Application Of Manual Statistics

User Manual:

Open the PDF directly: View PDF .
Page Count: 349

Download
Open PDF In Browser	View PDF

Unedited White Cover Version

��
�
List of acronyms

AMNE

activities of multinational enterprises

BD4

Fourth edition of the OECD Benchmark Definition of Foreign
Direct Investment

BPM6

Sixth edition of the Balance of Payments and International
Investment Position Manual

CAPI

computer assisted personal interviewing

CATI

computer assisted telephone interviewing

CAWI

computer assisted web interviewing

CIF

cost, insurance, freight

CPC

Central Product Classification

EBOPS

Extended Balance of Payments Services Classi fication

Eurostat Statistical Office of the European Union
FATS

foreign affiliates statistics (as used in thi s Manual )

FDI

foreign direct investment

FDIR

Framework for Direct Investment Relationships

FISIM

financial intermediation services indirectly measured

FOB

free on board

FTE

full-time equivalent

GATS

General Agreement on Trade in Services

GATT

General Agreement on Tariffs and Trade

GDP

gross domestic product

Harmonized Commodity Description and Coding Sy stem

ICFA

ISIC, Rev.4 Categories for Foreign Affiliates in services

ICSE

International Classification of Status in Empl oyment
(International Labour Organization)

ICT

information and communication technology

IDC

internet data collection

ILO

International Labour Organization

IMF

International Monetary Fund

IMTS

International Merchandise Trade Statistics

IRTS

International Recommendations for Tourism Sta tistics

ISCO

International Standard Classification of Occu pations
(International Labour Organization)

���
�

ISIC

International Standard Industrial Classificat ion of All
Economic Activities

ITRS

International Transactions Reporting Systems

MFN

most-favoured-nation (treatment)

MNE

multinational enterprise

NPISH

non-profit institution serving households

NPO

non-profit organization

OECD

Organisation for Economic Co-operation and De velopment

RSIM

recommendations on statistics of internationa l migration

SBR

statistical business register

SNA

System of National Accounts

SPEs

special-purpose entities

TFSITS Task Force on Statistics of International T rade in Services
TRIPS

Agreement on Trade-related Aspects of Intell ectual Property
Rights

TSA

tourism satellite account

UCI

ultimate controlling institutional unit (used in respect of
foreign affiliates)

United Nations

UNCTADUnited Nations Conference on Trade and Develo pment

�

W/120

Services Sectoral Classification List (see W orld Trade
Organization document MTN.GNS/W/120)

XBRL

Extensible Business Reporting Language report ing

����
�

TABLE OF CONTENTS
INTRODUCTION....................................................................................................................................... 1
A.

Background ....................................................................................................................... 2

Purposes and scope of the Guide ....................................................................................... 5

Organization of the Guide ................................................................................................. 6

PART I GENERAL FRAMEWORKS ..................................................................................................... 7
Chapter 1

Conceptual frameworks ................................................................................................. 7

Introduction ....................................................................................................................... 7

GATS, modes of supply and the statistical frameworks .................................................. 8

Statistics on resident/non-resident transactions in services ............................................. 10

Foreign Affiliates Statistics ............................................................................................. 13

Additional indicators for analysing the international supply of services ........................ 14

Integrated economic statistics ......................................................................................... 15

Chapter 2

Legal framework ........................................................................................................... 17

Summary of good practices ............................................................................................. 17

The importance of the legal framework .......................................................................... 17

Legal acts regulating institutional arrangements............................................................. 19

Legal acts governing content and availability of data sources ........................................ 20

Legal acts protecting confidentiality ............................................................................... 21

Country experiences ........................................................................................................ 22

Chapter 3

Institutional arrangements ........................................................................................... 24

Summary of good practices ............................................................................................. 24

The purpose and types of institutional arrangements ..................................................... 24

Characteristics of effective institutional arrangements .................................................. 25

Country experiences ........................................................................................................ 27

PART II

DATA COLLECTION ......................................................................................................... 30

���
�
Chapter 4
framework

Introduction and overview of data sources within the modes of services supply
......................................................................................................................................... 30

Data sources and data collection: an overview ............................................................... 30

Sources for modes of supply data ................................................................................... 33

Comparing data sources .................................................................................................. 33�

Chapter 5

Statistical business registers and survey frames ........................................................ 35

Summary of good practices ............................................................................................. 35

Statistical Business Registers .......................................................................................... 36

C.
A satellite register for trade in services and use of business registers to identify
potential mode 4 self-employed service suppliers ........................................................................... 42
D.
Chapter 6

Country experiences ........................................................................................................ 44
Enterprise and Establishment Surveys ....................................................................... 47

Summary of good practices ............................................................................................. 47

General purpose and description of enterprise or establishment surveys ........................ 48

Enterprise and establishment surveys for trade in services statistics purposes...... ...... ...50

Foreign affiliate statistics and the international supply of services ................................. 65

Chapter 7
A.

Surveys of persons and households, and population censuses .................................. 72
Introduction and summary of good practices .................................................................. 72

B.
General Purpose and Description of Population Censuses and Persons and
Households Surveys.…………………………………………………………………… ....... …….74
C.

Population census ............................................................................................................ 75

Household surveys .......................................................................................................... 76

Labour force surveys ....................................................................................................... 84

Border surveys ................................................................................................................ 87

Complementary surveys of persons ................................................................................ 94

Chapter 8
A.

International Transaction Reporting Systems............................................................ 97
Summary of good practices ............................................................................................. 97

��
�
B.

General description and purpose of ITRS ....................................................................... 97

Using the ITRS for collecting data on international trade in services ............................ 98

Country experiences ...................................................................................................... 101

Chapter 9

Administrative Records .............................................................................................. 104

A summary of good practices........................................................................................ 104

General purpose and description of administrative records ..........................................105

Examples of administrative records .............................................................................. 107

Country experiences ..................................................................................................... 113

Chapter 10

Other Data Sources ..................................................................................................... 116

Summary of compilation guidance ............................................................................... 116

Other data soures and Big Data ..................................................................................... 116

External party sources ................................................................................................... 125

Linked microdata .......................................................................................................... 130�

Chapter 11

Comparing Data Sources ............................................................................................ 133

Summary of good practices ........................................................................................... 133

Comparison of data sources for resident/non-resident transactions ............................. 133

Good collection practices and comparison of data sources for FATS ......................... 142

D.
Preferred collection practices and comparison of data sources for mode 2 and 4
numbers of persons ........................................................................................................................ 143
E.

Country experiences ...................................................................................................... 146

PART III DATA COMPILATION ..................................................................................................... 150
Chapter 12 Introduction and overview of data compilation within the modes of services
supply statistical framework .............................................................................................................. 150
A.
Data compilation within the modes of services supply statistical framework: issues
to consider………………………………………………………………………………… ....... ...150
B.

Focusing on an integrated approach .............................................................................. 151

A phased implementation of the recommendations on data compilation...................... 151

���
�
D.

Use of models and estimates ......................................................................................... 152�

Chapter 13

Integration of data from different sources................................................................ 154

Summary of good practices ........................................................................................... 154

Integrated business survey programme: an overview .................................................. 155

Integration, consolidation, and merging of data............................................................ 156

Country experience ....................................................................................................... 158

Chapter 14

Compilation of resident/non-resident trade in services statistics ............................ 160

Summary of good practices ........................................................................................... 160

Compilation of individual service categories ................................................................ 163
B.1.

Goods-related services……………………………………………………… ...... ……163

B.2.

Passenger transport and travel……………………………………………… ...... …...175

B.3.

Construction…………………………………………………………………… ....... …184

B.4.

Insurance, pension and financial services……………………………………....... …..189

B.5.

Services related to intellectual property products…..……………………… ....... ……204

B.6.

Education and Health services…………………..…………………………… ....... ……212

B.7.

Government goods and services n.i.e.…………..……………………………… ....... …212

Allocation of resident/non-resident trade in services to modes of supply ..................... 215

Service transactions between related (affiliated) enterprises ........................................ 223

Chapter 15
services

Compilation of Foreign Affiliates Statistics and the international supply of
....................................................................................................................................... 226

Summary of good practices ........................................................................................... 226

General Description ...................................................................................................... 227

Selecting data sources ................................................................................................... 229

General compilation issues............................................................................................ 234

Compiling FATS Variables ........................................................................................... 238

Attribution of FATS variables....................................................................................... 247

����
�
G.

Regional aggregation .................................................................................................... 253

Chapter 16

Compilation of additional indicators on international supply of services ............ 255

Introduction ................................................................................................................... 255

Summary of good practices ........................................................................................... 256

Mode 2 and mode 4 movements and stays .................................................................... 257

Linking services trade data with the business register ................................................. 267

Other indicators of interest for the global analysis of service industries ...................... 268

Chapter 17

Estimation and modelling of missing data, forecasting or back-casting ................ 272

Summary of good practices ........................................................................................... 272

Imputation for filling data gaps and for data editing purposes ......................................273

Forecasting, back-casting, and revising time series ...................................................... 275

Allocating to individual EBOPS items and trading partners .........................................276

Model-based estimates .................................................................................................. 277

PART IV CROSS-CUTTING TOPICS ............................................................................................. 283
Chapter 18

Metadata ...................................................................................................................... 283

Summary of good practices ........................................................................................... 283

Metadata: basic concepts and definitions, the role of SDMX ....................................... 284

Indicative lists of metadata items .................................................................................. 286

Metadata standards of international and regional organizations ...................................291

Country experiences ...................................................................................................... 295

Chapter 19

Quality management and quality reporting ............................................................. 297

Summary of good practices ........................................................................................... 297

Quality management: an overview of basic concepts .................................................. 298

Focusing on quality assurance ...................................................................................... .302

Quality measurement and reporting .............................................................................. 307

Country experiences ...................................................................................................... 312

�����
�
Chapter 20

Data and Metadata Dissemination ............................................................................ 315

Summary of good practices ........................................................................................... 315

Data and metadata dissemination: an introduction........................................................ 316

C.
Factors to consider in dissemination of data and metadata compiled within the
framework for describing the international supply of services ..................................................... 318
D.

Presentation of statistics by mode of supply ................................................................. 327

E.
Combined presentation of international merchandise and trade in services
statistics.. ....................................................................................................................................... 332

�
�

Chapter 21

Use of Information and Communication Technology............................................. 334

Summary of good practices ........................................................................................... 334

Use of ICT at the data collection stage ......................................................................... 335

Use of ICT at data compilation and dissemination stages............................................. 338 �

Introduction
1.
Rapid technological advances in the past few decades in transport, computing, and
telecommunications, including the development of the internet and electronic commerce, have
allowed enterprises to avail themselves of resources for production at more distant locations than ever
before and have enabled them to import or export services from ever wider markets. This trend
towards globalization, reinforced by liberalization policies and the removal of regulatory obstacles to
economic activities, has fuelled the steady growth of multinational enterprises, international
investment and trade in goods and services. Better communication and transport have also facilitated
the movement of people for the purposes of tourism, migration, employment and trade. The result has
been a growing economic interdependency among countries, be it through international trade, multinational enterprises, global value chains or the outsourcing of business functions. Particularly in the
field of trade in services, which are increasingly becoming the focus of trade liberalization
agreements, market access deregulation, and other policy initiatives, statisticians face a growing
challenge to produce statistics that measure these complex global transactions in a clear, accurate, and
timely manner.
2.
The services sector accounts for the largest share of business activity, employment, and
economic growth in most economies, yet the role of services trade, particularly in designing policies
and negotiating regional agreements, continues to be poorly understood. A major reason for this
disconnect is the absence of abundant, high-quality data on trade in services that is comparable across
economies – making it difficult to measure the impact of services trade on the economy and provide
useful information for negotiators and policy makers incoming up with market access or policy tools
that could facilitate trade in services.
3.
In this context the United Nations Statistical Commission adopted1 in 2010 at its 41st session
the Manual on Statistics of International Trade in Services 2010 (MSITS 2010) and endorsed the
implementation plan, while urging the Task Force on Statistics of International Trade in Services (TFSITS) to give high importance to the development of corresponding compilation guidance. To ensure
close coordination with the work of the International Monetary Fund (IMF) on the compilation
guidance for BPM6, the actual work on this Compilers Guide for MSITS 2010 started late in 2011
with the support of a UN Expert Group established for this purpose. The MSITS 2010 Compilers
Guide aims to help the compilers while staying consistent with the recommendations in MSITS 2010,
related statistical recommendations and the compilation guidance, which has been developed for the
Balance of Payment and International Investment Position Manual (BPM6).2
4.
MSITS2010 Compilers Guide serves the purpose of harmonizing and improving the ways in
which statisticians at the national level collect, compile and disseminate statistics of international
trade in services. The aim is to increase the availability and quality of such statistics in order to fulfill
the urgent needs and demands for statistics reflecting modalities through which services may be
supplied by policy makers, researchers, market analysts and the public in general. While the
international standard in economic statistics of the 2008 System of National Accounts (SNA 2008)3
and the BPM6 are in the process of being implemented, this Compilers Guide is meant to provide the
statistical community with guidelines, best practices, case studies, and practical advice on the
compilation of data in compliance with the MSITS 2010 framework for compilation of statistics on
the international supply of services.

