See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/350099474

Measuring Global Value Chains

Chapter · March 2021

DOI: 10.4337/9781788113779.00009

CITATIONS READS
6 1,078

1 author:

Timothy Sturgeon
Massachusetts Institute of Technology
92 PUBLICATIONS 14,355 CITATIONS

SEE PROFILE

All content following this page was uploaded by Timothy Sturgeon on 16 March 2021.

The user has requested enhancement of the downloaded file.

 Measuring Global Value Chains

Timothy Sturgeon, MIT Industrial Performance Center

Chapter 3 in the Handbook for Global Value Chains,

forthcoming from Edward Elgar,
edited by Stefano Ponte, Gary Gereffi, and Gale Raj-Reichert

Introduction
As the geographic fragmentation and of industries has increased and cross-border coordination
has improved, efforts to develop better measures and indicators of Global Value Chains (GVC)
have intensified. Producers of official and semi-official statistics and private market research
information have gradually improved data resources available for GVC research, but the process
is slow, piecemeal, and incomplete. At the same time, demand for timely, policy-relevant GVC
research is rising. As a result, two distinct streams of research have begun to come together to
grapple with shared methodological challenges and statistical gaps.

First is what might be called the social science stream: the interdisciplinary body of research
variously referred to as global commodity chains, global production networks, and global value
chains, pursued by like-minded social scientists from the fields of economic sociology, economic
geography, development studies, political science, international business and management,
history, radical economics, industry studies, area studies, and international political economy.
This research has not followed any set format or formal methodological approach but relied on a
mix that can include or combine qualitative field research, small sample surveys, secondary
sources, and mainly descriptive use of official statistics, market research studies, and press
reports. Often, the goal is to trace the vertical and horizontal business relationships in specific
industries, typically from the supply-side, “from conception to end use” (Gereffi et al, 2001),
while considering modes of chain governance and the relative power of actors in the chain, the
structuring effects and pressures of international agreements and institutions, and the regulatory
and institutional influences that prevail in the various jurisdictions where value addition takes

1
place. Many of these variables are difficult to quantify and highly uneven and idiosyncratic
across industries, time, and space. As a result, GVC researchers often rely on rich description.

To manage this complexity and increase the relevance of results, GVC scholars have often
limited the sectoral and geographic scope of their research to the intersection of specific global
industries and localities (e.g., cut flowers in Kenya, footwear in Brazil, eyeglasses in Veneto,
electronics contract manufacturing in Shenzhen, motorcycles in Vietnam, and so on). Deeper
insights require the time-consuming development of industry knowledge (e.g., lead firms,
suppliers, products, components, specialized services and logistics, markets, standards, and
technologies) and familiarity with the social, commercial, and regulatory characteristics of the
(sometimes multiple) places where the chain touches down. While diligent researchers can
develop cogent analysis of specific GVCs and draw conclusions about their impact on specific
places using these ad hoc techniques, the results are often received as anecdote and case study,
which can fall short of what is required for evidence-based policy-making. Still, the value of this
stream of GVC research has been high enough to foster active engagement with policy-makers
and spur the producers of official and semi-official statistics to develop more and better GVC-
relevant data resources.

Second, trade and development economists have produced an overlapping body of research in
parallel to the non-economics social science stream.1 While many have adopted the GVC label
recently, other descriptions used over the years have included: the new international division of
labor (Fröbel et al., 1980), multistage production (Dixit and Grossman, 1982), slicing up the
value chain (Krugman, 1995), the disintegration of production (Feenstra, 1998), fragmentation
(Arndt and Kierzkowski, 2001), vertical specialization (Hummels et al., 2001; Dean et al, 2007),
global production sharing (Yeats, 2001), offshore outsourcing (Doh, 2005), and integrative trade
(Maule, 2006). The main data source for economists have been international trade statistics, and
several complementary product groupings of intermediate and final goods have been published
to help researchers separate GVC-related international trade2 from more traditional arms-length

1
Denoting “social science” and “economics” streams of GVC research is admittedly awkward since both fall within
the social sciences.
2
There is some debate about the definition of “GVC trade,” or if such a category even exists. On one hand, some
economists see GVC trade as referring a general trend toward more elaborate and geographically extensive patterns

2
trade.3 As Chor (in chapter 16 in this volume) demonstrates, the main focus of this research has
been to investigate the determinants of firm-level outsourcing and offshoring decisions and to
estimate the aggregate impact of global production on specific economies.

These studies suffer from the opposite problem faced by the social science stream: narrowly
defined analysis conducted with (usually) aggregate statistics with little direct insight into the
strategies or motives of firms or other actors in the chain. The motives and priorities of firms are
mainly assumed from theory (e.g., based on theoretical trade-offs related to transaction costs or
property rights) rather than collected directly from actors in the chain. The methods are suitable
for econometric analyses that test the relationships between specific variables in a model, but the
identities, practices, and (critically) geographic footprint of key actors in the value chain remain
in a black box. Institutions and the role of technology are typically missing. Often, the price of
methodological “rigor” are lessons that are both abstract and narrow in scope. While the results
are seen as evidence-based, they too often fail the test of high policy relevance.

