Export Performance in Europe: What Do We Know from Supply Links?

Jesmin Rahman; Tianli Zhao

doi:10.5089/9781475516555.001.A001

Jump to Content Jump to Main Navigation

Export Performance in Europe

What Do We Know from Supply Links?

Author: Jesmin Rahman https://orcid.org/0000-0002-2324-3537 and Mr. Tianli Zhao¹

¹ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

Author’s E-Mail Address: jrahman@imf.org; tz49@cornell.edu

Publication Date:: 07 Mar 2013

eISBN:: 9781475516555

Language:: English

Keywords:: WP; VA export; export production; export structure; performance in Europe; sectoral export evolution; supply link export

Download PDF (1.5 MB)

One of the most important recent developments in international trade is the increasing interconnectedness of export production through a vertical trading chain network that streches across many countries, with each country specializing in particular stages of a good’s production. Using value added trade statistics, this paper tries to dissect and reshape understanding of European exports: where exports values are created, the role of vertical supply links in export growth, what is contributing to the growth in supply links, and how comparative advantages of countries are affected by supply links over time. Our analysis finds strong role of supply links in cross-country export performance in Europe, where these links between countries grew based on physical proximity, cost differential and similarity in export structure.

Abstract

I. Motivation¹

Over the past few decades, dramatic changes have taken place in the way international trade occurs between countries. Production processes have increasingly involved a sequential, vertical trading chain stretching across many countries, with each country specializing in one or more stages of production. This fragmentation of production has made intra-industry trade dominate world merchandise trade. As products cross border multiple times, this has also resulted in world trade growing faster than both global GDP and global value-added (VA) in manufacturing (Figure 1).

The increasingly fragmented production process in tradables has come with some data and policy challenges. As all official trade statistics are measured in gross terms, which include both intermediate inputs and final products, they “double count” a part of the value of goods: the part that crosses international borders more than once. As a result, official trade statistics are becoming increasingly less reliable as a gauge of value contributed by any particular country, reducing its reliability as a tool to measure export competitiveness and form policy advice.

To illustrate the point, suppose a German car maker ships $50,000 worth of car components to its subsidiary in Hungary. A factory in Hungary then assembles the car and sells it to a dealership in France at $55,000. The gross or official trade statistics would record $50,000 worth of exports from Germany to Hungary as well as $55,000 worth of export from Hungary to France (Figure 2). But in VA terms, Hungary’s exports to France would be only $5,000.

The shortcomings of gross trade statistics, as well as their inconsistency with the System of National Accounts standards, have been well recognized (Hummels and others, 2001; Ando and Kimura, 2003; Koopman and others, 2008 and 2011; Breda and others, 2008; and Bems and Johnson, 2012). But the problem goes well beyond that of data inconsistency. A bigger role of supply links in exports growth implies a large and possibly increasing role of foreign VA. However, if a country’s exports growth is driven mostly by value crossing borders rather than being produced domestically, its impact on growth and employment is negligible. To get a true picture of a country’s exports growth, we need to strip the foreign VA component from total exports. At the same time, given the importance of foreign VA or supply links as an engine for exports growth, we need to also understand the symbiotic relationship between foreign and domestic VA components of exports.

A recent paper by Koopman and others (2011) develops the first unified decomposition method that allows a full concordance between VA trade and gross trade statistics. In this paper, we adopt their framework to decompose gross exports data into VA measures using the newly released world input-output table. This enables us to (i) make a connection between gross/official statistics and VA statistics in merchandise and services trade, and (ii) distribute all VA embedded in a country’s exports to its original sources at the country and product level. By analyzing trends and developments in the decomposed flow data, this paper aims to reshape our understanding of international trade in Europe: where values are created, the role of vertical supply links in export growth, what factors contribute to the growth of supply links, and how comparative advantages of countries are affected by supply links over time.

Although our sample includes a total of 40 countries, we mainly focus on developments in Europe. Since mid-1990s, a number of Central European economies, such as Czech Republic, Slovak Republic, Hungary and Poland, experienced growth that was led by the export sector. At the same time, a number of other European countries, including some periphery countries in the Euro zone (EZ), travelled a different growth path that relied on domestic demand and fast credit growth. To what extent plugging into the pan-European supply chain helped the first group achieve its export success? What factors helped them to plug into supply links? For countries in the EZ periphery that are desperately looking to increase exports to rebalance their external position and bring back growth, what lessons can be learned from the supply link experience?

We start with a description of our decomposition methodology in Section II. In Section III, we use our decomposed exports statistics to look at the role of vertical supply chain in overall export growth and competitiveness. Section IV uses regression analysis to explore what factors contribute to a firm’s decision to locate a part of its production abroad. Section V takes a close look at a set of European countries to see which countries have successfully benefited from being part of the supply network. Finally, conclusions and related policy implications are discussed in Section VI.

II. Dissecting Gross Exports in Europe

We adopt the conceptual framework developed in Koopman and others (2011) to decompose sources of VA in exports. The methodology is described in Annex 1. As shown in Figure 3, we decompose gross exports into five main categories depending on the location of VA and stage of production: (1) domestic VA in final goods, (2) domestic VA in intermediate goods not processed for further exports, (3) domestic VA in intermediate goods processed for exports to third countries, (4) domestic VA that is exported to another country but returns back to the original country for exports to a third country, and (5) VA imported from abroad as inputs into exports, i.e. foreign VA.

We compute the 5-category VA decomposition for manufacturing and services exports respectively using data from the world input-output table (Annex 2). Components (1) through (4) give us the value of exports that is created domestically, while component (5) gives us the value of exports created abroad. Components (1-2) tell us how much of a country’s exports are created as stand-alone exports, i.e. outside any supply chain, while components (3-5) give us exports generated by supply links. Supply link related exports have two components: upstream, which include domestic VA intermediate exports that are processed for further exports (components 3-4), and downstream, which include foreign VA exports (component 5). A large share of foreign VA in a country’s exports signifies its position as a downstream processor or assembler.

Based on the above decomposition, we discuss some key developments in manufacturing and services exports during 1995–2008. Detailed tables are in Annex 3 (Tables 1a and 1b).²

The share of domestic VA has declined over time. During 1995–2008, the average share of domestic VA (components 1-4) in manufacturing exports in our sample countries declined from 72 to 62 percent (Figure 4). Similar declines were visible in Europe and sub-groups of countries in Europe, where the share of domestic VA in total exports declined by 9 to 13 percentage points. The decline in the share of domestic VA in services trade was less pronounced, reflecting a lower degree of fragmentation in international trade in services.
The role of supply links increased over time. During 1995–2008, the average share of world manufacturing exports produced via supply links (components 3-5) went up from 42 to 54 percent (Figure 4). Increases of similar magnitude were experienced by Europe and country sub-groups in Europe. For services, the average share of supply link related exports increased from 32 to 42 percent, implying a pace of increase that is similar to that in manufacturing.
Downstream activities dominate supply links. For example, on average European countries imported 41 percent of their manufacturing exports from abroad in 2008 (Figure 3). For several countries, namely, Slovak and Czech Republics, Hungary, and Lithuania, the share is over 50 percent. Such a high share of foreign VA components in exports indicates that downstream assembly plays a strong role in export growth. For advanced countries in Europe apart from the EZ periphery, the lesser dominance of downstream activities reflects a higher share of domestically produced stand-alone exports.

The above discussion shows that supply chains have dominated exports of European countries, with a high share of value being produced abroad. When such a high share of a country’s exports is created abroad, it is natural to ask how do these countries perform if the foreign VA added component is excluded? What role has foreign VA played in these countries’ overall export growth? Has the reliance on supply links been beneficial, or increasing fragmentation of export production simply shifted abroad a part of what was previously produced domestically as firms sought to reduce costs?

To answer these questions, we normalize gross exports, and its two main sub-components, domestic and foreign VA, by GDP. The change in the ratio of exports to GDP is often interpreted as “beyond the trend” growth, where an increasing exports to GDP ratio implies that a country’s growth is orienting itself more towards export, and less toward domestic demand. The percentage increase in gross exports over GDP is simply the sum of percentage increase in domestic VA over GDP and percentage increase in foreign VA over GDP.

Figure 5 compares growth in domestic VA exports and gross exports during 1995–2008. While the increase in domestic VA exports is lower than gross exports in all European countries, the difference between the two seems particularly large for Belgium, Bulgaria and Ireland, where large increases in gross exports to GDP ratio during 1995–2008 mostly reflected increasing foreign VA. However, these increases should be considered together with the level of domestic VA. For example, if an economy has a large domestic VA exports to GDP ratio (for example, Ireland), the room for catch-up increase may be less than an economy where this ratio is low (for example, Greece). Figure 5 therefore also shows the average ratio of domestic VA exports to GDP in European countries during 1995–2008.

We divide European countries into four groups based on the increase in the ratio of domestic VA exports of goods and services to GDP during 1995–2008 (Table 1, across columns). An increase in this share implies that a country increased its export orientation in growth during this time period, which is also a sign of increasing competitiveness. We find that most European countries in our sample increased their domestic VA exports to GDP ratio, i.e. their export orientation in growth during this time period. However, as mentioned before, this needs to be contextualized in terms of a country’s overall export orientation. Therefore, across rows in Table 1, we also show the average domestic VA exports to GDP ratio in European countries. A position in the north-western corner in Table 1 shows high and increasing export orientation, while a position in the south-eastern corner shows low and declining export orientation during 1995–2008.

Table 1.

Domestic VA Export Performance in Europe, 1995–2008

Source: Authors’ calculation using world input output table.

Table 1.

Domestic VA Export Performance in Europe, 1995–2008

	Domestic VA exports/GDP increase more than 10 ppts	Domestic VA exports/GDP increase more than 5 ppts	Domestic VA exports/GDP increase less than 5 ppts	Domestic VA exports/GDP decline
Average domestic VA exports/GDP greater than 30 percent	Austria and Hungary	Czech republic, Malta, Slovenia and Sweden.	Belgium, Bulgaria, Estonia, Ireland, and Netherlands
Average domestic VA exports/GDP between 20-30 percent	Germany and Slovak Republic	Denmark, Lithuania, Malta and Poland	Finland, Romania, Russia and the UK.	Latvia
Average domestic VA exports/GDP less than 20 percent		Greece	France, Portugal, Spain, and Italy	Cyprus and Turkey

Source: Authors’ calculation using world input output table.

Table 1.

Domestic VA Export Performance in Europe, 1995–2008

	Domestic VA exports/GDP increase more than 10 ppts	Domestic VA exports/GDP increase more than 5 ppts	Domestic VA exports/GDP increase less than 5 ppts	Domestic VA exports/GDP decline
Average domestic VA exports/GDP greater than 30 percent	Austria and Hungary	Czech republic, Malta, Slovenia and Sweden.	Belgium, Bulgaria, Estonia, Ireland, and Netherlands
Average domestic VA exports/GDP between 20-30 percent	Germany and Slovak Republic	Denmark, Lithuania, Malta and Poland	Finland, Romania, Russia and the UK.	Latvia
Average domestic VA exports/GDP less than 20 percent		Greece	France, Portugal, Spain, and Italy	Cyprus and Turkey

Source: Authors’ calculation using world input output table.