See Decision 41/104 of the 2010 Commission Report, E/CN.3/2010/34.
Balance of Payments and International Investment Position Compilation Guide. Available at:
http://www.imf.org/external/pubs/ft/bop/2007/bop6comp.htm.
3
System of National Accounts 2008, United Nations et al., 2009.
2

Box 1. Uses of terms statistics of international trade in services and statistics on the international
supply of services
The information needs described in the Manual on Statistics of International Trade in Services 2010 touch upon
a multi-purpose field of statistical work. As desc ribed in the box I.2 of the Manual, the conceptual framework
introduced in it covers a broad range of statistica l domains including (1) statistics on services tran sactions
between residents and non-residents (or resident/no n-resident trade in services statistics) broken down by
detailed service categories, partner countries and modes of supply, (2) foreign affiliates statistics (FATS)
covering supply of services through commercial presence broken down by type of services and partner countries
and related data on foreign affiliates (e.g., emplo yment, value added etc.) as well as (3) additional monetary and
non-monetary indicators for analysing the international supply of services. Therefore, the information needs
described in the MSITS2010 go beyond the concept of international services transactions or trade in se rvices as
covered in the balance of payments or in the rest of the world account of the SNA.
However, the exact terminology to refer to this complete set of statistical domains differs across var ious
statistical fora and users. In particular, compiler s should note that the term statistics of international trade in
services is an umbrella term which should be used in the br oad sense as outlined above. Indeed, these combined
components of the MSITS2010 are recognized by the UN Statistical Commission as a separate field of sta tistical
activity under this title. The broad use of the term initially referred to the fact that in the mid-90 s trade
agreements were increasingly covering services as w ell as goods. The most well-known and wide-reaching
agreement that triggered information needs as described in the MSITS2010 is the General Agreement on T rade
in Services (GATS).
The term statistics of international trade in services is often interpreted by users to refer only to statistics on
services transactions between residents and non-residents (i.e. as covered in the balance of payments goods and
services account). Therefore, the statistical frame work developed in MSITS2010 in order to compile
information on the modes of services supply uses the term statistics on the international supply of services to
make it more explicit that also components covered by the extended scope of international trade in services are
covered. For the purposes of this Guide, and in line with the MSITS2010 (in particular with the more
substantive chapters 3, 4 and 5), the term statistics of international trade in services (or trade in services
statistics) is used in more general contexts to describe the wide statistical domain, and the term statistics on the
international supply of services is used when explicit emphasis is placed on the in clusion of FATS and
additional monetary and non-monetary indicators for measuring the international supply of services.

Background

A.1.

The GATS: an overview

5.
In 1994 the Uruguay round of trade negotiations was completed by signing the Agreement
establishing the World Trade Organization (WTO). Annex 1B to the Agreement contains General
Agreement on Trade in Services known as the GATS4. The GATS established a set of rules and
disciplines governing the use by the WTO members of measures affecting trade in services. The trade
in services was defined in Article I of the GATS as “the supply of a service”. The same article
defined also four different ways in which a service can be supplied by a natural and juridical person
(service supplier) from the territory one of the WTO members to a service consumer in the territory of
another WTO member. Article XXVIII provided the elaboration of the meaning of the main of GATS
concepts for use in international negotiations5 on liberalizing trade in services and related analytical
work.
6.
The term “service” is not defined in the Agreement. However, the scope of services with
which the GATS is concerned was clarified by the GATT Secretariat in the document

Available at http://www.wto.org/english/docs_e/legal_e/26-gats_01_e.htm.
The regional economic integration agreements take their inspiration from the GATS definitions. However,
some of these agreements may define the modes in a somewhat different way. The compilers should be awa re of
this while interpreting data on value of the services supplied under different modes in a regional con text.
5

MTN.GNS/W/120 entitled “Services Sectoral Classification List” (W/120). The list was based on the
consultations with the WTO Members and issued in 1991. W/120 identifies relevant sectors and
subsectors and enabled members to undertake specific commitments. It should be noted that the WTO
Members have tended to avoid any major changes in the list to ensure the stability and comparability
of commitments over time, even though related international statistical classifications have been
revised (see Chapter 1, Section B for details).
7.
The GATS created a need in specific data leading eventually to establishing of a special
statistical domain – statistics of international trade in services. In view of a fundamental importance
of the GATS conceptual framework for these statistics its key elements are described in Chapter 1
(Section B) of this Guide.
A.2.

GATS related data needs

8.
The GATS negotiators and trade in services policy makers as well as business community,
research institutions and public at large need detailed and internationally comparable statistical
information on the supply of services by mode, type of service and trading partner. The policy makers
require such data to ensure an informed decision making process leading to specific commitments,
comparison of national commitments and a conduct of efficient negotiations as well as to assess the
extent of liberalisation reached in specific sectors/markets and to provide statistical background for
the settlement of disputes.
9.
The availability of timely and comparable data reflecting the international supply of services
would greatly benefit business community as such data would significantly facilitate the evaluation of
the importance of each type of internationally traded service and understanding of how those services
are provided in the respective economies thus helping to realize the competitive advantages in a
globalized world. The research institutions and public at large are also interested in obtaining such
data to assess the role of services in economic and social developments both in their respective
countries and globally and to participate more effectively in the formulation of their countries’ trade
policies.
10.
The conceptual and practical issues related to the compilation of data on value of
internationally supplied services occupy the most of the Guide. However, it is necessary to stress that
statistical needs related to the GATS go beyond the value of services as various non-monetary
indicators of modes of supply are critically important for assessing their economic and social impact.
Chapter 16 of the Guide is specifically focused on the conceptual and data compilation issues related
to the compilation of such indicators.
A.3.

MSITS as a statistical response to the GATS related data needs

11.
The GATS has highlighted that the scope of supply of services which is of a primary interest
to trade negotiators and trade policy makers is far broader than what statistics conventionally measure
(e.g., includes supply of services by commercial presence) thus making the statistical community
aware of an informational gap which had to be filled. Dealing with this issue was a serious challenge
as the GATS conceptual framework differs from statistical frameworks adopted internationally for
economic statistics. To meet this challenge successfully it was necessary to conceptualize the supply
of services by modes in a statistical context.
12.
In response to this challenge the UN Interagency Task Force on Statistics of International
Trade in Services drafted Manual on Statistics of International Trade in Services (MSITS) and the
United Nations Statistical Commission (UNSC) adopted it at its thirty-second session in March 2001.
The Manual established the statistical framework based on which measurement of supply of services
was made possible. The Manual was revised by the Task Force less than a decade after its adoption to
take into account publication of the revised international statistical standards including BPM6, SNA
3

2008, CPC version 2, ISIC Rev.4, IMTS 2010, IRTS 2008 and others as well as the necessity of
further elaboration of the modes of supply measurement. The revised version of the Manual was
adopted by the UNSC in 2010 and provides a clearer, more detailed and more comprehensive
statistical framework, which once implemented, will result in data crucial for statistical and economic
analysis of international supply of services and evidence-based decision making.
13.
The MSITS 2010 statistical framework was developed utilizing two main building blocks: (i)
BPM6 concepts and definitions describing transactions between residents and non-residents of
different economies and (ii) concepts and definitions developed in Foreign Affiliates Statistics
(FATS) on the basis of BPM6 and the OECD 4th edition of the Benchmark Definition of Foreign
Direct Investment. The use of these two building blocks was necessitated by the following
consideration. Interpreting supply of services between natural and juridical persons located in the
territories of different WTO members in terms of the services transactions between residents and nonresidents of different economies made possible not only estimation of value of supply of services by
all modes except for commercial presence, but also to provide details of such supply by service
categories and trading partners. However, since the BPM6 conceptual framework cannot cover supply
of services through commercial presence of service suppliers in the economy of service consumers a
new statistical framework for measuring the supply of services through that mode had to be
established. Such a framework was developed as part of foreign affiliates statistics. It should be
underlined that that framework, once implemented, will not only generate information needed for the
GATS related purposes, but also will result in statistics indispensable for a better understanding the
overall dynamics of the global economy and for assessing of the impact of globalization on
individual countries. Chapter 1 contains an overview of the main elements of these two components
of MSITS 2010 statistical framework as well as the main concepts underlying the compilation of
additional monetary and non-monetary indicators relevant for analyzing the international supply of
services. The rest of the Guide deals with various data collection, data compilation and data
dissemination issues.
14.
While stressing the importance of the MSITS 2010 conceptual framework it should not be
overlooked that MSITS recommendations and guidelines on the compilation of statistics by modes of
supply are laid out only for statistical purposes. They do not imply any attempt to interpret the
6
GATS. In this connection MSITS 2010 recognizes that a comprehensive statistical treatment of
modes of supply that would fully mirror the GATS legal definition and other GATS articles is beyond
its scope7. The same limitation applies to the present Guide.
A.4.

MSITS 2010 and the Compilers Guide for MSITS 2010

15.
Although the original version of the Manual was released in 2002, the international statistical
community did not produce compilation guidance to accompany its recommendations. However, with
the adoption of the 2010 edition of the Manual the UNSC specifically requested that the Task Force
develop appropriate compilation guidance. This is the reason why UNSD, with the assistance of the
Task Force, established in December 2011 an Expert Group on compilation of trade in services
statistics (EG-CSITS) to assist in the preparation of the Compiler's Guide for MSITS 2010. The expert
group included all members of the interagency Task Force, as well as compilers from developing and
developed economies, and was convened by UNSD. The guide represents the result of the work of
both Task Force and the Expert Group.
16.
EG-CSITS was established to ensure active country involvement in the development of the
Compilers Guide. In March 2012, the first meeting of EG-CSITS was held through virtual discussion.
The main objective of the meeting was to review the annotated outlines of all chapters of the

6
7

MSITS 2010, paragraph 5.32.
Ibid., paragraph 5.27.

compilers guide and to define the scope of the Guide and the content of each of the chapters. A
revised annotated outline was thereafter circulated for worldwide consultation in July 2012 and first
draft chapters were prepared as well. The outcome of the worldwide consultation and the draft
chapters were being discussed in a virtual meeting in October 2012. Further draft chapters were
discussed at a third virtual meeting in March 2013. In all three meetings about 65 representatives of
countries and international organizations actively participated. At the end of June 2013, the expert
group came together for a face to face meeting in Geneva to discuss the first full draft of the compilers
guide. A fully edited version of the draft Guide was subsequently prepared and presented to the Task
Force at its meeting in October 2013 and was discussed electronically by the Expert Group during
November 2013. The unedited draft version of this Guide was submitted to, and adopted by, the
Statistical Commission at its forty-fifth session, held in March 2014. The Task Force finalized the
draft guide over the course of 2014.
B.

Purposes and scope of the Guide

17.
The main purpose of the Guide is to assist countries in the production of high quality official
statistics on international trade in services in compliance with MSITS 2010. The Guide strives to
achieve this by (i) clarification and elaboration of a number of more difficult conceptual issues and
8
(ii) identification of good practices in the implementation of MSITS 2010 .Further, the Guide is
intended to better integrate the compilation of statistics of international trade in services in the context
of a global statistical work and, therefore, recognizes importance of not only such pillars as BPM6
and SNA2008, but also of the UNSC recommendations for other related statistical domains as well as
its guidance contained in Guidelines on Integrated Economic Statistics 9 and National quality
assurance framework.10
18.
In particular, the Guide aims to complement the BPM6 compilation guidance on balance of
payments statistics, focusing on the compilation of data by EBOPS 2010 categories, trading partner
and mode of supply. Also, it includes more detailed compilation guidelines for FATS as far as it
relates to the international supply of services, and provides guidance on how to compile statistics on
the modes of supply on the basis of the framework laid down in Chapter V of MSITS 2010. It is
expected that an increasing number of countries will begin compiling services data by modes of
supply following the good practices provided in the Guide. Further, the Guide recognizes the
importance of data quality, metadata and the use of IT tools in the statistical production process and
contains guidance and advice relevant to those areas of statistical work. In addition, the Guide is
intended to promote trade in services statistics by making available to users a comprehensive source
of information on its methodological foundations and on data compilation practices. This will enable
users to understand better the nature of these statistics and, consequently, to interpret and use them
correctly and more effectively.
19.
As it is not possible to present a single best practice applicable in all cases the Guide outlines
various options that may be suitable depending on a country circumstances and resources. For
countries that have well-developed compilation systems, the Guide contains criteria against which
these compilation systems may be compared and evaluated. For countries that may need to improve
parts of their compilation systems, the Guide presents examples of good practices and related country
experiences. For countries that do not have well-developed systems, the Guide contains advice on

It should be noted that clarification of the conceptual issues implies (i) explanation of the terms u sed in the
definitions of particular concepts and (ii) operationalization of the definitions by relating them to the statistical
procedures which might be used to obtain anticipated data. The term “good practice” is to be understoo d as a set
of activities contributing to the implementation of recommendations contained in MSITS 2010 and resulting in
production and dissemination of high quality SITS.
9
Available at: http://unstats.un.org/unsd/nationalaccount/ies/GuidelinesIES.pdf.
10
Available at: http://unstats.un.org/unsd/dnss/QualityNQAF/nqaf.as px.

good practices and other guidance that can be used as a starting point to develop appropriate data
collection and data compilation procedures. It should be noted that the identification of good practices
described in the Guide benefited from the input of national compilers which have gone through the
process of implementing the recommendations contained in MSITS 2010. The Guide focuses on the
main features of the relevant international recommendations and good practices, while some more
technical and country specific details are available on the dedicated UNSD website.
20.
Scope of the Guide was determined by the tasks which the trade in services compiler
typically has to deal with. These tasks range from the operationalisation of the conceptual framework
and defining the set of variables to be compiled to the setting in place effective institutional
arrangements, identification of appropriate data sources, data collection and data compilation
procedures, and to the organization of efficient quality assurance and data dissemination programmes.
The Guide elaborates challenges and good practices in the above areas of work, provides numerous
country experiences and provides guidance on how those practices can be applied under different
country circumstances.
C.

Organization of the Guide

21.
The organization of the Guide largely follows the logic of the statistical process beginning
with an overview of general frameworks underlying compilation of trade in services statistics
followed by the description of data collection and data compilation issues, good practices, data and
metadata dissemination and concluding with the elaboration of several key cross-cutting topics.
22.
Part I of the Guide provides an overview of general frameworks, namely conceptual
frameworks (Chapter 1), legal frameworks (Chapter 2) and institutional arrangements (Chapter 3),
which underpin statistics of international trade in services.
23.
Part II focuses on data collection and starts off with an introduction and overview of data
sources within the modes of supply framework (Chapter 4), which is followed by elaboration of
registers and survey frames (Chapter 5), enterprise and establishment surveys (Chapter 6), surveys of
persons and households (Chapter 7), international transaction reporting system (Chapter 8),
administrative records (Chapter 9) and other data sources (Chapter 10). Part II is concluded by
comparing data sources (Chapter 11).
24.
Part III elaborates various issues of data compilation. It begins with an introduction and
overview of data compilation within the modes of supply framework (Chapter 12) and elaboration of
challenges and good practices in the integration of data from different sources (Chapter 13). Specific
issues and good practices relevant to the compilation of particular data sets are elaborated in the
subsequent chapters as follows: resident/non-resident trade in services statistics (Chapter 14),
compilation of FATS and the international supply of services (Chapter 15), compilation of additional
indicators on the international supply of services (Chapter 16). Guidance on estimation and modeling
of missing data, forecasting or back-casting is contained in the concluding Chapter 17.
25.
Part IV focuses on cross-cutting topics and consists of four chapters which provide guidance
on metadata (Chapter 18), quality management (Chapter 19), data and metadata dissemination
(Chapter 20), and on use of information and communication technology (Chapter 21).
26.

