7/28/2019 854 Global Value http://slidepdf.com/reader/full/854-global-value 1/40 UNITED NATIONS CONFERENCE ON TRADE AND DEVELOPMENT INVESTMENT AND VALUE ADDED TRADE IN THE GLOBAL ECONOMY GLOBAL VALUE CHAINS AND Development Advance unedited version A preliminary analysis
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
The Division on Investment and Enterprise o UNCTAD is a global centre o excellence dealing with issuesrelated to investment and enterprise development in the United Nations System. It builds on three-and-a-hal decades o experience and international expertise in research and policy analysis, osters intergovernmentalconsensus-building, and provides technical assistance to developing countries.
The terms country/economy as used in this Report also reer, as appropriate, to territories or areas; thedesignations employed and the presentation o the material do not imply the expression o any opinionwhatsoever on the part o the Secretariat o the United Nations concerning the legal status o any country,territory, city or area or o its authorities, or concerning the delimitation o its rontiers or boundaries. Inaddition, the designations o country groups are intended solely or statistical or analytical convenience anddo not necessarily express a judgment about the stage o development reached by a particular country orarea in the development process. The major country groupings used in this Report ollow the classicationo the United Nations Statistical Oce. These are:
Developed countries: the member countries o the OECD (other than Chile, Mexico, the Republic o Koreaand Turkey), plus the new European Union member countries which are not OECD members (Bulgaria,Cyprus, Latvia, Lithuania, Malta and Romania), plus Andorra, Bermuda, Liechtenstein, Monaco and SanMarino.
Transition economies: South-East Europe and the Commonwealth o Independent States.
Developing economies: in general all economies not specied above. For statistical purposes, the data orChina do not include those or Hong Kong Special Administrative Region (Hong Kong SAR), Macao Special Administrative Region (Macao SAR) and Taiwan Province o China.
Reerence to companies and their activities should not be construed as an endorsement by UNCTAD o those companies or their activities.
The boundaries and names shown and designations used on the maps presented in this publication do notimply ocial endorsement or acceptance by the United Nations.
The ollowing symbols have been used in the tables:
• Two dots (..) indicate that data are not available or are not separately reported. Rows in tables havebeen omitted in those cases where no data are available or any o the elements in the row.
• A dash (–) indicates that the item is equal to zero or its value is negligible.
• A blank in a table indicates that the item is not applicable, unless otherwise indicated.
• A slash (/) between dates representing years, e.g., 1994/95, indicates a nancial year.
• Use o a dash (–) between dates representing years, e.g. 1994–1995, signies the ull period involved,including the beginning and end years.
• Reerence to “dollars” ($) means United States dollars, unless otherwise indicated.
• Annual rates o growth or change, unless otherwise stated, reer to annual compound rates.
Details and percentages in tables do not necessarily add to totals because o rounding.
The material contained in this study may be reely quoted with appropriate acknowledgement.
Figure 17. Key value added trade indicators, by quartile of inward FDI
stock relative to GDP, 2010............................................................................................ 19Figure 18. Correlation between growth in GVC participation and
GDP per capita .............................................................................................................. 21
Figure 19. GDP per capita growth rates by quartile of growth in GVC
Box 1. International efforts to map GVCs and the UNCTAD-Eora GVC Database
The growing importance o GVCs has led to the realization that the way international trade has traditionally been
accounted or may no longer be sucient. A growing body o work exists aimed at netting out the “double-counting”eect o GVCs on global trade, determining value added in trade, and mapping how value added moves betweencountries along GVCs beore nal consumption o end-products. Value added in trade can be estimated basedon international input-output (I-O) tables which illustrate the economic interactions between countries (see the Technical Annex). To date, and using dierent methodologies, several initiatives have sought to compile inter-countryI-O tables. A selection o the main initiatives is listed in the table below.
Project Institution Data sources Countries Industries Years Comments
UNCTAD-EoraGVC Database
UNCTAD/Eora
Nationalsupply-useand I-O tables,and I-O tablesrom Eurostat,IDE-JETRO and
OECD
187
25-500depending
on thecountry
1990-2010
“Meta” database drawingtogether many data sourcesand interpolating missingpoints to provide broad andconsistent coverage, even o data-poor countries
Inter-Country-Input-Outputmodel (ICIO)
OECD/WTONational I-Otables
40 182005,
2008, 2009
Based on national input-outputtables harmonised by theOECD
Asian InternationalI-O tables
Institute o DevelopingEconomies(IDE-JETRO)
Nationalaccounts andrm surveys
10 76
1975,1980,1985,1990,1995,2000,
2005
US-Asian tables. Also bilateral tables, includingChina-Japan.