Trade in value-added estimates from international input-output datasets

All research on social processes has limitations, and variation in questions, methods and results
across disciplines are to be expected, and can be complementary, though the differences can be
so vast that little conversation or learning takes place between research fields.4 This has begun
to change recently. In 2007, a team of industry studies researchers loosely associated with the
social science GVC stream published a rough geographic breakdown of intermediate inputs for
the 30 GB Apple iPod, including assembly costs (Linden et al, 2007). The analysis showed that
the iPod was (famously) designed in Cupertino, California, by Apple and assembled in

of trade, especially in intermediates. Others specifically define GVC trade as only trade where value-added
components that cross borders more than once. Related to this, but with a different focus on the governance of
interfirm networks, the social science stream tends to define GVC trade as trade that requires some level of “explicit
coordination.” For specific classifications that have sought to define GVC trade, see the following footnote.
3
Rauch (1999) identifies three product classes in SITC Rev.2: (i) homogeneous goods, which are traded on an
organized exchange, (ii) reference priced goods with published prices, and (iii) differentiated goods without
published prices. Sturgeon and Memedovic (2010) identify three sets of “specified” intermediate and final goods in
industry sectors with coordinated supply chains: apparel and footwear, motor vehicles, and electronics. Athukorala
(2010) provides a list of 525 “parts and components” drawn from six-digit HS commodities across a variety of
industries. These classifications have been used both descriptively and as controls in formal modeling.
4
In fact, a look at the bibliographies in the social science and economics literature on GVCs commonly shows little
or no overlap.

3
Guangdong Province, China, by Foxconn, a Taiwan-based contract manufacturer, mainly from
components imported from the USA, Japan, Europe, and South Korea. The study estimated that
only about $3.70, or 3% of the iPod’s wholesale price of $123.12 was added in China from
assembly and testing.5

This highlighted, if inadvertently,6 that the wholesale or “factory gate” price of the iPod (and
similar devices, it could be assumed), recorded on the export manifest and entered into official
trade datasets such as UN COMTRADE, might bear little relationship to how much value had
been added in China, much less by domestic Chinese companies. The paper received widespread
press attention, including from the New York Times, and Linden and his colleagues followed up
with similar studies of the Apple iPhone and other devices, including Hewlett Packard and
Lenovo notebook computers (Dedrick et al,, 2010). The approach was replicated and extended
by researchers at the Asian Development Bank (2010), the OECD (2011), and also by a series of
studies of various products by researchers at the Research Institute of the Finnish Economy (e.g.,
Ali-Yrkkö et al, 2011).

These “product-level GVC studies” (Sturgeon et al, 2012) had great impact because of their high
policy and political relevance. Prior to the emergence of GVCs, it was safe enough to assume
that most if not all of the value of an exported product or service was added in the exporting
country. This lent gross trade statistics a great deal of analytic power and policy relevance
because industrial capabilities could be judged by the scale and technological content of exports
(for example see Lall, 2001 and chapter 21 in this volume by Staritz and Whittaker). GVCs
challenge such assumptions and complicate the formation of trade and industrial policies. For
example, with GVCs, import restrictions traditionally used to protect domestic industries could
harm exporters that rely on imported inputs. Product-level GVC studies also highlighted
possible weaknesses in the way input-output tables estimate the import content of GDP

5
A similar observation was made eight years earlier by a reporter for the Las Angeles Times (Tempest, 1996), who
demonstrated that only 35 cents of the value of a $10 “Tea Party” Barbie doll was added in Mainland China, where
it was assembled largely from imported materials.
6
The original motivation of Linden and his colleagues was not to reveal shortcomings of analysis based on gross
trade statistics, it was to estimate the distribution of value capture (by firm and country) in Apple’s value chain.
Estimating the geographic distribution of value-added was only one component of a broader analysis that included
estimates of Apple’s distribution costs, overhead costs and profit margins.

4
(Houseman et al, 2014). Politically, the results were important because of their implications for
international trade negotiations based on calculated trade deficits. In the case of the iPod,
China’s official export statistics were overstated by nearly 4,000%. With the 2008-09 economic
crisis and subsequent slowdown in international trade, concerns about rising protectionism drove
the WTO, the World Bank, and influential trade economists to focus on the argument that
imports are essential to exports, and called for new research and data to underpin an evidence-
based response.

The question was, is the iPod case typical or atypical? By combining Chinese input-output
tables with information that separates export processing from normal trade, Koopman et al
(2008) estimated that about half of the gross value of total Chinese exports in 2006 was derived
from imported inputs, rising to 80% for technology-intensive sectors such as electronics. For
export-processing production as a whole, primarily consisting of products branded by non-
Chinese firms, foreign value-added was estimated to be 82% (p. 19). However, the processing
trade micro-data used by this research team was not collected by the Chinese government after
2008, so this study was unable to be extended for China, much less to other economies.

Still, these early trade-in-value-added studies stimulated important GVC-related questions. How
much foreign value-added value is embodied in a country’s exports? How does this differ across
industries? How far into a country’s industrial base do GVCs penetrate? To what degree are
services embodied in goods exports? In an extraordinary collaboration between teams of
economist and statisticians from the fields of national accounting and trade, a series of important
projects emerged in the late 2000s to link information in national accounts on intermediate input
use and domestic value-added, by sector, with data on international trade in goods and services.
The result was a series international (global and regional) input-output (IIOs) datasets.7

7
These projects, many of which are related and involve deep collaborations and cross-fertilizations, include: the
Asian International Input Output (AIIO) Table created by Japan External Trade Research Organization’s Institute of
Developing Economies (IDE-JETRO), initially released in 2006; the EORA dataset, developed by KMG Associates
(spun off from the University of Sydney by Purdue University’s Global Trade Analysis Project (GTAP); and the
World Input-Output Database (WIOD), a large-scale EU project under the umbrella of the 7th Research Framework
Program (FP7) centered at the University of Groningen in the Netherlands. These projects include various levels of
country coverage, pre-calculated indicators, and methodological transparency.

