a}[e_aa, e_a$, MC_{a→a}(e_aa, e_a$), D_{a→a}(e_ab, e_a$)]
- Empirical implications:
  - Results broadly consistent with baseline: backward integration increases export prices and reduces export quantities; forward integration increases import prices and quantities.
  - For average GVC integration: standalone import elasticity falls by about half, export elasticity declines by about a third (reported qualitative magnitudes preserved as in source).
  - GVC integration explains part of US dollar dominance but is not the sole determinant.

### Visualization, heterogeneity, and quantitative illustrations
- Figure 4 (10 percent depreciation vis-à-vis all currencies): economies at 25th vs 75th percentiles of GVC integration show:
  - Prices: greater GVC integration amplifies price effects.
  - Quantities: greater GVC integration dampens quantity effects.
  - Patterns hold in short and medium run.
- Figure 5: For a fixed level of trade openness, greater GVC participation mutes trade-balance response; for fixed GVC integration, greater trade openness increases trade-balance responsiveness (openness fixed at median economy; backward/forward GVC fixed at median).
- Figure 6: Partial correlation between trade openness and GVC integration shows greater integration into value chains is strongly associated with larger trade flows.
- Figure 7:
  - Distribution of medium-term trade balance elasticities shows significant heterogeneity.
  - For the average country, a 10 percent depreciation is estimated to increase the trade balance by about 1 percentage point of GDP.
  - Estimated trade-balance elasticity for the average economy has not changed much since early 2001 because rising GVC integration has been offset by rising trade openness.

### Robustness checks (selected)
- Appendix Table A.2: unweighted regressions replicate Table 3 results.
- Appendix Table A.3: includes importer’s real domestic demand growth as control.
- Appendix Table A.4: proxy for import demand shocks using estimated importer fixed effects.
- Appendix Table A.5: exercise with only the direct component of GVC integration.
- Appendix Table A.6: excludes intra-Euro Area trade; example coefficients preserved in source (Observations: 18,708; selected short-run Stand-alone: PX 0.572*** (0.0707), PM 0.625*** (0.0577), QX 0.113 (0.0762), QM -0.198*** (0.0627), TB/Y 0.388*** (0.151)).
- Appendix Table A.7: substitutes time FE with global controls (global GDP growth, inflation, real export shocks, VIX, real oil prices).
- Appendix Table A.8: exporter-time FE as supply control (Observations: 19,182; Stand-alone: PX 0.617*** (0.0845), PM 0.681*** (0.0629), QX 0.0758 (0.0814), QM -0.310*** (0.0719), TB/Y 0.483*** (0.140)).

### Interpretation and conclusions
- Greater GVC integration reduces the exchange rate elasticity of gross trade volumes in short and medium terms due to rigidities and low substitutability in supply chains.
- Greater GVC integration is associated with larger gross trade flows, which amplifies exchange rate effects on the trade balance and largely offsets the dampening of quantity elasticities.
- Overall conclusion: increased integration into global value chains has not materially altered the working of exchange rates; exchange rate flexibility remains beneficial for external adjustment.

*Source: IMF Working Paper WP/19/300 — "Global Value Chains and External Adjustment: Do Exchange Rates Still Matter?" (Sections I–IV as provided).*

### Section 1

### Global Value Chains and External Adjustment: Do Exchange Rates Still Matter? — Section 1

### Abstract — Core result
- The analysis indicates that greater integration into international value chains reduces the exchange rate elasticity of gross trade volumes.
- This result holds both in the short and medium term, pointing to the rigidity of value chains.
- Greater value chain integration is associated with larger gross trade flows, relative to GDP, which tends to amplify the effect of exchange rate movements.
- Combining the two results suggests that, for most countries, integration into global value chains does not materially alter the working of exchange rates and the benefits of exchange rate flexibility in facilitating external adjustment remain.
- Paper identifiers: WP/19/300; Research Department; December 2019.