Non-ocial dataset.Includes data on areas such asenergy volumes, land use, CO2emissions and internationalmigration.
World Input-Output Database(WIOD)
Consortium o 11 institutions.EU unded.
National supply-
use tables40 35 1995-2009
Based on ocial nationalaccounts statistics.Uses end-use classication toallocate fows across partnercountries
The UNCTAD-Eora GVC Database uses input-output tables to estimate the import-content ratio in exportableproducts and value added trade. Its value added trade data are derived rom the Eora global multi-region input-output (MRIO) table. The Eora MRIO brings together a variety o primary data sources including national input-output tables and main aggregates data rom national statistical oces; input-output compendia rom Eurostat,IDE (Institute o Developing Economies)–JETRO (Japan External Trade Organization) and OECD; national accountdata (the UN National Accounts Main Aggregates Database; and the UN National Accounts Ocial Data); and tradedata (the UN Comtrade international trade database and the UN ServiceTrade international trade database). Eoracombines these primary data sources into a balanced global MRIO, using interpolation and estimation in someplaces to provide a contiguous, continuous dataset or the period 1990-2010. The Eora MRIO thus builds onsome o the other eorts in the international community. Accompanying every data point in the results provided on
the Eora website (www.worldmrio.com) is an estimate o that data point’s standard deviation, refecting the extentto which it was contested, interpolated, or estimated, during the process o assembling the global MRIO romconstituent primary data sources. Further details on the EORA database can be ound in the Annex: “Technical noteon the UNCTAD-Eora GVC Database”.
The joint OECD-WTO project (see table), which recently published its rst results, is recognized as a comprehensiveeort to set a common standard or data on value added in trade. Placing signicant emphasis on methodologyit necessarily sacrices some coverage (o countries, industries and time series) or statistical rigor. In contrast,the primary objective o the UNCTAD-Eora GVC Database is extended coverage to provide a developing countryperspective. This explains the choice o the MRIO approach, the key innovation o which is the use o algorithmsthat put together unrelated data and minimize accounting discrepancies irrespective o the type o underlying data,allowing the inclusion o data-poor countries.
At the global level, the average oreign value added
in exports is approximately 28% (gure 3). That
means, roughly, that around $5 trillion o the $19
trillion in 2010 world exports o goods and services
has been contributed by oreign countries or
urther exports and is thus “double counted” in
global trade gures.2 The remaining $14 trillion isthe actual value added contribution o trade to the
global economy (or around one-th o global GDP).
These gures dier signicantly by country and by
industry, with important policy implications:
• At the country level, oreign value addedin exports indicates what part o country’sgross exports consist o inuts that have beenproduced by other countries, or the extent towhich a country’s exports are dependent onimported content. It is also an indication o the
level o vertical specialization o economies:
the extent to which economic activities in acountry ocus on particular tasks and activitiesin global value chains.
• At the industry level, the average oreignvalue added is a proxy or the extent to whichindustry value chains are segmented or“ne-sliced” into distinct tasks and activitiesthat generate trade, compounding thedouble counting eect. This is important orpolicymakers designing, or example, industrialdevelopment, trade and investment promotion
policies.
Which countries incorporate the most oreign value added in their exports?
Developed countries, as a whole, at 31% have
a higher share o oreign value added in exports
than the global average (gure 4), i.e. their import
dependence o exports appears higher. However,
this picture is distorted by the weight in global
gures o internal trade within the highly integrated
I. Value added trade patterns in the global economy 5
Box 2. Understanding value added trade data and indicators
A country’s exports can be divided into domestically produced value added and imported (oreign) value added that
is incorporated into exported goods and services. Furthermore, exports can either go to a oreign market or nalconsumption or as intermediate inputs to be exported again to third countries (or back to the original country). Theanalysis o GVCs takes into account both oreign value added in exports (the upstream perspective) and exportedvalue added incorporated in third-country exports (the downstream perspective). The most common indicators,which will also be used in this report, are as ollows:
1. Foreign value added (oreign value added as a share o exports) indicates what part o a country’s grossexports consists o inputs that have been produced in other countries. It is the share o the country’s exportsthat is not adding to its GDP.3
2. Domestic value added is the part o exports created in-country. It is the share o the country’s exportsthat contributes to GDP (domestic value added trade share). The sum o oreign and domestic value addedequates to gross exports. As a share o GDP, domestic value added measures the extent to which trade con-tributes to the GDP o a country.