The Guide has a number of Annexes which will be provided in the final version.

Part I General Frameworks
The Manual on Statistics of International Trade in Services 201011 provides an overarching statistical
framework to compile data required for measurement of the international supply of services and
elaborates the relationships between various components of that framework. This Compilers Guide
accompanies this reference publication. Part I contains an overview of the conceptual frameworks
(Chapter 1), the legal framework (Chapter 2) and the institutional arrangements (Chapter 3). Part II
and Part III focus on data sources/collection and data compilation. Part IV deals with cross-cutting
issues.

Chapter 1

Conceptual frameworks

1.1.
Scope. This chapter introduces the conceptual frameworks identified in MSITS 2010
(section A); describes the General Agreement on Trade in Services (GATS), modes of supply and
the statistical frameworks (section B); and then provides an overview of concepts relating to
statistics on resident/non-resident transactions in services (section C), foreign affiliates statistics
(section D) and additional indicators for analysing the international supply of services (section E).
The chapter concludes with a description of the guidelines on integrated economic statistics
(section F).
A.

Introduction

1.2.
The General Agreement on Trade in Services (GATS) has been a driving force in the
drafting of MSITS, both for the first version of the Manual at the end of the 90s and its current
2010 edition. The GATS establishes a set of rules and disciplines that govern the use of measures
affecting trade in services by the World Trade Organization (WTO) members. Under the GATS,
WTO members are committed to entering into successive rounds of trade liberalizing negotiations,
which are conducted by type of service and by mode of supply. Economic Integration Agreements
at the regional or bilateral level follow the same approach. Consequently, availability of statistics
on the international supply of services detailed by the services type, mode of supply and by partner
country is highly important for trade in services policy making and related analysis.
1.3.
Chapter II of MSITS 2010 focuses on conceptual framework for the development of
statistics of international trade in services and explains how existing statistical systems and
classifications can be used to progressively build these statistics to provide the necessary
information for measuring the international supply of services. It is consistent with concepts and
definitions included in the 2008 System of National Accounts (2008 SNA) and the sixth edition of
the Balance of Payments and International Investment Position Manual (BPM6). The compilation
guidance provided for these frameworks should be used as a starting point and this Guide should
be read as an extension of that guidance intended to cover specific compilation needs in the context
of MSITS2010 implementation.
1.4.
As the relevant conceptual frameworks are outlined in MSITS2010 and are detailed on
other international manuals, this chapter only briefly introduces the basic elements of these
frameworks – those that are essential in the context of compiling trade in services statistics, and
makes reference to relevant material in the manuals. It clarifies the information needs, in particular
those of the GATS, and links this with the compilation of statistics on:

MSITS 2010 is available at:
https://unstats.un.org/unsd/tradeserv/TFSITS/msits2010/docs/MSITS%202010%20M86%20(E)%20web.pdf .

Services transactions between residents and non-residents, broken down by services type
(according to Extended Balance of Payments Services Classification 2010 (EBOPS 2010)),
partner, by modes of supply (mainly 1, 2 and 4) and by relation between the parties (trade
between affiliated firms and between unaffiliated firms);

ii.

Foreign Affiliates Statistics (FATS), focusing on the variables and breakdowns of interest
in the context of the international supply of services while being consistent with the
frameworks described in the Organization for Economic Co-operation and Development
(OECD)'s Benchmark Definition of Foreign Direct Investment 4th edition and OECD's
Handbook on Economic Globalisation Indicators; and

iii.

Additional monetary and non-monetary (quantitative) indicators relevant for the
assessment of the importance various modes of supply (in particular, modes 2 and 4) in an
economic context.

GATS, modes of supply and the statistical frameworks

1.5.
The MSITS2010 describes the structure and guiding principles of GATS and its scope. For
making commitments in GATS negotiations, WTO members often make reference to the
MTN.GNS/W/120 entitled "Services Sectoral Classification List" (hereinafter referred to W/120).
This list should be viewed as an optional classification system of services sectors for negotiating
12
purposes, rather than a statistical classification like EBOPS 2010 which is used to compile and
13
publish resident/non-resident trade in services statistics.
1.6.
Following the GATS, the international supply of services can take place through four
different modes of supply which depend on the territorial presence of the supplier and the
consumer at the time of the transaction. The four modes can be described as follows:

i. Mode 1 (cross-border supply): the supply of services from the territory of one Member
into the territory of any other Member;

ii. Mode 2 (consumption abroad): the supply of services in the territory of one Member to
the service consumer of any other Member;

iii. Mode 3 (commercial presence): the supply of services by a service supplier of one
Member, through commercial presence, in the territory of any other Member; and

iv. Mode 4 (presence of natural persons): the supply of services by a service supplier of one
Member, through the presence of natural persons of a Member in the territory of any other
Member.14 Mode 4 is further described in Box 1.1.
1.7.
Existing statistical frameworks can be used as a basis for the production of data to satisfy
the GATS information needs regarding modes of services supply. For example, the value of the
international supply of services of a country can be established once statistics on the output of

MSITS 2010 Annex 1.
https://unstats.un.org/unsd/tradeserv/TFSITS/msits2010/docs/MSITS%202010%20M86%20(E)%20web.pdf .
13
A correspondence table has been developed between W/120 and EBOPS 2010 (See also MSITS2010
paragraphs 2.11-2.23).
14
A natural person of a Member is defined as a natio nal of that Member or to a non-national who has the right
to permanent residence in that Member. See article XXVIII of the GATS for further information.

foreign controlled affiliates have been developed for the services sector to accompany the BoP
services statistics. The combination of both datasets would cover all four modes, provided
compilers ensure that appropriate breakdowns are delivered (i.e., service type, mode of supply,
partner country and exchanges between related and unrelated trade). As MSITS 2010 explains, to
produce a complete picture of the international supply of services, additional data, including both
monetary and non-monetary indicators, are still necessary.
Box 1.1: Clarifying mode 4 from a statistical perspective
The MSITS 2010 chapter V introduces mode 4 in a sta tistical context, based on the description of GATS article
1 and the Annex on the Movement of Natural Persons under the Agreement. Mode 4 can, in general, be
described as covering foreign natural persons enter ing the host economy to:
(i) fulfil directly service contracts (contractu al service suppliers): this covers self-employed persons
(independent professionals) or employees of a forei gn service supplier;
(ii) work in a foreign affiliate that delivers s ervices (intra-corporate transfer or directly recru ited by the
affiliate); or
(iii) negotiate a service contract (services sell ers), negotiate the constitution/acquisition of an establishment
supplying services (persons responsible for setting up commercial presence), market a service etc.
The purpose of stay under mode 4 is to provide services under a service contract. Therefore, persons c rossing
the borders to access the employment market and per sons present for the production of goods or for the
provision of services supplied under governmental authority are excluded. Also, permanent migration is
excluded from the mode 4 movements (the GATS does n ot apply to measures affecting residence, citizensh ip
or employment on a permanent basis).
Persons covered by mode 4 are, therefore, not limi ted to foreign persons directly involved in the rendering of
services (as indicated in (i) above), but include a lso those persons whose presence abroad is instrume ntal to the
provision of a service (items (ii) and (iii)). Ther efore, data to be compiled in connection with mode 4 of supply
of services cover:
a.
The value of services supplied dire ctly by persons moving under the mode 4 commitments as
indicated under (i) above– that is the value which, in general, would be measured as trade in services in the
balance of payments. The exception would be for services supplied by self-employed persons staying more
than a year (still in the context of a service cont ract). However, normally such persons will represen t a small
proportion of the mode 4 population. In addition, although in principle the residence of the persons would
change, it may be difficult in practice for compile rs to identify (or even consider) that the residence has
changed if the length of stay is just for a few mon ths more (e.g. 2-3 months) and the value of supplied services
would still be recorded under services in the balance of payments. Compiling the mode 4 value of services is
not relevant for persons entering the host economy to perform activities described in points (ii) and (iii) above.
In fact, work of persons covered by point (ii) is c onsidered instrumental to the output of foreign affiliates
which supply services through commercial presence (mode 3). Persons covered by point (iii) are not
producing or delivering a service to consumers at the time of their presence in clients' country. They are rather
engaged in negotiations for an eventual future supply of services.
b.
Non-monetary (quantitative) data o n the number of natural persons crossing borders to supply
services (or the number of trips of such persons) f or the all three categories described above (see ch apter 16 of
this Guide).
See MSITS 2010 Chapter 5 for more information.

1.8.
For the compilation of data on the value of the international supply of services, MSITS
2010 develops a phased approach which includes (a) the compilation of resident/non-resident trade
in services statistics in the context of the Balance of Payments (suggesting a breakdown of
transactions according to EBOPS2010, by partner country, by mode of supply (mainly referring to
modes 1, 2 and 4) and by the relationship between the trading parties) and (b) the development of
foreign affiliates statistics (FATS), exclusively referring to mode 3. FATS should include data on
output (or sales/turnover if output is not possible) broken down by service type (or service activity
9

if type is not possible), by partner country (i.e., country of operation of affiliate for outward FATS,
or of the ultimate controlling unit (UCI) for inward FATS), and identifying to whom output is sold,
with a particular interest in the output sold within the country of establishment of the controlled
affiliate.
1.9.
MSITS 2010 also recommends the compilation of other indicators of interest to users (both
monetary and non-monetary), including other FATS variables (employment, value added, number
of enterprises etc.); non-monetary quantitative indicators for mode 4 (i.e. number individuals
crossing borders and temporarily abroad to provide services in the context of services contracts)
and mode 2 (individuals going abroad to consume services); and indicators such as foreign direct
investment (FDI), services sectoral indicators or services trade by enterprise characteristics.
1.10. For services negotiations, especially within the GATS framework, users need
comprehensive information on the international rendering of services and on their mode of supply,
as well as on the operators taking part in these operations and their main features. This need has
led to the development of the statistics on the international supply of services which goes beyond
the primary objective of the Balance of Payments. Proper legal and institutional arrangements are
extremely important for the development of statistics needed for analysis of the international
supply of services. In most cases, the compilation of such statistics is a cooperative effort of
several agencies such as the National Statistical Office, the Central Bank, the Ministry of Economy
and the national agency in charge of trade negotiations. This calls for a better coordination of data
collection and data compilation between different institutions (see chapters 2 and 3 for details).
1.11. In order to maximally exploit the potential of combining and comparing statistics on the
international supply of services within and between countries, compilers are encouraged to use
internationally-accepted concepts and methods and classification systems when developing their
data collection and compilation systems. In the case of trade in services, these concepts include the
MSITS2010 and related international statistical systems and frameworks, such as BPM6,
SNA2008 and BD4. For modes of supply, compilers should align with the framework developed in
chapter V of MSITS 2010. In terms of classifications for BOP services transactions, FATS, and
mode 4 (and mode 2) quantitative indicators, compilers should, to the extent possible, use the
EBOPS2010 classification for service categories, the ISIC Rev.4 for activities and the partner
country data classifications used in the Balance of Payments Data Structure Definition, which was
prepared in the context of the Statistical Data and Metadata Exchange (SDMX) initiative
sponsored by seven international organizations – Bank for International Settlements (BIS), the
European Central Bank (ECB), Eurostat, the International Monetary Fund (IMF), OECD, UN and
World Bank).15
1.12. If compilers deviate from these concepts, for example in order to reflect specificities of
their economies or to take into account certain statistical frameworks adopted in their countries,
they are encouraged to develop classification systems that are compatible with those listed above
and construct the appropriate conversion tables to international systems.
C.

Statistics on resident/non-resident transactions in services

1.13. Concepts recommended for use in statistics on services transactions between residents and
non-residents are based on the MSITS2010 chapter III, which in turn utilizes the BPM6 chapter
10. 16 Statistics compiled following these concepts and definitions reflect the value of services

In the Area Dimension (CL_Area) of the Balance of P ayments DSD. individual country coding follows the
ISO 3166 standard. The latest version of the BOP DS D is available on the SDMX website
(http://sdmx.org/?page_id=1747).
16
BPM6, Chapter 10.

mainly supplied via modes 1, 2 and 4. The collection of such data is, in general, the responsibility
of a country's balance of payments compilation agency. In many countries this agency will be the
Central Bank, but in others this responsibility is given to the agency in charge of compiling
economic statistics (usually the national statistical office (NSO)). Alternatively, some countries
have opted for a division of tasks (e.g., with the statistical office collecting trade in services data
and the central bank compiling and disseminating them). The collection and compilation of
resident/non-resident services transactions are further discussed in chapters 5, 6, 7, 8, 9, 10, 11, 13
and 14 of this Guide.
1.14.

For producing these statistics, compilers should:
i.
Follow the general BPM6 principles regarding institutional units, residence, centre
of predominant economic interest, economic transactions, valuation, market prices, accrual
17
accounting, gross recording, etc.), and, of course, the definition of services;
ii.
Breakdown of services transactions by EBOPS 2010 (on a step-by-step basis,
according to the needs of the compiling economy); compilers should be aware that EBOPS
2010 is completely consistent with the BPM6 classification of services but provides a
more detailed breakdown, and also suggests a number of complementary groupings for
compilation given the interest in particular sectors;
iii.
Breakdown of value of services by trading partner (again, to be implemented
gradually and in line with the needs of the compiling country), determined according to the
economy of residence of the respective trading partner; 18
iv.

