A partnership with academia

Building knowledge for trade and development

Vi Digital Library - Text Preview

Globalization and Trade Flows: What You See is Not What You Get!

Working paper by Maurer, Andreas, Degain, Christophe, 2010

Download original document

The trade collapse that followed the recent financial crisis has led to a renewed interest on the measurement issues affecting international merchandise trade statistics in the new globalized economy. The international fragmentation of industrial production blurs the concept of country of origin and calls for the production of new statistics on the domestic content of exports, with a view of estimating trade in value added. Alongside, the international statistical community has revised in 2010 the concepts and definitions on both, international merchandise trade and trade in services statistics. This paper discusses the various issues related to the concepts of "goods for processing" and "intra-firm trade" in trade statistics, and provides an overview of the method of analysing the impact of the fragmentation of production in international value chains.

Staff Working Paper ERSD-2010-12 Date: June 2010

World Trade Organization
Economic Research and Statistics Division

Globalization and trade flows: what you see is not what you get!

Andreas Maurer and Christophe Degain


Manuscript date: 22 June 2010

Disclaimer: This is a working paper, and hence it represents research in progress. This
paper represents the opinions of the authors, and is the product of professional research. It is
not meant to represent the position or opinions of the WTO or its Members, nor the official
position of any staff members. Any errors are the fault of the authors. Copies of working
papers can be requested from the divisional secretariat by writing to: Economic Research and
Statistics Division, World Trade Organization, Rue de Lausanne 154, CH 1211 Geneva 21,
Switzerland. Please request papers by number and title.

Globalization and trade flows: what you see is not what you get!

Andreas Maurer and Christophe Degain1

Abstract: The trade collapse that followed the recent financial crisis has led to a renewed
interest on the measurement issues affecting international merchandise trade statistics in the
new globalized economy. The international fragmentation of industrial production blurs the
concept of country of origin and calls for the production of new statistics on the domestic
content of exports, with a view of estimating trade in value added. Alongside, the
international statistical community has revised in 2010 the concepts and definitions on both,
international merchandise trade and trade in services statistics. This paper discusses the
various issues related to the concepts of "goods for processing" and "intra-firm trade" in
trade statistics, and provides an overview of the method of analysing the impact of the
fragmentation of production in international value chains.

Keywords: trade statistics, goods for processing, intra-firm trade, trade in intermediate products,

trade in value added

JEL Classification: C49, F13, F23, F49

1 With contributions from Florian Eberth, Hubert Escaith, Andreas Lindner and Yann Marcus.

Page 2

A. Introduction

The 2008 financial crisis, which translated into a global economic crisis, impacted on trade in a harsh
manner. As of the fourth quarter of 2008, trade flows started to deteriorate, although unevenly across
regions. While Europe recorded a 16 per cent year-on-year decline of its merchandise exports,
comparable rates in North America and Asia were only 7 and 5 per cent respectively. Estimates of
2009 point towards a merchandise trade to GDP elasticity of up to 5 at world level while a long-term
world average of annual values between 1960 and 2008 indicates an elasticity of 1.6. However, for
manufactured goods this value is slightly higher (2.1), that is, manufactures respond more to changes
in GDP than non-manufactures. Several explanations have been advanced to explain this
over-shooting of trade in relation to GDP movements, amongst them global production chains and
vertical specialization, decline of transport costs and tariffs, and improved infrastructure services.

Merchandise trade statistics record trade flows on a "gross basis" whereas the gross domestic product
(GDP) aggregate measures the value added created during the production of goods and services (i.e.
output less intermediate consumption). Thus, trade in intermediate products is only taken into
account in GDP through the additional value produced at each step of the production process.2 In
merchandise trade statistics, in contrast, intermediate goods are counted at full value each time they
cross the border. The share of intermediate manufactured products in non-fuel world trade was
around 40 per cent in 2008.3 This proportion can vary among countries depending on their export
specialization: for Brazil, China or India the share of intermediate goods in total flows in the
manufacturing sector was about 70 per cent in 2005.4

Through the phenomenon of international supply chains, "multiple gross counting" is amplified.
Indeed, the increase in trade elasticity observed after the late 1980s (see Chart 1) is probably due to a
change in business model, where specific segments of manufacturing production have been
progressively outsourced to other countries. Escaith et al. (2010) concludes that the related increase
in trade elasticity observed at global level is only temporary, with a progressive return to long term
average. The trade overshooting observed during the run-up to the 2008-2009 crisis is mainly due to
short-term composition and inventory effects.

2 WTO (2009), "International Trade Statistics 2009", p. 1
3 WTO (2009), "International Trade Statistics 2009", p. 2
4 Miroudot, S., Lanz, R., Ragoussis, A. (2009)

Page 3

Chart 1: World GDP Elasticity of Imports, 10-year-average
(GDP weighted at market exchange rates)







1990 1995 2000 2005 2009

Note: each year on the graph indicates the last year of the ten years' data sample.
Source: Escaith, H., Lindenberg, N., Miroudot, S. (2010)

The following paper tries to shed light on the relevance of official trade statistics in the face of these
economic transformations, focusing on "goods for processing", "intra-firm trade", and "trade in value

B. Trade statistics

Trade statistics of economies are available and used in different statistical frameworks: national
accounts, balance of payments, customs (national or international concepts), input-output tables, etc.
The objective of official merchandise trade statistics is to describe adequately economic phenomena
and provide decision makers with relevant indicators. Yet, the structural changes linked to
globalization are challenging the relevance of these traditional trade statistics. As mentioned in
Escaith (2008), what you see in these statistics is no more what you get. The issue is also important
for trade in services statistics, as these registered flows may also misrepresent current economic
phenomena. Indeed, international supply chains that characterise global manufacturing, particularly
since the 1990s, are also affecting trade in services, as firms in industrialised countries look beyond
the boundaries of the developed world to lower costs and access new talents. 5

In addition, users of trade statistics often face issues of consistency and interpretation, especially
when numbers are taken from different statistical frameworks.6 This happens mainly for conceptual

5 According to Gereffi and Fernandez-Stark (2010), firms are attracted to developing countries as
offshore destinations for their competitive advantages, such as low human resources costs,
technological skills, language proficiency, similar time zones, and geographic and cultural proximity
to major markets. As more sophisticated work and knowledge-intensive activities are performed
abroad, the supply of scientific, engineering and analytical talent offered by developing countries also
become key in attracting firms.
6 To identify "best values" of a country's merchandise trade, international agencies developed an
inter-agency "Common Data Set", available online, that displays officially reported data and "best
values" according to international concepts specified by the United Nations for international

Page 4

reasons, as frameworks follow different objectives and policy purposes. Although different concepts
may exist on justifiable grounds and the international statistical community strives for consistency in
defining common concepts, for users it may be difficult to be aware and understand them at times.
For example, the forthcoming revisions of the System of National Accounts (2008 SNA)7 and of the
Balance of Payments (BPM6)8 opt for a strict application of the change of ownership principle in
recording of international transactions of goods and services. The objective is to improve statistical
adequacy for economic and financial statistical aggregates in relation to the residency criterion. As a
consequence, BPM6 recommends to record trade flows in goods for processing where no change of
ownership takes place on a net basis (value added), and treat them as "manufacturing services on
inputs owned by others".

A major data source for these aggregates are merchandise trade statistics on customs basis, which,
after adjustments, flow into the balance of payments and national accounts. The forthcoming revision
of International Merchandise Trade Statistics – Concepts and Definitions (IMTS 2010)9 reinforces the
concept of "physical crossing of borders" in recording cross-border movements of merchandise by
customs, hence it recommends to record goods for processing on a gross basis.

These conceptual differences, in addition to the practical difficulty to strictly implement the "change
of ownership principle" when recording border crossings, may lead to significant divergences of
national trade statistics aggregates presented in respective frameworks and thus greater statistical
dissonances between individual economies' data. This may create another source of confusion for
users of these statistics. For example, it may change "statistically" a country's specialization with
respect to its relative importance of international trade in goods or services.