3. GVC participation4 indicates the portion o a country’s exports that is part o a multi-stage trade process,by adding to the oreign value added used in a country’s own exports also the value added supplied to othercountries’ exports. Although the degree to which exports are used by other countries or urther export gener-ation may appear less relevant or policymakers as it does not change the domestic value added contributiono trade, the participation rate is a useul indicator or the extent to which a country’s exports are integrated ininternational production networks and it is thus helpul in exploring the trade-investment nexus.This variablecorrects the limitation o the previous indicators in which countries at the beginning o the value chain (e.g.exporters o raw materials) have a low oreign value added content o exports by denition. It gives a morecomplete picture o the involvement o countries in GVCs, both upstream and downstream.
A country’s GVC participation, measured as a share o exports, eectively assesses the reliance o exportson GVCs. In this sense, it is also an indicator o how much hypothetical “damage” to GVCs (and global GDP)would occur i a country’s exports were blocked; alternatively, it represents the vulnerability o the GVC toshocks in the respective country.
GVC indicators can also be used to assess the extent to which industries rely on internationally integratedproduction networks. For example, a number o complex methods have been devised in the literature to measureGVC length.5 This report will use a simplication device by looking at the degree o double counting in industrieswhich, conceptually, can serve as a rough proxy or the length o GVCs.
Data on value added trade by industry can provide useul indications on comparative advantages and competitivenesso countries, and hence orm a basis or development strategies and policies. However, this short launch report willocus primarily on country-level indicators; WIR13 will explore industry value added trade data and its developmentimplications in greater detail.
Source: UNCTAD; additional reerences listed in the endnotes.
o EU originated exports. Japan and the UnitedStates show signicantly lower shares o “double
counting”.
Thus, while developing countries have a lower
share o oreign value added (25%) than the world
average (28%) their oreign value added share is
signicantly higher than in the United States and
Japan – or than in the EU, i only external trade is
taken into account. Among developing economies,
the highest shares o oreign value added in
trade are ound in East and South-East Asia andin Central America (including Mexico) where
processing industries account or a signicant part
o exports. Foreign value added in exports is much
lower in Arica, West Asia, South America and in
the transition economies, where natural resources
and commodities exports with little oreign inputs
tend to play an important role. The lowest share
o oreign value added in exports is ound in South
Asia, mainly due to the weight o services exports,
Note: GVC participation indicates the share o a country’ exports that is part o a multi-stage trade process; it is the oreign valueadded used in a country’s exports (upstream perspective) plus the value added supplied to other countries’ exports (downstreamperspective), divided by total exports. GVC participation growth here is the annual growth o the sum o the upstream and downstreamcomponent values (CAGR).
links) indicates that much o their exports are
processed and incorporated in third-country
exports – i.e. they operate at the starting point o
GVCs. South Asia remains the lowest ranked region
in terms o GVC participation. Much o the services
exports rom the region satises domestic demand
in importing countries and is not used to produce
urther exports.
However, South Asia is the region with the highest
GVC participation growth rate, albeit rom a low
base. Transition economies also show aster than
average growth. Nearly all developing regions
outpace the developed world in GVC growth.
Remarkable is the rapid growth rate o GVCs in the
least developed countries partly because o a low
base in terms o absolute values.
As noted above, GVC participation – or the role that
individual countries play in international production
networks – is driven by many dierent actors,
including size o the economy, industrial structure
and level o industrialization, composition o exports
and positioning in value chains, policy actors, and
others. As a result, countries with very dierent
characteristics may be very similar in the ranking o
Note: The GVC participation rate indicates the share o acountry’s exports that is part o a multi-stage trade process; it isthe oreign value added used in a country’s exports (upstreamperspective) plus the value added supplied to other countries’exports (downstream perspective), divided by total exports.