### Introduction — motivation and research questions
- Conventional Mundell-Fleming framework: exchange rate movements change relative prices and through expenditure-switching affect export and import volumes.
- Ongoing debate: whether increased complexities of international trade, notably global value chains (GVCs), alter how exchange rates operate and affect external balances.
- Key channel concerns:
  - Backward GVC integration (use of imported intermediate inputs in exports) can mute expenditure switching because trade prices and marginal costs move together.
  - Dominant currency pricing (e.g., US dollar) and price stickiness in that currency can mute short-term expenditure switching (reference to Gopinath, et al., Forthcoming).
- Empirical focus: responses of trade prices and quantities to exchange rate movements in a panel of bilateral manufacturing trade among 37 advanced and emerging market economies.

### Mechanisms — How GVCs alter exchange-rate effects
- Backward linkages (BWD):
  - Use of imported intermediates means exporter marginal costs (MC_a) depend on exchange rates vis-à-vis upstream suppliers.
  - A depreciation of currency a vis-à-vis suppliers raises marginal costs and dampens the export quantity response relative to the traditional channel.
  - Operational interpretation: a GVC-related exchange rate shock through backward linkages is akin to a supply shifter.
- Forward linkages (FWD):
  - If exports from a to b are reexported to downstream buyers, exchange rates of b relative to downstream buyers affect demand for b’s exports and thus demand for a’s intermediate inputs.
  - A depreciation can enhance competitiveness of re-exported goods and increase demand for imported inputs, dampening the import-volume response relative to the traditional case.
  - Operational interpretation: a GVC-related exchange rate shock through forward linkages operates as a demand shifter.
- General characterization of trade flows:
  - Trade flows (prices and volumes) are generically characterized to include: bilateral exchange rate effects, BWD supply shifter (MC_a→b), and FWD demand shifter (D_a→b).

### Thought experiment — depreciation vis-à-vis all other currencies
- For a unilateral depreciation (de_aall = depreciation of a against all others):
  - Exports: traditional channel improves competitiveness and boosts exports; BWD raises marginal costs and reduces export volume response.
  - Imports: traditional channel shifts demand away from imports; FWD can increase demand for imported inputs used for re-exports, counteracting the traditional import response.
- Table 1 summary of expected effects of a depreciation (prices in country a’s currency vs. quantities):
  - Exports: Stand-alone prices + ; BWD/FWD linkages prices + (BWD); Stand-alone quantities + ; BWD/FWD linkages quantities – (BWD)
  - Imports: Stand-alone prices + ; BWD/FWD linkages prices + (FWD); Stand-alone quantities – ; BWD/FWD linkages quantities + (FWD)

### Empirical approach — data and identification
- Panel estimation of exchange rate pass-through and trade elasticities at the country-pair level for both the bilateral and dollar exchange rates, following the approach in (Gopinath, et al., Forthcoming).
- Focus on bilateral trade flows to mitigate aggregation biases associated with weighted effective exchange rate measures.
- Use of newly constructed bilateral prices and quantities data (from (Gopinath, et al., Forthcoming)).
- Novel measures of international value-chain-related exchange rate shocks vis-à-vis both upstream and downstream trading partners (and the dominant currency) are constructed; dominant currency pricing explored explicitly.
- Cross-sectional differences in estimated elasticities are driven by countries’ degree of integration in trade and GVCs.

### Key empirical findings (as described)
- For a given level of trade openness:
  - Countries that use imported intermediate inputs intensively have lower export volume elasticities.
  - Countries that re-export a large share of imports have lower import volume elasticities.
- Despite lower elasticities due to GVCs, greater GVC integration is associated with greater trade openness, which amplifies effects of exchange rate movements on the trade balance as a share of GDP, largely offsetting the dampening GVC effect on trade volumes.
- Net implication in the data: increased integration into GVCs has not materially altered the working of exchange rates in facilitating external adjustment.