The fragmentation of international production (supply chains) and vertical integration also increase
the importance of intra-firm trade of multinationals. "The notion of a firm with a unique national
identity is quickly fading".10 The strict application of the change of ownership principle for goods
under processing and merchanting11 impacts on the recording of such international transactions,
necessitating perhaps, supplementary data collections. Foreign AffiliaTes Statistics (FATS) or trade
and business register links ("zero costs for respondents") are in question to help capture such
transactions. The complexity of corporate ownerships however may limit such data collections.

The valuation of intra-firm transactions itself is often mentioned as a particular concern when
measuring international trade flows as they may not reflect full market prices to take advantage of
fiscal or tax regulations. Transfer pricing is a complex issue. The concept of transfer pricing refers to
both the issue and the solution of a valuation problem in international transactions. On one hand, it
means the allocation of profits for tax and other purposes between affiliates of a multinational
enterprise, using artificial prices (over or under invoicing). On the other hand, transfer pricing refers
to the valuation methods used by tax authorities to avoid taxes.

merchandise trade statistics. See the inter-agency Common Data Set, Task Force on International
Merchandise Trade Statistics, at http://imts.wto.org/common_dataset_e.htm.
7 System of National Accounts (2008 SNA), see http://unstats.un.org/unsd/sna1993/snarev1.asp
8 Balance of Payments, 6th edition (BPM6),
see http://www.imf.org/external/pubs/ft/bop/2007/bopman6.htm
9 International Merchandise Trade Statistics: Concepts and Definitions 2010 (IMTS 2010), see
10 Desai, M.A. (2009)
11 While goods sent/received for processing for which no ownership takes place are classified as
manufacturing services on inputs owned by others, merchanting (the goods are not entering the country
of residence of the merchant) are classified in the trade in goods account as negative exports when
bought and positive exports when sold. Thus, only the balance is accounted for in the trade in goods

Page 5

However, it is difficult for statisticians to correct these distortions which arise from the companies'
shifting assets, income and profits for reducing tax burdens. International guidelines on such
transactions are available from either the OECD or the WCO, the question being to which extent these
are implemented, for example, outside the OECD region.12

Goods shipped for processing, increased intra-firm trade of multinationals or transfer pricing are inter-
related phenomena which are growing with an increasing fragmentation of production chains.
This adds to the difficulty to determine a product's origin and to reply to the question of "Who
produces for whom in the world economy?"13.

Vertical fragmentation of production is a phenomenon of globalization that affects both goods and
services-producing industries. It leads to an increase of trade flows in intermediate products (parts
and components) in the manufacturing sector and also to an increased services content in goods
("trade in tasks"). Industrial supply chains may blur the country of origin concept as part of the
commercial value of an imported good may not originate in the "country of origin" mentioned in the
custom documents. Consequently, what part of value is a country adding to an exported product and
which part is coming from an earlier imported product?

Users turn to international trade statistics for replying to such questions. The remainder of this paper
describes the decomposition of trade flows for analyzing the effects of goods for processing,
intra-firm trade and trade in value added.

C. Goods for processing – is gross reporting reflecting economic reality?

1. Economic importance

With multinational corporations fragmenting international production, sales of goods between firms
or multinational companies has increased. This is reflected in an increase of trade in intermediate
goods. Outsourcing is often used to describe this phenomenon, shifting production either to another
domestic company or a foreign firm abroad (offshore-outsourcing).14 When the company abroad is an
affiliate, involving foreign direct investment to establish it, trade between the two entities is
considered intra-firm.

Merchandise trade receives a "continued interest...due to its crucial role in economic development as
it binds producers and consumers located in different countries into a global economic system".15 The
underlying principle for it is to measure trade flows gross. Users are interested to analyse these trade
flows at global level as well as for trade in intermediate goods, shipments for processing, or intra-firm

Goods sent abroad for processing in particularly defined zones are usually benefiting special tariff and
tax treatments. Often, these activities are taking place in Export Processing Zones (EPZs) of
developing economies, where imported materials and other intermediate goods are processed or
assembled for export.16

12 OECD Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations, see also
Transfer Pricing and Customs Valuation, edited by A. Bakker and B. Obuoforibo, IBFD, 2009.
13 Daudin, G., Rifflart, C., Schweisguth, D. (2009)
14 Krugman, P., Obstfeld, M. (2005), p. 20
15 Draft IMTS 2010, see http://unstats.un.org/unsd/statcom/doc10/BG-IMTS2010.pdf
16 Tax concessions and subsidies offered to firms operating in EPZs are covered by the WTO
Agreement on Subsidies and Countervailing Measures.

Page 6

In addition to contributing decisively to a country's export diversification strategy, EPZs have also
been an increasing source of employment and income for developing economies.17 Based on balance
of payments (BOP) statistics, it is estimated that around a fifth of developing country exports are out
of export processing zones while the share on the import side is about 13 per cent.

A number of economies identify within their balance of payments data on goods for processing in the
compiling economy. This allows a partial (lower) approximation of imports of intermediate input for
processing, and the subsequent (re-) exports of finished products.18 Statistics available from
developing economies hint that the share of their exports resulting from their participation in global
supply chains have accounted for not less than 18 per cent of their total exports since 2000, probably
considerably more. Over the 2000-2008 period, China alone accounted for about 67 per cent of all
reported exports of "compensating products" resulting from inward processing, while Mexico
represented another 18 per cent.

Chart 2: The importance of goods for inward processing in developing economies total exports
and imports (minimum approximation), 2000-2008
(Billion dollars)

Source: IMF Balance of Payments statistics and WTO estimates.

The overall impact on economies in which these processing operations take place appears to be
substantial, as the resulting value added (approximated by the difference between exports of
processed products and imports of intermediate inputs) represents some 30 to 35 per cent of the
exported products.

EPZs are an important development factor, for example, China's processing exports account for nearly
half of its total export value19, but smaller economies also benefit from such zones (see Box 1 below).

17 See Milberg (2007) for a survey on EPZs prepared for the International Labour Organization.
18 Conceptually goods for processing include both processing in the compiling economy (inflows of
goods received from abroad for processing and their subsequent re-exports to the initial shipper) and
processing abroad (outward processing). At present, compilation of information on outward processing
is marginal among developing economies: such goods are generally not identified separately from
general merchandise trade. Thus statistics available on goods for processing for some 30 developing
economies widely reflect imports into and exports from their processing zone, which is an
approximation for inward processing.
19 Ma, A.C., Van Assche, A. & Hong, C. (2008)

Exports, BOP basis


2000 01 02 03 04 05 06 07 2008

Goods resulting from inward processing Other goods

Imports, BOP basis


2000 01 02 03 04 05 06 07 2008

Goods for inward processing Other goods

Page 7

While China's trade numbers indicate that in 2009 nearly half of its exports came out of an EPZ, and
one third of its imports were contracted in such zones, China further distinguishes domestic from
foreign-owned enterprises (FOE) in these zones. For example, in 2008/09, two thirds of China’s EPZ
trade were actually performed by foreign-owned enterprises.