Singapore
Belgium
Netherlands
United Kingdom
Hong Kong, China
Sweden
Malaysia
Germany
Korea, Republic of
France
China
Switzerland
Russian Federation
Saudi Arabia
Italy
Thailand
Japan
Taiwan, Province of China
Spain
Canada
United States
Mexico
Australia
Brazil
India
82%
79%
76%
76%
72%
69%
68%
64%
63%
63%
59%
59%
56%
56%
53%
52%
51%
50%
48%
48%
45%
44%
42%
37%
36%
Upstream component
Downstream component
For example, the United States and Mexico have
near identical GVC participation rates, but Mexican
exports include a signicant amount o processing
trade, with high oreign value added inputs, whereasUnited States exports are used more downstream
in value chains, as intermediate inputs in the exports
o other countries.
Again, GVC participation is a relative concept.
United States rms may dominate many value
chains in terms o absolute size, but in relative
terms the participation in GVCs o many smaller
economies is higher. In other words, United States
rms also export many nal products that are not
used downstream to generate urther exports.
The GVC participation rate is a useul metric or
examining the trade-investment nexus because it
indicates the extent to which countries’ exports are
integrated into international production networks.
Box 3. Estimating trade within the international production networks of TNCs
The estimates or trade taking place with the international production networks o TNCs in gure 15 are based on
evidence on investment-trade links o individual countries and regions:6
• In the United States, in 2010, aliates o oreign TNCs accounted or 20% o exports and 28% o imports o goods, while TNCs based in the United States accounted or 45% o exports and 39% o imports. Thus sometwo-thirds o both exports and imports o goods can be considered as within the international production net-works o TNCs.
• In Europe, in 2009, French TNCs accounted or some 31% o goods exports and 24% o imports, while oreignaliates in France accounted or 34% and 38%, respectively. Thus some 64% o total French exports and 62%o total French imports o goods in 2009 can be considered as within the international production networks o TNCs. Similar scattered evidence exists or other EU countries.
• In Japan, TNCs based there accounted or 85% o exports o goods and services, while oreign aliates con-tributed a urther 8%. Thus 93% o total Japanese exports o goods and services are linked to TNCs.
• In China, oreign aliates accounted or some 50% o exports and 48% o imports in 2012. Adding the trade
activities o Chinese TNCs, although perhaps not as large as the share o their French or United States coun-terparts given the lower (but growing) share o Chinese outward FDI, would lead to estimates o trade withininternational, production networks in excess o the United States share.
• In developing countries as a group it is likely that the share o trade within the production networks o TNCs ishigher, or two reasons: (a) the productivity curve o rms is steeper than in developed countries, meaning thattrade is likely to be even more concentrated in a small number o large exporters and importers with above-average productivity, i.e. predominantly TNCs and their aliates; (b) the share o extractive industries in theirexports (at around 25%) is signicantly higher than the world average (around 17%) and the extraction andtrade o natural resources generally involves TNCs.
A signicant share o this trade is intra-rm trade, the international fows o goods and services between parentcompanies and their aliates or among these aliates, as opposed to arm’s length trade between unrelated parties(inter-rm trade). For example, the share o exports by United States aliates abroad directed to other aliatedrms, including parent rms, remained high at about 60% over the past decade. Similarly, nearly hal o the exports
o goods by oreign aliates located in the United States are shipped to the oreign parent group and as much as70% o their imports arrive rom the oreign parent group. Japanese TNCs export 40% o their goods and servicesto their own aliates abroad. Although urther evidence on intra-rm trade is patchy, the general consensus is thatintra-rm trade accounts on average or around 30% o a country’s export, with large variations across countries.
The above explanations or the most part ocus on merchandise trade. There is evidence that TNC involvement inservices trade, with a growing share o intra-rm trade in services (e.g. corporate unctions, nancial services, etc.),is even higher. Where not in the orm o intra-rm trade, services trade oten takes place in NEM relationships (IT/ BPO, call centers, etc.). NEMs as a whole (including contract manuacturing activities) are estimated to be worthover $2 trillion (see World Investment Report 2011 ).
Arm’s length trade by TNCs (exports to and imports rom unrelated parties) is estimated to be worth around $6 trillion,the residual. Non-TNC-related trade includes all transactions between rms that have only domestic operations,anonymous transactions on commodity exchanges, etc.