Strong growth in export orientation. These countries experienced a double-digit increase in their domestic VA exports to GDP ratio. Four European countries managed to achieve such increase: Austria, Germany, Hungary and Slovak Republic. During 1995-2008, they also maintained an average domestic VA exports to GDP ratio between 26 to 32.
Moderate growth in export orientation. This group experienced an increase in the domestic VA exports to GDP ratio between 5 and 10 percentage points during 1995-2008. There are eight countries in this group: Czech Republic, Poland, Lithuania, Sweden, Denmark, Malta, Greece and Slovenia. However, the growth masks considerable heterogeneity in the importance of exports across these countries. For example, the average domestic VA exports to GDP ratio was 36 in Czech Republic, while this share was only 9 in Greece (the lowest in our European sample). For the rest, the range was between 22 to 32 percent. So while Czech Republic has a high and increasing export orientation in growth, Greece shows a low but increasing export orientation in growth during this period.
Mild growth in export orientation. This group contains countries where domestic VA exports to GDP ratio increased by less than 5 percentage points during 1995-2008. Thirteen European countries, or about a half of our sample, belong to this group. Just like the previous group, the members are heterogeneous in terms of the importance of exports in growth. For example, the average ratio of domestic VA exports to GDP was 47 in Ireland making it the most export-oriented economy in our sample. This ratio, on the other hand, was only 17 and 18 in Portugal and Spain, respectively reflecting larger room for export-led growth.
Declining growth in export orientation. Three European countries show a decline in domestic VA exports to GDP ratio during 1995-2008. The share of domestic VA exports remained flat during the boom years of 2000s in these countries, which reflect the stronger role played by domestic demand in growth throughout this period (Figure 5). Turkey and Cyprus also show a relatively low share of domestic VA exports in GDP at 14 and 15 percent, making the declining importance of exports more of a concern for export competitiveness than in Latvia, where the average share of domestic VA exports was above 25 percent.

Next, we try to see to what extent foreign VA, which has been a strong engine of exports growth in much of Europe, helped in increasing domestic VA exports, which is after all what counts for job and economic growth. We see a strong positive relationship between change in a country’s foreign VA and domestic VA exports expressed in percent of GDP (Figure 6). We also test for whether increasing foreign VA exports cause domestic VA exports to grow, i.e. the ability of downstream assembly function to create domestic jobs and growth. Specifically, we test the impact of foreign VA growth of up to 5-year lag on domestic VA:

\begin{matrix} 1 - year lag : log \frac{{DV}_{t}}{{DV}_{t - 1}} = β_{0} + β_{1} log \frac{{FV}_{t - 1}}{{FV}_{t - 2}} + Σ δ_{i} {year}_{i} & (1) \end{matrix}

\begin{matrix} 2 - year lag : log \frac{{DV}_{t}}{{DV}_{t - 1}} = β_{0} + β_{1} log \frac{{FV}_{t - 2}}{{FV}_{t - 2}} + Σ δ_{i} {year}_{i} & (2) \end{matrix}

\begin{matrix} 3 - year lag : log \frac{{DV}_{t}}{{DV}_{t - 1}} = β_{0} + β_{1} log \frac{{FV}_{t - 3}}{{FV}_{t - 4}} + Σ δ_{i} {year}_{i} & (3) \end{matrix}

\begin{matrix} 4 - year lag : log \frac{{DV}_{t}}{{DV}_{t - 1}} = β_{0} + β_{1} log \frac{{FV}_{t - 4}}{{FV}_{t - 5}} + Σ δ_{i} {year}_{i} & (4) \end{matrix}

\begin{matrix} 5 - year lag : log \frac{{DV}_{t}}{{DV}_{t - 1}} = β_{0} + β_{1} log \frac{{FV}_{t - 5}}{{FV}_{t - 6}} + Σ δ_{i} {year}_{i} & (5) \end{matrix}

We find a positive and statistically significant relationship between foreign VA and domestic VA export growth for all lag specifications (Table 2). Therefore, increasing foreign VA exports during 1995–2008 resulted in increasing domestic VA exports. As world GDP growth was driven by growth in world trade, and world trade growth was driven by supply links, foreign and domestic VA were complementary to each other creating a virtuous circle for countries that were able to plug into regional or global vertical supply chains.

Table 2.

Impact of Foreign VA Growth on Domestic VA Growth

Table 2.

Impact of Foreign VA Growth on Domestic VA Growth

	(1)	(2)	(3)	(4)	(5)	(6)
Foreign VA Growth in period t-1	0.015					0.078
	(0.47)					(2.2)
Foreign VA Growth in period t-2		0.101				0.089
		(3.26)				(2.6)
Foreign VA Growth in period t-3			0.135			0.116
			(4.32)			(3.64)
Foreign VA Growth in period t-4				0.110		0.099
				(3.55)		(3.33)
Foreign VA Growth in period t-5					0.153	0.116
					(4.95)	(3.79)
R-square	0.59	0.65	0.68	0.70	0.74	0.77

Table 2.

Impact of Foreign VA Growth on Domestic VA Growth

	(1)	(2)	(3)	(4)	(5)	(6)
Foreign VA Growth in period t-1	0.015					0.078
	(0.47)					(2.2)
Foreign VA Growth in period t-2		0.101				0.089
		(3.26)				(2.6)
Foreign VA Growth in period t-3			0.135			0.116
			(4.32)			(3.64)
Foreign VA Growth in period t-4				0.110		0.099
				(3.55)		(3.33)
Foreign VA Growth in period t-5					0.153	0.116
					(4.95)	(3.79)
R-square	0.59	0.65	0.68	0.70	0.74	0.77

We also notice that the increase in exports to GDP ratio during 1995–2008 was not particularly influenced by low initial values of exports to GDP ratio (Figure 6, second panel). Countries that had a high exports to GDP ratio in 1995, such as Ireland, Czech Republic, China and Taiwan, maintained or further strengthened their position over time.

These two findings, i.e. foreign VA exports contribute positively to domestic VA exports, and countries have retained/strengthened their competitive position in exports, are related. To the extent world trade is dominated by supply links and these links take time to establish, it is not surprising that countries which were already well linked in 1995 are the ones that benefitted disproportionately from growth in exports. What this implies is that success of an export-led growth strategy depends, among other things, on finding an appropriate position in the VA chain and nurturing this vertical relationship over time.

This poses an additional difficulty for countries that are not already well-linked in the European supply chains to increase the role of exports in growth (Box 1). The extent of integration with supply links, measured both by the numbers of links and volume of trade through these links, is low in some EZ periphery countries, such as Greece and Portugal. Given that supply links take time to establish, for these countries to benefit from such links would not be immediate even if conditions are conducive. In the following section, we investigate what factors help establish these supply links.

III. What Factors Help Countries Establish Supply Links?

The analysis in the previous two sections show that a group of European countries have increased their exports to GDP ratio during 1995-2008 through integration with supply links. Initially, these countries attracted hubs, such as Germany, Austria or Sweden, to locate a part of their downstream production in these countries. Over time, that created a virtuous circle whereby foreign and domestic VA increased hand in hand enhancing the role of exports in growth. To the extent success in export-led growth depends on plugging into this virtuous circle, it is important to investigate what factors contribute to a country’s decision to send a part of its production abroad.

We use an augmented gravity model to explore this question empirically. Following McCallum (1995), which has been a corner stone of gravity literature, we consider the following specification:

\ln (F V_{i j t}) = β_{0} + β_{1} \ln (Y_{i t} * Y_{j t}) + β_{2} \ln (G_{i t} * G_{j t}) + β_{3} \ln D i s t_{i j} + Σ λ_{k} C X_{k} + Σ α_{n} S_{n} + Σ μ_{t} T_{t} + ε_{i j t}

where i and j denote countries, and the variables are defined as follows:

FV_ij is the foreign VA from country i embodied in country j’s export,
Y is nominal GDP,
G is GDP per capita,
Dist_ij is the distance between countries i and j,
CX_k is the set of controlled gravity variables,
S_n is the set of structural variables,
T_t is the set of time control,
ε_ij is the error term.

We use the OLS model with time dummies as our baseline equation. To check for the robustness of our estimated results, we also use two other estimation strategies, namely, OLS with no control and two-way fixed effect with both time and country-pair dummies.³

Augmented Gravity Variables

The original gravity equation includes GDP, per capita GDP and the distance between each pair countries. Empirical applications of the gravity equation over time have expanded to cover a wide range of issues, such as the impact of free trade arrangements (Matyas and others, 1997; Egger and Egger, 2004), currency unions (Pakko and Wall, 2001; Glick and Rose, 2002), and common border (McCallum, 1995; Anderson and Wincoop, 2003) on trade.

Following the literature, we also include these variables, i.e. common language, common border, free-trade-agreement, and a dummy variable to capture whether a country is a resource exporter or not. The purpose is to control for as many variables as possible that may explain trade flows between two countries.⁴

Σ λ_{k} {CX}_{k} = λ_{1} {Comlang}_{ij} + λ_{2} {ComBorder}_{ij} + λ_{3} {FTA}_{ij} + λ_{4} {ResourceExporter}_{i}

Our estimation results show all gravity variables to be statistically significant with the expected signs (Table 3). Higher GDP level, lower distance, the presence of a common border and common language positively affect a country’s decision to locate a part of its export production in another country. Lower tariff and free trade agreements also influence this decision positively. For example, reducing distance between countries by 1 percent increases the value of foreign VA exports by 0.5 percent. Similarly, increasing host country’s market size (i.e. GDP) by 1 percent increases foreign VA exports by 0.6 percent.

Table 3.

Regression Results of Determinants of Foreign VA

Notes: 1 denotes source country and 2 denotes recipient country

^{^**}denotes 1% significance level

^*denotes 10% significance level

Number of observations: 17640

Table 3.

Regression Results of Determinants of Foreign VA

Variable	OLS with no Control	OLS with Time Control (Baseline)	Two-way Fixed Effects
	(1)	(2)	(3)
Log of GDP_1	0.8255^**	0.8272^**	0.6695^**
	(0.0057)	(0.0057)	(0.0957)
Log of GDP_2	0.6127^**	0.6100^**	−0.8357^**
	(0.0057)	(0.0058)	(0.0986)
Log of GDP per capita_1	−0.1052^**	−0.1018^**	−0.0127
	(0.0102)	(0.0104)	(0.0910)
Log of GDP per capita_2	0.1656^**	0.1807^**	1.2998^**
	(0.0120)	(0.0121)	(0.0934)
Log of Distance	−0.5378^**	−0.5370^**
	(0.0111)	(0.0112)
Common Language Dummy	0.6847^**	0.6731^**
	(0.0399)	(0.0398)
Common Border Dummmy	0.7629^**	0.7618^**
	(0.0370)	(0.0368)
Resource-rich Dummry	0.3089^**	0.3088^**
	(0.0262)	(0.0261)
FTA Dummy	0.3350^**	0.3507^**	0.0731^**
	(0.0245)	(0.0261)	(0.0115)
Down Stream Tariff	−0.0179^**	−0.0112^**	−0.0359^**
	(0.0035)	(0.0037)	(0.0020)
Exchange Rate Volatility	−1.5051^**	−1.6656^**	0.9182^**
	(0.3819)	(0.3851)	(0.1423)
Difference in Unit Labor Costs	0.8801^**	0.8872^**	0.5983^**
	(0.0908)	(0.0903)	(0.0801)
Industry Similarity	−1.7370^**	−1.8217^**
	(0.2458)	(0.2450)

Notes: 1 denotes source country and 2 denotes recipient country

^{^**}denotes 1% significance level

^*denotes 10% significance level

Number of observations: 17640

Table 3.

