## wp1918

## Source details

**Canonical URL:** [wp1918](https://www.imf.org/-/media/files/publications/wp/2019/wp1918.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wp1918.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wp1918.pdf.json)

---

### Traditional and GVC Trade — Introduction and Overview
- Over the last two decades, trade and production have become increasingly organized around global value chains (GVC) or global supply chains.
- Drivers: advances in information and transportation technologies and falling trade barriers enabling unbundling of production across locations (Feenstra and Hanson 1997; Grossman and Rossi-Hansberg 2008).
- Production fragmentation implies intermediate goods and services cross borders several times along the chain.
- Comparative concept: "Traditional trade" (exports produced in one country and absorbed in destination) versus "GVC Trade" (measured as percent of nominal world GDP).

### Theoretical Channels and Expected Effects
- For developed economies: access to competitively priced inputs, higher variety, and economies of scale (Baldwin and Lopez‐Gonzalez, 2013).
- For emerging economies: GVCs as fast track to industrialization (Baldwin 2011).
- Mechanisms raising productivity:
  - finer division of labor across countries (Grossman and Rossi-Hansberg 2008);
  - greater input varieties (Halpern, Koren, and Szeidl 2015);
  - increased competition, learning externalities, technology spillovers (Li and Liu (2014), Kee (2015)).
- Welfare gains can be larger in multiple-sector frameworks with input-output linkages (Caliendo and Parro 2015, Ossa 2015).

### Empirical Evidence and Data Developments
- Historical evidence: Hummels, Ishii, and Yi 2001; Hummels, Rapoport and Yi 1998 — GVCs accounted for large share of trade growth (1970-1990s); Johnson and Noguera (2012) and Baldwin and Lopez-Gonzales (2013) — accelerated growth in 2000-09.
- Expansion slowed after the global financial crisis, contributing to trade slowdown (IMF, 2016a).
- Methodological progress: decomposition of gross trade into origins of value-added (Koopman and others 2014; Wang and others 2013).
- Data initiatives enabling analysis: Trade in Value-Added Statistics (63 countries); World Input Output Database (43 countries); Eora MRIO database for 189 countries (Lenzen and others, 2013).
- Research shifting toward impacts of GVC participation on economic outcomes (Constantinescu and others, 2017; World Bank 2016, 2017).

---

### GVC and Income per Capita — Data, Measures, and Stylized Facts
- Core dataset: Eora MRIO covering 189 countries and 26 sectors over 1990-2013; indicators constructed for 180 economies at country, sectoral, and bilateral levels.
- Methodology: decomposition of gross exports into foreign value-added (FVA) and domestic value-added (DVA) following Koopman and others (2011) and Aslam and others (2017).
- GVC participation (vertical specialization) = sum of forward linkages and backward linkages.
- Position indices: Upstreamness index and Downstreamness index.
- Stylized facts:
  - Europe: share of exports involved in GVC trade increased from about 60 to 70 percent between 2000 and 2013.
  - Over 2000-2013, GVC trade share increased on average by 5 percentage points in Asian and Western Hemisphere countries.
  - Share of services exports in total world exports increased by 5 percentage points over 2000-13.
  - In value-added terms, share of services exports in world exports is almost twice as large as under gross exports concept.
  - Services: high forward linkages and limited backward linkages; higher upstream than downstream index values.
  - Major manufacturing: sizable backward linkages and higher downstream than upstream indices.

### Sector and Country Highlights (selected figures from Tables)
- Electrical and Machinery (2013): Value added concept share in world exports: 37.7; Gross concept: 46.7; Backward linkage: 32; Forward linkage: 15; Downstream Index: 2.8; Upstream Index: 2.4.
- Financial and Business Services (2013): Value added concept share: 20.7; Gross concept: 5.7; Backward linkage: 8; Forward linkage: 91; Downstream Index: 1.7; Upstream Index: 2.4.
- Transport Equipment (2013): Value added concept share: 5.2; Gross concept: 8.1; Backward linkage: 37; Forward linkage: 10; Downstream Index: 3.0; Upstream Index: 2.0.
- CHN Electrical and Machinery (2013): Share in World Exports (Value added concept) 1.43; Gross concept 3.12; Backward linkage 41; Forward linkage 23; Downstream Index 3.8; Upstream Index 2.9.
- USA Financial and Business Activities (2013): Share in World Exports (Value added concept) 3.9; Gross concept 1.7; Backward linkage 36; Forward linkage 61; Downstream Index 1.6; Upstream Index 2.2.