5
Perhaps most important is the joint OECD / WTO Trade in Value-added indicators (TiVA)
project, which is being managed long-term by the OECD and is emerging as a permanent fixture
in the international statistical system. The first release of TiVA indicators was made in January
of 2013, with several updates made since and many improvements planned over the medium and
long terms. TiVA estimates can provide a broad picture of how countries are integrated into
GVCs, overall and in specific industries. For example, foreign value-added is estimated to
account for about one-third of China’s total export value in both 1995 and 2011 (imports
embodied in exports is sometimes referred to as backward GVC participation), and for the
computer and electronic sector, China’s largest export sector, foreign value-added in 2011 is
estimated to be 55% of gross trade in the sector and 13% of all Chinese exports (see Figure 1).
Domestic value-added re-imported or sent on to a third country for further processing can also be
consistently estimated, as can the use of exported intermediates in the production of partner
economies (sometimes referred to as forward GVC participation). As subsequent TiVA
iterations are released, conclusions might be drawn about trends. Notably, and perhaps
surprisingly given China’s policy drive to increase domestic value-added, the foreign value-
added share of China’s total export value is estimated to have fallen by only 1.2% from 1995 to
2011. Less surprisingly, Chinese value-added sent to third countries is estimated to have
increased from 9.5% of gross export value to 15.6%, reflecting China’s growing role as an
exporter of intermediate inputs.

Figure 1. Trade in value-added estimates for China’s exports, 1995 and 2011.

Source: WTO, “Trade in value-added and global value chains: statistical profiles”, web access:
https://www.wto.org/english/res_e/statis_e/miwi_e/countryprofiles_e.htm

6
The ready availability of trade in value-added datasets opened the floodgates for quantitative
research on the integrated structure of the global economy. They have attracted the attention of
scores of trade economists who have produced many hundreds of papers.8 In general, TiVA-
type statistics are freely available and derived indicators are gradually becoming easier to access,
as evidenced by the availability of country ‘dashboards’ such as the one shown in Figure 1.
Analysis extends beyond the contribution of imports and domestic value-added to exports to the
average number of stages in value chains, the involvement of non-exporters in export chains, and
the contribution of services to exports, the effects of trade agreements on GVCs, and more
accurate estimates of the gains from trade and revealed comparative advantage, all with detail
possible by country and industry.

However, there are questions about the meaning and accuracy of IIO datasets, as they are,
essentially, estimations based on national input-output tables that are also estimations. Data on
domestically and internationally traded services tends to be especially thin (see chapter 14 in this
volume by Low). Most questionable are IIOs, such the 190 economy EROA dataset, which
include countries that do not produce input-output tables at all.9 Information on firms is
completely absent even in the most conservative IIO, so it is impossible to know if exports are
flowing from domestic firms or foreign affiliates of multinationals, or if they are based on
product innovation from domestic lead firms or on specifications embodied in orders from global
buyers. While the results of studies based on IIOs certainly have policy relevance, missing
variables such as ownership and firm size mean that the research generally does not touch on the
central concerns of the social science stream: the development (i.e., upgrading) opportunities,
challenges, and experiences introduced by GVCs in specific places and in very specific sectors
and segments of the chain.

The producers of IIOs, in general, have been fully transparent about these and other
shortcomings. They are aware that the ‘balancing’ estimates undertaken in many cases require

8
Google Scholar listed nearly 600 research papers that rely on TiVA by July, 2018. For WIOD, see Timmer et al,
(2014) and Los et al (2015) and papers listed on http://www.wiod.org/home.
9
According to the EORA website (http://worldmrio.com/documentation/faq.jsp), the proxy methodology is as
follows: “…for which there is no official IO table (e.g. Tunisia) we start with a 26-sector proxy IO table combining
diverse industries and products from the Australian, US, UK and Japanese economies, then scale it to match the
available data that for that country, such as GDP, GNE, imports bundle, exports bundle, and the like.”

7
heroic assumptions, and that industry and product detail, especially on services, is sorely lacking.
Efforts to improve IIOs have created new momentum for national statistical agencies to make
extensions to input-output tables in regard to heterogeneous enterprise characteristics, including
ownership (foreign or domestic control), trading status (importers, exporters, and two-way
traders), and firm size (small, medium and large). Such estimates could help to bring the social
science and economics streams of GVC research closer together by helping to characterize the
role of multinational firms in GVCs, and by extension to identify underperforming enclaves
exporting on the basis of imported technology and intermediate inputs that can limit spillovers to
the local economy to low wage employment (i.e., a low value-added trap). There is also work
on better integrating information on jobs, and including information on income (profit) flows,
which would help to reveal who captures the gains from GVCs and where those gains ultimately
flow, extending the analysis beyond value-added to value capture, and back to the original
interests of Linden and his colleagues.

Official and unofficial statistics

A fully fleshed out IIO, based on real data rather than estimates, publicly available and with
direct links to information on employment, wages, and career trajectories; sources of technology;
value capture; and including information on the identity or at least ownership and size class of
enterprises is an identifiable yet distant goal. In the meantime, steady progress is being made to
both adapt existing statistics to better capture the features and dynamics of GVCs (in a sense
IIOs fall into this category since they repurpose existing statistics), and to collect new data with
the specific aim of filling in knowledge gaps about GVCs. Not surprisingly, given their
significance and persistence, significant efforts are being made to fill statistical gaps related to
services. Private data is also becoming richer and more readily available (though typically at a
cost), so opportunities for their use in GVC studies is increasing. This section reviews a few of
these advancements but is not meant to be comprehensive.