Table 1: China's trade by customs regime and type of enterprise, 2008-09
(Billion dollars and percentage)

2008 2009
Exports Imports Exports Imports

1. Total 1429 1133 1202 1006
2. Processing and assembling 111 90 93 76
3. Procesing with imported material 565 288 494 246

of which FOE:
4. Processing and assembling 46 46 40 38
5. Procesing with imported material 388 207 341 179
Total processing trade / Total trade (2+3)/1 47.3 33.4 48.8 32.0
Processing trade (FOE) / Total trade (4+5)/1 30.4 22.3 31.7 21.6

Processing trade (FOE) / Total processing trade (4+5)/(2+3) 64.2 66.9 64.9 67.4

Source: China's Customs Statistics Dec. 2008-2009, Tables 5-7

Box 1: INTERVIEW - Special Kenyan export zones increase sales -regulator

Sales by factories in Kenya's special Export Processing Zones (EPZs) have increased nine fold
since the first such zone was set up 15 years ago, the regulator of the sector said on Wednesday.
Total sales from the zones have grown over 800 per cent to 28 billion shillings ($365.8 million)
last year from just three billion shillings in 1994. "The experience has been a success story. Since
we started the EPZ program, things have never been the same in the manufacturing sector," said
Joseph Kosure, acting chief executive of the Export Processing Zones Authority. The exclusive
zones were created to attract investment by giving incentives such as a 10 year tax holiday on both
corporate and income tax for foreign directors. Factories in the special regions also have
continuous supplies of power and water that normal industries are not assured of. "We attempt to
maintain world class standards for the operators," Kosure said. "We give them investor support
they don't have to suffer immigration, labour, customs issues." The zones, for example, have
custom officers and will soon enjoy the services of labour and immigration officers on site. "We
do not want fiscal incentives only. We are attempting to depend on procedural incentives also."
The introduction of the African Growth and Opportunity Act (AGOA), a programme by the
United States to encourage exports from the continent through duty and quota free trade has also
boosted EPZ output. "When AGOA came in 2001, there were leaps and bounds in terms of
attraction of investors from outside and from the technology transfer. Kenyan entrepreneurs also
believed that yes, there is something here that they could try," Kosure said.

(By Helen Nyambura-Mwaura NAIROBI, Aug 5, 2009 (Reuters))

Page 8

2. Statistical measurement

International merchandise trade statistics record international transactions when goods physically
cross borders. The forthcoming IMTS 2010 is reinforcing this principle, in that it says that "in all
cases goods for processing, as well as goods resulting from the processing (compensating products in
customs terminology) are to be included in merchandise exports and imports of the countries, as
applicable, at their full value".20

Goods sent abroad for processing can either be (i) returned to the sending company (no change of
ownership), (ii) cleared for home use in the processing country, or (iii) exported further to a third
country. For the second and third case, a change of ownership usually takes place which should result
in a consistent treatment between IMTS and the Balance of Payments (BPM) when recording these

The forthcoming BPM6, to be consistent with 2008 SNA, excludes from the goods account
transactions in "goods sent abroad for processing" for which no change of ownership is taking place.
The new Manual recommends that in that case a processing fee is recorded under "manufacturing
services on physical inputs owned by others". However, both inward and outward movements of such
goods should be tracked to assist in identifying the necessary processing fee.21 These values should
help identify cases where the goods are subsequently sold, rather than returned, in which case they are
identified as an export from the owner's economy at the time of sale. Also under 2008 SNA, the
processing fees by the outward processing economy would be recorded as imports of services, while
the value of goods for processing would be excluded from exports/imports of goods in the goods

Depending on the size and nature of a country's processing activities, the application of the new
BPM6 recommendations can lead to a significant shift in the goods-services split of an economy's
trade, while customs-based merchandise trade statistics still show goods for processing as trade in
goods. Thus, differences in published trade aggregates of a country may increase in future due to
conceptual differences between statistical frameworks and the [in] ability of measuring and allocating
properly the processing fee.

In the course of preparing the BPM6 decision on the treatment of goods for processing, a fictive
example (see table below) was calculated by the IMF for China's current account balance of 2007. It
argues that the new treatment is not changing the current account; it changes only an economy's split
into goods and services in favour of the services position.

20 Draft IMTS 2010, see http://unstats.un.org/unsd/statcom/doc10/BG-IMTS2010.pdf, p. 15.
21 Keeping track of the gross values may also prove useful when deriving price indices, or conciliating
SNA indicators with basic economic statistics, for example in the industrial sector.

Page 9

Table 2: Potential impact of the new BPM6 treatment of processing trade on China's current
account balance, 2007
(Billion dollars)

BPM5 Adjustment BPM6

Exports of goods 1220 -618 602
Imports of goods 905 -352 553
Goods trade balance 315 -266 49

Exports of services 122 266 388
Imports of services 130 130
Services balance -8 266 258

Balance on goods and services 307 307
Current account balance 372 372

Source: IMF, BOPCOM,-08/11

Although this example shows an identical current account balance before and after implementation of
BPM6 changes would only offset each other when the processing fee would be equal to the difference
between goods received for processing and goods returned after processing -- in practice, this does not
hold. In fact, the value of the assembled product may be greater than the cost of the intermediate
inputs plus the processing fee in that it includes for example additional costs for services (e.g.,
research and development, etc).22 Large exchange rate variations between the import and re-export
dates may also introduce significant doses of statistical noise.

3. Measuring the processing fee

The processing fee recorded in BPM6 as manufacturing services may include materials purchased by
the processor.23 For most cases, the value of the manufacturing services or the processing fee is not
simply the difference of the value of goods before and after processing. Some reasons for such a
difference may be the reliability of the customs value of goods as it may include profit margins or be
subject to other factors of pricing in respect of intra-firm trade; cross-period movements (goods are
supplied in one period but returned in another); price effects through holding gains/losses; no
reporting of exports if goods are destroyed during processing; inclusion of overheads such as
research, patents, finance, and marketing, etc.

For compilers, who use the merchandise trade values in developing estimates of processing services,
above factors may be a source of concern. For compilers, who decide to use data from enterprise
surveys or an International Transactions Reporting System, above factors may help explain
differences to values derived from merchandise trade.

22 2008 SNA, chapter 28, para 28.14
23 IMF, Balance of Payments 6th edition, chapter 10, p. 161

Page 10

4. Data collection, gaps and dissemination

For the implementation of the new statistical standards for goods for processing, the following
variables will be needed:

ƒ Value of the processing fees;
ƒ Value of raw materials/semi-manufactures used for processing (that were not imported from

the declaring country or that were imported from a third country);
ƒ Information on ownership status (with/without change of ownership).

Possible options for the data collection could consist of following methods:

ƒ Expanding the trade declaration documents accordingly (sub-categories for imports and

ƒ Survey on processing activities;
ƒ Applying new data models and imputations;
ƒ Other models, e.g. combinations of the methods above.

Each of the methods would come with advantages and disadvantages (trade declarations could deliver
most details and all additional data required but are the one with the highest burden for traders).

Under the new statistical standards, the value of external trade in goods shown under National
Accounts and Balance of Payments frameworks will be significantly different from that shown under
merchandise trade statistics since the latter statistics are not compiled based on the "change of
ownership" principle.

In dissemination of the respective trade aggregates, users need to be provided with the respective
metadata to interpret the numbers in the respective context. For example, releasing two different
figures on trade in goods will require appropriate bridge tables to explain the differences between the

In this context it might be interesting to mention that the OECD is experimenting a different
presentation of trade figures in a consolidated manner, e.g. a merged view of goods and services
trade through a product-oriented scope combining the ISIC, CPC and EBOPS classifications. First
exploratory steps have been undertaken by the OECD Statistics Department to produce such a

D. Intra-firm trade – trade flows and ownership

1. Economic importance

Multinational Enterprises (MNE) can move goods and services across borders without a change of
ownership; the choice is often linked to tax considerations. Fragmentation of production across
countries and vertical integration of multinationals have increased the importance of intra-firm trade.
For the US, these trade flows are estimated at between 45 and 48 per cent on the import side, and 31
to 32 per cent on the export side for the period 1992 to 2004.25 Multinationals may relocate parts of
their production processes to countries with a relative comparative advantage. Anecdotal evidence
reports that "one-third of all [merchandise] trade" is within companies (for individual sectors it might
be even higher), for example, a multinational shipping raw or intermediate goods to an affiliate abroad
for assembly. The country assembling the final product may then export it to a third country. For

24 Wistrom, B. (2010)
25 Akram, T., Khan, H.A., Holladay, J.S. (2007)

Page 11

example, Japanese multinationals shifted their production base to the East Asian region which in turn
exported final products to third countries. In 2006, intra-firm trade of US multinationals with their
majority-owned affiliates amounted to between 37 and 38 per cent of their total trade.26 The
complexity of corporate ownerships however makes it often difficult to identify intra-firm trade in
practice. Moreover, the national definition of related firms differ from country to country, in some
instances, only partial ownership of the affiliate is sufficient to qualify, while in other places,
controlling participation or full ownership is required.