Identify or estimate the modes of supply related to the services transactions; and

Specify the relationship between the parties involved in the services transactions.

1.15. When compiling statistics on the international supply of services as specified above, the
following considerations should be taken into account. The prices at which exchanges between
affiliated firms are valued may not represent market prices. However, the concept of market prices
is particularly important given that increased globalization is paired with (and often driven by)
enterprises headquartered in one country that establish affiliates in other countries to produce and
distribute goods and services -- see Boxes 1.2 and 1.3. International trade in services between such
parents and their foreign affiliates has rapidly increased. Therefore, within the framework of trade
in services statistics, identification of trading partners in transactions between parents and foreign
affiliates has an important analytical value.

17
18

MSITS 2010, paragraph 3.41.
See above footnote on partner country classification.

Box 1.2 Transfer and market prices
In some cases, the values against which transaction s between affiliated enterprises are priced
internally may not represent market prices. In general, when there is an international transaction
between two affiliated firms, it is expected that the value of the transaction to the exporting affiliate
will be equal to the value of the transaction for the importing affiliate and hence they will cancel out,
leaving the multinational enterprise (MNE)’s overall profits unchanged, no matter what price it
chooses to use to value the transaction. However, i n a world where there are taxes on international
transactions and where the rates of business income taxation differ across countries, the multinational
enterprise will have definite financial incentives to choose strategically the “transfer price” to
minimize the amount of tax paid to both jurisdictions. Transfer prices are the prices at which an
enterprise transfer physical goods and intangible property or provide services to associated
enterprises. Because transfer pricing might result in transactions between affiliated enterprises being
under or over invoiced compared to unrelated party transaction, adjustment should be made when
these exchange values do not represent market price s.
The OECD Transfer Pricing Guidelines state that, fo r income taxation and customs valuation
purposes, enterprises follow the arm’s length standard, i.e., set the transfer price equal to the pric e that
two unrelated parties would negotiate when trading the same or substantially similar products under
the same or substantially similar circumstances. Th e OECD guidelines propose five methods to adjust
transfer prices by arm’s length standard, which are based on the comparability of the transactions.
Transactions are considered comparable when their “ economically relevant characteristics” are the
same, or if they differ, the differences have no ma terial impact on the results. In practice internal and
external transactions are unlikely to be exactly comparable, therefore the OECD Guidelines
recommend that material differences are identified, quantified and adjusted for in determining the
arm’s length transfer price. Moreover, since transfer pricing is not an exact science, the Guidelines
recommend that transfer prices be set inside a rang e of acceptable arm’s length prices, called the
arm’s length range.
See the OECD Transfer Pricing Guidelines at http://www.oecd.org/ctp/transfer-pricing/transferpricing-guidelines.htm for more information.

Box 1.3 Trade in value added
With the increase of international fragmentation of production, intra- and inter industry trade,
creating so-called Global Value Chains (GVCs), has increased dramatically. Conventional measures
of international trade do not always reflect the flows of goods and services within these chains, and
are unable to answer policy questions related to the economic (and employment) impact of their
country’s embeddedness in GVCs. In response, OECD a nd WTO have developed an analytical data
set on trade in value added (TiVA), in order to provide insight into the value that is added by each
country in the production of goods and services that are consumed worldwide.
The estimates are based on a global input output table constructed from official national Supply and
Use tables (or input-output tables) and bilateral trade statistics (goods and services) made consisten t
with national accounts data. Since TiVA combines national statistics to develop a global analytical
toolkit requires high quality input data from national compilers, including information on bilateral
trade in services (by EBOPS category) and on the ch aracteristics of the firms involved in services
trade (notably their industry classification). The first aspect requires balance of payments compilers
to develop trade in services statistics by partner, the second aspect involves the development of data
sets by national compilers that allow for the analysis of services trade companies by economic
characteristics (known as services trade by enterpr ise characteristics (STEC)). These statistics are
developed through linking trade and business registers which allows identifying which firms are
responsible for trade in services.

1.16. The identification of trading partners is especially important in the case of service items,
such as manufacturing services on physical inputs owned by others, research and development,
computer services, audio-visual services, charges for the use of intellectual property n.i.e. and
professional and management consulting services, which are provided within global production
and marketing networks. It is recognized, however, that compiling these statistics by trading
partner is resource-intensive and difficult, owing to issues related to disclosure and incompleteness
of information (the issues related to the identification of trading partner are further discussed in the
context of data collection Part II and data compilation Part III).
1.17. A second consideration relates to the bundling of services with other services or with
goods (such as in the tourism sector). Some arrangers will bundle various types of services and
goods and the final client will only make one payment for a package (which will also cover the
19
service fee for arranging these products). In principle the payments for related services (or
20
goods) should be unbundled, but if this unbundling is not possible, it may be a strong indication
that the bundling has resulted in the creation of a new product that is supplied to client (i.e. the
client is buying a different product). In this case, an appropriate classification has to be found for
such a product.
1.18. A final consideration is the strong link between intellectual property products and services
trade. Services related to intellectual property rights, for example, creative services such as
computer or audio-visual services, take an ever increasing importance in today's business world;
therefore, information on these services becomes progressively more important for economic
analysis. Chapter 14 addresses the relevant EBOPS 2010 services components in one group,
including services related to intellectual property products and other business/personal services
(covering charges for the use of intellectual property n.i.e., telecommunication, computer, and
information services, other business services, and personal, cultural, and recreational services).
MSITS 2010 also suggests some relevant complementary groupings within EBOPS2010:
Computer software transactions, audio-visual transactions and cultural transactions.
D.

Foreign Affiliates Statistics

1.19. MSITS 2010 focuses on the foreign controlled subset of affiliates. The key concepts of
FATS (as described in MSITS 2010 chapter IV) are based on foreign direct investment
relationships and control, which are consistent with those specified in the BPM6, the OECD
Benchmark Definition of Foreign Direct Investment 4th edition 21 and the OECD Handbook on
Economic Globalisation Indicators.22 Details of FATS collection and compilation are discussed in
Chapters 5, 6, 10, 11, 13 and 15 of this Compiler's Guide.

BPM6 (paragraph 3.17) does not refer specifically t o services transactions, but it recommends to unbundle
two or more different transactions that appear as a single transaction from the perspective of the par ties
involved.
20
Travel agencies and tour operators function in that way. Besides the treatment of transactions of trav el agents
which is described, BPM6 and MSITS 2010 do not prov ide further information on unbundling. However both
manuals propose an alternative presentation of travel by type of products consumed which would encoura ge the
unbundling of transactions. This is actually supported by tourism statisticians in the context of the Tourism
Satellite Accounts, and more generally to better li nk the information with the needs for establishing supply/use
and input-output tables.
21
MSITS 2010 and this Guide adopt the definition of foreign direct investment as provided in the 4 th edition of
the Benchmark Definition of Foreign Direct Investment (BDM4). Available at:
http://www.oecd.org/fr/daf/inv/statistiquesetanalysesdelinvestissement/fdibenchmarkdefinition.htm.
22
These definitions are further described and adapted to the EU context in the Eurostat FATS
Recommendations Manual, available at http://ec.europa.eu/eurostat/en/web/products-manuals-and-guidelines//KS-RA-12-016.

1.20. A number of variables are recommended for compilation: sales/turnover or output,
employment, value added, trade, number of enterprises etc. Output is the variable recommended
for measuring the supply of services through mode 3, but sales or turnover can be used in the
absence of data on output. Variables should be broken down to the extent possible as follows: type
of service provided (if possible using a classification compatible with EBOPS 2010 definitions and
identifying the provision of services in the country of operations), by main economic activity, by
country of operation of affiliates for outward FATS and by country of the UCI for inward. The
BPM6 basic principles also apply as relevant in FATS: residence, determining control which is
necessary for determining the FATS universe, recording of information on an accrual basis and for
monetary information valuation at market prices. The valuation of transactions between affiliated
firms may not represent market prices. This question is further discussed in Box 1.3.
1.21. Given that at the time of writing this Guide, FATS is still a relatively new statistical
domain, it is suggested that compilers develop a FATS compilation system in stages. MSITS 2010
recommends, as an initial priority, that FATS be compiled on an activity basis, because this is
required for the compilation of some variables, as well as since data are currently most widely
available on that basis. However, as compilation of data on a product basis is recognized as a
longer-term goal, compilers are encouraged to work towards providing product detail for those
variables that lend themselves to this basis of attribution (namely, sales (turnover) and/or output,
exports and imports). If this level of specificity cannot be achieved, compilers may wish to
disaggregate sales or output in each industry between sales/output of goods and sales/output of
services, as a first step towards achieving a product basis.
1.22.
Inward FATS find often their place in structural business statistics (SBS), traditionally
compiled by NSOs, but they can also be collected through means used to collect foreign direct
investment statistics (or in conjunction with those, such as a dedicated survey). Outward FATS can
be produced within the scope of the foreign direct investment statistics (FDI or dedicated survey).
FATS can also be compiled from administrative records of a public agency in the economic or
commercial area related to foreign direct investment. For both inward and outwards FATS,
analysing the relationships between different data collection and compilation options yields a
number of advantages, including reduction of reporting burden and cost effectiveness. FATS
compilers may also consider accessing surveys based on a framework taken from business
23
registers of international groups prepared by supranational bodies. The current interest in the
phenomenon of economic globalisation and the new approach of international trade statistics in
terms of value added, and their relation to global value chains, have revealed the importance of
reinforcing these statistics.
E.

Additional indicators for analysing the international supply of services

1.23. MSITS 2010 also recommends the compilation of additional indicators in order to further
analyse the international supply of services beyond the value of such transactions. These indicators
are needed in order to respond to the numerous information needs of trade negotiators and other
users. Some of these indicators are directly related to the (GATS) negotiations, as they support
comparison of commitments, provide background information for the settling of disputes, or can
be used more generally for an in depth analysis of the international supply of services. The other
indicators range from data on the number of persons/trips relevant to modes 2 and 4, the use of
data on services production, prices, employment or sectoral indicators (e.g., tourism,
telecommunications, etc.). All of these additional sets of statistics do not require the development
of new statistical frameworks, but rather refer to existing ones. Their compilation particularly
indicates the potential offered by the linking of trade in services statistics and business registers.

See, for example, the Eurostat Euro Groups Register.

Data collection and compilation of these other indicators are described in Chapters 6, 7, 9, 10, 11,
13 and 16.
1.24. Frameworks of particular interest are those related to the movements of individuals across
borders; i.e., tourism and migration statistics, whether for supplying services (mode 4) or for
consuming services (mode 2). Obtaining information on the characteristics of these movements
and on the individuals involved is highly important for trade policy making and monitoring, for the
analysis of global supply of services, as well as broader tourism and migration policy, in particular
for mode 4. Data on the movement of natural persons should include detailed information on the
characteristics and activities of such persons as well as on the service they are rendering or
consuming while abroad.
F.

Integrated economic statistics

1.25. At its thirty-seventh session in 2006, the UN Statistical Commission (“the Commission)
recommended the operational use of an integrated approach to economic statistics in national
economic statistics programmes to ensure the efficiency of the statistical process. 24 The
Commission subsequently developed the Guidelines for Integrated Economic Statistics,25 as it
recognized the significance of an integrated approach for increasing the consistency and coherence
of economic statistics. It enhances the quality and analytical value of short-term, annual and
benchmark economic statistics and macroeconomic statistics.26
1.26. The guidelines for integrated economic statistics are based on internationally adopted
standards, including the System of National Accounts 2008 (2008 SNA), the Balance of Payments
Manual, sixth edition (BPM6) and more specialized technical manuals, such as those on the
measurement of prices, sectoral and business statistics and FDI. The guidelines use the
methodological standards, the recommendations and the emphasis on policy-relevant data in those
manuals as the basis for the organizing principles and detailed practices for preparing integrated
economic statistics. 27 The main characteristics of this integrated approach are described in Box 1.4.

Based on the report of the Secretary-General on integrated econo mic statistics (E/CN.3/2006/5).
United Nations Publication, Guidelines for Integrated Economic Statistics. ST/ESA/STAT/SER.F/108.
http://unstats.un.org/unsd/nationalaccount/docs/IES-Guidelines-e.pdf.
26
Ibid., preface.
27
Ibid., p. 1.
25

Box 1.4 An Integrated approach to economic statistics
Guidelines on integrated economic statistics identified the following features of an integrated approach:
i.

Common concepts, definitions and classifications. The use of harmonized terminology,
definitions, concepts, standards and classifications is necessary in a national statistical system so
that the various data collections are comparable an d can be related to each other;

ii.

Business registers and frames. Business registers h ave a central role in integrated economic
statistics in providing a central sampling frame for all business surveys.

iii. Standardization of surveys. Integration should be comprehensive and encompass survey design,
sample frame, and questionnaire design;
iv. Administrative data. Administrative source data ca n be integrated for statistical purposes;
concepts need to be matched with statistical records. The advantage of using administrative
records and various government data is to promote a more efficient use of data collections, and
reduce the burden of the respondents;
v.

Data editing, linkage and integration. Documentation of the editing process and transparency are
indispensable to ensure that the resulting data can be used by various statistical domains and will
be widely accepted and understood by users;

vi. Dissemination and communication. Integration may facilitate providing user-friendly
presentations of data, explanations of concepts, en sure consistent format across publications,
electronic sources and websites.

Chapter 2

Legal framework

2.1.
Scope. This chapter outlines the need for and importance of legal framework that regulates
activities of the institutions involved in the data collection and data compilation of in accordance with
the statistical framework recommended in MSITS 2010 for measurement of the international supply
of services, as well as the content of data sources and confidentiality of provided information. The
chapter consists of the following sections: summary of good practices (section A); the importance of
the legal framework (Section B); legal acts governing content and availability of data sources (section
C); legal acts regulating institutional arrangements (section D); legal acts protecting confidentiality of
reported data (section E); and country practices (section F).
A.