GVC-relevant complementary groupings in official statistics

Official statistics from national data agencies and semi-official statistics from international
agencies such as the United Nations, OECD, WTO, UNCTAD, World Bank, and the IMF have
been preferred sources of information for GVC researchers because they are free, transparent,

8
and objective. Vast progress has been in posting information on the internet where it can be
searched, organized, analyzed, and downloaded for free.10 Internationally comparable data on
trade, GDP, foreign investment, and tariffs, for example, are easily obtainable on line.11

However, official statistics commonly lack relevant information because they stop at national
borders, use incompatible classifications, and vary widely in regard to completeness and
coverage. There are no comprehensive and comparable international statistics on employment or
output by industrial sector, for example, especially at the level of detail that would be needed for
meaningful GVC research. With the exceptions of trade and FDI statistics — and now IIOs —
most official and semi-official data treat the global economy as a de facto collection of discrete
nation states. UNCTAD introduced new data on bilateral foreign investment in 201412, but it
does not include breakdowns by industry, much less value chain segment. Data on traded
services is difficult to collect, and this is reflected in the coarseness incompleteness of available
multi-country data resources.13

Again, a critical shortcoming of official and semi-official statistics, from a GVC perspective, is
opacity in regard to enterprise ownership. Information on publicly listed companies is easy to
obtain, but it does not include consistent or comparable non-financial variables such as sourcing
relationships, geographic sales, or facility locations.14 Otherwise, company names and other key
enterprise characteristics (e.g. nationality of ownership) are almost always excluded from public
datasets for reasons of confidentiality. For example, China is the world’s largest exporter of
mobile phone handsets, by far, but there is no consistent way of knowing from official trade

10
The World Bank’s Open Data Bank offers hundreds of development indicators for most of the world’s economies
(see https://data.worldbank.org), for example.
11
The World Integrated Trade Solutions (WITS) website, a multi-agency collaboration, aggregates some of these
country-level data, and provides a host of built-in analytical tools (see https://wits.worldbank.org) for example.
12
See: http://unctad.org/en/Pages/DIAE/FDI%20Statistics/FDI-Statistics-Bilateral.aspx
13
The most recent international classification for trade in services, the extended balance of payments systems
BMP6, includes only 71 distinct service categories (compared to more than 8,000 for merchandise trade in the
Harmonized Tariff System), and coverage at the level of individual countries is extremely spotty in datasets
published by international organizations such as Eurostat (see: http://ec.europa.eu/eurostat/web/balance-of-
payments/data/database) and OECD/WTO (see: http://www.oecd.org/std/its/international-trade-in-services-
statistics.htm).
14
For example, filings for US-listed companies can be found in the US Security and Exchange Commission’s
“Edgar” database, see: https://www.sec.gov/edgar.shtml

9
statistics if domestic Chinese firms or firms based in another country are responsible.15 To
reiterate, prior to the expansion of GVCs — roughly dated to the mid 1990s and especially to the
early to mid 2000s — it was generally safe enough to assume that a country’s exports were
largely driven by domestic lead firms and supply bases.

These data gaps, and many others, are persistent. Data agencies the world over are faced with
the twin challenges of limited resources and businesses’ complaints about the time required to
respond to surveys (i.e. respondent burden). As a result, data agencies have been trying to do
more with existing resources, including producing disclosable statistics and indicators from
enterprise and establishment-level administrative and other confidential “micro-data”.16 Four
examples are provided here.

A new classification of business functions. The concept of business functions is relatively new to
the statistical toolbox. Business functions offer statisticians and survey respondents a limited yet
relatively comprehensive set of generic, easy-to-understand categories that describe the various
functions carried out by enterprises, in addition to their main economic activity. Business
function statistics are needed because enterprises, in addition to producing the goods or services
from which they earn their revenues, typically require a variety of support functions to support
their primary line of business. Since support functions can be defined separately from the main
industry classification, they can be expressed as a generic list of business services. 17

The motivation for recent business function surveys stems from increased outsourcing — either
domestically or internationally — of support functions such as customer contact services,

15
There are exceptions. For example, the US Department of Transportation’s National Highway Traffic Safety
Administration publishes annual reports from Part 583 of the American Automobile Labeling Act that lists the
geographic content of all passenger vehicles sold in the United States by make and model, including the location of
final assembly and sourcing of engines and transmissions (see: https://www.nhtsa.gov/part-583-american-
automobile-labeling-act-reports#part-583-american-automobile-labeling-act-reports-listings).
16
An example is Eurostat’s Trade by Enterprise Characteristics dataset, which includes indicators such as revenues,
employment, and wages according to enterprise characteristics such as trading status (importers, exporters, or two-
way traders) and ownership status (domestic or foreign owned), and size class (see:
http://ec.europa.eu/eurostat/web/international-trade-in-goods/data/focus-on-enterprise-characteristics-tec).
17
From a statistical perspective, the core/primary function of the enterprise can be associated with the industry or
activity code(s) and associated products (goods or services), while support functions can be associated with the
various business services. Large scale or municipal services such as telecommunications or sewerage services are
generally excluded from business function lists.

10
software coding, and “back-office” functions such as payroll and document management. Most
recently, enterprises are experimenting with fragmenting and relocating the R&D process, with
various tasks interlinked across enterprises and/or geographic jurisdictions, even in the context of
contemporaneous projects. The growing use and capabilities of ICT systems have accelerated
both outsourcing and trade in such ICT-enabled services. Because business functions are generic
in respect to industry or output, results can be compared across different types of firms, and
aggregated to provide broader views of sourcing patterns within a specific economy or location.

For statistical purposes, business functions can be conceived as an alternative or complimentary

grouping of existing classifications of activities/industries (NACE, ISIC or NAICS), products
(CPA/HS/EBOPS), and/or occupations (ISCO/SOC) classifications. The core or primary
function of the enterprise can be associated with the main or secondary industry or activity
code(s) and associated products (either goods or services or both), while support functions can be
associated with the various business services that enterprises can either provide internally or
source externally.