---

### Main Empirical Findings on Income, Investment, Human Capital, and Productivity
- Participation in GVCs has a positive impact on:
  - Income per capita.
  - Investment.
  - Productivity.
- Effects are associated with GVC trade measures (intermediate goods and GVC-related trade) rather than conventional final-goods trade.
- Heterogeneity:
  - Upper-middle and high-income countries exhibit robust positive effects from GVC participation.
  - Effects for low and lower-middle income countries are less robust.
- Position (upstreamness) relationship with development is not straightforward; services can be upstream and high in value-added while manufacturing patterns differ.
- "Moving up" into higher-tech sectors occurs in some cases but is not universal; gains conditional on other factors.

### Empirical Framework and Key Regression Results (selected values from Table 3)
- Reduced form: log GDP per capita regressed on lagged trade and GVC participation variables with controls and fixed effects.
- Selected coefficient estimates (standard errors in parentheses):
  - Column (1) — Share of intermediate trade in GDP:
    - Coefficient on Share of Intermed. trade in GDP: 1.610*** (0.339).
    - Observations: 3,049; R-squared: 0.172.
  - Column (2) — Distinguishing GVC-related and non-GVC-related trade:
    - Coefficient on Share of GVC-trade in GDP: 2.738** (1.148).
    - Coefficient on Share of regular trade in GDP: 0.227 (0.458).
    - Observations: 3,049; R-squared: 0.183.
  - Channels:
    - Column (3) Log(I/N): GVC-related trade positively affects investment; Observations: 3,045.
    - Column (4) Human capital: Share of GVC-trade in GDP: 0.367*** (0.103); Observations: 2,777.
    - Column (5) Productivity: Share of GVC-trade in GDP: 1.366* (0.750); Observations: 2,625.
  - Heterogeneity by income (column (7), interactions):
    - GVC-trade*(Low Income): -0.947 (0.3875).
    - GVC-trade*(Low Middle Income): 1.413* (0.915).
    - GVC-trade*(High Middle Income): 2.143** (0.855).
    - GVC-trade*(High Income): 2.848*** (0.948).
    - Observations: 2,770; R-squared: 0.160.
  - Position indices (column (8)):
    - Upstream Index coefficient: -0.370 (0.2561).
    - Downstream Index coefficient: 0.102 (0.229).
    - Sample excludes commodity exporters; Observations: 1,165; R-squared: 0.275.
  - Instrumental Variables (column (9)):
    - IV constructed as average FVA from Germany, Japan, and the United States embodied in exports of three countries closest in income level (or geography).
    - Column (9) results broadly similar to OLS.
    - Observations: 2,320; R-squared: 0.132.
  - Small-sample robustness (column (6)):
    - Restricting to 50 countries typical in other databases fails to replicate column (1) results.
    - Observations: 970; R-squared: 0.170.

### Economic Significance and Interpretation
- An increase in the share of GVC trade in GDP by 10 percentage points (from median country to 75th percentile) would represent an increase in income levels by 15 to 30 percent (based on coefficients in columns (2) and (7) of Table 3).
- Mechanisms: task specialization and offshoring, access to better/cheaper inputs, knowledge spillovers, investment links (bilateral GVC participation positively related to bilateral FDI).
- Endogeneity addressed via lagged variables and an IV strategy; caveats remain.
- Gains are not automatic; depend on complementarities: institutions, contract enforcement, infrastructure quality. Low-income countries may require additional supporting factors.