Regression Results of Determinants of Foreign VA

Variable	OLS with no Control	OLS with Time Control (Baseline)	Two-way Fixed Effects
	(1)	(2)	(3)
Log of GDP_1	0.8255^**	0.8272^**	0.6695^**
	(0.0057)	(0.0057)	(0.0957)
Log of GDP_2	0.6127^**	0.6100^**	−0.8357^**
	(0.0057)	(0.0058)	(0.0986)
Log of GDP per capita_1	−0.1052^**	−0.1018^**	−0.0127
	(0.0102)	(0.0104)	(0.0910)
Log of GDP per capita_2	0.1656^**	0.1807^**	1.2998^**
	(0.0120)	(0.0121)	(0.0934)
Log of Distance	−0.5378^**	−0.5370^**
	(0.0111)	(0.0112)
Common Language Dummy	0.6847^**	0.6731^**
	(0.0399)	(0.0398)
Common Border Dummmy	0.7629^**	0.7618^**
	(0.0370)	(0.0368)
Resource-rich Dummry	0.3089^**	0.3088^**
	(0.0262)	(0.0261)
FTA Dummy	0.3350^**	0.3507^**	0.0731^**
	(0.0245)	(0.0261)	(0.0115)
Down Stream Tariff	−0.0179^**	−0.0112^**	−0.0359^**
	(0.0035)	(0.0037)	(0.0020)
Exchange Rate Volatility	−1.5051^**	−1.6656^**	0.9182^**
	(0.3819)	(0.3851)	(0.1423)
Difference in Unit Labor Costs	0.8801^**	0.8872^**	0.5983^**
	(0.0908)	(0.0903)	(0.0801)
Industry Similarity	−1.7370^**	−1.8217^**
	(0.2458)	(0.2450)

Notes: 1 denotes source country and 2 denotes recipient country

^{^**}denotes 1% significance level

^*denotes 10% significance level

Number of observations: 17640

A recent study (World Economic Forum, 2013) argues industry case studies show that non-tariff trade barriers, such as market access, border administration, telecommunications and transportation infrastructure and business environment, play an important role in hindering supply links. This may very well be the case: investors are likely to locate a part of their export production in a place where customs agencies work round the clock resulting in no delays in processing than in a place encumbered with interrupted service and frequent inspections. The lack of a long enough time-series prevents us from including this variable in our regression. Also, for our sample of countries which include a large number of European Union and OECD countries, the value of non-tariff barriers is likely to show low variability across countries. Nonetheless, a scatter plot of foreign VA and the value of World Economic Forum’s Enabling Trade Index, which is a composite of market access, border administration, telecommunication and transportation infrastructure and business environment, shows a mild positive relationship. In other words, lower non-tariff barriers help with supply links.

Structural Variables:

In addition, we include a list of structural variables that are commonly thought to drive fragmentation of export production. These include labor cost differential, initial level of similarities in industrial structure, and exchange rate volatility.⁵

Σ α_{n} S_{n} = α_{1} ({ULC}_{it} - {ULC}_{jt}) + α_{2} {Sim}_{ijt} + α_{3} {VolatilityEX}_{ijt}

Unit labor cost differential is equal to the unit labor cost in country i minus the unit labor cost in country j. Our estimation results show a statistically significant positive coefficient for this variable: countries with higher unit labor cost would locate more downstream production process to counties with lower unit labor cost (Table 3). This result implies that cross country differences in factor prices is effectively utilized in the formation of vertical production chains. This is consistent with Sinn (2004, 2006) which argues that Germany’s high wages and rigid labor market stimulated a wave of international relocation of production to seek lower cost, especially in the automotive sector, to neighboring eastern European countries in the early 1990s.

We also estimate the impact of industrial similarity on foreign VA export from country i and j.⁶ Since fragmentation within a product or intra-industry trade is an important driver of supply links, two countries with a similar initial export structure are more likely to link. In manufacturing trade, this may also be driven by the likely availability of skilled labor if two countries have a similar export or industrial structure.

The “similarity index” is calculated relative to Germany, where a lower value implies higher export similarity with Germany (Annex 4). This index shows a strong similarity between the export structure of Germany and four highly export-oriented central European countries in 1995, which grew stronger by 2008. For EZ periphery countries, while Spain and Portugal increased their similarities with Germany’s exports structure during 1995-2008, Ireland and Greece decreased theirs (Figure 8). Two caveats that need to be mentioned for this index. First, although Germany is the largest hub in Europe, which is why we choose Germany as a benchmark, it is not the only hub. Second, our disaggregation divides total exports of goods and services into only 35 sectors. Therefore, it does not take into account quality differences or level of refinements within a particular product. An index with more disaggregated product level data and taking into account different hubs may better capture the degree of industrial similarity between a hub and a host.

Our estimation shows a strong negative coefficient for initial industrial similarity index: vertical integration is likely to occur between countries with similar industrial structure (Table 3). This result is statistically robust across estimation methods.

In the voluminous literature on exchange rate volatility and trade, there is no consensus on the appropriate method for measuring such volatility. The most widely used measure of exchange rate volatility is the standard deviation of the first difference of log of the exchange rate. This measure has the property that it will equal zero if the exchange rate follows a constant trend, which presumably could be anticipated and therefore would not be a source of uncertainty. Clark and others (2004) argue that real rates are preferable on theoretical grounds. We measure exchange rate volatility by the standard deviation of the first difference of log of real bilateral exchange rate.

Our results show a negative and statistically significant relationship between foreign VA export and volatility of the bilateral exchange rate. One reason that cross-border joint production might be adversely affected by exchange rate volatility stems from the assumption that firms cannot alter factor inputs in order to adjust optimally to take account movements in exchange rates. We also find the same negative relationship in OLS estimation with no control, but not in the two-way fixed effects.

The strong significance of gravity variables and industrial similarity index imply that among EZ periphery countries, Spain holds the strongest potential to increase its links with European hubs, such as Germany, provided there is a similar industrial structure. For other countries, supply links provide limited prospects in the short run. However, more competitive wages may open up opportunities with other hubs in the region.

The variables in above regression analysis are measured in different units of measurement. For example, foreign VA trade is measured in million U.S. dollars whereas the Industry Dissimilarities are indices in the scale of 0.01. Therefore, it is difficult to compare which of the independent variables have a greater effect on the dependent variable from the results in Table 3.

To evaluate the contribution of each variable on foreign-value-added trade, we compute the standardized coefficient for our baseline model (OLS with time control).⁷ Standardized coefficients are the estimates resulting from an analysis carried out on independent variables that have been standardized so that their standard deviations are all one. Thus, standardized coefficients tell us how many standard deviations a dependent variable will change, per standard deviation increase in the independent variable.

For example, the standardized coefficient for ULC differential is 0.044, which means that one standard deviation increase in ULC differential results in 0.044 standard deviation increase in bilateral value-added trade (Table 4). We can see that the traditional gravity variables are dominant in explaining supply links as captured by foreign VA, compared to the effects of structural variables. In other words, without help from gravity variables, such as a large economic size or close distance to supply hubs, countries have to undergo large structural adjustments if they want to meaningfully increase the supply chain linkages.

Table 4.

The Standardized (Beta) Coefficients

Table 4.

The Standardized (Beta) Coefficients

Variable	Standardized Coefficient
Log of GDP_1	0.672
Log of GDP_2	0.496
Log of GDP per capita_1	−0.048
Log of GDP per capita_2	0.085
log of distance	−0.266
Common Language Dummy	0.066
Common Border Dummy	0.089
Resource-rich Dummy	0.052
FTA Dummy	0.071
Downstream Tariff	−0.015
Exchange Rate Volatility	−0.018
Unit Labor Costs differentials	0.044
Industry similarity Index	−0.029

Table 4.

The Standardized (Beta) Coefficients

Variable	Standardized Coefficient
Log of GDP_1	0.672
Log of GDP_2	0.496
Log of GDP per capita_1	−0.048
Log of GDP per capita_2	0.085
log of distance	−0.266
Common Language Dummy	0.066
Common Border Dummy	0.089
Resource-rich Dummy	0.052
FTA Dummy	0.071
Downstream Tariff	−0.015
Exchange Rate Volatility	−0.018
Unit Labor Costs differentials	0.044
Industry similarity Index	−0.029

IV. Supply Links and Revealed Comparative Advantage

The concept of revealed comparative advantage (RCA) proposed by Balassa (1965) has proven to be a useful technique in research and policy applications in international trade. In standard applications, it is defined as the share of a sector in a country’s total gross exports relative to the world average of the same sector in world exports. When the RCA exceeds one, the country is considered to have a revealed comparative advantage in that sector; when the RCA is below one, the country is considered to have a revealed comparative disadvantage in that sector. Koopman and others (2011) show that the problem of multiple counting of VA components in official trade statistics makes the traditional computation of RCA noisy and misleading. The VA decomposition of exports adopted in our study provides a way to remove the distortion of multiple counting by focusing on domestic VA in exports.

We look at the four successful central European countries that pursued export-led growth through greater integration with supply links to see how their tradable sector evolved during 1995–2008 in terms of comparative advantage. We disaggregate domestic VA exports into manufacturing and services, further dividing each category into labor-, capital- and knowledge-intensive sectors. The classification of sectors is documented in Annex 3, Table 2. Here we present a few key observations (Figure 9):

Central European countries enhanced their comparative advantage in manufacturing over time. In 1995, none of the four countries had a comparative advantage in knowledge-based manufacturing. By 2008, they all acquired such advantage in addition to retaining/improving their RCA in labor- and capital-intensive manufacturing. Strong and growing supply links with European hubs enabled these countries to move up the value ladder.
Enhanced comparative advantage in manufacturing in central Europe has not necessarily come at the expense of services. Some of these countries show strong RCA in services exports as well. For example, Hungary and Poland have a RCA higher than 1 in two of the three categories of services exports. Czech Republic and Slovakia, on the other hand, started with a RCA in all three services category in 1995 but over time moved to recreate comparative advantage in manufacturing. Over time, Czech Republic and Slovakia’s RCA became closer to that of Germany’s in line with their stronger supply link relationship. The harmonization of RCA reflects the dominance of intra-industry in supply links between these two countries and Germany.

We further zoom in on product level export data to see whether performance was driven by particular products (Figure 10, Annex 3 Tables 3a-3d). Indeed, we see the importance of transport equipment and machinery industries in the export success story of these countries. During 1995–2008, exports of all major categories more than doubled in these four countries.

But exports of machinery and transport equipment increased by 7–22 times. The dominance of machinery and transport equipment exports is overwhelming. The share of these products in total exports of goods and services increased from around 10 percent to over 20 percent during this time period in Hungary, Czech, and Slovak Republics. This attests to the role of finding a niche few sectors to secure success in a supply link driven trade environment.

We compare sectoral export evolution of four central European countries with that of EZ periphery countries which, apart from Ireland, are much less dependent on supply links than the central European countries.