Summary of good practices

2.2.
It is a good practice that national laws and regulations be established to define the rights and
responsibilities of all agencies involved in the collection, exchange, processing, compilation and
dissemination of data on services transactions between residents and non-residents, foreign affiliates
statistics and additional monetary and non-monetary indicators for analysing the international supply
of services, so that those agencies will be in a better position to establish the necessary institutional
arrangements detailing their involvement in the statistical process. More specifically, such laws
should give the compiler a clear mandate to collect necessary data from relevant sources and to
compile and disseminate these statistics in the formats deemed appropriate by the responsible agency.
This Guide encourages compilers to actively participate, whenever appropriate, in the preparation of
the necessary modifications to national legislation or relevant administrative regulations. Moreover, it
is a good practice that international recommendations for trade in services statistics are properly
reflected in national laws or regulations.
2.3.
It is further advised that additional legal acts establish more specific provisions for the data
content or the specific circumstances under which the data must be reported to other national agencies
as well as the frequency of dissemination. A legal basis for surveys related to resident/non-resident
trade in services statistics and FATS should also be well-established and should be cited on the survey
forms themselves. In the case of outward FATS, where the relevant company is located abroad, the
legal acts should provide a mandate for statistical compilers to require information from domestic
companies on their foreign affiliates located in other countries. The statistical compilers should also
be well aware of the contents of the legal provisions underpinning the international bank transaction
reporting system (ITRS) implemented in the country and make full use of them. It is further advised
that the national statistical act establishes a reminder and penalty system, including the possibility to
impose fines on data reporters, in cases of non-compliance in reporting required information.
2.4.
Finally, it is of extraordinary importance that the legal framework includes provisions
guaranteeing the protection of individual data, restricting the use of such data to official statistical
purposes only, and preventing identifiable enterprise-specific information from being disclosed in the
disseminated statistics.
B.

The importance of the legal framework

2.5.
The legal framework promoted by this Guide consists of the laws and legally-binding
regulations that govern the data provision relationship between members of the national statistical
system and data reporters as well as all aspects of data compilation and dissemination.
2.6.
A well-established legal framework is the foundation of an effective and well-functioning
national statistical system, because it:
i.
Gives the statistical compiler a clear mandate to collect data and to compile and
disseminate the statistics, and makes relevant information (e.g., supplementary data sources)
17

available to the compiler. In view of the multiplicity of data sources necessary for compilation
of statistics on the international supply of services, a strong and comprehensive legal
authority to collect data is of special significance.
ii.
Identifies and defines the national competent authority and the roles other
governmental agencies should play in particular statistical domains. (see also Chapter 3).
iii.
Is a prerequisite for establishing a high-quality system of official statistics. For
example, the first element of the quality assurance framework recommended by the UN
28
Statistical Commission calls for the availability of a sound legal framework , as does the
IMF Data Quality Assessment Framework (DQAF) for Balance of Payments and
International Investment Position Statistics.29
iv.
Is likely to increase public trust in the confidential treatment of reported data and
thereby to facilitate obtaining higher response rates and the submission of accurate
information. Statistical confidentiality is a key element in collecting, storing, and
disseminating statistics and related micro-data. Thus, the legal framework should include
provisions guaranteeing the protection of individual data and restricting the use of such data
to official statistical purposes only.
v.
Assists in the implementation of an integrated approach to statistics of trade in
services between residents and non-residents and FATS, as only well-coordinated efforts of
several agencies may guarantee a successful compilation and dissemination of such statistics.
The stronger the legal powers of coordination at the center of the statistical system, the
greater the chance of integrating the statistics effectively.30
vi.
Provides the possibility to impose penalties on data reporters, including enterprises, in
cases of non-compliance in reporting required information (including delays in reporting,
missing or unreported data, and errors, etc.).
2.7.
Legal acts relevant for statistics of international trade in services exist at different levels and
include legal acts regulating the collection, processing and dissemination of statistics; the working
relations of the concerned agencies; and the legal acts protecting the confidentiality of information.
The structure of the legal framework can vary from country to country. Often there is a national
statistical act establishing the mandate of the national statistical authority and governing general
statistical issues, such as establishment of a reminder and penalty system for enterprises failing to
comply with the reporting obligation. The national act could also identify statistical areas that are part
of the mandate. In addition to the general act, additional acts can establish more specific provisions
for a single statistical area like trade in services or FATS, for the data content or for the specific
circumstances under which the data must be reported to other national agencies and/or defining the
valid media and frequency of reporting and dissemination.
2.8.
If a country becomes a party to an international convention or adopts international
recommendations, it then needs to incorporate the binding provisions and non-binding
recommendations of those conventions in its national laws and regulations. In this connection, it
should be noted that in the field of international trade, many international agreements exist to govern
transactions in goods and services. For example, the results of the Uruguay Round of Multilateral
Trade Negotiations contain numerous legal obligations of the WTO members and are incorporated in
the national legislation. In a similar sense, international recommendations for statistics on the

NQAF, page 9. https://unstats.un.org/unsd/dnss/docs-nqaf/2012-13-NQAF-E.pdf.
IMF, DQAF for Balance of Payments and Internationa l Investment Position Statistics (2012), page 1.
http://dsbb.imf.org/images/pdfs/dqrs_bop.pdf.
30
Guidelines on IES, paragraph 4.5.
29

international supply of services should be properly reflected in national laws or regulations. The
international recommendations, including the MSITS2010, the Balance of Payments and International
Investment Position Manual, and associated implementation guidelines, provide the foundation for the
data to be collected. These recommendations must be supported by national legal acts in order to give
the statistical office a clear mandate for collecting the necessary data elements.
2.9.
In some regions there may exist legally-binding regional agreements that impact the national
definition and/or scope of statistics on the international supply of services and regulations pertaining
to the national agencies involved in the statistical process. Sometimes there are separate acts
regulating different areas of statistics. For example, in the European Union, “Regulation (EC) No
184/2005 of the European Parliament and of the Council of 12 January 2005 on Community statistics
31
concerning balance of payments, international trade in services and foreign direct investment”
regulates the data to be submitted to Eurostat on trade in services between residents and non-residents
and “Regulation (EC) No 716/2007 of the European Parliament and of the Council of 20 June 2007
on Community statistics on the structure and activity of foreign affiliates”32 regulates the FATS data
to be submitted to Eurostat.
2.10. In many instances, the introduction of necessary improvements to the quality of official
statistics may be significantly facilitated by developing appropriate legal provisions, or, if such
provisions already exist, by amending them with regards to the collection of administrative data to be
used for statistical purposes. The national agency or agencies responsible for the overall compilation
and dissemination of statistics on the international supply of services should, whenever appropriate,
actively participate in the discussion of respective national legislation or relevant administrative
regulations in order to establish a solid foundation for the high quality and timeliness of these
statistics.
C.

Legal acts regulating institutional arrangements

2.11. As often more than one national institution is involved in the compilation of data in
accordance with the statistical framework recommended in MSITS 2010 for measurement of the
international supply of services, it is either the national statistical office or the central bank that is the
responsible national authority for coordination of such activities as well as for data dissemination. In
many countries, a statistics act or formal legal arrangement exists which determines the national
authority for data collection (for instance through surveys), for data compilation and for data
dissemination. In some economies countries, these responsibilities may be shared among two or more
agencies. For example, central banks may have responsibility for obtaining data from financial
institutions, while the national statistical agency has responsibility for collection from other institutes
and for the data compilation and dissemination. Or, an investment approval agency or a financial
supervisor may be a very important source of information about cross-border transactions in services.
Under such circumstances, it is important that the legal or other arrangements are in place allowing
the agency responsible for the compilation of a particular data set to receive or get access to the
relevant data sources.
2.12. When the compiling statistical authority is dependent upon other national institutions for data,
close cooperation and coordination is needed. Such cooperation is facilitated by the appropriate
legislation. In this context, it is a good practice that compilers actively participate, whenever
appropriate, in the necessary modifications to national legislation or relevant administrative
regulations in order to establish a solid foundation for enhancing the quality and timeliness of the
statistics. It is a good practice that the national laws and regulations define the rights and

Regulation No 184/2005.
http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2005:035:0023:0055:EN:PDF.
32
Regulation No 716/2007.
http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2007:171:0017:0031:EN:PDF.

responsibilities of all agencies involved in the collection, exchange, processing, compilation and
dissemination of the statistics, so that those agencies will be in a better position to establish the
necessary institutional arrangements detailing their involvement in the statistical process. For
instance, if adequate legal provisions are in place, the responsible agency could establish, faster and
more efficiently, a working arrangement with the organizations keeping records relevant to statistics
on the international supply of services. See chapter 3 for further discussion of institutional
arrangements.
D.

Legal acts governing content and availability of data sources

2.13. If the statistics are based on administrative records, it is advisable to include a paragraph in
the legal text that gives the statistical office a legal right to access these data sources. Such language
would facilitate the cooperation between the institutions involved, thereby contributing to timely data
deliveries and ensuring transparency.
2.14. One of the administrative sources relevant to trade in services statistics is customs records.
The World Customs Organization (WCO) is the international platform at which countries reach legal
agreements on customs regulations, including those relevant to compilation of data on international
freight and insurance related to the transportation and customs clearance of goods. Among the most
relevant WCO conventions is the revised international convention on the simplification and
33
harmonization of customs procedures (known as the revised Kyoto Convention), which provides
standards for various customs procedures (for example, procedures on inward and outward processing
of goods which have to be understood when compiling, for example, such items as manufacturing
services on physical inputs owned by others). The WCO also participates in setting legal provisions
for the customs valuation of goods, which has significant consequences on the valuation of the related
international transactions in services. In this context, it should be noted that it is the WTO’s
responsibility to formally adopt such provisions and amend them as necessary.34
2.15. Much of the data need for measuring the international supply of services come from statistical
surveys or from ITRS, formerly known as foreign exchange record systems. Most ITRS evolved as
by-products of foreign exchange control acts which were, and in many instances still remain, legally
binding. The statistical compilers should be well aware of the contents of the legal provisions
underpinning the national ITRS and make full use of them. Also, as countries are increasingly
planning to remove certain legal obligations, statistical compilers should prepare themselves well in
advance to make necessary adjustments in their data collection arrangements.
2.16. The importance of a well-established legal basis for statistical surveys is recognized and
promoted by all concerned international and regional organizations. For example, according to the
IMF, good legal authority needs to state that reporting of statistical information is mandatory,
especially for large enterprises.35 It is a good practice to explicitly refer to such legislation in the
resident/non-resident trade in services statistics and FATS survey forms.
2.17. Compared to outward FATS, inward FATS is often easier to collect because the relevant
enterprises are present in the compiling economy. Often only information about the foreign-owned
share of a domestically-located enterprise needs to be collected, which, in turn, can be linked to
existing information in the national business statistics. In the case of outward FATS, where the
33
The International Convention on the Simplification and Harmonization of Customs procedures entered into
force in 1974. The revised Kyoto Convention (RKC) was adopted in June 1999. The provisions contained in the
RKC aim at the facilitation of trade but at the sam e time make customs records a highly standardized a nd
reliable data source for trade statistics across co untries. The RKC comprises a Body, a General Annex and
Specific Annexes.
34
Statistical compilers should take note of the WTO Agreement on Customs Valuation; see IMTS 2010, Annex
D.
35
See BPM6 Compilation Guide, paragraphs 2.5.

relevant economic entity is located abroad, the legal acts should provide a mandate for statistical
compilers to require information from domestic enterprises on their foreign affiliates located in other
countries. This aspect might cause difficulties for compilers because of the limited data availability
and sensitivity to reporting such data. In either case, it is important that the compilers have a clear
legal mandate for collecting the data.
E.

Legal acts protecting confidentiality

2.18. According to the UN fundamental principles of official statistics, individual data collected by
statistical agencies for statistical compilation are to be strictly confidential and used exclusively for
36
statistical purposes (UN Statistical Commission, 14 April 1994). Statistical confidentiality is a key
element in producing reliable statistics as it is important to gain the trust of the data providers. Thus,
legal provisions should be established to ensure confidentiality and proper use of data and prevent
identifiable enterprise-specific information from being obtained in the disseminated statistics.
2.19. The legal acts on confidentiality should also address the use of administrative data for official
statistical purposes. In particular, it is important that legal acts state that administrative data
forwarded to the statistical authority become statistical data upon receipt, which implies that data
validated and edited by the statistical agency should not be shared with the administrative authority
which initially provided the administrative data. Examples of legal acts on data confidentiality are
provided in boxes 2.1 and 2.2.
2.20. Confidentiality of customs declarations. In general, customs declarations are not subject to the
same level of confidentiality measures as are other statistical instruments. By design, customs
declarations are used to assess tariffs, fees and taxes, and to enforce multiple agencies’ requirements
for admissibility of goods into the country or to enforce the country’s exports laws and regulations.
Once transmitted to the agency responsible for the compilation of international trade statistics, in
many cases, that agency treats the information as confidential. However, in most cases, the compiling
agency does not subject all data to rigorous disclosure reviews, and rather applies ‘passive disclosure’
methods by which importers/exporters inform the agency of possible situations for investigation and
for some form of statistical suppression. This fact should be well noted by the statistical compilers
while considering access to customs declarations for obtaining, for example, information on crossborder trade in services.
2.21. The compiling agency may also establish appropriate regulation to safeguard confidentiality
in the exchange of basic information among agencies. However, regardless of the legal status of
confidential information, whether personal or commercial, this information should not be excluded
from the statistics and should be reported in aggregate form so that the confidential aspects of these
operations cannot be identified (see Chapter 20). It is further desirable that national legislation defines
rights and responsibilities regarding access to the micro-data, highlighting the appropriate principles
and procedures. The responsible agency should cooperate with the national legislature to establish
such laws.

UN Statistical Commission, 14 April 1994. http://unstats.un.org/unsd/dnss/gp/fundprinciples.aspx and
http://unstats.un.org/unsd/dnss/gp/FP-New-E.pdf. In addition, the General Assembly adopted Resolut ion
A/RES/68/261 entitled "Fundamental Principles of Official Statistics (without a vote) thereby endorsin g the
Fundamental Principles of Official Statistics, as adopted by the Statistical Commission in 1994 and reaffirmed
in 2013, and endorsed by the Economic and Social Council in its resolution 2013/21". The draft text, which was
adopted
without
any
changes,
can
be
found
in
A/68/L.36
and
at
http://www.un.org/en/ga/search/view_doc.asp?symbol=A/68/L.36.