---

### Economy-Wide Upstreamness — Sectoral Composition and Bilateral Supply Chains
- Indicator: change in share of high-tech sectors in value-added exports used to assess "moving up."
- Trends:
  - Share of labor-intensive manufacturing in value-added exports has decreased over time for many countries in Asia and Latin America.
  - For European countries, labor-intensive share in total manufacturing of GVC exports relatively stable.
  - Rise in share of services exports in total value-added exports, most drastic in US, Japan, Germany, and China; more modest elsewhere.
- Bilateral value-added export analysis (example: Germany’s auto supply chain):
  - Panel analysis decomposes FVA in Germany's auto exports by participant country and sector.
  - Normalized ratios of hi(low) tech services (manufacturing) by contributor country mostly cluster around 1 (relative importance stable), but with large heterogeneity:
    - China and Denmark: shift towards more high-tech services in German auto supply chain.
    - Russia: became more intensive in low-tech manufacturing (mining and quarrying).
    - Czech Republic, Hungary, Poland, Slovak Republic: ratios close to 1.
    - Romania: shifted away from low-tech to more high-tech manufacturing.

---

### Share of Services in Value-Added Exports — Determinants of Bilateral GVC Participation
- Shift toward services and high-tech manufacturing observed overall, with heterogeneity across countries.
- Structural gravity specification for bilateral FVA (equation (6)):
  - log(FVA_ijt) = β X_ijt + η_it + η_jt + ε_ijt
  - X_ij includes distance, common language, common currency, colonial ties; policy variables include preferential trade agreement and exchange rate volatility.
- Fixed-effects decomposition (equation (7)) used to analyze time-invariant country characteristics explaining deviations from fundamentals.

### Key Determinants (selected coefficient estimates from Table 3, Table 5, Table 6)
- Bilateral determinants (Table 3; coefficients with standard errors):
  - Log (distance): -1.010*** (0.019); -0.752*** (0.013); -0.579*** (0.013); -0.574*** (0.013).
  - Common border: 1.823*** (0.071); 1.390*** (0.096); 1.386*** (0.078); 1.386*** (0.078).
  - Common language: 0.349*** (0.025); 0.266*** (0.022); 0.261*** (0.022).
  - Common colonial history: 0.402*** (0.097); 0.494*** (0.096); 0.513*** (0.095).
  - Common currency: 1.513*** (0.107); 1.416*** (0.106).
  - Preferential Trade Agreement: 0.648*** (0.024); 0.627*** (0.024).
  - Exchange rate volatility: -1.183*** (0.207).
  - Observations: 544,170; 544,170; 544,170; 537,775. R-squared: 0.899; 0.907; 0.913; 0.913.
  - Interpretation: proximity and standard country-pair characteristics strongly influence bilateral GVC participation; same currency and lower exchange rate volatility increase participation.
- Exporting-side time-invariant characteristics (Table 5; selected coefficients):
  - Log(GDP): 0.740*** (0.024); 0.753*** (0.022); 0.705*** (0.028); 0.742*** (0.025); 0.742*** (0.025); 0.841*** (0.037); 0.826*** (0.042).
  - Contract enforcement: 0.085*** (0.027).
  - Rule of law: 0.167*** (0.062).
  - Business entry procedures: -0.024* (0.014).
  - Unit labor costs (lag): -0.411** (0.169).
  - Human capital (lag): 0.564** (0.243).
  - Infrastructure quality: 0.075 (0.085).
  - Observations vary (e.g., 531,527; 482,021; 528,316); R-squared range: 0.854 to 0.934.
  - Interpretation: larger GDP, stronger contract enforcement and rule of law, fewer business entry procedures, lower unit labor costs, higher human capital, and better infrastructure associated with higher exporting-country GVC participation.
- Importing-side time-invariant characteristics (Table 6; selected coefficients):
  - Log(GDP): 0.782*** (0.035); 0.770*** (0.031); 0.685*** (0.035); 0.783*** (0.033); 0.721*** (0.032); 0.781*** (0.066); 0.725*** (0.035).
  - Contract enforcement: 0.231*** (0.040).
  - Rule of law: 0.501*** (0.079).
  - Business entry procedures: -0.088* (0.020).
  - Human capital (lag): 0.926*** (0.121).
  - Infrastructure quality: 0.412*** (0.064).
  - Observations vary (e.g., 531,581; 481,785; 528,390); R-squared range: 0.765 to 0.854.
  - Interpretation: contract enforcement, rule of law, human capital, and infrastructure quality strongly positively associated with importing-country GVC participation.