Chart 3: Intra-firm trade in US total private services exports, 1997-2007
(Billion dollars)









1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Total private services

Source: Bureau of Economic analysis, US Department of Commerce

While the globalization of business and firms' increasing fragmentation of different stages of the
production process has dramatically increased the role of intermediate goods in merchandise trade
over the last decade, a similar pattern has been emerging in the field of services.

The bulk of this increase is not represented in cross-border transactions, but in the activity of foreign
affiliates installed closer to their customers. Based on Foreign Affiliates Statistics (FATS), the United
States has recorded an increasing share of trade between multinationals and their majority-owned
foreign affiliates, growing from 21.5 per cent to more than a quarter of its total trade (27.5 per cent) in
2007 (covering only non-bank affiliates, and excluding transportation services and travel).27

2. Statistical measurement

IMTS 2010 emphasizes the inclusion of trade transactions between related parties (whatever the
definition of "related"). These transactions should be "included in international merchandise trade
statistics the same way as if these transactions would take place between unrelated parties..." These
transactions should be valued at arm's-length prices; however, if both trading partners belong to the
same multinational enterprise, risks and profits may be transferred in such a way to benefit of tax
regimes. "Transfer" prices may be used which would in principle require adjustments to reflect arm's
length prices. Such adjustments are however not easy.

26 Akram, T. et al. (2007)
27 WTO (2009), "International Trade Statistics 2009", p. 6

Page 12

The collection of data on intra-firm trade could theoretically also be implemented by respective
modifications (extensions) of the trade declaration documents (additional variable). Again, this would
involve a high burden on respondents and the necessity of the respective knowledge whether traders
concerned are affiliated or not (definition of affiliation?). Collecting these data via the FATS
framework or in particular via data linkages (combining trade and business registers) might be more
appropriate alternatives. As a consequence, transfer prices could be better recognized as such and be
adjusted (e.g. towards "arm's length" prices).

The BPM6 recognizes the importance of exchanges of goods and services between affiliated
enterprises. Many cross-border movements in goods are between affiliated enterprises. The goods
may be moved for processing, resale, and other purposes. The question may arise as to whether there
has been a change of economic ownership, especially when EPZs are involved. Whether there has
been a change in economic ownership is determined according to the usual principle that the
economic owner is the party that bears the risks and rewards of ownership. In cases where there has
been a change of possession of goods between affiliated enterprises, but it is not known whether there
has been a change in ownership, the following factors should be considered:

ƒ When affiliated enterprises28 are separate legal entities, their transactions should be treated

according to the parties' own arrangements as to whether there is a change of ownership or not.

ƒ Between a quasi-corporation29 and its owner, legal title is not usually available as evidence of the

nature of the movement of goods. The preferred treatment in this case is to identify which part of
the legal entity assumes the risks and rewards of ownership, based on evidence such as which
location has the goods recorded in its accounts and is responsible for the sale of the goods. The
treatment should be consistent with reporting by the branch in business accounts and enterprise or
establishment surveys.

In some cases an estimate of a market-equivalent price may need to be made. For example, barter
trade, aid goods, provision of goods and services between affiliated enterprises, under or
over-invoicing, goods on consignment or for auction, or where goods change ownership but a final
price is determined later may require adjustment to the goods value. Such adjustments may also
require corresponding financial account items, such as trade credit; in the case of goods supplied by
direct investors to their direct investment enterprise below cost or without charge, the corresponding
entry is direct investment equity.

On the services side, a direct investor may temporarily provide equipment to its direct investment
enterprise, which would be recorded as an operating lease. Management and ancillary services may
also be provided. (BPM6, para.10.150): services for the general management of a branch, subsidiary,
or associate provided by a parent enterprise or other affiliated enterprise are included in other business
services, often under professional and management consulting services. However, reimbursements of
ancillary services supplied by affiliated enterprises, such as transport, purchasing, sales and
marketing, or computing, should be shown under the relevant specific heading. Management fees are
included in other business services. However, disproportionately large values of services between
affiliated enterprises should be examined for signs that they are disguised dividends, for example,
indicated by large fluctuations that do not reflect actual changes in the services provided.

Compilers in each of the economies involved are encouraged to cooperate and exchange information
in order to avoid asymmetrical recordings of bilateral data.

28 Affiliated enterprises are enterprises related through direct investment ownership structures, such as
branches, subsidiaries, associates, and joint ventures.
29 A quasi-corporation is an unincorporated business that operates as if it were an entity separate from
its owner(s).

Page 13

E. Trade in value added: what is the country of origin in an interconnected world?

1. Economic importance

International production sharing through supply chains is a phenomenon of globalization in both
goods and services-producing industries. It leads to an increase of trade flows in intermediate
products (parts and components), especially in the manufacturing sector. In parallel, the services
content of manufactured goods is increasing, due to the need of linking the global production chain
and to guarantee increased product diversity and customization of products.30

Today, for advanced industrial countries, services (R&D, transportation and insurance) and
intellectual property rights may take a much larger share in the domestic content of a manufactured
product than manufacturing per se. In addition, the distinction between providers of services and
providers of intermediate goods in the productive chain is increasingly blurred. Indeed, "services
provide the mortar that holds international supply chains together".31 Traditional trade-related services
such as transportation are gaining importance in the final price of merchandise, and research has
shown that the rate of protection due to transport costs is higher than due to tariffs.32

The often-quoted iPod example shows this increased service content: an iPod retails for $299 while its
production costs are $144 resulting in a makeup of $155.33 This difference is due to retailers and
Apple's costs for design, research, development, etc. However, the production costs of the iPod
manufactured abroad enter into the trade balance of the countries' bilateral exchanges as traditional
merchandise trade statistics measure cross-border trade flows gross. If the iPod's assembling country
receives further inputs from other countries to assemble their product, it imports intermediate goods.
Depending on the size of fragmentation, these intermediate goods may cross frontiers several times.
Traditional trade statistics may therefore "overstate the degree of import competition that comes from
one's trading partners."34.

Industrial supply chains add to the "blurriness" of the country of origin concept as the value of an
imported good does not necessarily fully originate from the "country of origin" mentioned in custom
documents. For example, if the value of an iPod imported by the US from China is about $150, "the
value added to the product through assembly in China is probably a few dollars at most".35 Indeed,
when strictly analyzing country of origin, i.e., combining the value added content with the country in
which it was produced, bilateral trade balances may be altered (see Table 5). Again, the Apple iPod
example is a model case as only a small fraction is value added originating from China – the
remainder stems from the US, Japan and other countries where conception and production of critical
components, as well as marketing efforts are located.

The electronics sector is an industry particularly concerned. "It is characterized by a high value to
weight ratio and production of parts and components can easily be separated in time and space".36

Comparing trade in value added instead of gross flows in trade may change the analysis of trade. For
example, intra-regional trade flows and trade of regional blocs such as NAFTA or ASEAN show a
higher growth than trade with third countries. Analysing trade in value added may change this trend
and reveal that intra-trade growth may mainly be based on flows of intermediate goods within

30 Nordas, H. K. (2005)
31 Nordas, H. K. (2005)
32 Limao, N., Venables, A. J. (1999)
33 Ikenson, D. (2010)
34 Lamy, P. (2010)
35 Linden, G., Kraemer, L., Dedrick, J. (2007)
36 Nordas, H. K. (2005)

Page 14

increasingly integrated productive networks. It is therefore important to develop trade measures that
complement gross recording with the value added analysis.