Figure 15. Sector composition of global gross exports, value added inputs to exports, and FDI stock, 2010
production. The parallel with FDI is clear: more than
60% o global FDI stock is allocated to services
activities, a signicant part o which is linked to
GVCs (gure 15). The share o services FDI is stillmore than 35% i only non-nancial sector FDI is
considered (although nancial sector FDI is not
only a value chain in its own right but also provides
crucial services to other GVCs).
This picture is almost the same in both developed
and developing countries. Developing country gross
exports o primary sector output (commodities) and
primary sector value added in trade are only around
4 percentage points higher than the average or all
countries, driven by slightly higher primary sector
inward FDI stock (8% compared to the 7% average).
How does the presence o TNCs aect countries’ GVC participation?
The involvement o TNCs in generating value added
trade is conrmed by the statistical relationship
between FDI stock in countries and their GVC
participation rates (gure 16). The correlation is
strongly positive, and increasingly so over time,
especially in the poorest countries, indicating thatFDI may be an important avenue or developing
countries to gain access to GVCs and grow their
participation.
Ranking countries by the ratio o FDI stock over
GDP and grouping them in quartiles (gure 17)
shows that the group o countries with most FDI
relative to the size o their economies tend to have:
• higher oreign value added in their exports(oreign aliates o TNCs producing or exportstend to use value added produced by otherparts o the TNC production network);
• higher GVC participation (oreign aliateso TNCs not only use oreign inputs in theirproduction, but also supply to other parts o the TNC network or urther exports); and
• a higher contribution o value added trade totheir GDP.
Source: UNCTAD-Eora GVC Database, UNCTAD FDI Database, UNCTAD FDI Database.Note: The sectoral breakdown o gross exports is based on ISIC, rather than SITC (normally used or merchandise trade), or
consistency with the classication employed or value added trade and FDI. Thus, rened oil/petroleum products and ood andbeverages are classied under manuacturing.
Note: data or 187 countries over 20 years. The regression between the annual GVC Participation growth and annual FDI Inward(stock) growth, in logs, shows a positive and signicant correlation, at the 5% level. This relation also holds, at the 5% level, dividing
the sample in developed and developing countries, and in two time periods (1990-2000 and 2001-2010). All regressions use lagged(one year) inward FDI stock growth rates.
Figure 17. Key value added trade indicators, by quartile of inward FDI stock relative to GDP, 2010
Note: data or 180 countries, ranked by inward FDI stock relative to GDP and grouped in quartiles (o 45 each); data reported aremedian values or each quartile.
Figure 16. Correlation between levels of inward FDI stock and GVC participation
9
1 3
1 7
2 1
0 5 10 15
FDI Stock
9
1
3
1 7
2 1
0 5 10 15
FDI Stock
1990−2010 Fitted values
GVC Participation vs FDI Inward Stock
Developed Countries - logsGVC Participation vs FDI Inward Stock
Note: the regression between the annual GDP per capita (in PPS) growth and annual GVC participation index growth, in logs, shows
a positive and signicant correlation, at the 5% level. This relation also holds, at the 5% level, dividing the sample in developed anddeveloping countries, and in two time periods (1990-2000 and 2001-2010). To avoid picking-up a compositional eect resulting romthe correlation between a country’ s total value added (used as a component to calculate the GVC participation index) and its percapita GDP, all regressions use lagged (one year) GVC growth rates.
Clearly the optimal policy outcome is depicted in the
top right hand quadrant, where countries increase
GVC participation through growth in the domestic
value added in exports. Examples o countries in
the top right quadrant include China, Indonesia,
Thailand and Peru. While increasing oreign value
added content in exports may be a short-term
trade-o or policymakers, longer term the creation
o domestic productive capacity yields the better
results.
Are there dierent GVC development paths?
The dierent outcomes in each o the combinations
o GVC integration and domestic value added
suggest that there may be a set o distinct “GVC
development paths” or evolutionary lines in
countries’ patterns o participation in GVCs.
Although the matrix is a simplication o reality that
cannot capture all the dynamics o development,
broadly, a number o GVC development paths can
be hypothesized (gure 21), each with a set o
prevalent trade and investment patterns:
• Engaging in GVCs. Developing countries
may see imports o intermediate goods,components and services increase, as wellas the importance o processing exports. This pattern oten coincides with an infux o processing FDI and the establishment o NEM-relationships (e.g. contract manuacturing) with TNCs.