During 1995–2008, the role of manufacturing decreased in some EZ periphery countries. These countries showed RCA in some manufacturing categories in 1995. For example, Greece showed RCA in labor and capital-intensive manufacturing, and Ireland in capital and knowledge-intensive manufacturing in 1995. Over time, both of these countries lost RCA in manufacturing. On the other hand, Spain and Portugal increased their RCA in labor-intensive manufacturing.
EZ periphery has a higher comparative advantage in services production. As of 2008, all four countries show a RCA value higher than one in two out of three categories of services exports. The dominance of services industry is most relevant for Greece, which has run a deficit in all major manufacturing export sub-categories during 1995–2008 (Figure 11, Annex 3 Tables 3e). Spain and Portugal, on the other hand, show some comparative advantage in manufacturing products.

What lessons can be learnt from the analysis of RCA? Supply links are more dominant in manufacturing and successful linking often involves finding niche manufacturing sectors. Although Ireland’s experience would testify that successful linking could take place through services as well. Most EZ periphery countries have RCA in services. Improving exports would need to lever this services sector RCA. For this to happen through intra-Europe trade, further liberalization of services trade in Europe and finding niche sectors in services trade would be helpful. For example, Germany does not demonstrate a RCA in services. Liberalizing services trade in surplus countries like Germany would be one channel through which service-heavy periphery countries could benefit in terms of improving trade balance and growth.

V. Conclusion

One of the most important recent changes in the global economy involves increasing interconnectedness of production processes in a vertical trading chain that stretches across many countries, with each country specializing in particular stages of a good’s production sequence. Because of vertical production linkages, intermediate products move across borders multiple times before being assembled to a final good. This phenomenon challenges the traditional wisdom of international trade theory, as well as the ability for gross trade data to provide an accurate picture regarding countries’ export performance and competitiveness.

In this paper, we use a newly released world input-output table to characterize the development of vertical integration as well as to investigate its impact on countries’ exports. Our purpose is to reshape our understanding of trade in Europe based on VA trade statistics. To accomplish this, we adopt a cutting-edge framework developed by Koopman and others (2011) to decompose each country’s gross exports according to its VA sources. We try to see the performance on domestic VA exports over time, the role of supply links in export performance, and factors determining success in a trade set up dominated by supply links.

Our analysis shows that strongest export performance globally and in Europe during 1995-2008 has been the result of successful integration with supply links. This integration often relied on a few niche sectors, rather than the entire spectrum of tradable products. Our empirical investigation shows that the ability to link depends on gravity variables, such as the size of the GDP, per capita income, and distance from the hub country, but also cost differential and similarlities in industrial structure. This result is consistent with our conjecture that firms have incentive to unbundle the production process and putting fragments of it abroad to take advantage of low-cost foreign factors of production. Our analysis also shows that successful linking helps countries in Europe move up the value chain.

What prospects or lessons do we have for Euro zone periphery countries? Success in exports growth would depend on successfully linking to supply chains. Greater links with upstream export hubs in Europe can greatly help these countries improve their export prospects.

Benefiting through supply links hold the strongest prospects for Spain among these countries, because of its larger size, sizable existing links, geographical proximity to Germany as well as an export structure that is similar, and perhaps the weakest for Greece, due to its small size, service-heavy export structure, low level of links and geographic location. For the latter, further liberalization of services trade in Europe, in addition to finding niche sectors and maintain competitive wages, would offer some prospects of stronger export-led growth.

Box 1.

Supply Links in Europe

The bilateral flows in VA exports allow us to have a glimpse of regional joint-production networks and individual countries’ participation (Box Figure). The size of the dot in the figure below for each coutnry depends on the country’s total participation in the VA network, which is captured by the sum of downstream and upstream supply link exports. The arrows represent the flows of VA exports between two countries. To make the plot informative, only bilateral value-added flows above 2 billion U.S. dollar are plotted.¹ For example, an arrow from Germany to Poland indicates that Germany’s value-added embodied in Poland’s exports is above 2 billion U.S dollar.

As shown in the Figure above, Germany is the most important hub in the export supply network of Europe in terms of value of trade, followed by Italy, Netherlands, the UK and France. Moreover, Germany also has the largest number of arrows in both directions linking with other countries (Table 1). Germany provides upstream inputs to 33 countries and receives inputs from 33 countries as well making it the most connected country in world trade above China and the USA. Italy, France, the UK, Netherlands, Belgium, Spain are Poland are other big hubs in Europe with strong upstream and downstream links. Russia also plays an important role in the region’s network but only because of its role as a supplier of oil and gas.

Box Table.

Degree of Interconnectedness in Europe

Source: Authors’ calculation using world input output Table.

Box Table.

Degree of Interconnectedness in Europe

	VA export flow >=$200m		VA export flow >=$800m		VA export flow >=$2000m		VA export flow >=$4000m
Country	Degree (Input)	Degree (Out)	Degree (Input)	Degree (Out)	Degree (Input)	Degree (Out)	Degree (Input)	Degree (Out)
DEU	25	26	22	22	17	17	11	15
ITA	23	24	14	19	8	10	4	3
FRA	21	23	15	17	8	11	7	6
GBR	21	23	13	15	8	10	3	7
BEL	22	21	15	13	8	6	4	3
NLD	21	22	13	14	7	10	5	4
ESP	20	22	11	15	6	8	3	2
AUT	19	20	12	9	2	2	1	1
POL	19	20	11	14	4	3	1	1
RUS	13	25	1	20	0	13	0	4
SWE	20	16	11	9	7	2	2	1
HUN	19	16	9	2	1	1	1	0
CZE	18	16	11	7	3	1	1	1
IRL	16	16	9	7	6	3	2	0
DNK	17	14	10	5	4	2	1	1
TUR	13	16	5	5	2	1	0	0
ROM	12	13	2	2	0	0	0	0
SVK	13	12	6	5	2	1	1	0
FIN	16	7	4	4	2	2	0	0
LUX	12	8	7	2	6	0	2	0
PRT	8	8	3	3	1	0	0	0
GRC	8	6	3	1	0	0	0	0
SVN	8	5	2	1	0	0	0	0
BGR	7	4	1	0	0	0	0	0
EST	3	3	0	0	0	0	0	0
LTU	3	3	1	0	1	0	0	0
LVA	2	1	0	0	0	0	0	0
MLT	1	0	0	0	0	0	0	0
CYP	0	0	0	0	0	0	0	0

Source: Authors’ calculation using world input output Table.

Box Table.

Degree of Interconnectedness in Europe

	VA export flow >=$200m		VA export flow >=$800m		VA export flow >=$2000m		VA export flow >=$4000m
Country	Degree (Input)	Degree (Out)	Degree (Input)	Degree (Out)	Degree (Input)	Degree (Out)	Degree (Input)	Degree (Out)
DEU	25	26	22	22	17	17	11	15
ITA	23	24	14	19	8	10	4	3
FRA	21	23	15	17	8	11	7	6
GBR	21	23	13	15	8	10	3	7
BEL	22	21	15	13	8	6	4	3
NLD	21	22	13	14	7	10	5	4
ESP	20	22	11	15	6	8	3	2
AUT	19	20	12	9	2	2	1	1
POL	19	20	11	14	4	3	1	1
RUS	13	25	1	20	0	13	0	4
SWE	20	16	11	9	7	2	2	1
HUN	19	16	9	2	1	1	1	0
CZE	18	16	11	7	3	1	1	1
IRL	16	16	9	7	6	3	2	0
DNK	17	14	10	5	4	2	1	1
TUR	13	16	5	5	2	1	0	0
ROM	12	13	2	2	0	0	0	0
SVK	13	12	6	5	2	1	1	0
FIN	16	7	4	4	2	2	0	0
LUX	12	8	7	2	6	0	2	0
PRT	8	8	3	3	1	0	0	0
GRC	8	6	3	1	0	0	0	0
SVN	8	5	2	1	0	0	0	0
BGR	7	4	1	0	0	0	0	0
EST	3	3	0	0	0	0	0	0
LTU	3	3	1	0	1	0	0	0
LVA	2	1	0	0	0	0	0	0
MLT	1	0	0	0	0	0	0	0
CYP	0	0	0	0	0	0	0	0

Source: Authors’ calculation using world input output Table.

¹ If all VA flows are plotted, there would be arrows between almost every pair of countries. The threshold level of 2billion U.S dollar is the authors’ arbitrary choice.

Annex 1. Decomposition Methodology

A. Decomposing Gross Trade Statistics

We adopt the conceptual framework developed in Koopman and others (2011) to decompose the sources of VA in global production of tradables. The decomposition methods are summarized below.

Assume an m-country world, in which each country produces goods in n differentiated tradable sectors. The m-country production and trade system can be written as an Inter-County Input-Output model in the form of block partitioned matrix

\begin{matrix} (1) & [\begin{matrix} X_{1} \\ ⋮ \\ X_{m} \end{matrix}] = [\begin{matrix} A_{11} & \dots & A_{1 m} \\ ⋮ & ⋱ & ⋮ \\ A_{m 1} & \dots & A_{mm} \end{matrix}] [\begin{matrix} X_{1} \\ ⋮ \\ X_{m} \end{matrix}] + [\begin{matrix} Y_{11} & + & \dots Y_{1 m} \\ ⋮ \\ Y_{m 1} & + & \dots Y_{mm} \end{matrix}] \end{matrix}

where X_m is the n×1 gross output vector of country m, Y_ij is the n×1 final demand vector that shows demand in country j for final goods produced in country i, and A_ij is the n×n IO coefficient matrix, giving intermediate use in country j of goods produced in country i.

Deriving the Leontief inverse matrix from equation (1) and pre-multiplying it with the final demand matrix, we get:

\begin{matrix} (2) & {[\begin{matrix} I - A_{11} & \dots & - A_{1 m} \\ ⋮ & ⋱ & ⋮ \\ - A_{m 1} & \dots & I - A_{mm} \end{matrix}]}^{- 1} [\begin{matrix} Y_{11} & \dots & Y_{1 m} \\ ⋮ & ⋱ & ⋮ \\ Y_{m 1} & \dots & Y_{mm} \end{matrix}] \end{matrix} = [\begin{matrix} B_{11} & \dots & B_{1 m} \\ ⋮ & ⋱ & ⋮ \\ B_{m 1} & \dots & B_{mm} \end{matrix}] [\begin{matrix} Y_{11} & \dots & Y_{1 m} \\ ⋮ & ⋱ & ⋮ \\ Y_{m 1} & \dots & Y_{mm} \end{matrix}] = [\begin{matrix} X_{11} & \dots & X_{1 m} \\ ⋮ & ⋱ & ⋮ \\ X_{m 1} & \dots & X_{mm} \end{matrix}]

where B_ij denotes the n×n block Leontief inverse matrix, which is the total requirement matrix giving the amount of gross output produced in country i required for a one-unit increase in final demand in country j. It follows that, X_ji is the output of country j used to produce goods eventually consumed in country i.