### Construction of GVC shifters — methodology summary
- Backward GVC (supply-side) shifters:
  - For exporter a, a backward GVC shifter is the weighted sum of bilateral exchange rate movements relative to upstream suppliers.
  - Weights are import-content shares coming from each upstream trading partner in a’s exports.
  - Computation is at country-sector level and mapped to bilateral level using sectoral composition of bilateral trade flows.
  - Imported intermediate inputs include direct and indirect components; import-content weights are computed using global input-output tables (WIOD).
  - Matrix representation: GVC_C a,s abcd = (I_S − A_a,s)^(−1) M_a,s (definitions: G matrix is S×C; I_S identity S×S; A_a,s is S×S expenditure-share matrix; M_a,s is S×C matrix of sector shares of expenditures on intermediate inputs from country c).
  - Change in marginal costs for sector s in exporting economy a: ∆lnMC_G a,s = GVC_a,s abcd · ∆ln e_a,s (where ∆ln e_a,s is C×1 vector of bilateral exchange rate movements; positive values represent a depreciation of a).
  - Mapping to bilateral export relevance: bilateral ∆lnMC_a→b computed as weighted average across sectors using X_a,s→b,s sectoral export values.
- Forward GVC (demand-side) shifters:
  - Forward linkages described as a selling of intermediate inputs by a to b, where b uses inputs to produce goods re-exported to d; exchange rates vis-à-vis downstream buyers affect demand for those final goods and thus demand for a’s intermediate inputs.
  - (Section I ends before full formal forward-shifter formulas; methodology continues in subsequent sections not included here.)

*Source: IMF Working Paper WP/19/300 — Global Value Chains and External Adjustment: Do Exchange Rates Still Matter? — Section I*

### Section 2

### wpiea2019300-print-pdf - Section 2

### Forward integration and the forward GVC shifter
- Concept:
  - A depreciation of a’s currency relative to b’s unambiguously increases exports from a to b only if country b’s currency also depreciates relative to country d’s; otherwise, downstream demand from d can fall and reduce b’s demand for intermediates from a.
  - The proposed forward integration measure captures changes in an importer’s demand due to exchange rate fluctuations of downstream buyers.
- Formal definition:
  - GVC forward shifter:
    - GGVF_aas_fbb = x_ab,t^I × (I_S − Ã_b,t)^−1 R_b,t
    - x_ab,t^I is an S×S matrix; element x_{aa, s}^{I, sb} reflects exports from country-sector (aa s) used as intermediate inputs by country-sector (br), expressed as a share of total exports from country a to country b.
    - Ã_b,t is an S×S matrix; element Ã_{a,s}^{s b} represents the share of country-sector (b a)’s output used as intermediate inputs in country-sector (b r).
    - R_b,t is an S×C matrix; element R_{a,s}^{s b} represents the share of gross output in country-sector (b a) that is exported to country d.
    - GGVF_aas_fbb is an S×C matrix; element GGVF_{aas_fbb, s b} reflects exports from country-sector (a s) to country b that are reexported to country d, expressed as a share of total exports from a to b.
  - Aggregation:
    - GGVF_aas_fbb (1×C) vector obtained by summing across rows of GGVF_aas_fbb matrix; d-th element is share of exports from a to b reexported to d.
  - Forward linkage demand shifter:
    - ∆lnD_aas_abba = GGVF_aas_fbb · ∆ln e_a,s
    - Interprets exchange-rate-induced demand shocks through forward linkages; assumes b’s downstream demand is sensitive only to changes in its bilateral exchange rates vis-à-vis trading partners that source intermediates from it.

### GVC integration: Stylized facts
- Main points:
  - Most economies have become increasingly integrated into global value chains, with large cross-country differences.
  - In some small economies the import content of exports (backward integration) can reach one-third to one-half.
  - Examples: Belgium, Czech Republic, Hungary, Slovak Republic heavily integrated into European value chains.
  - Large systemic economies (China, Japan, United States) remain dominated by traditional trade.
- Data sources and coverage:
  - Figure 3: Integration into Global Value Chains, 2001–14 (Manufacturing, trade-weighted average across trading partners).
  - Bilateral GVC measures constructed using WIOD 2016 (Timmer, Dietzenbacher, Los, Stehrer, & Vries, 2015).
  - Overlap covers 37 economies over 2000-14.