• Preparing or GVCs. Some developingcountries may see exports remainpredominantly within sectors and industrieswith domestic productive capacity (with limitedneed or imported content). FDI infows help
produce intermediate goods and services or
G V C
g r o w t h
− 1
− . 5
0
. 5
1
−.2 −.1 0 .1 .2GDPpc growth
− 1
− . 5
0
. 5
1
G V C
g r o w t h
−.2 −.1 0 .1 .2GDPpc growth
2001−20101990−2000
GVC growth vs GDP per Capita growthDeveloped Countries
GVC growth vs GDP per Capita growthDeveloping Countries
Note: data or 120 countries, ranked by GVC participation growth and grouped in quartiles (o 30 each); growth rates reported aremedian values or each quartile.
Figure 20. GDP per capita growth rates for developing countries with high/low growth in GVC participation,
and high/low growth in domestic value added share, 1990-2010
Note: data or 123 developing countries, ranked by growth in GVC participation and domestic value added share; high includes thetop two quartiles o both rankings, low includes the bottom two; GDP per capita growth rates reported are median compound annualgrowth rates or countries in each quadrant.
Median of GDP per capita growth 1990-2010
1st quartile
(Countires with rapidlygrowing GVC participation)
4th quartile
(Countires not increasingtheir GVC participation)
2nd quartile
3rd quartile
3.3%
2.1%
1.2%
0.7%
GVC participationgrowth rate
Growth of the domestic value addedshare of exports
shape GVCs through their investments in productive
assets worldwide account or some 80% o global
trade.
UNCTAD’s data show that almost all developingcountries, including the poorest, are increasingly
participating in GVCs. Evidence on GVC links
in developing countries – based on the data
presented here and on UNCTAD’s wider research
on GVCs – suggests that they can have important
development benets:
• GVCs can acilitate access to global marketsand integration in the global economy ordeveloping countries, which no longer have todevelop an entire industry to generate exports,
but can ocus on ewer tasks within industryvalue chains.
• Participation in GVCs generates employmentand may result in aster GDP and incomegrowth.
• Moreover, GVCs can be an important avenueor developing countries to build productivecapacity, including through technologydissemination and skill building, openingup opportunities or longer-term industrialupgrading.
However, GVCs can also entail risks or developingcountries:
• Many o the potential development benetso GVCs — in particular technologydissemination, skill building and upgrading —are not automatic. Developing countries canremain locked into relatively low value addedactivities.
• The location o tasks and activities within GVCsis determined by dynamic actors — includingrelative labour productivity and cost — and
as such can shit around the internationalproduction networks o multinational rms (theycan be ootloose).
• The sustainability impact o GVCs can besignicant, starting rom the environmentalimpact o moving goods along internationallydispersed value chain segments, to therisk o rms moving activities with greaterenvironmental impact to less regulatedlocations. Similarly, the social and labourimpact o GVCs must be taken into account.
This balance o opportunities and risks makes a
well-inormed policy debate on the development
impact o GVCs o paramount importance. The
raison d’être or the UNCTAD-Eora GVC Database
on value added trade and investment is to stimulate
and contribute to such debate.
UNCTAD will, in the coming months, deepen the
analysis o the data, ocusing in particular on the
development impact and policy implications or
developing countries. Questions that UNCTAD will
aim to answer include:
• What are the implications o new insights onGVCs or investment and trade theory?
• What are the prospects or urther evolution o GVCs and their role in global investment andtrade dynamics?
• What are the drivers and determinants o the location or re-location o cross-borderproductive activity via (equity and non-equity)investment in GVCs?
• Should developing countries adopt specic
policies in their development strategy toincrease GVC participation? I so, under whatcircumstances, based on what criteria?
• How can developing country policymakerspromote upgrading over time? Is the middle-income trap a real challenge or policymakers?