The first official survey to introduce the concept of business functions in a statistical context was
the European Survey on International Sourcing, initially carried out in 2007 and repeated again
in 2012 and 2017/2018. Statistics Canada used a similar approach in 2009 and 2012 in its
Survey of Innovation and Business Strategy. Unofficially, the approach was pilot tested in the
United States in the National Organizations Survey (NOS) in 2011 (see Brown et al, 2013).
Currently, the International Sourcing/GVC survey is being made a permanent part of the
European Statistical System, a manual is being written, and an updated list of business functions
proposed. The results from this work will be taken up as a starting point for an internationally-
agreed upon classification under the auspices of the United Nations Statistical Commission. The
proposed list for Europe and comparative list of business functions used in prior surveys can be
found in Table 1.

11
Table 1. Examples of business function lists used in statistical surveys
2012 International 2010 National 2009/2012 Survey of
2007 International PROPOSED LIST
Sourcing/Global Organizations Survey Innovation and
Sourcing Survey 2017/2018 International Sourcing/
Value Chains Survey (US: Brown & Business Strategy
(Eurostat) Global Value Chains Survey (Eurostat)
(Eurostat) Sturgeon) (Statistics Canada)

(7 functions: 1 core and (6 functions: 1 core and (8 functions: 1 core and (14 functions: 2 core and
(9 functions: 2 core and 7 support)
6 support) 5 support) 7 support) 12 support)

Provision of goods Core business functions

Core business function Core business function Primary business function
• Production of goods (for the market)
Production of services • Provision of services (for the market)

Transportation, logistics, Transport, logistics, and distribution support

Distribution and logistics Distribution and logistics Distribution and logistics
and distribution functions

Marketing, sales and after Marketing, sales services Customer and after-sales Call centres and help
sales services including and after sales services, service centres Marketing, sales, after sales service support
help desks and call incl. help desks and call Marketing, sales and after function
Sales and marketing
centres centres sales service

Data processing

Information technology Software development

ICT services ICT services IT services and software support functions
systems
Information & comm. tech.
(ICT) services

Legal services

Accounting and book-

Management,
Administrative and Administrative and keeping Management, administration, and back-office
administration, and back
management functions management functions Human resource support functions
office functions
management

Financial management

Research and R&D, Engineering and related technical

Research & Development Research and
R&D, engineering and development (R&D) services and R&D support functions
Development of Products,
Engineering and related related technical services Engineering and related
Services, or Technology • Research & development services (R&D)
technical services technical services • Engineering and related technical services
(except R&D)
Facilities Maintenance
Other Other Other Other business functions
Other

Source: based on Nielsen and Sturgeon, 2014.

A new classification of ICT-enabled services. While data on trade in ICT services is not
universally available, the definition is generally agreed upon. ICT services consist, in the main,
of telecommunications (also known as communications or network) services, IT services, and
software. A more significant gap exists in the realm of ICT-enabled services, which has been
defined as any service task, business process, or business function that can be delivered
remotely. The list of tasks, processes and functions that fit this definition are growing along with

12
the computerization of work and maturing of the Internet. Because of this, ICT-enabled services
are generating significant new opportunities for developing countries to participate in
knowledge-intensive segments GVCs, and also creating concern about job loss in advanced
economies. ICT-enabled services lack an internationally agreed upon definition. Conceptually,
ICT-enabled services include activities that can be specified, performed, delivered, evaluated and
consumed electronically. Previously, jobs related to ICT-enabled services have been defined as
those which make extensive use of ICT and highly codifiable knowledge, and thus require a
minimal degree of face-to-face contact (van Welsum and Reif, 2009). Because it covers inputs,
outputs and work processes, this concept is too broad to be used for the measurement of
international trade, which focuses on products (outputs) that cross borders.

To help fill the data gap related to newly tradable services, an internationally agreed-upon
complementary grouping was recently developed under the auspices of the Partnership on
Measuring ICT for Development and its Task Group on Measuring Trade in ICT Services and
ICT-enabled Services (TGServ)18 and accepted by the UN Statistics Commission (see UNCTAD,
2015). The method was to identify services that did and did not have the potential to be traded
electronically. Using one of the more detailed product classifications (CPC 2.1), correspondences
to the main international classifications for trade in services (EBOPS 2010) and industry
(activity) (ISIC rev.4) were made. Services deemed to be without high potential for being
delivered remotely were travel and transportation services and in person and on-site services.

As Table 2 shows, the main categories of ICT services make up the first two subcategories of
potentially ICT enabled products, and seven additional categories round out the category of
(potentially) ICT-enabled services. This framework has already been used by the US Bureau of
Economic Analysis to estimate imports and exports of in ICT-enabled services in the United
States (US BEA, 2016), and has recently been deployed in pilot surveys by the statistical offices
of Costa Rica, India, and Thailand.

18
TGServ members are UNCTAD, UNSD, WTO, OECD, UNESCWA, ITU and the World Bank.

13
Table 2. Classification to identify potentially ICT-enabled services
Main services types and sub-categories
Type 1: Potentially ICT-enabled services •
1.1 ICT services – Telecommunications
1.2 ICT services – Computer services (including computer software)
1.3 Sales and marketing services, not including trade and leasing services
1.4 Information services
1.5 Insurance and financial services
1.6 Management, administration, and back office services
1.7 Licensing services
1.8 Engineering, related technical services and R&D
1.9 Education and training services •
Type 2: Not ICT-enabled services
2a. Transportation and travel services
2.1 Transportation and travel services - transport industry services
2.2 Transportation and travel services - freight services
2.3 Transportation and travel services - passenger services
2b. In person and on-site services
2.4 In person and on-site services - trade and leasing services
2.5 In person and on-site services - utilities and infrastructure-related services
2.6 In person and on-site services - agricultural, forestry, fishing and mining services
2.7 In person and on-site services - construction services
2.8 In person and on-site services - health and social services
2.9 In person and on-site services - in-person and recreational services
2.10 In person and on-site services - maintenance and repair services
2.11 In person and on-site services - manufacturing services
2.12 In person and on-site services - public and membership organization services
Source, UNCTAD, 2014

A new classification of specified intermediate goods and services. The rise of GVCs has made
the analytical distinction between GVC-related trade and arms-length important. Researchers
and policymakers can use trade in specified, customized, or differentiated intermediate goods
and services as a proxy for GVC trade. This is because specified intermediates are highly
dependent on the industry for which they are made and are commonly produced to the
specifications of one or a small number of buyers. Generic intermediate goods are typically
more commodified. They lie farther upstream in the value chain, are used across multiple
industries, have well known reference prices, and are sometimes sold on auctions. They are
therefore indicative of arm’s length trade, rather than explicitly coordinated trade.