### Econometric analysis — Baseline specification
- Estimation aims:
  - Estimate exchange rate pass-through and trade volume elasticities in a panel at the country-pair level, with existing GVC links given.
  - Focus on exchange rate movements associated with depreciation of both exporter and importer currencies.
  - Include contemporaneous and lagged regressors to gauge short-term and medium-term effects.
- Key variables and notation:
  - ∆_{t−l} ln e_aa denotes change in log of bilateral nominal exchange rate.
  - alll M C_{a→a,s,abba.} is bilateral metric of backward integration.
  - alll D_{a→a,s,abba.} is bilateral measure of forward integration.
- Controls:
  - (i) country-pair fixed effects; (ii) time fixed effects; (iii) exporter’s PPI growth; (iv) importer’s CPI and GDP growth.
- Trade balance effect (generic expression):
  - dT/GDP_bs = X/Y_bs [de × (β_PPP + β_QPP) + de × ∆lnMC_bs (γ_P + γ_Q)] − M/Y_bs [de × (β_PPP + β_QPP) + de × ∆lnD_bs (δ_P + δ_Q)]
  - Stand-alone short-run response: contemporaneous β-coefficients; medium-run: sum of contemporaneous and three lags.

### Data and estimation details
- Bilateral GVC measures: WIOD 2016.
- Bilateral trade price and volumes: COMTRADE-based constructions (Boz, Cerutti, & Pugacheva; Gopinath et al.).
- Sample overlap: 37 economies, period 2000-14.
- Estimation features:
  - Observations: 18,708
  - R-squared: 0.308, 0.341, 0.385, 0.385, 0.462 across dependent variables
  - Lags: 3
  - Dyad FE: YES
  - Year FE: YES
  - Excluding intra-EA trade
- Table summaries (Baseline specification; weighted regression):
  - Note: *** p<0.01, ** p<0.05, * p<0.1. Standard errors clustered at the dyadic level in parenthesis. 1/ Assumed 10 percent depreciation and 0.15 openness.
  - Reported coefficient blocks (short-run and long-run elasticities; columns labeled PX PM QX Q MTB/Y):
    - Stand-alone: 0.4416*** (0.0640); 0.5584*** (0.0640); 0.2112*** (0.0658); -0.2112*** (0.0658); 0.458*** (0.135)
    - GVC (for average BWD and FWD level): 0.277*** (0.0658); 0.149*** (0.0476); -0.176** (0.0723); 0.09531* (0.0535); -0.215** (0.0999)
    - Stand-alone + GVC: 0.719*** (0.0337); 0.708*** (0.0396); 0.0350 (0.0399); -0.116** (0.0491); 0.243** (0.100)
    - Stand-alone: 0.400*** (0.0802); 0.600*** (0.0802); 0.561*** (0.106); -0.561*** (0.106); 1.382*** (0.214)
    - GVC (for average BWD and FWD level): 0.251*** (0.0736); 0.0832 (0.0936); -0.183* (0.102); 0.329** (0.140); -0.518*** (0.195)
    - Stand-alone + GVC: 0.650*** (0.0537); 0.683*** (0.0662); 0.378*** (0.0887); -0.232** (0.0925); 0.865*** (0.180)

### Key empirical findings
- Prices:
  - A depreciation of the exporter’s currency leads to increased export prices.
  - Additional increase in export price elasticity due to backward integration: 0.277 in the short run and 0.255 in the medium run (compared with stand-alone effects 0.442 and 0.400 respectively).
  - Amplification effect on import prices is also observed.
- Quantities:
  - GVC integration amplifies price responses but dampens quantity responses.
  - Stand-alone short-run export (import) volume elasticity: about 0.211 (−0.211).
  - For an economy with average GVC integration, export (import) volume elasticity in short run: about 0.04 (−0.12).
  - Dampening is persistent into the medium run, indicating rigidities in value chains over 0-3 years.
- Trade balance sensitivity:
  - For an economy with average degree of trade openness, greater GVC integration implies lower sensitivity of the trade balance to the exchange rate in both short and medium run.
  - The combination of GVC effects on prices tends to offset across imports and exports; the dominant channel reducing trade-balance sensitivity is the response of trade quantities.