• Can we measure the “ootloose” nature o some o the links in the chain? What kind o shocks and vulnerabilities might threaten thegains rom GVC participation? Is trade morevolatile within GVCs?
added embodied in gross trade fows. To calculatethe latter, we start rom a row vector v with each
element representing the share o value added
per unit o output by country (that is v 1 = V 1 / X 1 ),
combined with the Leontie inverse and a vector
e summarizing aggregate exports by country as
retrieved by the sum o the intermediate inputs
exported abroad and exports o nal goods10
The rst matrix T is the key matrix o our analysis,
and or ease o readability it is replicated in the next
Figure A.2. The matrix essentially describes how
the value added contained in the exports o each
country (and industry) is generated (by column)
and distributed (by row) across countries. The rst
column o the matrix describes the value added
contained in the export o country 1.11 This is
composed o two parts:
• the term T v 11 (in the matrix multiplication we
have that T v 11 = v 1L11e1 ) denotes the Domestic
Value Added (DVA) content o exports o country 1;
• the generic term T v k1 (in matrix notation T v
k1 =v k L k1e1 ) denotes the Foreign Value Added (FVA)content o exports o country 1 generatedby country k (with k ≠ 1 ). Recall that theproduction o output by country 1 (part o which is exported) requires inputs rom othercountries. In producing these inputs, the othercountries also generate value added. Hence,
this term represents the share o value addedthat has been generated in country k (v k ) andthat has been imported by country 1 (L k1 ) inorder to produce its exports (e1 ).
The (column) sum o Domestic and Foreign Value
Added, by construction, will yield the total exports
o country 1.
The other columns o the T matrix replicate the
exercise or the other countries. So in column 2 o
the matrix we will nd the term T v 22, which denotes
the DVA content o exports o country 2, as well
second column) or its exports. More specically,
in matrix expression we have T v 12 = v 1L12e 2: hence
this term represents the share o exports o country
2 (e 2 ) that depends on the value added sourced
by country 1 (v 1L12 ). The same would be true or
a country 3, in which the term T v 13 in the third
column indicates how much country 3 is sourcing
in terms o value added rom country 1. Hence,
by reading the matrix along the row, rather than
along the column (and excluding the diagonal term
T v kk ), we would have an indication o how much o each country’s domestic value added enters as an
intermediate input in the value added exported by
other countries. The latter terms is what Koopman
et al. (2011) call “indirect value added exports”
(DVX). Clearly, by construction what each country
contributes to all the others in terms o indirect
value added exports has to be equal at the world
level to what each country sources rom all the
others in terms o oreign value added, that is at
the world level FVA = DVX. The latter gives a rough,
though not perect, proxy o the double countingembedded in the gross (ocial) trade gures.
More precisely, part o the DVA exported and
incorporated in third countries’ export can itsel
return home and thus generate some urther
double counting, as the original DVA measure
would include a share o domestic value added that
is returned home ater being processed abroad.12
However, given the complexity o computing all
these terms or a MRIO with 187 countries, and
since a perect decomposition o gross exports in
as the generic term T v k2, which denotes the FVA
content o exports o country 2 generated by
country k , and so on. Hence, the DVA can be read
on the diagonal o the matrix as the generic termT v
kk or any country k in the dataset.
Now, consider country 1 and country 2. As we have
seen, country 1 is sourcing some value added rom
country 2 or its exports (the term T v 21 which we have
1 The Eora Project, originally unded by the AustralianResearch Council, based at the University o Sydneyand comprising an international team o researchers,developed the so-called “world multi-region input-output database” that is the basis or the generationo the value added trade estimates in the GVCDatabase discussed in this paper. For urther details,see http://www.worldmrio.com/.
2 Equating oreign value added with the doublecounting in global trade gures is a simplication.Some urther double counting takes place withinoreign value added as exported value added canre-enter countries to be incorporated in urther
exports, and so orth. Such circular double countingcan be signicant in some countries and someindustries, but is marginal in most.
3 This variable is related to an active literature onmeasuring vertical specialization, with the rstindicator calculated being the value o importedinputs in the overall (gross) exports o a country. Therenement to this indicator o vertical specializationcorrects or the act that the value o (gross)imports used by country A to produce exports(as retrieved rom ‘standard’ I–O tables) in realitymight incorporate the domestic value-added o the same country A that has been used as aninput by a oreign country B, rom which the samecountry A then sources. Allowing instead only orthe oreign value-added o country B to enter in thecalculation o country A’s inputs nets out this eect.See: Hummels, D., J. Ishii and K.-M. Yi (2001) “Thenature and growth o vertical specialization in worldtrade”, Journal o International Economics 54(1),75-96; and Johnson, R.C. and G. Noguera (2012)“Accounting or intermediates: Production sharingand trade in value-added”, Journal o InternationalEconomics 86(2), 224-236.