Regarding exports, let E_ij be the n×1 vector of gross exports from i to j. Gross exports from i to j is divided into final good Y_ij and intermediates A_ijX_j. The intermediates are further divided into goods that are processed and consumed by country j (A_ijX_jj), goods that are processed and re-exported by j to third countries (Σ_k≠i,j A_ijX_jk), and intermediate goods exported from i to j then processed and exported back to j (A_ijX_ji):

\begin{matrix} (3) & E_{ij} = Y_{ij} + A_{ij} X_{j} = Y_{ij} + A_{ij} X_{jj} + Σ_{k \neq i, j} A_{ij} X_{jk} + A_{ij} X_{ji} \end{matrix}

Equation (3) traces the downstream use of exports from country i to country j, however, it does not provide information on the upstream contribution from other countries to the exports of country i. Thus, we still need to compute the upstream VA of country i’s exports in order to derive a complete picture of supply links and disaggregation of VA.

Formally, we define V_i to be the 1×n direct VA coefficient vector. Each element of V_i gives the share of direct domestic VA in total output. This is equal to one minus the intermediate input share from all countries (including domestically produced intermediates):

\begin{matrix} (4) & V_{i} = u (I - Σ_{j} A_{ji}) \end{matrix}

Where, u is a 1×n unity vector.

Combining the VA coefficient vector with the partitioned Leontief inverse matrix provides information regarding the VA share. For example, each element in the 1×n vector V_iB_ii gives the domestic VA share of a particular sector in country i. Similarly, the corresponding element in vector V_jB_ji is the share of country j’s VA in the same sector produced in country i

Let E_i* be the total export from i, i.e. E_i* = Σ_j≠i E_ij = Σ_j≠i (A_ijX_j + Y_ij)

The gross exports from country i can be divided into domestic VA export (DV_i) and foreign VA export (FV)_i.

\begin{matrix} (5) & E_{i *} = {DV}_{i} + {FV}_{i} \end{matrix}

Using the derived information on VA share, Koopman and others (2011) shows that:

\begin{matrix} (6) & {FV}_{i} = Σ_{j \neq i} V_{j} B_{ji} E_{i *} \end{matrix}

\begin{matrix} (7) & {DV}_{i} = V_{j} B_{ii} E_{i *} \end{matrix}

Combining the downstream use of export in equation (3) with the VA decomposition in equation (5), we can decompose gross exports into five VA categories (^{Figure 3}):

\begin{matrix} (8) E_{i *} & = {DV}_{i} + {FV}_{i} \\ = V_{i} B_{ii} \underset{j \neq i}{Σ} Y_{ij} + V_{i} B_{ii} \underset{j \neq i}{Σ} A_{ij} Y_{jj} + V_{i} B_{ii} \underset{j \neq i}{Σ} \underset{k \neq i, j}{Σ} A_{ij} X_{jk} + V_{i} B_{ii} \underset{j \neq i}{Σ} A_{ij} X_{ji} + {FV}_{i} \end{matrix}

For country i, the terms in equation (8) correspond to the following, respectively:

(A: V_iB_ii Σ_j≠i Y_ij): DV in the form of final goods and services consumed by the direct importer;

(B: V_iB_ii Σ_j≠i A_ij Y_jj): DV in the form of intermediate inputs used by the direct importer to produce its domestically consumed products;

(C: V_iB_ii Σ_j≠i Σ_k≠i,j A_ij X_jk): DV in the form of intermediate exports used by the direct importer to produce goods for third countries

(D: V_iB_ii Σ_j≠i A_ij X_ji): DV in the form of intermediate exports used by the direct importer to produce goods shipped back to source country;

(E: FV_i): VA by foreign countries embodied in country i’s gross exports.

B. Measuring Vertical Integration

In previous literature, measures of vertical integration have been developed. Most of these proposed measures are easily taken to the data, specifically with the use of the input-output tables.

Earlier literature such as Feenstra and Hanson (1996 and 1999), Feenstra (1998), Campa and Goldberg (1997), use the share of imported intermediate input (in total input or in gross output) to measure the level of outsourcing. However, these measures fail to fully capture the supply links as countries are grouped either as producers in intermediate stages or as exporters of final goods while in reality the links are more complex.

Hummels and others (2001) suggest a measure of vertical specialization, focusing on those imported goods that are used as inputs to produce a country’s exports. (Hummels and others, 2001) Their measure emphasizes the twin ideas that the production sequence of a good involves at least two countries, and that, during this sequencing, the good-in-process crosses at least two international borders. The same approach is followed in Chen and others (2005), European Central Bank (ECB, 2005a), Breda and others (2008), and Koopman and others (2010).

Following the more recent group of literature originated from Hummels and others (2001), we define vertical integration or supply links as occurring when two or more countries provide VA in a good’s production sequence; at least one country must use imported inputs in its production process, and the resulting output must be exported.

Note that the notion of vertical integration is only sensible in at least a bilateral context. Thus, it has both an upstream side and a downstream side. The upstream supplier exports intermediate goods to a downstream producer who uses these intermediates to add value for further export. As an upstream supplier, a country’s participation in the global production chain depends on its VA to other countries’ exports. As a downstream assembler, a country’s participation in the global production chain depends on the foreign VA in its exports.

To evaluate this bilateral relation in supply links, we need to measure, for all country-pairs, the embedded foreign VA from one country in another country’s export. Koopman and others (2011) has shown that the matrix of VA by source in gross exports (VAS_E) can be specified as:

VAS \underline{} E = [\begin{matrix} V_{1} B_{11} E_{1 *} & \dots & V_{1} B_{1 m} E_{m *} \\ ⋮ & ⋱ & ⋮ \\ V_{m} B_{m 1} E_{1 *} & \dots & V_{m} B_{mm} E_{m *} \end{matrix}]

The elements of this matrix provide VA by source in gross exports between each country pair. For example, the element VAS_E_ij = V_iB_ijE_j* gives country i’s VA embodied in country j’s export. Therefore, diagonal elements of VAS_E matrix correspond to the domestic VA in each country’s exports. Off-diagonal elements give the foreign VA embodied in each country’s exports.

To link this bilateral VA relation with the country-level decomposition of export, note that the sum of off-diagonal elements along a column is the measure of VA from foreign sources embodied in a particular country’s gross exports, which is just equal to FV defined in equation (8). Here, we call it Downstream Participation (DP) and use it to measure a country’s participation in global VA chain as a downstream producer:

DP_i = FV_i = Σ_j≠i V_j B_ji E_i*

Similarly, the sum of off-diagonal elements along a row provides information on a country’s VA embodied as intermediate inputs in all other countries’ gross exports. It can be used to measure the country’s participation in global VA chains as an upstream supplier. We call it Upstream Participation (UP):

UP_i = Σ_j≠i V_i B_ij E_j*

C. Measuring Contribution of Vertical Integration to Overall Exports Growth

Note that the above analysis attempts to measure countries’ overall importance to global/regional VA networks. To capture the “intensity” of a country’s vertical specialization, we normalize country’s DP and UP by its gross exports.

Once we normalize DP with the country’s gross exports, we can use it to measure the degree of a country’s participation in global VA chain as a downstream producer (denoted by DDP).

{DDP}_{i} = \frac{{DP}_{i}}{E_{i *}} = \frac{Σ_{j \neq i} V_{j} B_{ji} E_{i *}}{E_{i *}}

Similarly, we normalize UP with the country’s gross export, which can be used as a measure of a country’s participation in upstream global VA chain (denoted by DUP).

{DUP}_{i} = \frac{{UP}_{i}}{E_{i *}} = \frac{Σ_{j \neq i} V_{i} B_{ij} E_{j *}}{E_{i *}}

Previous literature that looks at “import content of exports” as a proxy to measure a country’s degree of vertical specialization only takes into account participation as a downstream producer, giving an incomplete analysis.⁸ If a country is a major upstream supplier in the global VA network, it may have a fairly low foreign VA share in its export (e.g. Japan and Germany for example). In this case, the foreign content of exports will understate the country’s participation in supply links. To avoid this problem, in our paper, we measure a country’s degree of vertical integration in global VA chains by summing up DDP_i and DUPⁱ. In doing so, we essentially look at the country’s role in global production from both angles.

Annex 2: The World Input-Output Table

The World Input-Output Table used in our study is based on a newly released world Input-Output Table (WIOT) by Timmer and others (2012). The database covers 27 EU countries and 13 other major countries in the world for the period 1995 to 2009.⁹ The 40 countries included in our world input-output table cover more than 85 percent of world GDP.

Differing from previous databases such as GTAP, OECD and IDE-JETRO, the construction of WIOT relies on the national supply and use tables (SUTs) rather than input-output tables as its basic building blocks. Timmer and others (2012) argues that SUTs are a more natural starting point as they provide information on both products and (using and producing) industries.¹⁰ Moreover, the input-output table is often constructed on the basis of an underlying SUT, requiring additional assumptions.

Besides national SUTs, the construction of the WIOT also uses National Accounts time series data for industry output and final use, and bilateral international trade data in goods and services.

In the first step of the construction process, time-consistent output and final consumption series in the national accounts are used to benchmark national SUTs to ensure meaningful analysis over time.¹¹ In the second step, the national SUTs are combined with information from international trade statistics to construct so-called international SUTs. Basically, a split is made between use of products that were domestically produced and those that were imported. Finally, the international SUTs for each country are combined into a world input-output table.

For services trade, no standardized database on bilateral flows exists. These have been collected from various sources (including OECD, Eurostat, IMF and WTO), checked for consistency and integrated into a bilateral service trade database. As services trade is taken from the balance of payments statistics it is originally reported at Balance of Payments codes.

Annex 3: Background Tables

Table 1a.

Value-Added Decomposition of Exports in Manufacturing

Table 1a.