Box 2.1 The Confidentiality Act of the Philippines
Confidentiality of data in the Philippines is based on provisions of the Commonwealth Act 591, Section 4. The
Act states, inter alia, that data furnished to the National Statistics Office by an individual, corpor ation,
partnership, institution or business enterprise shall not be used in any court or in any public office either as
evidence for or against the individual, corporation, association, partnership, institution or business enterprise
from whom such data emanates; nor shall such data o r information be divulged to any person except authorized
employees of the National Statistics Office, acting in the performance of their duties; nor shall such data be
published except in the form of summaries or statistical tables in which no reference to an individual ,
corporation, association, partnership, institution or business enterprise shall appear. Any person vio lating the
provisions of this section shall, upon conviction, be punished by a fine or by imprisonment, or by both.
See: http://www.census.gov.ph/content/commonwealth-act-no-591

Box 2.2 Legal act on confidentiality - example of the European Union
Member States of the European Union have to comply with the regulation on European Statistics (Regulation
(EC) No 223/2009 of the European Parliament and of the Council) which contains provisions on data
confidentiality in its Article 20. The article stat es, in particular, that cconfidential data obtained exclusively for
the production of European statistics shall be used by the NSIs and other national authorities and by the
Commission (Eurostat) exclusively for statistical purposes unless the statistical unit has unambiguously given its
consent to the use for any other purposes” (Article 20, point 2), and that statistical results which m ay make it
possible to identify a statistical unit may be disseminated by the NSIs and other national authorities and the
Commission (Eurostat) in the following exceptional cases: (a) Where specific conditions and modalities are
determined by an act of the European Parliament and of the Council acting in accordance with Article 2 51 of the
Treaty and the statistical results are amended in s uch a way that their dissemination does not prejudice statistical
confidentiality whenever the statistical unit has so requested; or (b) Where the statistical unit has unambiguously
agreed to the disclosure of data”. (Article 20, poi nt 3 and 3(a))
See:. http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2009:087:0164:0173:en:PDF

Country experiences
F.1.

Country experience: the United States

2.22. The authority to collect statistics of international trade in services in the United States is
ultimately delegated to the Bureau of Economic Analysis (BEA), a statistical agency within the U.S.
Department of Commerce. The primary legal provision enabling BEA to collect information in a
timely manner from relevant institutions is known as the International Investment and Trade in
Services Survey Act ("the Act") (P.L. 94-472, 90 Stat. 2059, 22 U.S.C. 3101-3108, as amended)37 .
Provisions in the Act ensure that US entities (i.e., businesses, universities, hospitals, etc.) engaged in
international investment and trade in services are required to report their international transactions on
a periodic basis to the BEA. The Act specifies that the survey data may only be used for statistical and
analytical purposes. Access to the data is limited to officials and employees (including consultants and
contractors and their employees) of government agencies that are designated by the President to
perform functions under the Act. Certain other government agencies may be granted access to the data
under the Foreign Direct Investment and International Financial Data Improvements Act of 1990, but
only for limited statistical purposes. BEA is prohibited from granting another agency access to the
data for tax, investigative, or regulatory purposes. BEA cannot publish or otherwise release the data
collected on its surveys in a form that would allow the transactions of an individual reporter to be
identified.
2.23. The other primary legal provision governing the collection of resident/non-resident trade in
services statistics and FATS data is the Paperwork Reduction Act of 1995. By provisions of this act,

See A Guide To BEA’s Services Surveys ; available at: http://www.bea.gov/surveys/pdf/surveysu.pdf

US government surveys must undergo an approval process in which the agency in charge of collecting
the data is required to demonstrate to the approving authority (the Office of Management and Budget)
the following three conditions: (1) the intended data are necessary, (2) they cannot be obtained from
an existing source, and (3) their collection does not place an unreasonable burden on respondents. As
part of the survey design and clearance process, BEA publishes notices about proposed surveys in the
Federal Register. In these notices, BEA requests comments from users and respondents on all aspects
of the data collection, including BEA’s estimate of the burden imposed by the reporting requirements.
BEA considers all comments before making final decisions on the scope and design of its surveys.
BEA makes every effort to balance the needs of data users for complete, accurate, detailed, and timely
data and the concerns of respondents about the burden imposed by the reporting requirements. By
convention, approval of a given survey must be renewed within 3 years.
F.2. Country experience: Costa Rica
2.24. In Costa Rica, trade in services statistics are compiled and disseminated in accordance with
the terms and conditions of the National Statistical System (SEN) Law No. 7839 of November 4 ,
1998 and the Central Bank of Costa Rica (BCCR) Organic law 7558 of November 3, 1995 (with
subsequent revisions). The Central Bank of Costa Rica (BCCR)’s Organic Law does not specifically
assign the task of compiling trade in services statistics to the BCCR. Article 15(d) of the INEC/SEN
law officially assigns to the BCCR the task of preparing the basic statistics needed to prepare the
national accounts and other macroeconomic accounts. However, the BCCR has no legal basis to
request information from the non-financial private sector in order to compile trade in services
statistics. Thus, it is not permitted to impose sanctions on entities that do not provide data. In view of
this limitation, staff of External Sector Statistics Area (AESE) and the Economic Surveys
Area (AEE) have adopted procedures to motivate data providers to report their information, including
letters explaining to respondents the importance of providing the requested data and sending them
reports based on the statistical contribution as well as periodical consultations when visiting the
companies. The average level of response to surveys is 90 percent.38
F.3. Country experience: India
2.25.
In India, the collection of data foreign liabilities and assets (FLA) of Indian companies is
mandated by the Foreign Exchange Management Act (FEMA) of 1999. It is mandatory for Indian
companies which have received foreign direct investment (FDI) and/or made overseas investment in
any of the previous year(s), including the current year, to submit an annual return on FLA to the
Reserve Bank of India (RBI).39 Non-filing of the return before the due date is treated as a violation of
FEMA.
2.26. Initially, the objective has been to collect comprehensive information on the operations of
Indian companies having foreign participation in equity capital, including counterparty country
information for India’s participation in the IMF’s annual global Co-ordinated Direct Investment
Survey (CDIS). However, beginning in 2012, the scope of FLA has been extended to collect
information for the compilation of inward Foreign Affiliates Trade Statistics (FATS).

International Monetary Fund’s Special Data Dissemi nation Standards website for Costa Rica
http://dsbb.imf.org/Pages/SDDS/CtyCtgList.aspx?ctycode=CRI
39
Further information on Reserve Bank of India’s FLA return is available here
http://www.rbi.org.in/scripts/FAQView.aspx?Id=95

Chapter 3

Institutional arrangements

3.1.

Scope. This chapter describes the need for and purposes of institutional arrangements in
the context of the implementation the statistical framework recommended in MSITS 2010 for
measuring the international supply of services. It elaborates on characteristics of effective institutional
arrangements and on good practices in setting up such arrangements under different country
circumstances. The chapter consists of the following sections: summary of good practices (section A);
the need and purposes of institutional arrangements (section B); characteristics of effective
institutional arrangements (section C); and country practices (section D).
A.

Summary of good practices

3.2.

It is a good practice that agencies involved in collection and compilation of data on the
international supply of services establish institutional arrangements that involve the key producers and
data users, properly reflecting the country's needs, priorities and resources. Such institutional
arrangements should present a clear division of responsibilities and work between the agencies
involved in the compilation of statistics on services transactions between residents and non-residents,
FATS as well as additional monetary and non-monetary indicators for analysing the international
supply of services. It is essential that appropriate channels of communication and mechanisms of
coordination.

3.3.

This Guide underscores that institutional arrangements lay the groundwork for effective
process management, from the identification of data sources to the dissemination of output and for
promoting communication between the staff of the different institutions involved. Institutional
arrangements should be based on the legal framework and elaborate upon it as necessary. Institutional
arrangements should be periodically reviewed and improved as necessary to keep them relevant. An
advisory committee, or a similar body, should be established in order to support sound decisionmaking and to take into account the interests of stakeholders.
B.

The purpose and types of institutional arrangements

3.4.

The production of official statistics typically requires the participation of several agencies
and, consequently, the establishment of institutional arrangements. In the area of trade in services
statistics, agencies that may be involved in the statistical process may include the National Statistical
Office (NSO), the Central Bank (CB), the Ministry of Trade, the Ministry of Economy, the Ministry
of Finance, the Tax authorities, Immigration authorities, the National Tourism Administration, Border
protection agencies, Customs administration, Financial markets regulators, Chambers of Commerce
and Investment promotion bodies.

3.5. The institutional arrangements are generally understood as a set of agreements on the division
of the responsibilities between the agencies involved in the collection, compilation and dissemination
of data pertaining to a given statistical domain. These arrangements ensure that official statistics meet
users’ needs, follow quality standards and are compiled and disseminated in the most efficient way.
Meeting users’ needs is especially important in a relatively young and wide statistical domain like
statistics of international trade in services. The scope of institutional agreements may range from
determining the complete process of the statistical production and dissemination to regulating certain
parts of this process.
3.6.
National statistical systems are organized on the basis of the statistical and other
applicable national laws and regulations, which, to different degrees, specify the rights and
responsibilities of the involved agencies thus defining main features of the county’s statistical system.
Usually, two main types of national statistical systems are differentiated - centralized and
24

40
decentralized statistical systems . A national statistical system is referred to as centralized if the
management and operations of the statistical programmes are predominantly the responsibility of a
single autonomous government agency. A national statistical system is commonly referred to as
decentralized if the statistical programmes are managed and operated under the authority of different
institutions with coordination normally provided by a single body.

3.7.

Institutional arrangements should ideally complement the legal framework (see chapter
2). Especially in cases in which the legal framework is weak, institutional arrangements can play an
important role in improving it and addressing the current constraints. Different types of institutional
arrangements can equally provide for statistics on the international supply of services that follow
internationally recognized methodological guidelines, utilize all available statistical sources in an
efficient way and apply appropriate compilation procedures. Practical examples from countries with
different statistical systems are provided to help in setting up institutional arrangements in a most
effective way at the end of this chapter.
C.

Characteristics of effective institutional arrangements

3.8.
Institutional arrangements should involve the key producers and users of the statistics and
can be set up in different ways depending on each country's needs, priorities and resources. They
should contribute to establishing appropriate channels of communication and mechanisms of
coordination to ensure efficiency in statistical production. Pre-existing institutional arrangements in
related statistical domains (e.g. in international merchandise trade and tourism statistics) should be
taken into account and built upon when possible.
3.9.
As part of the balance of payments, resident/non-resident trade in services statistics are
important to monetary policy analysis and, together with the national accounts and other economic
statistics, describe a country’s economic development. Therefore, NSOs and CBs often share
responsibilities for the collection and compilation of these statistics.
3.10.

Another important analytical aspect of which statistical compilers should be aware is the
need of countries’ economic agencies (such as Ministries of Economic Affairs and the Chambers of
Commerce, among others) to conduct assessments of the competitiveness of countries and economic
sectors for the design of economic policies and international trade negotiations based on GATS.
Furthermore, the compilation of statistics on the international supply of services in some areas
coincides with other mandatory regulations. For example, Financial Market Authorities are typically
responsible for banking and insurance supervision. Moreover, NSOs, NCBs, and other institutions
may play very important roles in collecting, producing and disseminating foreign affiliate statistics
(FATS), in accordance with the legal mandate in the statistics area assigned to each institution.

3.11.

The success of the institutional arrangements, in most cases, depends on the existence of
a clear division of the responsibilities and mutually beneficial cooperation between NSOs and CBs
(and possibly other agencies involved), which have historically developed in different ways. In view
of the growing need for information going hand in hand with rising cost consciousness, NSOs and
CBs should seek cooperation in focusing on their respective expertise, making use of existing data
and ensuring consistency in statistical production.

3.12.

This process is further stimulated by the planned expiration in many countries of the bank
settlements systems (traditionally the main data source of central banks for balance of payments
statistics, which cover trade in services). These systems are generally being gradually replaced by
enterprise surveys as the main data source for the compilation of trade in services statistics. This
40
For details see: Guidelines on Integrated Economic Statistics, United Nations publication,
ST/ESA/STAT/SER.F/108, 2013, paragraphs 3.33 – 3.39.

transition leads to a more evident need for closer cooperation and coordination between the CBs and
the NSOs, as the latter are normally conducting enterprise surveys and maintaining the business
registers.

3.13.

NSOs and CBs should also look for close cooperation with the Chamber of Commerce
(or other representatives of a country’s private sector) to get support in producing high-quality data.
On the one hand, Chamber of Commerce and/or other representatives of a country’s private sector can
inform enterprises about the importance of timely and accurate reporting, the use of electronic media
and support the communication between statistics producers and enterprises in survey design as well
as in the interpretation of results. On the other hand, NSOs as well as NCBs can make customized
datasets available to the representatives of private sector which would reflect their specific economic
interests.

3.14.

CBs and Financial Market Authorities work closely together in fulfilling their respective
tasks for financial market stability, sometimes centralized in a single institution. Therefore, they
normally cooperate closely in mutual assistance and no additional legal act is needed for the sharing
of information. However, besides institutional arrangements, CBs as well as NSOs should ensure they
have legal access to administrative data on banking and insurance transactions that are collected for
supervisory purposes.

3.15.

The characteristics of effective institutional arrangements are as follows:
i.
They should take account of the respective responsibilities of the institutions involved
that cover all stages of the statistical process – from the identification of user needs and the
collection of raw data through data compilation, dissemination and evaluation;
ii.
The rights and responsibilities of the institutions involved should be clearly defined to
avoid misunderstandings, duplication of work or omission of significant elements of work;
iii.
The terms of cooperation should be laid out in a legal document so that any changes
to administrative procedures or statistical processes that could affect data compilation become
an integral part of the terms of cooperation and could be dealt with in advance;
iv.
At the same time, institutional arrangements should leave room for necessary
flexibility in every-day statistical production;
v.
Given the legal foundations of the cooperation of various agencies, one institution
(the national statistical office, the central bank or a specially established interagency body)
should have a clear mandate to monitor and coordinate aspects of the production process of
statistics on the international supply of services, as well as disseminate the data and keep in
contact with international organisations and other data users;
vi.
The main user groups should be included in the institutional arrangements and their
needs should be accounted for given the legal obligations and resource restrictions of the
institutions that produce statistics.