2. Statistical measurement

To be able to quantify the part of imported inputs in an economy's exports, different statistical
approaches are used. The most accurate results are achieved through tracing value added in
individual case studies. Well known examples are the analysis of iPod's supply chain or Mattel's
Barbie dolls. However, these samples were chosen more for their illustrative potential than for their
statistical representativeness.37 Another data source available is foreign trade statistics. While trade
in intermediate goods might deliver an estimate at product level, the main issue is to define what to
consider intermediate. For example, is software an intermediate product or is it for final use? The
third approach is to link foreign trade with production statistics, through the input-output (I-O)
statistical framework. International input-output (II-O) tables extend this concept to present linkages
between several countries and allow compiling estimates on the value added content of trade flows.

2.1 Case studies – can links of trade and business registers help to derive microeconomic data?

Some examples of such case studies date back to 1998, when a WTO Annual Report described that
only 37 per cent of the production value of a typical American car is generated in the US.38 The
recent release of Apple's iPad, also produced through a global supply chain, was taken up by press in
very much the same way as earlier examples of Apple's iPod or Mattel's Barbie doll, referring to an
increase of the US-China trade gap through imports of iPad parts manufactured in China and priced at
$287 while in fact only some $12 value was added in China.39

Ideally, to trace intermediate consumption of enterprises for measuring value added produced by
individual sectors of an economy, one would need micro data to determine for each enterprise input,
output and how much of the output is bound for domestic use or exportation. While there are of
course access and confidentiality issues to these data, there is a new, promising data source that
statistical offices experiment with – the linkage of trade and business registers, that is, the use of
administrative data.

Micro data may actually help get around some of the assumptions that are inherently used in input-
output analysis. For example, all firms within a sector are homogenous...in "that the intensity in the
use of imported inputs is the same between production of exports and production for domestic
sales."40 Thus, input-output analysis to date focuses on aggregated, industry level data under the
assumption of homogeneity. This is a typical assumption applied in I-O analysis as well. However,
companies are specializing (trade in tasks), that is, diversifying their production and related tasks such
as accounting, marketing, etc. through offshore outsourcing around the globe. Enterprises may
specialize in export markets while others do not export but produce for the domestic market only, thus
eventually providing inputs to other companies that trade abroad.

Through linking the registers, a Canadian approach for example attempts to identify "who is engaged
in international trade".41 Linking both importer and exporter registers allows to identify the
throughput, that is, the import content of exports at firm level. Such, linking registers is an ideal
addition to analyse trade in value added measures at micro level to verify results received from
input-output analysis for the macro level. Such measures at case level have the advantage of being
very precise. But they probably remain reserved for large statistical institutes in developed countries

37 Escaith, H. (2008)
38 Grossmann, G.M., Rossi-Hansberg, E. (2006)
39 Corcoran, T. (2010)
40 Koopmann et al. (2008), p.3
41 Armstrong, Ph. (2009), p. 4

Page 15

and are still at the stage of pilot projects. An alternative is to work from existing trade statistics to
disentangle trade in final goods and trade in intermediate inputs, on both the export and the import

2.2 Trade in intermediate products

Trade in intermediate goods is generally measured through a breakdown of trade statistics according
to the UN classification of Broad Economic Categories (BEC)42. A rough measure of trade in
intermediate goods based on existing trade statistics is the sum of trade in classification headings
having the words "part" or "components" in their descriptors (BEC 22+42+53) as a share in world

Some example results show that in the case of the US and China, the share of intermediate goods in
non-fuel merchandise is 54 and 38 per cent respectively.43 Other studies prove that the share of trade
in intermediate goods in world manufacturing exports increased from 18.5 to 22 per cent during the
period 1992-2003.44 The WTO estimated the share of intermediate goods in non-fuel merchandise
trade at around 40 per cent in 2008, with wide differences for individual countries.45

Chart 4: Share of intermediate goods in non-fuel merchandise trade, 2008

0% 10% 20% 30% 40% 50% 60% 70% 80%



China, Hong Kong

Costa Rica

Czech Rep.




Taipei, Chinese



Source: UNSD Comtrade database

However, trade statistics remain arbitrary as product definitions do not always allow to distinguish
between intermediate and final use, for example, is packaged software included in IMTS serving as an

42 This product classification helps to define classes of goods, for example, intermediate, capital or
consumption goods. From a national account perspective, capital and consumption goods are for final
43 Nordas, H. K. (2005)
44 Athukorala, P. C., Yamashita, N. (2006)
45 WTO (2009), "International Trade Statistics 2009", p. 2

Page 16

intermediate or as a final product? This difficulty can eventually be solved when working at
disaggregated level. Another limitation is that the methodology is limited to goods (trade in "parts
and components"). There is no similar classification for services, and current statistics on trade in
services are not detailed enough to allow a distribution of services according to their final or
intermediate use. In the context of offshoring, computer and information services as well as other
business services are often used as proxies of intermediate services.

2.3. International input-output approach

Another method for estimating trade in value added relies on the use of international input-output
(II-O) tables which gather national accounts and bilateral trade data on goods and services into a
consistent statistical framework. This approach is often referred to as indirect measurement resulting
in estimates while case studies or analysing trade in intermediate goods is more considered a direct

International input-output tables are commonly used for the analysis of global supply chains when
assessing the import content of exports. The measure of imported inputs that are embedded in exports
for a typical industry is also known as vertical specialization46. The rationale behind vertical
specialization is for companies to benefit from comparative and productivity advantages at different
stages of the production process. It is clearly associated to trade in inputs, caused by international
production chains and the development of outsourcing and offshoring strategies from multinational
enterprises. The domestic content of exports is a measure of exports net of imported inputs47. It
corresponds to the accumulation of the value added incorporated by each of the various domestic
sectors that contributed to the supply chain.

One of the most effective components of the II-O table is the intermediate demand matrix which
enables to identify the origin and use of intermediate goods and services produced and traded amongst
countries and industries covered by the table. The II-O analysis is often associated with the
application of the Leontief inverse matrix.48 The input-output model created by Wassily Leontief is
demand-driven and the elements of the Leontief inverse matrix enable to estimate both direct and
indirect impacts of a change in final use. This is of high interest for trade in value added analysis as
the Leontief approach enables to take into account all backward linkages between countries and

46 Hummels, D., Ishii, J., Yi, K-M. (2001)
47 Economic literature uses “domestic content of exports” and “domestic value added” interchangeably.
48 Leontief, W. (1951)

Box 2: Trade in services -- anecdotic data to show evidence

Estimating the size of offshore outsourcing is suffering of definitional and statistical challenges, and
existing classifications do not allow distinguishing the supply of services to end-use clients or, in the
context here, intermediate, or business-to-business services. Consequently, estimates vary according to
the kind of services included. For example, accounting all IT-enabled services, the OECD estimated a
total value of US$836 billion in 2005. A recent report from NASSCOM asserts that the global sourcing
market has increased threefold over the period 2004 to 2008 and estimates that the addressable market is
more than five times the current market size. Business Process Outsourcing, the report affirmed, has
become an integral part of the global delivery chain and a market in which outsourcing is a key driver.
Regional shifts in spending were observed, such as the increasing shares of emerging markets such as
CEMA (Central and Eastern Europe, Middle East and Africa), Latin America, and Asia Pacific
(excluding Japan). Healthcare was predicted to present emerging opportunities for the future, according to
another NASSCOM report. It estimated that outsourced healthcare may represent approximately
US$17 billion in market opportunity in the US alone. Source: (WTO document S/C/W/304)

Page 17

sectors present in the II-O table and to capture the value of imported inputs used directly and
indirectly (at all the stages of a country's production) in the manufacturing of exported goods.

The use of II-O tables presents some constraints and limitations which need to be considered when
compiling the data or interpreting the results obtained.

A strong assumption when dealing with II-O tables is that of the homogeneity of firms working for
the domestic or the export sector. In other words, the intensity in the use of imported inputs is
assumed to be the same between production for exports or for domestic consumption. While such a
hypothesis may be suitable for a majority of developed economies, this is not a valid one for
economies with pervasive processing trade. Indeed, the value added content of processing exports is
likely to be lower than that of similar goods produced for the domestic market -- see for example the
estimate later on including a preliminary adjustment for China's processing trade.