Value-Added Decomposition of Exports in Manufacturing

Country or Region	Decomposition in export of Manufactoring Good 1995							Decomposition in export of Manufactoring Good 2008
Country or Region	Gross manufacturing export (million $)	DVA in direct exports of final goods	DVA in intermediates absorbed by direct importer	Indirect DVA exports to third countries	Returned DVA	FVA	Total	Gross manufacturing export (million $)	DVA in direct exports of final goods	DVA in intermediates absorbed by direct importer	Indirect DVA exports to third countries	Returned DVA	FVA	Total
		(1)	(2)	(3)	(4)	(5)	(6)		(1)	(2)	(3)	(4)	(5)	(6)
CHN	136325	50.6%	22.6%	8.9%	0.4%	17.5%	100%	1803921	35.4%	19.8%	11.8%	1.5%	31.4%	100%
Rest of the world	418354	30.3%	25.9%	9.7%	2.4%	31.8%	100%	1647992	19.9%	21.2%	11.1%	3.6%	44.2%	100%
DEU	511451	36.3%	28.3%	14.3%	2.7%	18.3%	100%	1434387	29.5%	22.2%	16.5%	2.1%	29.8%	100%
USA	479211	35.6%	31.4%	11.0%	8.9%	13.1%	100%	1001641	30.9%	27.5%	14.8%	6.8%	20.1%	100%
JPN	399313	39.3%	38.0%	14.3%	1.6%	6.7%	100%	723610	29.2%	28.7%	21.1%	1.3%	19.8%	100%
FRA	261160	37.1%	25.6%	13.8%	1.4%	22.1%	100%	576334	32.8%	19.7%	15.5%	1.2%	30.9%	100%
ITA	219179	41.7%	25.0%	11.9%	0.8%	20.6%	100%	515486	34.3%	21.1%	15.6%	0.9%	28.1%	100%
KOR	119391	29.5%	29.6%	13.5%	0.4%	26.9%	100%	424822	18.8%	20.7%	16.2%	0.4%	43.9%	100%
GBR	225865	36.3%	25.6%	13.8%	1.3%	23.0%	100%	403728	32.3%	21.5%	16.4%	1.1%	28.6%	100%
NLD	148442	28.9%	21.2%	12.9%	0.7%	36.4%	100%	329863	24.0%	15.5%	13.8%	0.5%	46.2%	100%
CAN	158479	27.6%	34.4%	7.1%	0.8%	30.2%	100%	300671	26.3%	30.9%	9.9%	1.0%	32.0%	100%
ESP	86279	37.8%	24.9%	12.9%	0.6%	23.7%	100%	269583	30.3%	19.7%	14.5%	0.9%	34.7%	100%
BEL	134844	24.0%	19.2%	12.3%	0.5%	44.0%	100%	267975	18.0%	14.7%	13.4%	0.3%	53.7%	100%
TWN	108406	28.7%	24.8%	10.4%	0.2%	35.9%	100%	254976	10.6%	19.8%	20.8%	0.2%	48.6%	100%
MEX	54824	31.5%	25.4%	7.6%	0.3%	35.2%	100%	199372	28.3%	23.4%	9.4%	0.5%	38.5%	100%
IND	32005	40.8%	36.1%	10.8%	0.1%	12.2%	100%	176800	32.9%	24.3%	14.1%	0.4%	28.3%	100%
SWE	77872	29.1%	27.7%	15.3%	0.4%	27.5%	100%	176392	23.7%	20.7%	16.1%	0.3%	39.2%	100%
POL	23293	36.4%	27.7%	16.4%	0.2%	19.3%	100%	168624	28.6%	17.7%	16.9%	0.4%	36.4%	100%
AUT	47262	27.1%	27.9%	16.1%	0.5%	28.4%	100%	156202	23.1%	19.0%	17.7%	0.4%	39.8%	100%
BRA	41011	24.4%	50.0%	16.3%	0.3%	9.0%	100%	145530	32.8%	32.6%	18.6%	0.4%	15.6%	100%
TUR	24664	50.2%	23.4%	11.2%	0.1%	15.1%	100%	136191	31.7%	22.1%	13.9%	0.2%	32.1%	100%
CZE	16880	23.5%	25.5%	15.4%	0.7%	34.9%	100%	135925	19.8%	14.5%	15.0%	0.2%	50.5%	100%
IRL	40338	33.2%	16.6%	8.5%	0.1%	41.6%	100%	117948	26.1%	14.3%	9.2%	0.1%	50.3%	100%
RUS	27595	14.7%	47.9%	24.8%	0.6%	12.0%	100%	111251	13.2%	43.5%	32.1%	1.0%	10.3%	100%
FIN	39298	23.6%	34.4%	17.1%	0.3%	24.7%	100%	102544	18.5%	25.9%	18.4%	0.2%	37.0%	100%
IDN	36721	30.4%	38.6%	11.7%	0.2%	19.1%	100%	97085	19.3%	38.9%	20.3%	0.3%	21.2%	100%
AUS	30659	30.7%	39.4%	13.9%	0.2%	15.8%	100%	94421	21.6%	33.9%	20.8%	0.5%	23.1%	100%
DNK	48727	44.8%	19.1%	9.9%	0.2%	26.1%	100%	94058	35.6%	17.0%	13.3%	0.2%	33.8%	100%
HUN	8703	29.4%	22.0%	13.2%	0.1%	35.4%	100%	86360	18.7%	12.1%	12.9%	0.2%	56.1%	100%
SVK	7467	18.1%	27.5%	17.7%	0.5%	36.1%	100%	60929	18.7%	13.7%	15.8%	0.2%	51.6%	100%
PRT	21379	40.0%	19.3%	9.3%	0.2%	31.2%	100%	51345	28.7%	19.1%	13.0%	0.3%	38.9%	100%
ROM	6475	30.1%	28.5%	14.4%	0.1%	26.9%	100%	38436	25.7%	20.9%	18.4%	0.3%	34.7%	100%
SVN	7850	30.8%	21.3%	11.1%	0.0%	36.7%	100%	26238	24.2%	15.9%	15.0%	0.1%	44.9%	100%
GRC	4282	42.1%	22.4%	10.9%	0.2%	24.4%	100%	17847	26.7%	18.6%	12.3%	0.3%	42.2%	100%
BGR	3444	28.7%	22.8%	9.7%	0.0%	38.8%	100%	17499	17.5%	16.0%	13.2%	0.1%	53.2%	100%
LTU	1660	24.9%	23.3%	10.9%	0.0%	40.9%	100%	14435	20.3%	15.6%	11.3%	0.2%	52.7%	100%
LUX	6635	11.2%	21.5%	16.5%	0.1%	50.8%	100%	13346	11.6%	15.3%	17.5%	0.1%	55.5%	100%
EST	1380	28.7%	20.3%	10.7%	0.1%	40.2%	100%	8650	20.9%	21.1%	16.2%	0.1%	41.7%	100%
LVA	871	29.5%	28.4%	13.5%	0.0%	28.6%	100%	5598	24.4%	21.6%	14.9%	0.3%	38.8%	100%
MLT	1363	15.9%	11.7%	7.3%	0.0%	65.1%	100%	2763	17.5%	13.5%	12.7%	0.0%	56.4%	100%
CYP	848	42.9%	13.9%	3.9%	0.0%	39.2%	100%	1457	29.7%	18.3%	10.7%	0.1%	41.2%	100%

Table 1a.

Value-Added Decomposition of Exports in Manufacturing

Country or Region	Decomposition in export of Manufactoring Good 1995							Decomposition in export of Manufactoring Good 2008
Country or Region	Gross manufacturing export (million $)	DVA in direct exports of final goods	DVA in intermediates absorbed by direct importer	Indirect DVA exports to third countries	Returned DVA	FVA	Total	Gross manufacturing export (million $)	DVA in direct exports of final goods	DVA in intermediates absorbed by direct importer	Indirect DVA exports to third countries	Returned DVA	FVA	Total
		(1)	(2)	(3)	(4)	(5)	(6)		(1)	(2)	(3)	(4)	(5)	(6)
CHN	136325	50.6%	22.6%	8.9%	0.4%	17.5%	100%	1803921	35.4%	19.8%	11.8%	1.5%	31.4%	100%
Rest of the world	418354	30.3%	25.9%	9.7%	2.4%	31.8%	100%	1647992	19.9%	21.2%	11.1%	3.6%	44.2%	100%
DEU	511451	36.3%	28.3%	14.3%	2.7%	18.3%	100%	1434387	29.5%	22.2%	16.5%	2.1%	29.8%	100%
USA	479211	35.6%	31.4%	11.0%	8.9%	13.1%	100%	1001641	30.9%	27.5%	14.8%	6.8%	20.1%	100%
JPN	399313	39.3%	38.0%	14.3%	1.6%	6.7%	100%	723610	29.2%	28.7%	21.1%	1.3%	19.8%	100%
FRA	261160	37.1%	25.6%	13.8%	1.4%	22.1%	100%	576334	32.8%	19.7%	15.5%	1.2%	30.9%	100%
ITA	219179	41.7%	25.0%	11.9%	0.8%	20.6%	100%	515486	34.3%	21.1%	15.6%	0.9%	28.1%	100%
KOR	119391	29.5%	29.6%	13.5%	0.4%	26.9%	100%	424822	18.8%	20.7%	16.2%	0.4%	43.9%	100%
GBR	225865	36.3%	25.6%	13.8%	1.3%	23.0%	100%	403728	32.3%	21.5%	16.4%	1.1%	28.6%	100%
NLD	148442	28.9%	21.2%	12.9%	0.7%	36.4%	100%	329863	24.0%	15.5%	13.8%	0.5%	46.2%	100%
CAN	158479	27.6%	34.4%	7.1%	0.8%	30.2%	100%	300671	26.3%	30.9%	9.9%	1.0%	32.0%	100%
ESP	86279	37.8%	24.9%	12.9%	0.6%	23.7%	100%	269583	30.3%	19.7%	14.5%	0.9%	34.7%	100%
BEL	134844	24.0%	19.2%	12.3%	0.5%	44.0%	100%	267975	18.0%	14.7%	13.4%	0.3%	53.7%	100%
TWN	108406	28.7%	24.8%	10.4%	0.2%	35.9%	100%	254976	10.6%	19.8%	20.8%	0.2%	48.6%	100%
MEX	54824	31.5%	25.4%	7.6%	0.3%	35.2%	100%	199372	28.3%	23.4%	9.4%	0.5%	38.5%	100%
IND	32005	40.8%	36.1%	10.8%	0.1%	12.2%	100%	176800	32.9%	24.3%	14.1%	0.4%	28.3%	100%
SWE	77872	29.1%	27.7%	15.3%	0.4%	27.5%	100%	176392	23.7%	20.7%	16.1%	0.3%	39.2%	100%
POL	23293	36.4%	27.7%	16.4%	0.2%	19.3%	100%	168624	28.6%	17.7%	16.9%	0.4%	36.4%	100%
AUT	47262	27.1%	27.9%	16.1%	0.5%	28.4%	100%	156202	23.1%	19.0%	17.7%	0.4%	39.8%	100%
BRA	41011	24.4%	50.0%	16.3%	0.3%	9.0%	100%	145530	32.8%	32.6%	18.6%	0.4%	15.6%	100%
TUR	24664	50.2%	23.4%	11.2%	0.1%	15.1%	100%	136191	31.7%	22.1%	13.9%	0.2%	32.1%	100%
CZE	16880	23.5%	25.5%	15.4%	0.7%	34.9%	100%	135925	19.8%	14.5%	15.0%	0.2%	50.5%	100%
IRL	40338	33.2%	16.6%	8.5%	0.1%	41.6%	100%	117948	26.1%	14.3%	9.2%	0.1%	50.3%	100%
RUS	27595	14.7%	47.9%	24.8%	0.6%	12.0%	100%	111251	13.2%	43.5%	32.1%	1.0%	10.3%	100%
FIN	39298	23.6%	34.4%	17.1%	0.3%	24.7%	100%	102544	18.5%	25.9%	18.4%	0.2%	37.0%	100%
IDN	36721	30.4%	38.6%	11.7%	0.2%	19.1%	100%	97085	19.3%	38.9%	20.3%	0.3%	21.2%	100%
AUS	30659	30.7%	39.4%	13.9%	0.2%	15.8%	100%	94421	21.6%	33.9%	20.8%	0.5%	23.1%	100%
DNK	48727	44.8%	19.1%	9.9%	0.2%	26.1%	100%	94058	35.6%	17.0%	13.3%	0.2%	33.8%	100%
HUN	8703	29.4%	22.0%	13.2%	0.1%	35.4%	100%	86360	18.7%	12.1%	12.9%	0.2%	56.1%	100%
SVK	7467	18.1%	27.5%	17.7%	0.5%	36.1%	100%	60929	18.7%	13.7%	15.8%	0.2%	51.6%	100%
PRT	21379	40.0%	19.3%	9.3%	0.2%	31.2%	100%	51345	28.7%	19.1%	13.0%	0.3%	38.9%	100%
ROM	6475	30.1%	28.5%	14.4%	0.1%	26.9%	100%	38436	25.7%	20.9%	18.4%	0.3%	34.7%	100%
SVN	7850	30.8%	21.3%	11.1%	0.0%	36.7%	100%	26238	24.2%	15.9%	15.0%	0.1%	44.9%	100%
GRC	4282	42.1%	22.4%	10.9%	0.2%	24.4%	100%	17847	26.7%	18.6%	12.3%	0.3%	42.2%	100%
BGR	3444	28.7%	22.8%	9.7%	0.0%	38.8%	100%	17499	17.5%	16.0%	13.2%	0.1%	53.2%	100%
LTU	1660	24.9%	23.3%	10.9%	0.0%	40.9%	100%	14435	20.3%	15.6%	11.3%	0.2%	52.7%	100%
LUX	6635	11.2%	21.5%	16.5%	0.1%	50.8%	100%	13346	11.6%	15.3%	17.5%	0.1%	55.5%	100%
EST	1380	28.7%	20.3%	10.7%	0.1%	40.2%	100%	8650	20.9%	21.1%	16.2%	0.1%	41.7%	100%
LVA	871	29.5%	28.4%	13.5%	0.0%	28.6%	100%	5598	24.4%	21.6%	14.9%	0.3%	38.8%	100%
MLT	1363	15.9%	11.7%	7.3%	0.0%	65.1%	100%	2763	17.5%	13.5%	12.7%	0.0%	56.4%	100%
CYP	848	42.9%	13.9%	3.9%	0.0%	39.2%	100%	1457	29.7%	18.3%	10.7%	0.1%	41.2%	100%