3.16.

Further, the characteristics of effective cooperation can be described as:
i.
Operationalization of the relevant international statistical standards and good country
practices;

ii.
Development and implementation of the work programme for collection and
compilation of data in accordance with the statistical framework for describing the
26

international supply of services, including establishment of appropriate interagency data
compilation arrangements;
iii.
Establishment of close contact and regular consultations with the user community to
guarantee the analytical and policy relevance of the data compiled and disseminated;
iv.
Promotion of an integrated approach to data compilation as well as of appropriate
quality management to ensure that high-quality data on the international supply of services
meet ever-increasing user demands and are made available to users despite limited resources;
v.
Dissemination of easily accessible and detailed statistics on the international supply
of services to users, both domestically and internationally, and
vi.
Consultations with enterprises on questions concerning reconciliation and exchange
of data.

3.17.
Finally, some guidelines for creating effective institutional arrangements and terms of
cooperation are:
i.
Adopt a strategic approach to multi-level planning for advancing the integration of
economic statistics;
ii.
Implement effective process management from the identification of data sources to
the dissemination of outputs;
iii.
Periodically review institutional arrangements and initiate necessary adjustments in
order to keep them relevant in the light of evolving user needs and emerging new data
sources;
iv.
Establish an advisory committee so that the interests of all stakeholders are taken into
account and the committee members assist in the development of data by supporting sound
decision-making;
v.
Promote communication between the staff of the different institutions involved to
develop an understanding of the entire production process of statistics compiled within the
framework for describing the international supply of services as recommend in MSITS 2010.
D.

Country experiences
D.1.

Country experience: Austria

3.18.

Statistics on resident/non-resident transactions in services. In Austria, the
Oesterreichische Nationalbank (OeNB) is legally responsible for compiling and disseminating of
resident/non-resident trade in services statistics (Foreign Exchange Act 2004 § 6 (1)). In order to keep
the costs of data compilation low – both in reporting for as well as producing statistics, the OeNB
opted for a close cooperation with Statistics Austria. The two institutions have signed a basic
cooperation framework agreement to guide their cooperation in all fields of statistics related to the
Austrian economy and based on international and/or national legislation. The basic principle of
cooperation is that both institutions concentrate on their respective expertise and data access, which
governs the determination of specific areas of activity. The basic agreement defines the compilation
of services trade statistics as an area of “intensive cooperation” as the OeNB and Statistics Austria
work jointly on a single statistical product, in contrast to areas of “vital interest,” where one party is
responsible for data production and the other is a main user of the data. The details of cooperation in
this area are laid down in an individual service contract where one institution provides input for a
statistical product for which the other institution bears the legal obligation and responsibility. The
contract mainly specifies: (a) individual production stages, dates and interfaces, (b) classifications,
27

revisions and quality measures, (c) access to non-published data, (d) utilization and reconciliation of
existing registers and administrative data, (e) evaluation of new statistical, technical, financial and
legal requirements as well as (f) coordination of the stance to be taken in national, EU- and
international bodies.

3.19.

FATS: In contrast to resident/non-resident trade in services statistics, the NSO is
responsible for compiling and disseminating FATS in Austria. These statistics are compiled in close
cooperation with the OeNB, which has a service provider contract with the NSO. Concerning Inward
FATS, the OeNB determines which enterprises are foreign-controlled and the NSO collects the
respective variables as part of the structural business statistics. Concerning Outward FATS, the
respective variables are collected by the OeNB as part of the survey on FDI. By applying this
approach, FATS can be compiled with minimal additional reporting burden for enterprises in Austria.

3.20.
In addition to regular information sharing in the course of statistics production, the OeNB
and Statistics Austria have set up a committee consisting of at least two higher voting members of
each institution. The committee meets at least once each quarter to monitor the joint activities and
recommend enhancements where possible. It further initiates extensions and updates to the existing
areas of cooperation including the prolongation and amendments to the basic cooperation framework
agreement.
3.21.

With the introduction of direct reporting in compiling external statistics the OeNB also
initiated close cooperation with the Austrian Chamber of Commerce to support contacts with the
business sector. The two institutions decided to enter into a basic cooperation framework agreement
with the main focus on enhancing efficiency in producing and dissemination of up-to-date and
relevant statistical information. On the one hand, the OeNB aims at minimizing costs of statistical
production on both sides – in the data reporting by enterprises as well as in the Central Bank – by
making utmost use of administrative and register information. On the other hand, the OeNB makes
efforts to disseminate easily accessible and detailed statistical information in order to cut costs of
accessing the data. The Chamber of Commerce assists the OeNB in fulfilling its tasks by supporting
communication with enterprises in various ways; e.g., by offering access to internal media and events
and by arguing for the importance of statistical information in general. At the same time, the Chamber
of Commerce fosters the use of administrative and existing statistical information to keep reporting
obligations to a minimum and advocates the simplification of reports; e.g., by using electronic media.
D.2.

Country experience: Malaysia

3.22.

In Malaysia all statistics collected and published by the Department of Statistics Malaysia
(DOSM) are governed by the Statistics Act, 1965 (Revised 1989). Under the terms of this Act, DOSM
has the independence to determine the coverage, contents, methodology and periodicity of data
collection. For the collection, compilation and dissemination of trade in services statistics (including
FATS), DOSM has the lead responsibility. Memoranda of Understandings (MOUs) were established
between DOSM and the Central Bank of Malaysia (BNM) and Tourism Malaysia (TM) to facilitate
and improve cooperation with these agencies. BNM, through its Statistical Services Department, is
responsible for compiling BOP statistics from International Transactions Reporting Systems (ITRS),
while Tourism Malaysia collects tourism-related data to be used for statistics on the international
supply of services. In addition to these agencies, there are 14 other agencies that also contribute.

3.23.

To coordinate the dissemination of trade in services statistics, DOSM chairs the quarterly
meetings of the Inter-Agency Planning Group (IAPG) to brief members on the data prior to
publication. The members include the Economic Planning Unit (EPU), Treasury, BNM, Ministry of
International Trade and Industry (MITI), the Malaysia External Trade Development Corporation
(MATRADE) and the Malaysia Industrial Development Authority (MIDA). Technical meetings with
BNM and other agencies are held as necessary. In addition, The high-level committee structure of the
28

Industrial Master Plan for Malaysia includes a Working Group on Services Statistics (WGSS)
particularly to monitor the development of trade in services statistics.
D.3.

Country experience: Spain

3.24. In Spain, the Bank of Spain (BE) is responsible for the compilation and dissemination of the
resident/non-resident trade in services statistics and the National Institute of Statistics (INE) is
responsible for production and dissemination of FATS. The BE estimates travel services transactions
between residents and non-residents based on information from surveys developed by the different
institutions, including INE and the Institute of Tourism Studies (IET). A Cooperation Agreement
between INE, IET and BE was established to outline the exchange of information needed to estimate
the Travel item. For its part, the BE estimates the Other Services item of the BoP on the basis of a
bank settlements system combined with direct reporting for some large declarants.

3.25.
The planned expiration (set for 2014) of the bank settlements system currently used by
BE for compiling the Other Services item of BoP led to a 2004 agreement with INE for the joint
design of the International Trade in Services Survey (ITSS). INE is responsible for the design of the
sample, the fieldwork and the tasks subsequent to data collection (i.e., editing, imputing and grossingup of the sampling data), while BE contributes the main framework of the survey and the reporters of
international services transactions obtained from its bank settlements system. In 2004 BE signed an
agreement with INE for the joint design of the International Trade in Services Survey (ITSS). INE is
responsible for the design of the sample, the fieldwork and the tasks subsequent to data collection
(i.e., editing, imputing and grossing-up of the sampling data), while BE contributes the main
framework of the survey and the reporters of international services transactions obtained from its bank
settlements system.
3.26.
Since 2004 INE also has had a Cooperation Agreement with the Spanish Tax
Administration Agency (AEAT), which annually forwards the micro-data of VAT records for
international goods and services transactions to INE as well as the micro-data of VAT records
specifically on services transactions within the EU. Within this Agreement, INE also receives
annually the micro-data on the INTRASTAT/EXTRASTAT operators that dispatch or bring in goods
for processing or repair, which is useful for identifying the population that engages in the processing
and repair services category of EBOPS 2010.
3.27.

For the compilation of Outward FATS, INE has a Cooperation Agreement with the
Ministry of the Economy and Competitiveness under which every year the latter forwards micro-data
of the Foreign Investments Register (FIR) to INE, which means that there is no need for a specific
survey.

3.28.

Inward FATS are compiled based on the business structural surveys conducted by INE in
relation to the industrial, services and R&D sectors, to which is added information from the Ministry
of Development on the construction sector, information from the Ministry of the Economy and
Competitiveness (DG for Insurance and Pension Funds) on the insurance sector, and BE information
on the financial sector. There are no ad hoc cooperation agreements with these administrative bodies
for this exchange of information.

Part II

Data Collection

Scope. Part II focuses on data collection, while Part III deals with data compilation. It is recognized
that the boundary between data collection and data compilation is not always clearly defined.
However, efforts were made to discuss in this part main aspects of data collection, such as data
sources, their advantages and shortcomings, as well as various aspects of data collection process. Part
II begins with an introductory Chapter 4 providing an overview of data sources used within the modes
of services supply framework, which is followed by the description of registers and survey frames
(Chapter 5) and continues with elaboration of enterprise and establishment surveys (Chapter 6),
surveys of persons and households (Chapter 7), international transaction reporting system (Chapter 8),
administrative records (Chapter 9), and other data sources (Chapter 10). Part II is concluded by a
comparison of various data sources (Chapter 11).

Chapter 4

Introduction and overview of data sources within the modes of
services supply framework

4.1.
Scope. This chapter serves as an introduction to the other chapters of Part II by briefly
describing the main data sources which countries are encouraged to use in collecting information
necessary for the compilation of statistics on services transactions between residents and nonresidents, FATS and additional indicators on the international supply of services. Part II is followed
by Part III, which focuses on the data compilation issues.
A.

Data sources and data collection: an overview

4.2.
Part II describe the main data sources used for compiling statistics of international supply of
services. While the international transaction reporting system was in the past the most prevailing data
source of information for resident/non-resident trade in services statistics, it is being increasingly
supplemented by other sources of data as more detailed information on services categories and trading
partners is required. Nevertheless, for the balance of payments services statistics, the ITRS remains an
important data source in many countries and it should be utilized. Increasingly, statistical business
registers, which serve both as a source of information and the basis for the organization of various
surveys, are being recognized for their central role in the implementation of an integrated approach of
statistics and are a prerequisite for building a forward-looking programme of data collection in this
statistical domain. The use of administrative sources and other sources is another essential part of
such a programme, as these sources can complement statistical surveys and/or provide information
when, for example, surveys are not cost effective or difficult to organize.
4.3.
Possible data sources for compilation of statistics, according to the recommendations of
MSITS2010, and their definitions are listed below (each data source is covered more in depth in each
of the corresponding chapters cited):

i. Statistical business registers (chapter 5). The statistical business register (SBR) is commonly
understood as a register of economic entities active in the national economy. If various types
of those entities and their characteristics are taken into consideration then the definition can
be further elaborated.41

ii. Survey (sampling) frame (chapter 5). The survey frame (also called the sampling frame) is the
statistical tool used to gain access to the target population that is to all economic entities
41

See Eurostat publication Business registers: Recommendations manual at:
http://unstats.un.org/unsd/EconStatKB/Attachment174.aspx.

which are intended to be surveyed. There are two types of frames: list frames and area frames.
42
The frame is the backbone of the statistical system. It represents what must be regularly
measured by the statistical system. Its coverage must be as complete as possible and reflect
the organizational structure of all statistical units of the economy.

iii. Enterprise and Establishment Surveys (chapter 6). These surveys can be one of the following
types:
i.

Census: Includes all members of the population;

ii.

Partial coverage collection survey: Includes all enterprises above a certain threshold
measured in terms of their dimensions (e.g. nominal capital) or other variables
(e.g., significant cross-border activity);

iii.

Random sample survey: Includes enterprises that are preferably selected according to
rigorous sampling procedures, with the results “grossed up” for the whole
population;

iv.

Stratified random sample: groups population components according to the size of
selected activity so that enterprises within different strata have different
probabilities of selection. Usually, this is a combination of the partial coverage
and random sample options but is more sophisticated and might produce a high
level of coverage while remaining relatively cost-effective.

iv.

Surveys of Persons and Households (chapter 7). Surveys of individuals or households,
with household being defined as a group of persons who share the same living
accommodation, who pool some, or all, of their income and wealth and who consume
certain types of goods and services collectively, mainly housing and food.43

International Transactions Reporting System (chapter 8). A system of collecting data of
individual international settlements and/or transactions from banks, enterprises and
individuals. In most countries that maintain ITRS, the reporting is mandatory and
settlement data have been collected.

vi.

Administrative records (chapter 9). Administrative records are compiled for regulatory
purposes or to support and document the administration of various government
programmes (such as immigration regulation, social security benefits, education, and
health).

vii. Other data sources (chapter 10). These sources may include: credit card records,
mobile phone records, records of business associations, financial statements of
companies, reports of chambers of commerce, records of investment promotion
agencies, private databases and data compiled by trading partners. Some of these
sources are part of the body of information referred to as Big Data.
4.4.
Chapters 5-10 describe the roles and characteristics of these data sources (such as their design
and maintenance); how they can be used in the context of data collection of resident/non-resident
trade in services, FATS and modes of supply; their advantage and limitations, when relevant; and
country experiences in using such sources in statistical compilation.