### GVC integration and dominant currency pricing (USD)
- Extension to account for pricing in a dominant currency (US dollar):
  - Include exchange rates vis-à-vis trading partner and vis-à-vis the US dollar in regressions.
  - Construct USD-based backward-supply and forward-demand shifters.
- USD-based shifters:
  - ∆lnMC_a,s^US$ = GGV_a,t_bbdd × ∆ln e_{a$,s}  (∆ln e_{a$,s} is country a’s bilateral exchange rate vis-à-vis the US dollar)
  - ∆lnMC_{a→a,s}^{US$} = Σ [X_{a,s→a,s} / Σ X_{a,s→a,s s}] × ∆ln m c_{a s, s}^{US$}
  - ∆lnD_{a→a,s}^{US$} = GGVF_{a a,s_fbb} × ∆ln e_{a$,s}
- Exports and imports can be expressed as functions:
  - T_{a→a} = f_{a→a}[e_aa, e_a$, MC_{a→a}(e_aa, e_a$), D_{a→a}(e_ab, e_a$)]
  - Exchange rate changes operate on exports and imports both directly and through backward and forward linkages, and through USD-based GVC shifters.
- Refined econometric specification:
  - Includes bilateral and USD-based exchange rates and GVC shifters (bilateral and USD).
  - For brevity, reported results sum bilateral and USD coefficients (combined effect of depreciation vis-à-vis all other currencies).
- Empirical implications:
  - Results broadly consistent with baseline.
  - Backward integration: increases export prices and reduces export quantities.
  - Forward integration: increases import prices and quantities.
  - For countries integrated through both backward and forward linkages, trade volume elasticities fall while pass-through estimates increase.
  - With average GVC integration: standalone import elasticity falls by about half, export elasticity declines by about a third, other things equal.
- Conceptual note:
  - Use of imported intermediate inputs is a potential determinant of dominant currency pricing; these features are interrelated and not independent.

*Source: wpiea2019300-print-pdf - Section 2*

### Section 3

### wpiea2019300-print-pdf - Section 3

### Effects of GVC Participation on Trade Volume Elasticities and Trade Balance
- Greater integration into global value chains (GVCs) dampens the exchange rate elasticity of gross trade volumes in both the short and medium run, indicating low substitutability in international supply chains among downstream and upstream trading partners.
- For a country with an average degree of GVC participation and trade openness, the trade balance response (in percent of GDP) drops:
  - by about half in the short term, relative to a country with no GVC integration;
  - by about one third in the medium term, relative to a country with no GVC integration.
- Including GVC shifters in the empirical framework:
  - reduces the stand-alone pass-through estimates;
  - leads to somewhat higher trade volume elasticities;
  - lowers (in absolute terms) the coefficients on the US dollar exchange rate relative to a model without GVC shifters, although their sign and statistical significance remain unchanged.
- Interpretation: GVC integration explains part of the US dollar dominance, while other factors also determine that phenomenon.