Table 1b.

Value-Added Decomposition of Service Exports

Table 1b.

Value-Added Decomposition of Service Exports

	Decomposition in export of Service 1995							Decomposition in export of Service 2008
Country or Region	Gross service export (million $)	DVA in direct exports of final goods	DVA in intermediates absorbed by direct importer	Indirect DVA exports to third countries	Returned DVA	FVA	Total	Gross service export (million $)	DVA in direct exports of final goods	DVA in intermediates absorbed by direct importer	Indirect DVA exports to third countries	Returned DVA	FVA	Total
		(1)	(2)	(3)	(4)	(5)	(6)		(1)	(2)	(3)	(4)	(5)	(6)
USA	248726	16.6%	61.2%	14.7%	4.4%	3.2%	100%	617547	14.5%	52.6%	21.2%	5.1%	6.5%	100%
Rest of the world	89683	14.6%	54.4%	14.1%	3.4%	13.6%	100%	442483	10.7%	49.2%	15.8%	6.0%	18.4%	100%
CHN	20901	23.2%	48.2%	18.3%	0.4%	9.8%	100%	441337	17.2%	47.0%	16.5%	1.7%	17.5%	100%
GBR	58304	16.3%	55.9%	18.3%	1.4%	8.1%	100%	287800	14.3%	47.1%	27.8%	1.8%	8.9%	100%
DEU	58908	20.9%	49.7%	19.3%	3.0%	7.1%	100%	220775	17.6%	45.3%	22.5%	1.9%	12.6%	100%
RUS	32298	16.3%	55.2%	23.3%	0.7%	4.5%	100%	167853	10.6%	38.0%	43.9%	0.8%	6.7%	100%
NLD	49993	18.3%	43.9%	15.1%	0.6%	22.0%	100%	131606	15.3%	40.4%	21.5%	0.6%	22.1%	100%
JPN	83224	21.7%	58.0%	14.0%	1.9%	4.2%	100%	131166	21.5%	47.2%	20.7%	1.5%	9.1%	100%
FRA	57946	18.6%	55.6%	14.7%	1.2%	10.0%	100%	102687	13.8%	49.2%	23.7%	1.3%	12.0%	100%
BEL	38945	17.3%	39.9%	20.8%	0.9%	21.2%	100%	94687	14.9%	36.6%	23.5%	0.5%	24.5%	100%
IRL	6008	17.5%	48.5%	13.6%	0.1%	20.3%	100%	94096	12.4%	40.4%	15.4%	0.1%	31.7%	100%
ITA	40154	26.8%	47.4%	15.5%	0.9%	9.4%	100%	94062	22.4%	40.5%	23.7%	1.0%	12.4%	100%
ESP	15773	20.0%	50.9%	19.4%	0.7%	9.1%	100%	79944	15.9%	44.2%	24.8%	1.0%	14.1%	100%
LUX	12602	10.3%	35.6%	11.7%	0.1%	42.4%	100%	77065	7.4%	21.1%	8.7%	0.0%	62.8%	100%
SWE	17722	19.3%	45.8%	16.4%	0.4%	18.1%	100%	68015	16.0%	42.3%	21.7%	0.3%	19.7%	100%
DNK	11646	13.9%	41.7%	13.9%	0.2%	30.3%	100%	67967	9.9%	28.7%	13.7%	0.2%	47.5%	100%
KOR	27502	20.4%	51.4%	15.7%	0.4%	12.1%	100%	67097	16.5%	38.7%	16.4%	0.4%	28.0%	100%
CAN	27100	21.7%	57.4%	10.7%	0.5%	9.7%	100%	62590	21.8%	52.6%	15.5%	0.6%	9.5%	100%
IND	5733	22.9%	56.0%	14.6%	0.1%	6.3%	100%	59792	27.0%	46.1%	17.5%	0.4%	9.0%	100%
AUT	22111	17.3%	50.0%	17.9%	0.4%	14.4%	100%	59598	16.6%	43.2%	21.3%	0.4%	18.5%	100%
AUS	17511	27.8%	49.9%	12.6%	0.3%	9.4%	100%	43318	22.1%	46.5%	18.9%	0.4%	12.1%	100%
GRC	3913	27.1%	43.0%	12.9%	0.1%	16.9%	100%	41407	18.2%	38.6%	18.9%	0.1%	24.2%	100%
POL	5512	32.5%	42.8%	13.2%	0.1%	11.3%	100%	35846	20.5%	36.8%	22.5%	0.4%	19.8%	100%
MEX	16658	16.4%	58.8%	17.2%	0.2%	7.5%	100%	32927	17.4%	46.2%	28.5%	0.4%	7.5%	100%
BRA	8202	21.5%	54.4%	19.8%	0.3%	4.0%	100%	28666	15.9%	44.7%	32.9%	0.5%	6.0%	100%
HUN	6428	24.5%	40.5%	13.2%	0.1%	21.7%	100%	25789	20.6%	36.6%	19.2%	0.2%	23.3%	100%
TWN	16556	32.1%	38.8%	12.5%	0.2%	16.5%	100%	21942	17.2%	37.4%	17.2%	0.1%	28.1%	100%
CZE	7290	19.8%	43.9%	15.6%	0.4%	20.4%	100%	21259	21.2%	37.5%	20.8%	0.4%	20.2%	100%
ROM	2176	20.4%	50.1%	14.6%	0.0%	14.9%	100%	19966	19.1%	43.0%	21.9%	0.2%	15.8%	100%
FIN	6703	17.2%	49.8%	15.9%	0.2%	17.0%	100%	19763	16.6%	39.8%	24.7%	0.3%	18.6%	100%
PRT	5680	17.1%	49.9%	17.4%	0.2%	15.4%	100%	19373	17.2%	43.4%	21.1%	0.2%	18.2%	100%
IDN	8146	19.8%	54.6%	14.6%	0.2%	10.8%	100%	13715	21.4%	42.3%	22.0%	0.2%	14.1%	100%
SVK	2512	16.1%	45.6%	18.7%	0.4%	19.2%	100%	11242	22.6%	33.7%	21.9%	0.3%	21.5%	100%
LTU	864	27.9%	39.8%	12.8%	0.0%	19.5%	100%	8576	20.5%	38.9%	22.1%	0.1%	18.5%	100%
BGR	1804	32.1%	33.7%	10.0%	0.0%	24.2%	100%	8526	19.2%	32.0%	19.1%	0.1%	29.7%	100%
LVA	884	15.2%	46.5%	16.1%	0.0%	22.2%	100%	5876	21.4%	36.3%	22.8%	0.1%	19.3%	100%
SVN	1296	23.9%	44.6%	13.7%	0.0%	17.8%	100%	5442	17.2%	39.6%	19.7%	0.1%	23.4%	100%
EST	659	16.1%	26.4%	22.6%	0.1%	34.9%	100%	3997	12.3%	35.7%	22.5%	0.1%	29.5%	100%
CYP	865	32.6%	39.3%	12.0%	0.0%	16.1%	100%	3221	21.8%	34.4%	18.2%	0.0%	25.6%	100%
MLT	813	18.8%	39.3%	14.7%	0.0%	27.2%	100%	3051	16.5%	33.9%	17.2%	0.0%	32.4%	100%
TUR	183	22.1%	52.6%	17.8%	0.2%	7.4%	100%	1830	30.1%	26.2%	19.4%	0.3%	24.1%	100%

Table 1b.