14
While the Classification of Broad Economic Categories (BEC) within the United Nations
Statistical Division’s Commodity Trade Statistics Database (known as UN COMTRADE)19
identifies groupings of unprocessed and processed intermediate goods, many processed
intermediates are essentially commodities and therefore unsuitable as proxies for GVC trade
(Sturgeon and Memodovic, 2010).

In addition to other ongoing changes,20 the latest revision of the BEC (Rev.5) includes a new
specification dimension that identifies generic and specific intermediates with the expressed
purpose of identifying GVC trade. A manual for BEC Rev.5 describing the approach was
accepted by the UN Statistics Commission in January of 2016 (see UN Statistical Commission,
2015), and the product coding is being carried out in 2018. Once the classification is
implemented, users will have the option of sorting all historical and future statistics in
COMTRADE accessed according the BEC Rev. 5 classification. As such the specified
intermediates category will provide a more accurate proxy for GVC-related trade than the current
BEC categories for intermediate goods.

Private data sources and their limits

GVC researchers are not alone in their interest in investment and control across thin slices of
industry value chains. Data from company websites, press reports (especially trade press),
market research firms and industry associations are GVC research staples because they can often
be used to identify key firms, sourcing relationships, market positions, and facility locations. For
private companies, a prime fee-based source is Moody’s Bureau van Dyck “Orbis” database,
which contains information on about 275 million companies worldwide, including ownership
structure and subsidiary trees.21 Detailed, firm-level information on foreign investment can be

19
This widely used database contains import and export statistics reported by the statistical authorities of nearly 200
countries, from 1962 to the most recent year, currently 2012 to 2014, depending on the country.
20
It adds services and therefore refers to products rather than goods, it provides new top level of broad economic
categories, based on the main outputs of corresponding industries, to facilitate analyses of trade and production, and
it identifies end-use as separate dimension.
21
For Orbis see https://www.bvdinfo.com/en-gb/our-products/data/international/orbis. Similar sources include
Dunn and Bradstreet’s commercial database, which includes information on 285 million commercial entities,
including their subsidiary hierarchies (see: http://www.dnb.com/solutions/master-data/deepen-your-knowledge-of-
business-entities.html).

15
gleaned from datasets on mergers and acquisitions, which vary widely in quality, cost, and
coverage.22

Such data sources are problematic in several ways. Inaccuracies abound, time series are missing
or fragmented, classifications across data sources are incompatible — often intentionally so23 —
and the cost of data produced for the corporate market can be prohibitive for researchers.
Because datasets are primarily used for commercial purposes, such as contact lists and market
intelligence, even the most comprehensive and consistent market research datasets can exclude
information critical to GVC analysis such as employment, wages, and the location and volume of
production and sourcing. Companies are much more interested in the geography of markets than
the geography or production and sourcing, while for GVC researchers, the opposite is true.
Moreover, while large industries with small numbers of clearly identifiable lead firms such as
motor vehicles and ICT hardware tend to be consistently and comprehensively tracked in market
research datasets, 24 most industries and GVC segments are thinly and sporadically covered.

Very recently, there has been great interest in using data produced by and residing on the
internet. Examples include data traffic, web searches, browser clickstreams, and “scraping”
information from company websites. Some of this information is available for free, and some for
a fee. Web scraping has led to some significant insights. For example, Nathan and Rosso (2016)
set out to identify firms in the UK engaged in the digital economy. By combining contextual
phrases related to digital services found on company websites with administrative data from the
UK statistical agency (using the UK SIC classification), they estimate that about 70,000
companies were ‘missing’ from the UK’s official classification of the ICT industry. Based on
this, employment in the UK’s information economy was estimated to be 50% larger than
suggested by official statistics. However, most data harvested from the internet relates to

22
Examples include Bloomberg, Thompson-Reuters, and Dealogic.
23
Consulting and market research companies purposefully use their own, proprietary industry and product
breakdowns and company identifiers to differentiate their analysis from rivals.
24
For example, the International Organization of Motor Vehicle Manufacturers (OICA) publishes free firm-level
sales and production data by country and type of vehicle from 1999 (http://www.oica.net), and the US
Semiconductor Industry Association publishes monthly semiconductor consumption data by world region from 1972
(https://www.semiconductors.org/industry_statistics/global_sales_report/).

16
individual users, rather than the practices of companies and linkages between them (Sturgeon,
2017).

Final comments

A mix of research methods will always be needed to develop a complete picture of how global
industries are driven, structured, and intersect with local economies. In the social science stream,
at least, GVC researchers typically want to know how, where and by whom GVCs are governed,
how key firms in an industry divide work and exert power across ownership and national
boundaries, how information is exchanged, where work is carried out, and what the economic
and social impacts of these strategic choices are. How are sourcing requirements set,
communicated, and enforced across the supply chain by distant buyers and headquarters of key
multinational firms? The importance of such questions has increased as more industries have
become organizationally and spatially fragmented, with the locus of power shifting from the
unipolar world of the vertically integrated and nationally rooted firm to the multipolar world of
GVCs. However, little of this information is directly available in official statistics, especially at
the level of specificity in regard to ownership, location, and value chain segmentation that give
GVC analysis its power. This points to the enduring value of an eclectic, and often qualitative
approach to GVC research. Still, while development and social outcomes can be gauged by
traditional measures, such as GDP, employment, wages, exports, and various indices (e.g. GINI
or GDP per capita), making statistical linkages between these outcomes and the dynamics of
GVCs has proven to be difficult if not impossible with existing data resources.