### Empirical Estimates (Selected Table Results)
- Table 4 (weighted regression) reports coefficients for stand-alone, GVC, and stand-alone + GVC specifications across dependent variables including PX, PM, QX, QM, TB/Y with short-run and long-run elasticities.
- Selected coefficient examples (preserve reported values exactly):
  - Stand-alone: 0.571***, 0.690***, 0.130, -0.323***, 0.500*** (standard errors in parentheses reported in the source).
  - GVC (for average BWD and FWD level): 0.192***, 0.0643, -0.116, 0.170***, -0.238**.
  - Stand-alone + GVC: 0.763***, 0.754***, 0.0138, -0.153***, 0.263***.
  - Observations: 18,708; R-squared: 0.315, 0.348, 0.389, 0.389, 0.464 across specifications; Lags: 3; Dyad FE: YES; Year FE: YES.
- Table A.3 (with importer real domestic demand as control) selected coefficients:
  - Stand-alone: 0.559***, 0.698***, 0.148***, -0.298***, 0.460***.
  - GVC (for average BWD and FWD level): 0.181***, 0.0928***, -0.170***, 0.117**, -0.297***.
  - Stand-alone + GVC: 0.741***, 0.791***, -0.0211, -0.180***, 0.163*.
  - Observations: 19,220; R-squared: 0.270, 0.279, 0.205, 0.205, 0.233; Lags: 3; Dyad FE: YES; Year FE: YES.
- Alternative weighted-regression specification examples (from Appendix):
  - Stand-alone: 0.573***, 0.686***, 0.123, -0.211***, 0.331**.
  - GVC: 0.191**, 0.0608, -0.100, 0.188***, -0.238**.
  - Stand-alone + GVC: 0.763***, 0.747***, 0.0230, -0.0226, 0.0935.
  - Observations: 18,708; R-squared: 0.316, 0.349, 0.402, 0.402, 0.479; Lags: 3; Dyad FE: YES; Year FE: YES.

### Visualization and Heterogeneity (Figures 4–7)
- Figure 4 (response to a 10 percent depreciation vis-à-vis all currencies, weighted regression):
  - Shows that economies at the 25th and 75th percentiles of the distribution of GVC integration display materially different responses to exchange rates:
    - Prices: greater integration amplifies price effects.
    - Quantities: greater integration plays a dampening effect.
  - These patterns hold in both the short run and the medium run.
- Figure 5: For a given level of trade openness, greater GVC participation entails a more muted response of the trade balance to the exchange rate in the short and medium term. Conversely, for a given level of integration, greater trade openness increases the overall responsiveness of the trade balance in terms of percentage points of GDP.
  - Note: Openness fixed at the level of the median economy; backward and forward GVC integration fixed at the level of the median economy.
- Figure 6: Partial correlation between trade openness and GVC integration indicates greater integration into value chains is strongly associated with larger trade flows.
- Figure 7:
  - Panel 1: Distribution of medium-term trade balance elasticities shows significant variance across countries; heterogeneity is considerable but responses are economically meaningful for most cases.
  - For the average country (in terms of GVC integration and trade openness), a 10 percent depreciation is estimated to lead to an increase in the trade balance of about 1 percentage point of GDP.
  - Panel 2: Estimated trade balance elasticity for the average economy has not changed much since early 2001 because increasing GVC integration has been largely offset by accompanying increases in trade openness.

### Robustness Checks
- Robustness checks confirm qualitative and quantitative stability of results:
  - Appendix Table A.2: unweighted regressions replicate Table 3 results.
  - Appendix Table A.3: includes importer’s real domestic demand growth as control (instead of real GDP growth).
  - Appendix Table A.4: constructs a proxy for import demand shocks using estimated importer fixed effects from a regression of changes in trade volumes on importer and exporter fixed effects.
  - Appendix Table A.5: exercise with only the direct component of GVC integration.
  - Appendix Table A.6: results when intra-Euro Area trade is excluded due to lack of bilateral nominal exchange rate movements.
  - Appendix Table A.7: specification modeling global factors more granularly by substituting time fixed effects with a vector of global controls — global GDP growth, inflation, real export shocks, VIX, and real oil prices.
  - Appendix Table A.8: proxy for export supply shocks constructed using estimated exporter fixed effects from a regression of changes in trade volumes on importer and exporter fixed effects.

### Interpretation and Conclusions
- Key conclusions:
  - Greater integration into international value chains reduces the exchange rate elasticity of gross trade volumes in both the short and medium term.
  - Greater value chain integration is associated with larger gross trade flows, which tends to amplify the effect of exchange rates on the trade balance.
  - Overall, greater integration into global value chains has not materially altered the working of exchange rates, and the benefits of exchange rate flexibility in facilitating external adjustment remain.