Additionally, the construction of national I-O tables and their inter-country version is highly
resource-demanding and involves multiple methodological and compilation issues. For instance,
linking product-based trade data (HS, SITC) and activity-based industrial classifications (ISIC), or
setting up conversion procedures for price valuation (e.g. from FOB to producers' price or from CIF to
customs' price...) involve complex data processing. Also II-O tables' compilers, whose goal is to
integrate trade data and domestic production structures, have to face numerous issues related to
international trade statistics (e.g. non-matching mirror statistics, unallocated (confidential) trade,
treatment of re-exports...). As a result of this, they can opt for assumptions or estimations techniques
which may distort the original information. Another consequence of the tremendous effort required to
compile national I-O tables and to reconcile them within a common framework, is that the publication
of II-O tables is seriously delayed. Practically, II-O tables are usually compiled for benchmark years,
having year N table released year N+5... Thus, given the changes in production and trade structures,
II-O data may not always be appropriate for the analysis of current trade flows in value-added.
However, to overcome this weakness, procedures were developed to estimate II-O tables or related
multipliers between benchmark years.49

Finally, II-Os are only available at a relatively aggregated sectoral level (76 sectors maximum for the
Asian input-output (AIO) tables from IDE-JETRO) and thus are not suited for research on trade in
value added at product level. Moreover heterogeneity is too important. As mentioned by a US-NRC
report "Even with the most finely disaggregated import and export data, there are large differences in
unit values of exports and imports across countries, suggesting quality differences that cannot be
eliminated by disaggregation".50

The above remarks confirm that the evaluation of trade in value added through II-Os is indirect.
Therefore it is important to underline that obtained figures are estimates rather than actual measures.

While looking at the various studies and economic literature on the estimation of trade in value added,
one can briefly observe that results obtained vary widely depending on the database used or the
methodology applied and related assumptions or adjustments. A review of the studies on the domestic
value added of Chinese exports from the US Congressional Budget Office51 illustrates this remark.
Indeed, the report highlights that the value added share in Chinese exports can vary roughly from
65 per cent to 95 per cent depending on the studies carried out on standard I-O data (assuming
homogeneity of final production for domestic market and for export). Whereas studies that attempt to

49 The "RAS" method is commonly applied to update the technical coefficients in input-output analysis.
This procedure has been recently used and further developed by the Bank of Japan (Mori, T., Saski, H.,
2007), the ECB (Pula, G., Peltonen, T., 2009) or IDE-JETRO.
50 National Research Council (2006)
51 See http://www.cbo.gov/ftpdocs/95xx/doc9506/AppendixA.pdf

Page 18

distinguish firms whose output is bound for domestic consumption or for export estimate the domestic
value added share in Chinese exports at 35 per cent to 55 per cent.

Data used by the authors stem from IDE-JETRO AIO tables and provisional estimates which cover
nine Asian economies (China, Indonesia, Japan, Malaysia, Philippines, Singapore, South Korea,
Taipei Chinese and Thailand) as well as the United States.

The current methodology (see Appendix for a detailed description) focuses on the estimation of the
import content of exports, or vertical specialization, by economy and sector. It corresponds to the
measure of the share of imported inputs used directly and indirectly (through the use of the Leontief
inverse matrix) to produce exports, including not only the value of inputs imported but also the
foreign content of inputs domestically purchased.

The schematic presentation of the formula retained to compute the vector of sectoral import contents
is the following:

Import content of exports = M * (I-A)-1 * X

M is the import coefficients vector, I is the identity matrix, A is the matrix of technical coefficients
and X is the vector of exports. (I-A)-1 represents the Leontief inverse matrix which captures the
intermediate inputs used at any stage of the production process.

Once the vertical specialization is compiled, the domestic (value added) content of sectoral exports is
derived from:

Domestic content = Total exports – Import content of exports

Domestic content = [ I – M * (I-A)-1 ] * X

Averaging over all sectors, the average domestic content of total exports is:

Average domestic content = u * [ I – M * (I-A)-1 ] * X / [u*X]

Where u is the summation unit vector.

The table below presents the 2000 and 2008 estimates of the average import content and the domestic
content of exports for the ten AIO countries. The results obtained for 2000 are comparable to OECD
estimates52 or figures obtained through the application of the Hummels' formula53 to AIO 2000 data.

52 Araujo, S., 2009
53 See Hummels, D., Ishii, J., Yi, K-M. (2001)

Page 19

Table 3: Import content and domestic content of total exports, 2000 and 200854

China Indonesia Japan Korea Malaysia
2000 2008 2000 2008 2000 2008 2000 2008 2000 2008

Import content of exports 19.0 19.7 15.9 13.5 11.3 16.9 32.3 37.4 49.2 41.0

Domestic content of exports 81.0 80.3 84.1 86.5 88.7 83.1 67.7 62.6 50.8 59.0

Philippines Singapore Taipei, Chinese Thailand USA
2000 2008 2000 2008 2000 2008 2000 2008 2000 2008

Import content of exports 34.8 17.5 53.2 57.9 38.4 46.7 36.6 35.0 11.6 15.2

Domestic content of exports 65.2 82.5 46.8 42.1 61.6 53.3 63.4 65.0 88.4 84.8

Source: WTO estimates based on IDE-JETRO AIO tables

The developed economies of Japan and the US present comparable shares of imported inputs in
exports for 2008 (16.9 per cent and 15.2 per cent). Those shares increased significantly between 2000
and 2008, most probably due to the expansion of off-shoring and intra-firm activities of Japanese and
US multinational companies. The derived domestic value added content of these economies' exports
is inversely high, respectively 83.1 per cent and 84.8 per cent in 2008, reflecting the high content of
national inputs and services embedded in their manufacturing exports as well as the increasing weight
of commercial services exports.

Surprisingly, the vertical specialization observed in 2008 for Indonesia (13.5 per cent) is lower than
that of Japan and the US. The reason for this low figure lies with Indonesia's export structure which is
mainly composed of primary products that do not require intensive use of foreign inputs (agricultural
and oil exports of Indonesia amounted to 61 per cent of total exports in 2008).

Conversely, the exports of goods and services originating from Singapore, Taipei Chinese and
Malaysia are the most intensive in imported content amongst the AIO countries (respectively 57.9 per
cent, 46.7 per cent and 41 per cent in 2008), thus leading to a low magnitude of their trade in value

The estimates for China's import content of exports turn out to be low (around 20 per cent). As
previously mentioned, this is because standard II-Os do not apply specific treatment for processing
zones trade. For some economies, such as China, the share of exports from processing zones in total
exports is high (see Table 1), and the prior measure of import content of exports is obviously
underestimated as China's export processing zones employ much more imported inputs than exports
stemming from non-processing zone trade.

Assuming that all products imported in China's processing zones are embedded in the exports of these
zones, the import content of exports from Chinese processing zones can be estimated at 56 per cent in
2008 (see Table 4 below). The revised estimate for China's vertical specialization can then be
calculated through a weighted average combining the two types of trade. It amounts to 37 per cent
instead of 19.7 per cent with standard AIO data. This rough estimate confirms that any adjustment for
export processing zone trade may have a significant impact.55

54 2000 data for Indonesia and Philippines exclude the "Construction" and the "Electricity, gas and
water" sectors. Japan figures exclude the "Construction" sector. 2008 AIO estimates exclude the
"Construction" sector. The "Electricity, gas and water" sector is only partially covered.
55 Koopman et al. (2008), by decomposing Chinese trade into processed and non-processed trade,
estimates that the import content of Chinese exports is about 50 per cent.