4 This indicator was rst introduced in Koopman, R.,W. Powers, Z. Wang and S.-J. Wei (2011) “Givecredit to where credit is due: tracing value added in
global production chains”, NBER Working PapersSeries 16426, September 2010, revised September2011.
5 See Fally, T. (2011). “On the Fragmentation o Production in the US”, University o Colorado-Boulder, July.
6 Estimates are based on data rom the UnitedStates Bureau o Economic Analysis (“U.S. Aliateso Foreign Companies and U.S. (Ministry o Commerce) Multinational Companies”, 2012); ChinaMOFCOM; OECD; IDE-JETRO. Data or Europerom Altomonte, C., F. Di Mauro, G. Ottaviano, A.Rungi and V. Vicard. 2012. “Global Value Chains
during the Great Trade Collapse: A Bullwhip Eect?”ECB Working Paper Series No. 1412.
7 Detailed technical inormation on the constructiono Eora can be ound in M. Lenzen, K. Kanemoto, A. Geschke, D. Moran (2012) “Mapping theStructure o the World Economy.” EnvironmentalScience & Technology 46 (15): 8374–8381.Twomore approachable summaries are also due tobe published soon: “Tracing Embodied CO2in Trade Using High-Resolution Input-Output Tables”, chapter in Computationally Intelligent Data Analysis or Sustainable Development. ed. T. Yu.and The Eora MRIO, chapter in The Sustainability
Practitioner’s Guide to MRIO. ed. J. Murray and M.Lenzen.
8 United Nations (1999). “Handbook o input–outputtable compilation and analysis”. Studies in MethodsSeries F, No 74. Handbook o National Accounting.New York.
9 See Leontie, W. (1970). “EnvironmentalRepercussions and the Economic Structure: AnInput-Output Approach”. The Review o Economicsand Statistics, 52(3), 262–271.
10 Starting with the seminal work o Hummels, Ishii& Yi (ibid.), variations o this methodology haverecently been used in a number o recent papers.Johnson & Noguera (ibid.), Timmer, M., B. Los, R.Stehrer, and G. de Vries (2012) “Fragmentation,Incomes and Jobs. An analysis o Europeancompetitiveness”, WIOD Working Paper 9,Groningen; and Stehrer, R., N. Foster and G. de Vries (2012) “Value Added and Factors in Trade: A Comprehensive Approach”, WIIW Working paper80, Vienna ultimately reapportion worldwide naldemand across countries, rather than exports,allowing them to disentangle the value addedcreated in one country due to consumption inother countries (‘trade in value added’). The OECD-WTO exercise instead ollows an approach entirelysimilar to the one presented here and originally
proposed by Koopman et al. (ibid.): this approachdisentangles the domestic and oreign value addedembodied in a country’s gross exports (reerred toin this report as ‘value added (in) trade’). Details onthe OECD-WTO dataset and method can be oundin OECD - WTO (2012), “Trade in Value-Added:Concepts, Methodologies and Challenges”, www.oecd.org; and in De Backer and Miroudout (2012).“Mapping Global Value Chains”, Paper prepared orthe WIOD Conerence: Causes and Consequenceso Globalization, Groningen, The Netherlands, 24-26 April 2012.
12 For a precise decomposition, see Koopman et al.(ibid.).
13 To get an idea o the complexity o the exercise,
each yearly MRIO contains tens o millions o observations, that is around 4GB o data, buttogether with the superior variables needed tobalance the MRIO table at the world level, eachdataset to be used by the optimization algorithmgrows to 70 GB, and thus requires 2 to 10x asmuch in RAM capability to run. The Australian NCIsupercomputing acilities have been used by theEora team to retrieve value added trade data.
14 The GTAP dataset employs a dierent balancingalgorithm with respect to other existing world I–Otables (including Eora), as the balancing algorithmprioritize the correspondance between gross vs.SUT-derived trade fows rather than domestic value
added. See Koopman et al. (ibid.).
15 See Stehrer et al. (ibid.).16 The Eora data have been validated against the
WIOD data as the latter dataset gives direct access
to the original national Supply-use tables. Assuch, it allows to exactly replicate the underlyingmethodology in the construction o the indicators tobe compared across datasets.
17 Two outliers have been excluded rom thecomparison, that is Bulgaria and Luxembourg. Inboth cases FVA shares in WIOD were some 20%larger than the measure retrieved in Eora. Theaverage (absolute) dierence across the remainingcountries is instead around 6 percentage points.