*Source: wpiea2019300-print-pdf - Section 3 (IMF Working Paper content as provided).*

### Section 4

### wpiea2019300-print-pdf - Section 4

### Table A.6: Excluding intra-Euro Area Trade (weighted regression)
- Dependent variables: Short-run elasticty (PX, PM, QX, QM, TB/Y 1/) and Long-run elasticty (PX, PM, QX, QM, TB/Y 1/).
- Columns (1)–(5) reported coefficients (standard errors in parentheses). Observations: 18,708. R-squared: 0.351 / 0.382 / 0.450 / 0.450 / 0.473 (varies by specification). Lags: 3. Dyad FE: YES. Year FE: YES.
- Stand-alone (short-run / long-run where reported):
  - PX: 0.572*** (0.0707)
  - PM: 0.625*** (0.0577)
  - QX: 0.113 (0.0762)
  - QM: -0.198*** (0.0627)
  - TB/Y 1/: 0.388*** (0.151)
- GVC (for average BWD and FWD level):
  - PX: 0.187*** (0.0674)
  - PM: 0.137*** (0.0393)
  - QX: -0.0892 (0.0621)
  - QM: 0.0262 (0.0492)
  - TB/Y 1/: -0.0967 (0.101)
- Stand-alone + GVC:
  - PX: 0.760*** (0.0362)
  - PM: 0.762*** (0.0383)
  - QX: 0.0239 (0.0464)
  - QM: -0.172*** (0.0425)
  - TB/Y 1/: 0.291*** (0.101)
- Alternate specification (second block, same table) — Observations: 18,708:
  - Stand-alone:
    - PX: 0.532*** (0.0751)
    - PM: 0.727*** (0.0854)
    - QX: 0.508*** (0.106)
    - QM: -0.577*** (0.114)
    - TB/Y 1/: 1.335*** (0.243)
  - GVC:
    - PX: 0.199*** (0.0719)
    - PM: 0.0617 (0.0768)
    - QX: -0.1930 (0.134)
    - QM: 0.321** (0.109)
    - TB/Y 1/: -0.565*** (0.213)
  - Stand-alone + GVC:
    - PX: 0.730*** (0.0612)
    - PM: 0.789*** (0.0604)
    - QX: 0.315*** (0.0910)
    - QM: -0.256*** (0.0973)
    - TB/Y 1/: 0.769*** (0.206)

- Notes: Standard errors clustered at the dyadic level. 1/ Assumed 10 % depreciation and 0.15 openness.