### Country Fixed Effects and Institutional Quality
- Country fixed effects from regression (6) identify countries participating above or below fundamentals (e.g., among European countries, Turkey, Albania, Bosnia and Herzegovina tend to have lower GVC participation than implied by fundamentals).
- Strong institutional characteristics (business environment, infrastructure, contract enforcement, rule of law) facilitate participation on both exporting and importing sides.
- Survey indicators used include World Economic Forum Global Competitiveness Index (Quality of Infrastructure) and Doing Business Indicators (Contract enforcement); contract enforcement normalized 0-10 for 2013.

---

### Concluding Findings and Policy Recommendations
- Measurement: Eora MRIO Database enables consistent measurement of GVC participation and position.
- Documented patterns:
  - Manufacturing and services participate differently in GVCs; both forward and backward measures matter.
  - Substantial heterogeneity in GVC participation and position across countries and industries.
  - Participation in GVCs (rather than conventional trade) can positively affect economic performance; gains are heterogeneous and stronger for upper middle and high-income countries.
  - "Moving up" to hi-tech sectors is not automatic; many countries show little change in sectoral composition of participation.
  - Upstream sectors and services are more sensitive to trade barriers.
- Policy recommendations:
  - Improve infrastructure, connectivity, and institutions (contract enforcement, rule of law, ease of doing business).
  - Given rising role of services in GVC trade, better understand barriers to services trade and design reforms and trade agreements to facilitate services participation in GVCs.

*Source: IMF Working Paper — Authors’ calculations using the Eora MRIO Database (selected chapters and tables).*

### 1. Traditiona l and  GVC Trade ______________________________________________ 3

### 1. Traditional and GVC Trade

### Introduction and overview
- Over the last two decades, trade and production have become increasingly organized around what is commonly referred to as global value chains (GVC) or global supply chains.
- Advances in information and transportation technologies and falling trade barriers have allowed firms to unbundle production into tasks performed at different locations to take advantage of different factor costs (Feenstra and Hanson 1997; Grossman and Rossi-Hansberg 2008).
- Production fragmentation implies intermediate goods and services cross borders several times along the chain, often passing through many countries more than once.
- Figure 1 contrasts "Traditional trade" (exports of goods and services produced in one country and absorbed in the destination) and "GVC Trade" (measured as percent of nominal world GDP).

### Theoretical channels and expected effects
- For developed economies, GVCs provide access to more competitively priced inputs, higher variety, and economies of scale (Baldwin and Lopez‐Gonzalez, 2013).
- For emerging economies, GVCs are viewed as a fast track to industrialization: internationally fragmented production allows emerging economies to join existing supply chains instead of building them (Baldwin 2011).
- Theoretical mechanisms by which offshoring and GVCs raise productivity include:
  - finer division of labor across countries (Grossman and Rossi-Hansberg 2008);
  - availability of greater input varieties (Halpern, Koren, and Szeidl 2015);
  - increased competition, learning externalities, and technology spillovers (Li and Liu (2014), Kee (2015)).
- Welfare gains can be larger within a multiple-sector framework that considers input-output linkages (e.g., Caliendo and Parro 2015, Ossa 2015).

### Empirical evidence and data developments
- Empirical work documented a considerable rise in fragmentation of production:
  - Seminal works (Hummels, Ishii, and Yi 2001; Hummels, Rapoport and Yi 1998) show GVCs accounted for a large share of trade growth in world trade from 1970-1990s.
  - Johnson and Noguera (2012) and Baldwin and Lopez-Gonzales (2013) show growth accelerated further in 2000-09.
- The pace of expansion of supply chains slowed after the global financial crisis, contributing to the trade slowdown in that period (IMF, 2016a).
- Methodological progress: Koopman and others (2014) and Wang and others (2013) proposed methodologies to decompose gross trade flows into origins of value-added.
- Data initiatives enabling GVC analysis:
  - Trade in Value-Added Statistics (covering 63 countries)
  - World Input Output Database (43 countries)
  - Eora Multi-Region Input-Output (MRIO) database (Eora) for 189 countries (Lenzen and others, 2013).
- With expanded data coverage, research focus is shifting toward impacts of GVC participation on economic outcomes (Constantinescu and others, 2017; World Bank 2016, and 2017).