Page 20

Table 4: Adjusting the import content of Chinese exports for processing trade, 2008
(Billion dollars and percentage)

Exports Imports

Total 1429 1133

Processing and assembling 111 90
Processing with imported materials 565 288
Total processing trade 676 378

Total non-processing trade 753 755

Import content of Chinese exports:
1. From processing zones = 56%
2. From non-processing zones = 19.7%
3. Incl. processing and non-processing exports = 37%

Source: China's Customs Statistics, Dec. 2008 (Table 5 "Imports and exports by Customs Regime")

Nevertheless, above estimates might be affected by the implementation of the BPM6 and 2008 SNA
recommendation on the exclusion of all goods imported with no change in ownership from
merchandise trade statistics. For example, assuming that the customs regime "Processing and
assembling" refers to goods for which no change in ownership takes place, the application of the new
principle would lead to a decrease of the import content of Chinese exports by about 5 per cent points,
that is, to 32 per cent instead of 37 per cent. The adjustment applied to calculate this share is similar
to that described in the motor vehicle industry example at the end of the section (see Table 6).

Trade in value added can alter bilateral trade imbalances observed through traditional trade statistics,
with potential implications on trade analysis and policy making. This is clearly illustrated with Table
5 on the US-China trade balance. The US deficit vis-à-vis China would be lowered by 21 per cent
when estimated in value added terms. The amplitude of this deficit is all the more reduced (more than
40 per cent) when considering China's domestic content of exports adjusted for processing trade.
Besides, global trade balances remain unchanged when measured in value added terms.56

56 Daudin, G., Rifflart, C., Schweisguth, D. (2009)

Page 21

Table 5: US-China trade balance, 2008 - Comparison between traditional and value added
measurement, incl. adjustment for processing trade
(Billion dollars and percentage)

No adjustment for
China processing

Adjusted for
China processing

2008 * 2008 *

USA exports to China (traditional statistics) 71 71
USA's share of domestic content in exports (%) 84 84
USA exports to China (in value added terms) 60 60

USA imports from China (traditional statistics) 356 356
China's share of domestic content in exports (%) 80 63
USA imports from China (in value added terms) 285 224

Trade balance (traditional statistics ) -285 -285
Trade balance (in value added terms) -225 -165
Ratio trade balance (value added / traditional) (%) 79 58

* Excluding the "Construction" sector.

Source: WTO estimates

These estimates are based upon standard II-O tables. However, as described below, the strict
implementation of the criteria of change of ownership, which is contemplated in the new 2008 SNA
for the compilation of input-output tables, could lead to a bias in the compilation of trade in value
added multipliers.

Recent studies indicate that intermediate inputs occur in particular in high technology sectors such as
transport equipment or electronics..."the automobile industry is one of many that transcend national
boundaries and is only one example...".57

When aligning the treatment of goods for processing with the "change of ownership" principle in
2008 SNA, the fear is that information for the construction of input output tables is lost.

For example, Hummels et al. (2001) defines vertical specialization (VS) indices for country k and
industry i as:

VSki = (imported intermediateski / gross outputki) * exportski

Ideally, VS would be calculated for individual products. As these data do not readily exist, input-
output tables are used as these provide industry level data on imported inputs, gross output, and

The example of a motor vehicle industry producing two types of cars, one for domestic use and one
for exports, may illustrate a possible impact of the change of the statistical treatment of goods for
processing for which no change of ownership takes place. In this case, an international supply chain
uses a US affiliate installed in Mexico to produce a car which is sold domestically but also exported to
the US. The cars' electronic parts are imported from the US parent company (ownership of these parts

57 Ikensen, D. (2009)

Page 22

is not transferred from the US parent to the US affiliate in Mexico). The exported car contains a 50
per cent higher share of electronic equipment than the car produced for domestic consumption.

The table below recaps the different values of the car depending on its destination (domestic market
or export) and its recording ("average car") in the respective national accounts frameworks of 1993
and 2008.

Table 6: Simulation of the new treatment of goods for processing – 1993 and 2008 SNA

1993 SNA 2008 SNA

Domestic car Exported car Average car Domestic car Exported car Average car

Output (in dollars) 10000 12000 11000 10000 9000 9500

of which:
Domestic inputs 2000 2000 2000 2000 2000 2000
Imported inputs 2000 3000 2500 2000 0 * 1000
Value added 6000 7000 6500 6000 7000 6500

* No change in ownership

Source: WTO

At aggregate level ("average car"), the 1993 SNA would show imported inputs at 2500, while the
2008 SNA would only show imported inputs at 1000. The new treatment of goods for processing
would abandon the registration of the 3000 imported inputs for the exported car as in 1993 SNA,
however, input-output tables would have to include a processor's export of a service (processing fee)
(provided national authorities are able to record these flows). It is here where there is a doubt as to
which countries can implement the 2008 SNA goods for processing treatment across all involved
statistical frameworks. Thus, although attempting to increase transparency of economic relationships
of an economy with the rest of the world, it may actually impede the calculation of the domestic value
added content of exports.58

The 2008 SNA therefore suggests in its para. 28.20 that for reasons of continuity with past practices
the 1993 SNA treatment could be presented as a supplementary item.

F. Conclusion

In the past decades, increased vertical integration of multinational enterprises and the expansion of
processing zones, mostly in developing economies, led to significant changes in trade patterns. One
of the most noticeable features of this evolution is the increasing trade in intermediate goods in the
manufacturing sector. Intermediate inputs are intensively exchanged within international production
chains and imported in processing zones for the production of goods to be exported. The boundary
between goods and services is not always apparent, the magnitude of trade flows therefore at times
misleading. In short, current trade recording systems struggle with the adequate reporting of
globalisation phenomena in respect to goods for processing, merchanting, intra-firm trade, valuation
(transfer pricing) which may introduce some bias in these aggregates.

Revisions of international statistical standards were approved for correcting some of these aspects and
to account for new driving forces observed in international exchanges. But the application of new

58 Escaith, H. (2010)

Page 23

recommendations such as on goods for processing will raise challenges regarding trade data collection
and compilation (e.g. processing fees). Therefore, it is important for international organizations to
setup compilation guidance to ensure a proper application of new BPM6 and IMTS 2010
recommendations at national level.

In addition to changing statistical concepts, the notion of trade in value added receives an increasing
attention as it raises the question of who produces for whom? Analysing trade according to value
added may resize trade imbalances and thus impact on trade policy strategies as indicated recently in a
speech of WTO's Deputy Director General A. Jara..."the design of national policies needs also to be
adapted. Old “mercantilist” policies, based on the vision that trade is a competition between “us” and
“them”, becomes not only sub-optimal...but also a complete anachronism".59

Estimating trade in value added provides complementary tools for the analysis of production and trade
linkages. However, each of the three methods to estimate value added trade flows, as described in
Section E, has its own advantages and limitations. While firm surveys provide a single-case
evaluation, their availability and scope is still limited. Trade flows of intermediates goods provide a
rough estimate without linking up trade and production. II-O tables are most widely used to link trade
and production for deriving estimates on the import content of exports. However, these contain a
number of technical restrictions and are based on assumptions which impact on the estimate of the
share of imported inputs that are used for the production of exports. The improvement of II-O data
quality is therefore important but constitutes a time-consuming task.60

59 DDG Jara urges another way of looking at trade statistics,
see http://www.wto.org/english/news_e/news10_e/devel_26may10_e.htm
60Currently, a consortium of research institutes, funded by the European Commission in its 7th
Framework Programme, is developing a world input output database (WIOD). It aims at constructing
an internationally harmonized database that integrates detailed data on national production structures,
international trade, socio-economic issues, and environmental pressures. This database will be used for
empirical analysis, scenario studies and simulation of a wide range of policy measures (from WIOD
Conference: Industry-Level Analyses of Globalization and its Consequences, Vienna, May 26-28,

Page 24


AKRAM, T., KHAN, H.A., HOLLADAY, J.S. (2007), "US Intra-firm International Trade", Working
paper, SSRN

ARAUJO, S. (2009), "Vertical specialization and global value chains", OECD,

ARMSTRONG, PH. (2009), "Policy Applications of Linked Trade Data – the Canadian Experience
Using the Exporter/Importer register Databases", OECD STD/SES/WPTGS (2009) 17, p. 3 and p. 4

ATHUKORALA, P. C., YAMASHITA, N. (2006), "Production fragmentation and Trade Integration:
East Asia in a Global Context", North American Journal of Economics and Finance, 173, pp. 233-56