Value-Added Decomposition of Service Exports

	Decomposition in export of Service 1995							Decomposition in export of Service 2008
Country or Region	Gross service export (million $)	DVA in direct exports of final goods	DVA in intermediates absorbed by direct importer	Indirect DVA exports to third countries	Returned DVA	FVA	Total	Gross service export (million $)	DVA in direct exports of final goods	DVA in intermediates absorbed by direct importer	Indirect DVA exports to third countries	Returned DVA	FVA	Total
		(1)	(2)	(3)	(4)	(5)	(6)		(1)	(2)	(3)	(4)	(5)	(6)
USA	248726	16.6%	61.2%	14.7%	4.4%	3.2%	100%	617547	14.5%	52.6%	21.2%	5.1%	6.5%	100%
Rest of the world	89683	14.6%	54.4%	14.1%	3.4%	13.6%	100%	442483	10.7%	49.2%	15.8%	6.0%	18.4%	100%
CHN	20901	23.2%	48.2%	18.3%	0.4%	9.8%	100%	441337	17.2%	47.0%	16.5%	1.7%	17.5%	100%
GBR	58304	16.3%	55.9%	18.3%	1.4%	8.1%	100%	287800	14.3%	47.1%	27.8%	1.8%	8.9%	100%
DEU	58908	20.9%	49.7%	19.3%	3.0%	7.1%	100%	220775	17.6%	45.3%	22.5%	1.9%	12.6%	100%
RUS	32298	16.3%	55.2%	23.3%	0.7%	4.5%	100%	167853	10.6%	38.0%	43.9%	0.8%	6.7%	100%
NLD	49993	18.3%	43.9%	15.1%	0.6%	22.0%	100%	131606	15.3%	40.4%	21.5%	0.6%	22.1%	100%
JPN	83224	21.7%	58.0%	14.0%	1.9%	4.2%	100%	131166	21.5%	47.2%	20.7%	1.5%	9.1%	100%
FRA	57946	18.6%	55.6%	14.7%	1.2%	10.0%	100%	102687	13.8%	49.2%	23.7%	1.3%	12.0%	100%
BEL	38945	17.3%	39.9%	20.8%	0.9%	21.2%	100%	94687	14.9%	36.6%	23.5%	0.5%	24.5%	100%
IRL	6008	17.5%	48.5%	13.6%	0.1%	20.3%	100%	94096	12.4%	40.4%	15.4%	0.1%	31.7%	100%
ITA	40154	26.8%	47.4%	15.5%	0.9%	9.4%	100%	94062	22.4%	40.5%	23.7%	1.0%	12.4%	100%
ESP	15773	20.0%	50.9%	19.4%	0.7%	9.1%	100%	79944	15.9%	44.2%	24.8%	1.0%	14.1%	100%
LUX	12602	10.3%	35.6%	11.7%	0.1%	42.4%	100%	77065	7.4%	21.1%	8.7%	0.0%	62.8%	100%
SWE	17722	19.3%	45.8%	16.4%	0.4%	18.1%	100%	68015	16.0%	42.3%	21.7%	0.3%	19.7%	100%
DNK	11646	13.9%	41.7%	13.9%	0.2%	30.3%	100%	67967	9.9%	28.7%	13.7%	0.2%	47.5%	100%
KOR	27502	20.4%	51.4%	15.7%	0.4%	12.1%	100%	67097	16.5%	38.7%	16.4%	0.4%	28.0%	100%
CAN	27100	21.7%	57.4%	10.7%	0.5%	9.7%	100%	62590	21.8%	52.6%	15.5%	0.6%	9.5%	100%
IND	5733	22.9%	56.0%	14.6%	0.1%	6.3%	100%	59792	27.0%	46.1%	17.5%	0.4%	9.0%	100%
AUT	22111	17.3%	50.0%	17.9%	0.4%	14.4%	100%	59598	16.6%	43.2%	21.3%	0.4%	18.5%	100%
AUS	17511	27.8%	49.9%	12.6%	0.3%	9.4%	100%	43318	22.1%	46.5%	18.9%	0.4%	12.1%	100%
GRC	3913	27.1%	43.0%	12.9%	0.1%	16.9%	100%	41407	18.2%	38.6%	18.9%	0.1%	24.2%	100%
POL	5512	32.5%	42.8%	13.2%	0.1%	11.3%	100%	35846	20.5%	36.8%	22.5%	0.4%	19.8%	100%
MEX	16658	16.4%	58.8%	17.2%	0.2%	7.5%	100%	32927	17.4%	46.2%	28.5%	0.4%	7.5%	100%
BRA	8202	21.5%	54.4%	19.8%	0.3%	4.0%	100%	28666	15.9%	44.7%	32.9%	0.5%	6.0%	100%
HUN	6428	24.5%	40.5%	13.2%	0.1%	21.7%	100%	25789	20.6%	36.6%	19.2%	0.2%	23.3%	100%
TWN	16556	32.1%	38.8%	12.5%	0.2%	16.5%	100%	21942	17.2%	37.4%	17.2%	0.1%	28.1%	100%
CZE	7290	19.8%	43.9%	15.6%	0.4%	20.4%	100%	21259	21.2%	37.5%	20.8%	0.4%	20.2%	100%
ROM	2176	20.4%	50.1%	14.6%	0.0%	14.9%	100%	19966	19.1%	43.0%	21.9%	0.2%	15.8%	100%
FIN	6703	17.2%	49.8%	15.9%	0.2%	17.0%	100%	19763	16.6%	39.8%	24.7%	0.3%	18.6%	100%
PRT	5680	17.1%	49.9%	17.4%	0.2%	15.4%	100%	19373	17.2%	43.4%	21.1%	0.2%	18.2%	100%
IDN	8146	19.8%	54.6%	14.6%	0.2%	10.8%	100%	13715	21.4%	42.3%	22.0%	0.2%	14.1%	100%
SVK	2512	16.1%	45.6%	18.7%	0.4%	19.2%	100%	11242	22.6%	33.7%	21.9%	0.3%	21.5%	100%
LTU	864	27.9%	39.8%	12.8%	0.0%	19.5%	100%	8576	20.5%	38.9%	22.1%	0.1%	18.5%	100%
BGR	1804	32.1%	33.7%	10.0%	0.0%	24.2%	100%	8526	19.2%	32.0%	19.1%	0.1%	29.7%	100%
LVA	884	15.2%	46.5%	16.1%	0.0%	22.2%	100%	5876	21.4%	36.3%	22.8%	0.1%	19.3%	100%
SVN	1296	23.9%	44.6%	13.7%	0.0%	17.8%	100%	5442	17.2%	39.6%	19.7%	0.1%	23.4%	100%
EST	659	16.1%	26.4%	22.6%	0.1%	34.9%	100%	3997	12.3%	35.7%	22.5%	0.1%	29.5%	100%
CYP	865	32.6%	39.3%	12.0%	0.0%	16.1%	100%	3221	21.8%	34.4%	18.2%	0.0%	25.6%	100%
MLT	813	18.8%	39.3%	14.7%	0.0%	27.2%	100%	3051	16.5%	33.9%	17.2%	0.0%	32.4%	100%
TUR	183	22.1%	52.6%	17.8%	0.2%	7.4%	100%	1830	30.1%	26.2%	19.4%	0.3%	24.1%	100%

Table 2:

Classification of Merchandise and Services Exports

Table 2:

Classification of Merchandise and Services Exports

Categories	Sector number	Sector name
Primary and Natural resources	1,2	Agriculture, Hunting, Forestry and Fishing
Primary and Natural resources	1,2	Mining and Quarrying
labor-intensive manufacuring	4,5,6,16	Textiles and Textile Products
		Leather, Leather and Footwear
		Wood and Products of Wood and Cork
		Manufacturing, Nec; Recycling
capital-intensive manufacuring	3,7,8,10,11,12	Food, Beverages and Tobacco
		Pulp, Paper, Paper, Printing and Publishing
		Coke, Refined Petroleum and Nuclear Fuel
		Rubber and Plastics
		Other Non-Metallic Mineral
		Basic Metals and Fabricated Metal
knowledge-intensive manufacuring	9,13,14,15	Chemicals and Chemical Products
		Machinery, Nec
		Electrical and Optical Equipment
		Transport Equipment
labor-intensive service	18,19,20,21,22,26,35	Construction
		Sale, Maintenance and Repair of Motor Vehicles and Motorcycles; Retail Sale of Fuel
		Wholesale Trade and Commission Trade, Except of Motor Vehicles and Motorcycles
		Retail Trade, Except of Motor Vehicles and Motorcycles; Repair of Household Goods
		Hotels and Restaurants
		Other Supporting and Auxiliary Transport Activities; Activities of Travel Agencies
		Private Households with Employed Persons
capital-intensive service	17,23,24,25,27,29	Electricity, Gas and Water Supply
		Inland Transport
		Water Transport
		Air Transport
		Post and Telecommunications
		Real Estate Activities
knowledge-intensive service	28,30	Financial Intermediation
knowledge-intensive service	28,30	Renting of M&Eq and Other Business Activities
health/education/public service	31,32,33,34	Public Admin and Defence; Compulsory Social Security
		Education
		Health and Social Work
		Other Community, Social and Personal Services

Table 2:

Classification of Merchandise and Services Exports

Categories	Sector number	Sector name
Primary and Natural resources	1,2	Agriculture, Hunting, Forestry and Fishing
Primary and Natural resources	1,2	Mining and Quarrying
labor-intensive manufacuring	4,5,6,16	Textiles and Textile Products
		Leather, Leather and Footwear
		Wood and Products of Wood and Cork
		Manufacturing, Nec; Recycling
capital-intensive manufacuring	3,7,8,10,11,12	Food, Beverages and Tobacco
		Pulp, Paper, Paper, Printing and Publishing
		Coke, Refined Petroleum and Nuclear Fuel
		Rubber and Plastics
		Other Non-Metallic Mineral
		Basic Metals and Fabricated Metal
knowledge-intensive manufacuring	9,13,14,15	Chemicals and Chemical Products
		Machinery, Nec
		Electrical and Optical Equipment
		Transport Equipment
labor-intensive service	18,19,20,21,22,26,35	Construction
		Sale, Maintenance and Repair of Motor Vehicles and Motorcycles; Retail Sale of Fuel
		Wholesale Trade and Commission Trade, Except of Motor Vehicles and Motorcycles
		Retail Trade, Except of Motor Vehicles and Motorcycles; Repair of Household Goods
		Hotels and Restaurants
		Other Supporting and Auxiliary Transport Activities; Activities of Travel Agencies
		Private Households with Employed Persons
capital-intensive service	17,23,24,25,27,29	Electricity, Gas and Water Supply
		Inland Transport
		Water Transport
		Air Transport
		Post and Telecommunications
		Real Estate Activities
knowledge-intensive service	28,30	Financial Intermediation
knowledge-intensive service	28,30	Renting of M&Eq and Other Business Activities
health/education/public service	31,32,33,34	Public Admin and Defence; Compulsory Social Security
		Education
		Health and Social Work
		Other Community, Social and Personal Services

Table 3a.

Trade Balance by Product, Merchandise and Services: Czech Republic, 1995–2009

Table 3a.

Trade Balance by Product, Merchandise and Services: Czech Republic, 1995–2009

	1995			2000			2008
	Export	Import	Balance	Export	Import	Balance	Export	Import	Balance
Agriculture, Hunting, Forestry and Fishing	894	853	41	722	692	29	2368	2665	−297
Mining and Quarrying	586	842	−256	398	1236	−839	1706	6750	−5045
Food, Beverages and Tobacco	495	457	38	495	482	13	2173	2105	68
Textiles and Textile Products	664	482	182	619	472	147	1298	1188	110
Leather, Leather and Footwear	126	136	−10	71	126	−54	192	321	−129
Wood and Products of Wood and Cork	243	126	118	300	141	159	1132	442	690
Pulp, Paper, Paper, Printing and Publishing	399	546	−148	516	605	−90	1911	1551	360
Coke, Refined Petroleum and Nuclear Fuel	246	236	10	161	341	−180	360	1059	−699
Chemicals and Chemical Products	618	1117	−499	817	1205	−388	2796	4039	−1242
Rubber and Plastics	190	427	−237	558	528	30	3180	1481	1700
Other Non-Metallic Mineral	483	279	205	727	308	419	1913	929	984
Basic Metals and Fabricated Metal	1692	1379	313	1632	1491	141	7655	4872	2783
Machinery, Nec	816	1305	−490	1132	1332	−200	5522	4410	1113
Electrical and Optical Equipment	621	1616	−995	1487	1999	−512	7400	4368	3032
Transport Equipment	548	640	−92	1329	771	558	6955	3112	3843
Manufacturing, Nec; Recycling	393	274	119	474	297	177	1902	1093	809
Electricity, Gas and Water Supply	733	481	251	478	451	27

Topics

Countries

Series

About

Resources

Contributor Notes

Abstract

I. Motivation1

II. Dissecting Gross Exports in Europe

III. What Factors Help Countries Establish Supply Links?

Augmented Gravity Variables

Structural Variables:

IV. Supply Links and Revealed Comparative Advantage

V. Conclusion

Annex 1. Decomposition Methodology

A. Decomposing Gross Trade Statistics

B. Measuring Vertical Integration

C. Measuring Contribution of Vertical Integration to Overall Exports Growth

Annex 2: The World Input-Output Table

Annex 3: Background Tables

I. Motivation¹