### Role for growth and convergence
- EORA's broad country coverage allows a more comprehensive study of GVC impacts on income growth and convergence across different income levels.
- The narrative that supply chains facilitate convergence and growth (notably for China and some Asian countries) exists, but uncovering causal relations has been challenging due to prior data limitations and smaller country samples in other databases (discussion deferred to section IV).

*Source: IMF Working Paper (chapter "1. Traditional and GVC Trade")*

### 1. GVC and Income per Capita 2. GVC and Convergence

### 1. GVC and Income per Capita 2. GVC and Convergence

### Contribution and scope
- Constructs indicators of GVC participation and position for 180 economies at the country, sectoral, and bilateral levels.
- Core dataset: Eora Multi-Regional Input-Output (MRIO) database covering 189 countries and 26 sectors over the period 1990-2013.
- Methodology: decomposition of gross exports into foreign value-added (FVA) and domestic value-added (DVA) following Koopman and others (2011) and Aslam and others (2017).
- GVC participation (vertical specialization) defined as the sum of forward linkages and backward linkages unless otherwise stated.
- Position indices: Upstreamness index (number of additional production stages before final consumer) and Downstreamness index (number of previous stages embodied in production).

### Stylized facts on GVC trends and sectoral patterns
- Between 2000 and 2013, the share of exports involved in GVC trade in Europe increased from about 60 to 70 percent.
- Over 2000-2013, the GVC trade share increased on average by 5 percentage points in Asian and Western Hemisphere countries.
- The share of services exports in total world exports increased by 5 percentage points over 2000-13.
- When measured in value-added terms, the share of services exports in world exports is almost twice as large as under the gross exports concept.
- Services (e.g., financial and business services, wholesale trade) exhibit high forward linkages and limited backward linkages; services tend to have higher upstream index values than downstream index values.
- Major manufacturing sectors tend to have sizable backward linkages and higher downstream than upstream indices.

### Key statistics (sector and country highlights)
- Table 1 excerpts (GVC Participation and Position by Sector, 2013):
  - Electrical and Machinery (Value added concept share in world exports): 37.7; Gross concept: 46.7; Backward linkage: 32; Forward linkage: 15; Downstream Index: 2.8; Upstream Index: 2.4.
  - Financial and Business Services (Value added concept share): 20.7; Gross concept: 5.7; Backward linkage: 8; Forward linkage: 91; Downstream Index: 1.7; Upstream Index: 2.4.
  - Transport Equipment (Value added concept share): 5.2; Gross concept: 8.1; Backward linkage: 37; Forward linkage: 10; Downstream Index: 3.0; Upstream Index: 2.0.
- Table 2 highlights (largest supply chains, 2013) examples:
  - CHN Electrical and Machinery: Share in World Exports (Value added concept) 1.43; Gross concept 3.12; Backward linkage 41; Forward linkage 23; Downstream Index 3.8; Upstream Index 2.9.
  - USA Financial and Business Activities: Share in World Exports (Value added concept) 3.9; Gross concept 1.7; Backward linkage 36; Forward linkage 61; Downstream Index 1.6; Upstream Index 2.2.

### Main empirical findings
- Participation in GVCs has a positive impact on:
  - Income per capita.
  - Investment.
  - Productivity.
- These gains are associated with GVC trade measures (intermediate goods and GVC-related trade) rather than conventional final-goods trade, and results appear robust to endogeneity and reverse causality concerns.
- Heterogeneity: upper-middle and high-income countries exhibit robust positive effects from GVC participation; effects for low and lower-middle income countries are less robust.
- Upstreamness relationship with development: not straightforward — financial and business services are upstream and high in value-added, but the link is less clear in manufacturing.
- Moving up value chains occurs in some cases (movement into more high-tech sectors) but is not universal; gains are conditional on other factors.

### Empirical framework and channels
- Theoretical motivation: Cobb-Douglas production function decomposing gross output into domestic value-added and foreign value-added; GDP per capita decomposition into capital intensity, human capital, and productivity (equations (1)–(4) in the text).
- Reduced form empirical specification (equation (5)) regresses log GDP per capita on lagged shares of trade and GVC participation variables, controlling for country characteristics (population, area), country and year fixed effects.
- Channels examined: investment, human capital, and productivity (productivity measured as residual from decomposition; human capital index from Barro and Lee (2010)).