BALANCE OF PAYMENTS, 6th edition (BPM6), see

BRUNO, G., ESPOSITO, G., IANDOLI, L., RAFFA, M. (2004), "The ICT service industry in North
Africa and role of partnerships in Morocco", Journal of Global Information Technology Management,
vol. 7, no. 3, pp.5–26

CAVE, W. (2008), "Summary of cross-cutting issues", OECD, OECD/STD/SES/WPTGS(2008)20

statistical standardsfor goods for processing and merchanting in Hong Kong, China", IMF BOPCOM

CORCORAN, T. (2010), "The iPad lesson for China trade", in: Financial Post, Tuesday, April 6,

DAUDIN, G., RIFFLART, C., SCHWEISGUTH, D. (2009), "Who produces for whom in the world
economy?", OFCE, No. 2009-18

DEAN, J., FUNG, K.C., WANG, Z. (2007), "Measuring the vertical specialization in Chinese trade",
US International Trade Commission

DESAI, M.A. (2009), "The decentring of the global firm", Working Paper, 09-054, Harvard Business

ENGMAN, M. (2007), "Expanding International Supply Chains: The Role of Emerging Economies
In Providing IT and Business Process Services", OECD Work Paper No. 52

ESCAITH, H. (2008), "Measuring trade in value added in the new industrial economy: statistical
implications," Actes du 12ème colloque de l'Association française de Comptabilité nationale, INSEE
Méthodes N°122 and MPRA Paper 14454, Munich

ESCAITH, H. "Communication to the Working Group on the Impact of Globalisation on National
Accounts" (unpublished, January 2010)

ESCAITH, H., LINDENBERG, N., MIROUDOT, S. (2010), "International Supply Chains and Trade
Elasticity in Times of Global Crisis", WTO Working Paper ERSD-2010-08, February

FREUND, C. (2009), "The Trade Response to Global Downturns: Historical Evidence"

Page 25

GEREFFI, G., FERNANDEZ-STARK, K. (2010), "The Offshore Services Global Value Chain"
CORFO and the Center on Globalization, Governance & Competitiveness, Duke University.

GIRARD, M. (2008), "The treatment of goods sent abroad for processing in the context of the input-
output framework: work in progress", Statistics Canada, IMF BOPCOM 08/13

GROSSMANN, G.M., ROSSI-HANSBERG, E. (2006), "The Rise of Offshoring: It's Not Wine for
Cloth Anymore, Princeton University

HUMMELS, D., ISHII, J., YI, K-M. (2001), "The nature and growth of vertical specialization in
world trade "

IKENSEN, D. (2009), "Made on Earth – How Global Economic Integration Renders Trade Policy
Obsolete", December 2009, No. 42, CATO Institute

IDE-JETRO (2006) "How to make the Asian input-output Tables", Institute of Developing
Economies, March

IMF STATISTICS DEPARTMENT (2008), "Impact on BOP data of changes in international
standards for processing and merchanting", BOPCOM 08/11

2010), see http://unstats.un.org/unsd/statcom/doc10/BG-IMTS2010.pdf

ISAKSON, H. (2007), "Adding value to the European economy", Swedish National Board of Trade

JANSEN, R. (2008), "Goods for processing and the world-wide consultation for IMTS, revision 3",

JOHNSON, R. C., NOGUERA, G. (2009), "Accounting for intermediate production sharing and trade
in value added"

KOOPMAN, R., WANG, Z. WEI, S.J. (2008), "How much of Chinese exports is really made in
China? Assessing domestic value added when processing trade is pervasive"

KRUGMAN, P., OBSTFELD, M. (2005), "International Economics: Theory and Policy", 7th edition,
p. 20

LAMY, P. (2010), "Facts and Fictions in International Trade Economics", Conference of Trade and
Inclusive Globalization, Paris School of Economics

LEONTIEF, W. (1951), "Input-Output Economics", Scientific American

LIMAO, N., VENABLES, A.J. (1999), "Infrastructure, geographical disadvantage and transport
costs", World Bank Policy Research Working Paper No 2257

LINDEN, G., KRAEMER, L., DEDRICK, J. (2007), "Who captures value in a global innovation
system? The case of Apple's iPod"

MA, A.C., VAN ASSCHE, A., HONG, C. (2008), "Global Production networks and China's
Processing Trade", Journal of Asian Economics

MILBERG, W. (2007), "Export Processing Zones, Industrial Upgrading and Economic Development:
A Survey", New School for Social Research

Page 26

MIROUDOT, S., LANZ, R., RAGOUSSIS, A. (2009), "Trade in intermediate goods and services",
OECD, TAD/TC/WP(2009)1

MORI, T., SASAKI, H. (2007), "Interdependence of production and income in Asia-Pacific
economies: an international input-output approach", Bank of Japan Working Paper Series

NASSCOM (2009), "The IT-BPO Sector in India" Strategic Review

NATIONAL RESEARCH COUNCIL (2006), "Analyzing the US Content of Imports and the Foreign
Content of Exports", The National Academies Press, Washington DC

NORDAS, H. K. (2005), " International Production Sharing: a case for a coherent policy
framework", WTO Discussion Papers No.11

OU-YANG, L. (2008), "The changed treatment of goods for processing and merchanting in the
revised system of national accounts (2008 SNA) and BPM6", Census and Statistics Department,
Hong Kong SAR, OECD/STD/SES/WPTGS (2008)19

PULA, G., PELTONEN, T. (2009), "Has emerging Asia decoupled? An analysis of production and
trade linkages using the Asian international input-output table", ECB


WILLCOCKS, L., GRIFFITHS, C., KOTLARSKY, J. (2009), "Beyond BRIC: Offshoring in
non-BRIC countries: Egypt – a new growth market", LSE Outsourcing Unit report

WISTROM, B. (2010), "Trade by Product and Industry: An estimated example with actual data",
OECD (inter-agency Task Forces on Statistics of International Trade in Services and International
Merchandise Trade Statistics)

YI, K.M. (2003), "Can Vertical Specialization Explain the Growth of World Trade?", Journal of
Political Economy, 111(1), 52–102

Page 27


Annex 1 Preliminary WTO process for estimating the import content and the domestic content
of exports. Data source: IDE-JETRO Asian input-output (AIO) table.

Calculation step Fomula Unit Definition/Meaning

1. Total exports by

Total exports (X) =
exports to AIO partner
countries + exports to the
rest of the world (ROW)
Derived from the AIO

values in thousands $ Total exports in goods and
services by reporting

2. Leontief inverse (I-A)-1
I is the identity matrix
A corresponds to the
whole intermediate
demand matrix from the
AIO table

coefficient Each coefficient from the
Leontief inverse matrix
measures the production
of direct/indirect/induced
inputs in goods and
services needed in supply
country to produce one
unit of output in the
demand country

3. Multiplication of the
Leontief inverse matrix by
the exports vector

(I-A)-1 * X values in thousands $ Amount of inputs
(domestic and imported)
required from each
country/sector to produce
the exported goods and

4. Imports coefficients
from ROW and AIO
partners, by country/sector

Sum of total imports from
ROW and AIO partners
divided by output
Sourced from the AIO

coefficient Share of goods and
services imported from
ROW and AIO partners in
total output
Imports include "Freight
and insurance" as well as
"Duties and import
commodity taxes"

5. Total import content of
exports by country/sector

MROW+AIO * (I-A)-1 * X

values in thousands $ Amount of imported
inputs required from
ROW and AIO partners to
produce the exported
goods and services

6. Domestic content of
exports by country/sector

reported exports (see 1.) -
imported content of
exports (see 5.)

values in thousands $ Amount of domestic
goods and services
required to produce the
exported goods and

7. Shares of import
content and domestic
value added content in
total exports by

Ratio of "Import content"
(see 5.) or "Domestic
value added content" (see
6.) on reported total
exports (see 1.)


Page 28

Annex 2 The IDE-JETRO Asian input-output (AIO) table