Improvements in data resources and new estimates such as TiVA have come about, in part, from
the overlapping interests of economists and non-economist social scientists. So far, these
linkages have proceeded in a largely ad hoc manner. However, the United National Statistics
Division is now coordinating the production of a new Handbook on Accounting for Global Value
Chains, due to be released in late 2018, that will include state-of-the art concepts and instructions

17
for statisticians seeking to upgrade their national statistical systems to better reflect the reality of
GVCs.25 Going forward, there is scope for much more deliberate collaboration of this kind.

References

Ali-Yrkkö, J., Rouvinen, P., Seppälä, T. and Ylä-Anttila, P. (2011) 'Who Captures Value in
Global Supply Chains? Case Nokia N95 Smartphone', Journal of Industry, Competition
and Trade, 11, 263-278.
Arndt, S. and H. Kierzkowski (eds.) 2001. Fragmentation: New Production Patterns in the
World Economy, Oxford: Oxford University Press.
Athukorala, Prema-Chandra, 2010. “Production Networks and Trade Patterns in East Asia:
Regionalization or Globalization?” Asian Development Bank, Working Paper Series on
Regional Economic Integration, #56, August.
Brown, Clair; Sturgeon, Timothy and Cole, Connor. 2013. "The 2010 National Organizations
Survey: Examining the Relationships Between Job Quality and the Domestic and
International Sourcing of Business Functions by United States Organizations." UC
Berkeley: IRLE Working Paper 156-13 and MIT IPC Working Paper 14-001. Web
access:
http://ipc.mit.edu/sites/default/files/documents/NOS%20Working%20Paper%20IPC%20
14-001.pdf
Dean, Judith; Fung, K.C.; and Wang, Zhi. 2007. “Measuring the Vertical Specialization in
Chinese Trade.” U.S. International Trade Commission, Office of Economics Working
Paper, No. 2007-01-A.
Dedrick, Jason; Kraemer, Kenneth; and Linden, Greg. 2010. “Who Profits from Innovation in
Global Value Chains?: A Study of the iPod and Notebook PCs.” Industrial and Corporate
Change, 19(1), 81-116.
Dixit, Avinash and Grossman, Gene. (1982), Trade and Protection with Multistage Production,
Review of Economic Studies, Vol. 49, No. 4, pp. 583-594.
Doh, Jonathan. (2005). “Offshore outsourcing: implications for international business and
strategic management theory and practice”, Journal of Management Studies, 42(3), pp.
695-704.
Feenstra, Robert. 1998. “Integration of trade and disintegration of production in the global
economy.” Journal of Economic Perspectives, 12(4): 31-50.
Fernandez-Stark, Karina; Bamber, Penny; and Gereffi, Gary. 2011. “The offshore services value
chain: upgrading trajectories in developing countries.” International Journal of
Technological Learning, Innovation and Development, 4 (1): 206-234.
Fröbel, Folker; Heinrichs, Jürgen and O. Kreye, Otto. 1980. The New International Division of
Labor, Cambridge: Cambridge University Press.

25
For current news on this Handbook, see: https://unstats.un.org/unsd/trade/events/2018/rome/default.asp

18
Gereffi, Gary; Humphrey, John; Kaplinsky, Raphael, and Sturgeon, Timothy. 2001.
“Introduction: Globalisation, value chains and development.” IDS bulletin 32 (3), 1-8.
Global Value Chain Development Report. 2017. Measuring and Analyzing the Impact of GVCs
on Economic Development. A joint report from the World Bank, OECD, WTO, and
UIBE. International Bank for Reconstruction and Development/The World Bank, pp. 1-
14. Web access: https://www.wto.org/english/res_e/booksp_e/gvcs_report_2017.pdf
Houseman, Susan; Bartik, Timothy; and Sturgeon, Timothy. 2014. “Measuring Manufacturing:
How the Computer and Semiconductor Industries Affect the Numbers and Perceptions.”
Upjohn Institute working paper; 14-209.
Hummels, J. Ishii and Yi, K-M. 2001. “The Nature and Growth of Vertical Specialization in
World Trade”, Journal of International Economics, 54(1): 75–96.
Koopman, R., Z. Wang and S.-J. Wei. 2008. “How Much of Chinese Exports is Really Made in
China? Assessing Domestic Value-Added When Processing Trade Is Pervasive.” NBER
Working Paper Series, No. 14109, Cambridge, MA.
Krugman, Paul. 1991. “Increasing returns and economic geography”, Journal of Political
Economy, 99, pp. 483–99.
Lall, Sanjaya (2001) Competitiveness, technology and skills, Cheltenham: Edward Elgar.
Linden, Greg, Kraemer, Kenneth., & Dedrick, Jason. 2007. “Who captures value in a global
innovation system: the case of Apple’s iPod.” Irvine CA: Personal Computing Industry
Center.
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.419.2289&rep=rep1&type=pdf
Los, Bart; Timmer, Marcel; and de Vries, Gaaitzen. 2015. "How Global are Global Value
Chains? A New Approach to Measure International Fragmentation", Journal of Regional
Science, 55 (1), 66-92.
Maule, Christopher. 2006. “Integrative trade: issues for trade analysis, statistics and policy”,
Paper prepared for the CTPL Conference on Integrative Trade between Canada and the
United States – Policy Implications. http://www.dfait-
maeci.gc.ca/eet/research/TPR_2006/Chapter_2_Maule-en.pdf
Nathan, M. and Rosso, A. 2016. “Mapping digital businesses with big data: Some early findings
from the UK.” Research Policy 44:9, pp. 1714–1733
Nielsen, Peter Bøegh; and Sturgeon, Timothy. 2014. “A Revised List of Business Functions for
Statistical Surveys.” Prepared for: the Eurostat Task Force on subcontracting and Global
Value Chains multi-annual survey, June 11.
Nielsen, Peter Bøegh. 2008. “International sourcing: Moving business functions abroad.”
Statistics Denmark. Web access: www.dst.dk/globalisation and
http://epp.eurostat.ec.europa.eu/statistics_explained/index.php/International_sourcing_sta
tistics
Online_services_191015.pdf
Rauch, James E. (1999). "Networks Versus Markets in International Trade," Journal of
International Economics 48(1) (June 1999): 7-35.