### Table A.7: Global Control Variables (removing time FE)
- Dependent variables: Short-run elasticty and Long-run elasticty for PX, PM, QX, QM, TB/Y 1/.
- Sources and notes: Same data sources as Table A.6. Note: *** p<0.01, ** p<0.05, * p<0.1. Standard errors clustered at the dyadic level. 1/ Assumed 10 % depreciation and 0.15 openness.
- Key coefficients (observations: 15,575; R-squared: 0.330 / 0.364 / 0.383 / 0.383 / 0.458; Lags: 3; Dyad FE: YES; Year FE: YES):
  - Stand-alone:
    - PX: 0.601*** (0.0741)
    - PM: 0.740*** (0.0701)
    - QX: 0.151* (0.0901)
    - QM: -0.340*** (0.0829)
    - TB/Y 1/: 0.528*** (0.137)
  - GVC (for average BWD and FWD level):
    - PX: 0.187** (0.0761)
    - PM: 0.0409 (0.0500)
    - QX: -0.1250 (0.0854)
    - QM: 0.198*** (0.0664)
    - TB/Y 1/: -0.265** (0.106)
  - Stand-alone + GVC:
    - PX: 0.788*** (0.0412)
    - PM: 0.781*** (0.0408)
    - QX: 0.0258 (0.0508)
    - QM: -0.143*** (0.0535)
    - TB/Y 1/: 0.263** (0.102)
- Alternate block (observations: 18,708; R-squared: 0.315 / 0.348 / 0.389 / 0.389 / 0.464; Lags: 3; Dyad FE: YES; Year FE: NONO):
  - Stand-alone:
    - PX: 0.571*** (0.0798)
    - PM: 0.690*** (0.0647)
    - QX: 0.130 (0.0904)
    - QM: -0.323*** (0.0754)
    - TB/Y 1/: 0.500*** (0.137)
  - GVC:
    - PX: 0.192** (0.0765)
    - PM: 0.0643 (0.0458)
    - QX: -0.1160 (0.0794)
    - QM: 0.170*** (0.0591)
    - TB/Y 1/: -0.238** (0.100)
  - Stand-alone + GVC:
    - PX: 0.763*** (0.0392)
    - PM: 0.754*** (0.0377)
    - QX: 0.0138 (0.0500)
    - QM: -0.153*** (0.0499)
    - TB/Y 1/: 0.263** (0.102)
  - Alternate specifications within this block:
    - Stand-alone:
      - PX: 0.531*** (0.0764)
      - PM: 0.722*** (0.110)
      - QX: 0.533*** (0.105)
      - QM: -0.583*** (0.133)
      - TB/Y 1/: 1.388*** (0.222)
    - GVC:
      - PX: 0.183** (0.0654)
      - PM: 0.0225 (0.115)
      - QX: -0.188* (0.112)
      - QM: 0.299* (0.162)
      - TB/Y 1/: -0.490* (0.206)
    - Stand-alone + GVC:
      - PX: 0.714*** (0.0592)
      - PM: 0.745*** (0.0672)
      - QX: 0.345*** (0.0860)
      - QM: -0.284*** (0.105)
      - TB/Y 1/: 0.898*** (0.190)

### Table A.8: Exporter-time estimated FE as supply control
- Dependent variables: Short-run elasticty and Long-run elasticty for PX, PM, QX, QM, TB/Y 1/.
- Observations: 19,182. R-squared: 0.343 / 0.374 / 0.442 / 0.442 / 0.482. Lags: 3. Dyad FE: YES. Year FE: YES.
- Stand-alone:
  - PX: 0.617*** (0.0845)
  - PM: 0.681*** (0.0629)
  - QX: 0.0758 (0.0814)
  - QM: -0.310*** (0.0719)
  - TB/Y 1/: 0.483*** (0.140)
- GVC (for average BWD and FWD level):
  - PX: 0.186** (0.0662)
  - PM: 0.0667 (0.0432)
  - QX: -0.0620 (0.0683)
  - QM: 0.162*** (0.0518)
  - TB/Y 1/: -0.158** (0.100)
- Stand-alone + GVC:
  - PX: 0.803*** (0.0382)
  - PM: 0.747*** (0.0368)
  - QX: 0.0138 (0.0408)
  - QM: -0.147*** (0.0492)
  - TB/Y 1/: 0.325* (0.106)
- Alternate specification block (same table):
  - Stand-alone:
    - PX: 0.664*** (0.0642)
    - PM: 0.704*** (0.103)
    - QX: 0.396*** (0.0778)
    - QM: -0.537*** (0.124)
    - TB/Y 1/: 1.340*** (0.176)
  - GVC:
    - PX: 0.161*** (0.0650)
    - PM: 0.0404 (0.114)
    - QX: -0.1820 (0.0961)
    - QM: 0.254** (0.168)
    - TB/Y 1/: -0.473** (0.196)
  - Stand-alone + GVC:
    - PX: 0.825*** (0.0543)
    - PM: 0.744*** (0.0636)
    - QX: 0.214** (0.0845)
    - QM: -0.283 (0.103)
    - TB/Y 1/: 0.867*** (0.184)

- Notes: Standard errors clustered at the dyadic level in parenthesis. 1/ Assumed 10 % depreciation and 0.15 openness.

*Source: wpiea2019300-print-pdf - Section 4*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019300-print-pdf.pdf_