### Regression evidence (selected results from Table 3)
- Column (1) — Share of intermediate trade in GDP:
  - Coefficient on Share of Intermed. trade in GDP: 1.610*** (0.339).
  - Observations: 3,049; R-squared: 0.172.
- Column (2) — Distinguishing GVC-related and non-GVC-related trade:
  - Coefficient on Share of GVC-trade in GDP: 2.738** (1.148).
  - Coefficient on Share of regular trade in GDP: 0.227 (0.458).
  - Observations: 3,049; R-squared: 0.183.
- Channels (columns (3)–(5)):
  - Column (3) Log(I/N) (investment per capita): GVC-related trade positively affects investment; Share of Intermed. trade results reported generally positive.
  - Column (4) Human capital: Share of GVC-trade in GDP: 0.367*** (0.103).
  - Column (5) Productivity: Share of GVC-trade in GDP: 1.366* (0.750).
  - Observations vary: 3,045 (investment), 2,777 (human capital), 2,625 (productivity).
- Heterogeneity by income class (column (7), interaction of GVC-trade with income dummies):
  - GVC-trade*(Low Income): -0.947 (0.3875).
  - GVC-trade*(Low Middle Income): 1.413* (0.915).
  - GVC-trade*(High Middle Income): 2.143** (0.855).
  - GVC-trade*(High Income): 2.848*** (0.948).
  - Observations: 2,770; R-squared: 0.160.
- Position indices (column (8)):
  - Upstream Index coefficient: -0.370 (0.2561).
  - Downstream Index coefficient: 0.102 (0.229).
  - Sample excludes commodity exporters for this specification; Observations: 1,165; R-squared: 0.275.
- Instrumental Variables (column (9)):
  - IV for share of GVC-trade in GDP constructed as the average FVA from Germany, Japan, and the United States embodied in exports of three countries closest in income level (or geography) to the country in question.
  - Column (9) results broadly similar in magnitude and significance to OLS estimates.
  - Observations: 2,320; R-squared: 0.132.
- Sample robustness:
  - Column (6) Small Sample (restricting to 50 countries typical in other value-added databases) fails to replicate column (1) results, highlighting the importance of the larger Eora database. Observations: 970; R-squared: 0.170.

### Economic significance
- An increase in the share of GVC trade in GDP by 10 percentage points (from median country to 75th percentile) would represent an increase in income levels by 15 to 30 percent (based on coefficients in columns (2) and (7) of Table 3).

### Mechanisms, endogeneity, and interpretation
- Possible mechanisms: productivity effects from task specialization and offshoring, access to better or cheaper inputs, knowledge spillovers, and investment attraction via GVC links.
- Bilateral GVC participation and bilateral FDI volumes are positively related (Figure 8), suggesting FDI could be an important channel.
- Endogeneity concerns addressed via lagged trade variables and an IV strategy that aims to isolate exogenous variation in GVC participation driven by foreign value-added patterns in similar-income (or neighboring) countries.
- Caveats: gains from GVC participation are not automatic; they depend on country-specific complementarities such as institutions, contract enforcement, and infrastructure quality. For low-income countries, effects are not robust and may require additional supporting factors to materialize.

### Moving up value chains
- “Moving up” (e.g., shifting to upstream or higher-tech sectors) is not unambiguously linked to higher income: position index does not show a straightforward relationship with income level.
- High-income countries may participate across many stages of production, including manufacturing; position indices alone are insufficient to gauge “moving up.”
- Evidence suggests some shifts into higher-tech sectors occur with GVC participation, but such shifts are not universal and are conditional on other factors.

*Sources: Authors’ calculations and Eora database; excerpted content from the IMF working paper chapter titled "1. GVC and Income per Capita 2. GVC and Convergence."*

### 1. Economy-Wide Upstreamness

### 1. Economy-Wide Upstreamness

### Changes in sectoral composition of value-added exports
- A useful indicator to explore the “moving up” concern is how the share of high-tech sectors in value-added exports of a country has changed over time.
- The share of labor-intensive manufacturing in value-added exports has decreased over time for many countries in Asia and Latin America.
- For European countries, the labor-intensive share in total manufacturing of GVC exports has been relatively stable over time, closer to the 45-degree line in Figure 10, panel 1.
- For many countries there has been a rise in the share of services exports in total value-added exports; this rise is most drastic in the US, Japan, Germany, and China, while changes in the rest of the countries have been more modest.