19
Sturgeon, Timothy and Memedovic, Olga. 2010. “Mapping Global Value Chains: Intermediate
Goods Trade and Structural Change in the World Economy”. UNIDO Working Paper
05/2010
Sturgeon, Timothy; Nielsen, Peter Bøegh; Linden, Greg; Gereffi, Gary; and Brown, Clair. 2013.
“Direct Measurement of Global Value Chains: Collecting Product- and Firm-Level
Statistics on Value-added and Business Function Outsourcing and Offshoring.” Chapter
11 in Mattoo, Aaditya; Wang, Zhi and Wei, Shang-Jin (eds): Trade in value-added:
developing new measures of cross-border trade. Washington, DC; The International
Bank for Reconstruction and Development/The World Bank.
Sturgeon, Timothy. 2017. “Measuring the ‘New’ Digital Economy.” Power from Statistics
Outlook Report, European Commission. Web access:
https://powerfromstatistics.eu/outlook-report
Tempest, Rone. 1996. ‘‘Barbie and the World Economy,’’ Los Angeles Times, September 22,
A1 and A12.
Timmer, Marcel; Erumban, Abdul; Los, Bart; Stehrer, Robert; and de Vries, Gaaitzen. 2014.
"Slicing Up Global Value Chains", Journal of Economic Perspectives, 28(2), 99-118.
UK.” Research Policy 44:9, pp. 1714–1733
UN Statistical Commission. 2016. “Manual for the Classification by Broad Economic
Categories, Revision 5” United Nations Department of Economic and Social Affairs,
Statistics Division, Statistical Papers, Series M, No.53, Rev.5. Web access:
http://unstats.un.org/unsd/statcom/47th-session/documents/BG-2016-11-Manual-of-the-
Fifth-Revision-of-the-BEC-E.pdf
UNCTAD. 2014. “Toward a Taxonomy of ICT-related Services,” Prepared for the Expert
Meeting on Assessing the Feasibility of Measuring Trade in ICT Services and ICT-
enabled Services, UNCTAD, Geneva, Switzerland, March 25.
UNCTAD. 2015. “International Trade in ICT Services and ICT-enabled Services; Proposed
Indicators from the Partnership on Measuring ICT for Development.” Division on
Technology and Logistics; Science, Technology and ICT Branch; ICT Analysis Section,
Technical Note No.3, TN/UNCTAD/ICT4D/03, October. Web access:
http://unctad.org/en/PublicationsLibrary/tn_unctad_ict4d03_en.pdf
United Nations Conference on Trade and Development (UNCTAD), 2009. Information Economy
Report 2009: Trends and Outlook in Turbulent Times, United Nations, New York and
Geneva.
United Nations. 2016. “Classification by Broad Economic Categories, Rev.5” Department of
Economic and Social Affairs, Statistics Division. Statistical Papers Series M No.53,
Rev.5 (ST/ESA/STAT/SER.M/53/Rev.5) Web access:
http://unstats.un.org/unsd/statcom/47th-session/documents/BG-2016-11-Manual-of-the-
Fifth-Revision-of-the-BEC-E.pdf
US Bureau of Economic Analysis. 2016. “New BEA Estimates of International Trade in
Digitally Enabled Services.” Economics & Statistics Administration, United States
Department of Commerce, Bureau of Economic Analysis. May 25. Web access:
http://esa.doc.gov/economic-briefings/new-bea-estimates-international-trade-digitally-

20
 enabled-services Data:
http://www.bea.gov/scb/pdf/2016/05%20May/0516ICT_tables.pdf
US Bureau of Economic Analysis. 2016. “New BEA Estimates of International Trade in
Digitally Enabled Services.” Economics & Statistics Administration, United States
Department of Commerce, Bureau of Economic Analysis. May 25. Web access:
http://esa.doc.gov/economic-briefings/new-bea-estimates-international-trade-digitally-
enabled-services Data:
http://www.bea.gov/scb/pdf/2016/05%20May/0516ICT_tables.pdf
van Welsum, Desireé; and Reif, Xavier. 2009. "We Can Work It Out: The Globalization of ICT-
Enabled Services," NBER Working Paper No. 12799, Chapter in: International Trade in
Services and Intangibles in the Era of Globalization National Bureau of Economic
Research, Inc.
Welsum, Desireé van and Reif, Xavier. 2009. "We Can Work It Out: The Globalization of ICT-
Enabled Services." National Bureau of Economic Research, Working paper No. 12799.
http://www.nber.org/papers/w12799
Yeats, Alexander. 2001. “Just how big is global production sharing?” in S. Arndt and H.
Kierzkowski (eds.), Fragmentation: New Production Patterns in the World Economy,
Oxford: Oxford University Press.

21

View publication stats