### Bilateral value-added export analysis and global supply chains
- Using bilateral value-added export data, participation in major global supply chains (listed in Table 3) is examined to assess effects on sectoral composition of participant countries.
- Example: Germany’s auto supply chain is broken down by FVA in Germany’s auto exports to value-added by participant country and sectors.
- Panel 1 of Figure 11 shows contributions of different low- and high-tech manufacturing and services to Germany’s auto supply chain in 2013.
- For each contributing country, the ratio of hi(low) tech services (manufacturing) in its contribution is calculated and normalized by the value of that ratio in year 2000.

### Empirical patterns across participant countries
- On average, the normalized ratios are clustered around 1, reflecting that the relative importance of each sector has on average remained the same over time.
- There is large heterogeneity across countries and categories:
  - China’s and Denmark’s contributions to the German auto supply chain show a shift towards more high-tech services.
  - Russia’s contribution to the German auto supply chain became more intensive in low-tech manufacturing (due to mining and quarrying sector).
  - The Czech Republic, Hungary, Poland, and Slovak Republic have ratios close to 1, reflecting broadly similar sectoral composition over time.
  - Romania has shifted away from low-tech to more high-tech manufacturing.

_Authors’ calculations of GVC position index at country-sector level; and World Economic Outlook._

### 2. Share of Services in Value-Added Exports

### 2. Share of Services in Value-Added Exports

### Services and sectoral shifts in GVC participation
- Moving "up" the value-chain toward services or more high-tech manufacturing has taken place overall, with large heterogeneity across countries.
- The shift to high-tech sectors may also occur within narrower sub-sectors not visible in the 26-sector aggregation used here; more granular data could provide additional insights.
- Notable moves to high-tech services are documented for the US, China, Germany, and Japan; larger sectoral transformations are observed in some Asian countries and are more modest in European countries.

### Empirical strategy for determinants of bilateral GVC participation
- Structural gravity specification:
  - log(FVA_ijt) = β X_ijt + η_it + η_jt + ε_ijt  (equation (6))
    - X_ij includes country-pair characteristics: distance, common language, common currency, colonial ties.
    - η_it and η_jt are time-varying source- and destination-country fixed effects.
    - Policy variables included: indicator for a preferential trade agreement and exchange rate volatility.
- Time-invariant characteristics investigation:
  - η_it (fixed effects from equation (6)) are regressed as:
    - η_it = log(GDP_it) + α X_i + η_t + ε_it  (equation (7))
  - Source- and destination-country fixed effects are used to study which country characteristics explain deviations from fundamentals.

### Key regression results (Table 3: Determinants of Bilateral GVC Participation)
- Coefficients (standard errors in parentheses):
  - Log (distance): -1.010*** (0.019); -0.752*** (0.013); -0.579*** (0.013); -0.574*** (0.013)
  - Common border: 1.823*** (0.071); 1.390*** (0.096); 1.386*** (0.078); 1.386*** (0.078)
  - Common language: 0.349*** (0.025); 0.266*** (0.022); 0.261*** (0.022)
  - Common colonial history: 0.402*** (0.097); 0.494*** (0.096); 0.513*** (0.095)
  - Common currency: 1.513*** (0.107); 1.416*** (0.106)
  - Preferential Trade Agreement: 0.648*** (0.024); 0.627*** (0.024)
  - Exchange rate volatility: -1.183*** (0.207)
- Additional regression details:
  - Country-Year Fixed Effects: Yes (in all columns)
  - Observations: 544,170; 544,170; 544,170; 537,775
  - R-squared: 0.899; 0.907; 0.913; 0.913
- Interpretation:
  - Physical proximity and standard country-pair characteristics (common border, language, colonial links) are important determinants of GVC participation measured via foreign value-added.
  - Having the same currency or lower exchange rate volatility increases bilateral value-chain participation.
  - Industry-level estimates for 2013 (with source country-sector and destination country-sector fixed effects) show geographical proximity matters more for manufacturing; upstream industries are more sensitive to distance.

### Time-invariant country characteristics (exporting side, Table 5)
- Selected coefficients (standard errors in parentheses):
  - Log(GDP): 0.740*** (0.024); 0.753*** (0.022); 0.705*** (0.028); 0.742*** (0.025); 0.742*** (0.025); 0.841*** (0.037); 0.826*** (0.042)
  - Contract enforcement: 0.085*** (0.027)
  - Rule of law: 0.167*** (0.062)
  - Business entry procedures: -0.024* (0.014)
  - Unit labor costs (lag): -0.411** (0.169)
  - Human capital (lag): 0.564** (0.243)
  - Infrastructure quality: 0.075 (0.085)
- Observations and fit:
  - Observations: 531,527; 482,021; 528,316; 288,499; 288,499; 114,067; 98,350
  - R-squared: 0.854; 0.905; 0.860; 0.843; 0.843; 0.924; 0.934
- Interpretation:
  - Larger GDP, stronger contract enforcement and rule of law, fewer procedures to set up a business, lower unit labor costs, higher human capital, and better infrastructure are associated with higher exporting-country GVC participation.
  - High labor costs in the exporting country decrease competitiveness and thus participation in GVCs.

### Time-invariant country characteristics (importing side, Table 6)
- Selected coefficients (standard errors in parentheses):
  - Log(GDP): 0.782*** (0.035); 0.770*** (0.031); 0.685*** (0.035); 0.783*** (0.033); 0.721*** (0.032); 0.781*** (0.066); 0.725*** (0.035)
  - Contract enforcement: 0.231*** (0.040)
  - Rule of law: 0.501*** (0.079)
  - Business entry procedures: -0.088* (0.020)
  - Unit labor costs (lag): 0.016 (no standard error reported in table cell)
  - Human capital (lag): 0.926*** (0.121)
  - Infrastructure quality: 0.412*** (0.064)
- Observations and fit:
  - Observations: 531,581; 481,785; 528,390; 288,605; 462,710; 113,306; 371,482
  - R-squared: 0.765; 0.836; 0.815; 0.785; 0.841; 0.821; 0.854
- Interpretation:
  - For importing countries, contract enforcement, rule of law, human capital, and infrastructure quality are strongly positively associated with GVC participation.
  - High labor costs are not a significant determinant on the importing side in these specifications.

### Country fixed effects and institutional quality (Figures 12 and 13)
- Country fixed effects from regression (6) reveal which countries participate in GVCs above or below fundamentals; for example, among European countries, Turkey, Albania, and Bosnia and Herzegovina tend to have lower GVC participation than implied by fundamentals.
- Strong institutional characteristics (business environment, infrastructure, contract enforcement, rule of law) facilitate participation in GVCs on both exporting and importing sides.
- Survey-based indicators used:
  - World Economic Forum Global Competitiveness Index (Quality of Infrastructure)
  - Doing Business Indicators (Contract enforcement)
  - Data is for 2013; contract enforcement score normalized 0-10 (strongest).

### Concluding findings and policy implications
- The Eora MRIO Database is used to measure GVC participation and position consistently with theoretical methodology.
- Key documented patterns:
  - Manufacturing and services participate differently in GVCs; both forward and backward measures matter.
  - There is substantial heterogeneity in GVC participation and position across countries and industries.
  - Participation in global value chains, rather than conventional trade, can positively affect economic performance; gains are heterogeneous.
  - Upper middle and high-income countries appear to benefit, while robust effects are not found for low and lower-middle income countries.
  - "Moving up" to more hi-tech sectors is not automatic or frequent; many countries show little change in sectoral composition of participation.
  - Upstream sectors and services are more sensitive to trade barriers.
- Policy recommendations:
  - Improve infrastructure, connectivity, and institutions (contract enforcement, rule of law, ease of doing business).
  - Given the rising role of services in GVC trade, better understand barriers to services trade and design reforms and trade agreements that could facilitate services participation in GVCs.

*Source: IMF Working Paper — Authors’ calculations using the Eora MRIO Database (content from "2. Share of Services in Value-Added Exports" and related sections).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wp1918.pdf_