42
43

Guidelines on IES, paragraph 5.69.
SNA 2008, paragraph 24.12.

4.5.
In the case of enterprise surveys, chapter 6 describes specific types of surveys used for
compilation of certain service categories are discussed in detail, including transportation surveys,
manufacturing services, insurance and financial corporations, R&D surveys, surveys of tourism
activities, construction surveys, structural business surveys for FATS collection, foreign direct
investment surveys in the context of FATS (by incorporating key FATS variables), and separate
foreign affiliates surveys. Country experiences are also presented on the use of enterprise surveys to
collect data for modes of supply, manufacturing services, construction, education, computer software
and IT services, and FATS, among others.
4.6.
In the case of surveys of persons and households, chapter 7 highlights specific types of
surveys and how they can be used to collect data on the international supply of services. In particular,
population censuses can be used to gather benchmark information (in particular for compiling
outgoing mode 4 data) or used for household sampling; household and border surveys can be used to
collect data related to international travel, consumption of other services (e.g., internet
purchase/consumption of services, mode 4), and international provision of services of members of
households, in particular for self-employed persons (modes 1 or 4); and labour force surveys can be
used to collect information on the mode 4 variable “number of persons or trips” and on outbound
business travel.
4.7.
Chapter 8 highlights the advantages and limitations associated with using an ITRS to compile
trade in services statistics. It also presents issues to consider regarding ITRS reporting thresholds
(different approaches for dealing with missing values due to ITRS thresholds are discussed in chapter
17 in Part III of this Guide).
4.8.
Administrative records covered include tax records, customs records for estimating value of
trade related transport and insurance services, Immigration records and entry/departure cards, work
permits collected by the immigration authority, and population records. Administrative records or
commercial database on foreign affiliates are also useful for FATS but their forms and questionnaires
cannot be easily designed to provide data useful for statistical and analytical purposes. Thus, they
could rather be used as supplements to the survey on foreign affiliates. The chapter on administrative
records (chapter 9) also covers in detail how such sources can be used to collect data on modes 2 and
4.
4.9.
Chapter 10 on other data sources discusses the potential uses of Big Data; payment card data;
mobile phone data; external party sources; and linked microdata. The chapter also presents country
experiences in exploring the potential uses of such data sources to compile trade in services statistics.
4.10. Compiling statistics by partner country. There are a number of external sources that may also
provide information by partner country to which the compiler will have access, which can in
particular be useful to improve the quality of compiled statistics. These sources include bilateral data
compiled by compilers in other economies that represent the counterparts to transactions of residents
of the compiling economy.44 A similar approach can be adopted for other sets of data covered in this
compilers guide, such as for FATS.
4.11. When other sources are used, the compiler should make every effort to ensure that partner
country information is classified correctly. The compiler must also decide on the principle of
classification and the list of countries or country groupings to be shown. The country classification
should be based on the country of residence of the provider and the recipient of the service.45 The
same country classification should be used across all sources of data collection. If it is not possible to
44
45

Ibid., A5.21.
Ibid., A5.3.

obtain correctly classified data from the source, the compiler should, at least in significant cases,
46
investigate alternative sources to obtain supplementary information. If countries provide and/or
receive services from international organizations a separate partner group – international
organizations – should be used as such organizations are resident in an economic territory of their
own, and not the economy in which they are physically located (see MSITS 2010, paragraph 3.28).
B.

Sources for modes of supply data

4.12. When considering sources for modes of supply data, compilers should carefully evaluate how
residence is defined in their country, be aware of the laws and regulations in place for migration and
trade in services, and assess how this can be used in a statistical context. Mode of supply data for
trade in services can be either collected through a direct reporting (surveys) or be estimated/modelled.
In many cases a combination of the two approaches should be used to obtain the required aggregates.
Using surveys to collect modes of supply data has the advantage that the compiler has a significant
control over the data that can be obtained and that they can be tailored to answer specific policy
questions. Modes of supply data are unlikely to change significantly in the short-term, so these extra
questions may not need to be included each time the survey is run. However, this advantage needs to
be weighed against other considerations, such as cost, respondent burden, and difficulty in
implementing the survey (such as respondents understanding the concepts so that they will report
correctly).
4.13. Sources for mode 2 movements of persons are likely the same as those used for the collection
of travel information, such as household, border, and labour surveys. For incoming mode 2 persons,
specific surveys targeting students, medical personnel and tourists could be used as well as border
surveys. Values calculated from household surveys need to have sufficient metadata accompanying
them as there are likely to be large sample errors associated with such numbers Often the data will
need to be combined with other sources, such as administrative data on border counts or entry and
departure cards, to obtain relevant data on the number of mode 2 movements/persons.
4.14. Regarding mode 4 types of persons, border and household surveys, enterprise surveys, and
potentially inward FATS surveys (if they include variables measuring exports and import of services
of the foreign affiliate) are an important source both for incoming and outgoing employees and of
self-employed providing services under a contract. As for mode 2, the data will most probably need to
be combined, in particular for compiling breakdowns, to obtain relevant data on the number of mode
4 movements/persons, e.g. through the use of data on characteristics sourced from surveys of persons
or households, combined with counts of those crossing borders. Use of data from partner countries
may also be relevant, in particular that it may be easier to capture the relevant information for sending
countries.
4.15. For compilers using an International Transaction Reporting System (ITRS), modelling the
relevant mode of supply may be the only option. Data on Mode 3 (commercial presence) will need to
be collected in addition to existing data on resident/non-resident services trade as this mode of
services supply falls outside the scope of service transactions measured under the Balance of
Payments framework.
C. Comparing data courses
4.16. Chapter 11 provides comparison tables on the advantages and disadvantages of each type of
data source in the context of compiling specific service categories, FATS and the number of mode 2
and mode 4 movements. The chapter addresses the coverage of each data source, the accuracy of
reporting, the timeliness and frequency of the source, the relevance, and the burdens of reporting and
46

Ibid., A5.23.

processing the data involved. In guiding compilers in determining which data sources are most
appropriate to use on a case-by-case basis, the chapter also lays the foundation for the discussion of
integration of data from different sources, which is introduced in Part III: Data Compilation in chapter
13. Integrating data from different sources is advised as the principle way to ensure the production of
more detailed and more comprehensive statistics, as well as to reduce the burden on survey
respondents. Compilers are advised, therefore, to keep this goal in mind when reading Part II.

Chapter 5

Statistical business registers and survey frames

5.1.
Scope. A Statistical Business Register (SBR) is commonly understood as a register of
economic units resident in the national economic territory. An SBR is established and maintained for
statistical purposes; for example it provides a central sampling frame for business surveys and plays a
central role in the integrated collection and compilation of economic statistics, from which detailed
and interlinked indicators can be derived. This chapter focuses on how business registers and survey
frames can be used in the context of data collection of resident/non-resident trade in services, FATS
and modes of supply. It consists of four main parts: a summary of good practices (section A), roles
and characteristics of SBRs and survey frames (section B), a satellite register for trade in services and
use of business registers to identify potential mode 4 self-employed service suppliers (section C), and
some country examples (section D).
A.

Summary of good practices

5.2.
Compilers of data in accordance with the statistical framework for measurement of the
international supply of services are advised to use SBR as the central sample frame for their survey
programme in order to obtain better coverage, harmonisation of surveys, integration of trade in
services statistics with other economic statistics, reduction of costs, reduction of response burden,
prevention of double counting of statistical information, and above all, better quality and more
coherence in official trade in services statistics. A high-quality SBR helps to improve the efficiency
and coherence of the national statistical system.
5.3.
International guidelines for statistical business registers have recently been developed and are
being continuously updated47. Compilers are advised to refer to these guidelines for more detailed
information on setting up and maintaining a business register. This Compilers Guide focuses mainly
on the role of SBR as a central sample frame for surveys related to international trade in services.
5.4.
The maintenance and development of SBR should be well coordinated. It is a good practice to
have a clear agreement on the responsibilities of the agencies involved in the maintenance and
development of the SBR in the national statistical system. It should be noted that in some countries
this process is handled by one dedicated unit in the national statistical system, instead of several units
within the various substantive areas. This Guide advises that an appropriate legal framework is put in
place for enabling statistical authorities to access and use administrative records, especially tax and
social security records, to maintain SBR.
5.5.
It is a good practice that the SBR maintenance process is based for a large part on the use of
administrative sources such as the administrative company register, register of sole proprietors,
register of government units, the VAT register, tax records and records of the social security
administration. To ensure that SBR provides a solid basis for collection of data needed for measuring
the international supply of services, it is further advised to update SBR should be based also on
sources such as the settlements and international payments databases; trade registers; the FDI register;
balance sheet information; and other specific registers available from trade associations or regulatory
bodies.

47
See Guidelines on Integrated Economic Statistics, United Nations publication, ST/ESA/STAT/SER.F/108,
2013, Chapter 5, section C; see also materials avai lable on the website of the UNECE the Task Force
established to draft international guidelines on statistical business registers at:
http://www.unece.org/task_force_business_registers.html

5.6.
It is very important that SBR uses the SNA 2008 definitions for its statistical units. To ensure
more efficient collection of data on the international supply of services by various modes, this Guide
advises to include in SBR additional indicators providing information on international transactions of
the registered entities, as wells as on turnover, economic activity, number of employees, balance sheet
variables, and data on foreign ownership, in order to properly carry out selection and stratification of
the statistical units, sampling and estimation. It is also a good practice to use SBR to identify the
mode 4 self-employed population as merchants and small manufacturers (including self-employed and
one-man companies) should be registered in it.
5.7.
While it is a good practice that the national statistical system aims at maintaining one
multipurpose SBR, in practice, it may be useful to have a tailored-made satellite register consisting of
all enterprises which are engaged in the international supply of services. If established, it is a good
practice that such a satellite register is maintained by a dedicated unit within the agency responsible
for the compilation of trade in services statistics. Such a unit should, in particular, systematically
update this satellite register using the information available in the central SBR and actively participate
in the harmonization of the contents of these two registers.
B.

Statistical Business Registers
B.1.

Roles of the SBR

5.8.
The UN Statistical Commission recommends the SNA 2008 to obtain a comprehensive and
coherent set of statistics of the national economy. Implementation of SNA 2008 begins with setting up
a basic data collection system for all economic activities. This effort goes hand in hand with setting up
a system for integrated economic statistics, which is based for a large part on using an SBR as the
central frame for all business surveys. Therefore, a number of regional and international agencies
48
have been actively engaged in establishing and improving SBRs, especially in Africa , Asia and
Europe. These improvements will benefit the collection of data by means of business survey (and
some administrative data as well) for the compilation of trade in services statistics.
5.9.
The SBR is a means to an end rather than an end in itself. As such, it is a vital component of
an integrated program of economic surveys. The ultimate goal is the production of comprehensive,
coherent, and high-quality economic statistics. Figure 5.1 illustrates the various roles of the SBR. An
important role of the SBR is to maintain and keep track of the changes of statistical units and their
characteristics due to real life events (see Section B.3 for details). This is a continuous process in
which constant modifications of a collection of statistical units occur in time and space. The
frequency of modifications depends on the update strategy of the SBR. In this respect the SBR can be
considered as a kind of ‘Live Register’ in which the composition of units constantly changes over
time49 .

48
Guidelines for Building Statistical Business Registers in Africa: Laying the foundation for the harmon ization
of economic statistics programs, African Development Bank.
http://www.afdb.org/en/knowledge/publications/guidelines-for-building-statistical-business-registers-in-africa/.
49
Further details on the European experience in this respect can be found in the Eurostat manual on business
registers available at: http://ec.europa.eu/eurostat/ramon/statmanuals/files/KS-32-10-216-EN-C-EN.pdf

Figure 5.1 The roles of the Statistical Business Register

B.2.

The SBR as the central sample frame

5.10. The primary benefits of maintaining one sample frame are better coverage, harmonisation of
surveys, integration of survey data, reduction of costs, reduction of response burden, prevention of
double counting of statistical information, and above all better quality and more coherence in official
statistics. Of course, these benefits can only be realized when one central register is used to derive a
sample frame.
5.11. There are three reasons why construction and use of an SBR as the central sample frame is
desirable. First, if survey frames are independently created and maintained, there is no means of
guaranteeing that they are harmonized. As a result there may be unintentional duplication and/or
omission of activities. Second, an SBR enables practical application of standard statistical units and
their classifications, which is a crucial requirement for survey outputs to be integrated. Third, it is
more efficient for a single organizational unit to maintain the SBR as a source of frames for all
business surveys than for each survey team to be independently maintaining its own frame.
5.12. An up-to-date survey frame is required for each repetition of a regularly conducted survey. It
is more efficient to maintain a frame so that it can support the sequence of repetitions of a survey than
it is to create the frame afresh with each repetition. This is particularly true in the case of sub-annual
surveys, where overlap of sampled units from period to period is essential. Survey frame maintenance
is best achieved through the development of a single statistical business register (SBR) and its use as
the source of frames for all business surveys.
5.13. The SBR serves also as the basis for grossing up the results from these surveys to produce
estimates for the entire business population, and is the main source for data on business populations
and their demography. The SBR should preferably cover as much national economic activity as
37

possible. However, the high cost-benefit ratio involved in covering the smallest units means that some
sort of cut-off is usually applied in practice. In addition to its role as sampling frame, a high-quality50
SBR can also improve the efficiency of the national statistical system by coordinating and spreading
the samples of the various statistical surveys, which should help reduce the response burden, and
improve coverage and congruence of the survey results. Finally, a comprehensive and up-to-date SBR
has a central role in achieving integration of economic statistics, and is essential for the full coordination of source data that use the same basic information about business units51.
B.3.

Characteristics of a SBR

5.14. The SBR typically list the economic entities that are of interest for economic statistics.
Economic entities have numerous characteristics, but some of the most important ones include their
classification by (a) institutional sector (as defined in SNA 2008), (b) economic activity (as defined in
ISIC, Rev.4) and (c) location. Most countries provide laws that enable economic entities to define and
register themselves as legal entities - entities that are recognized by law or society, independently of
the persons or institutions that own them. A legal entity can own goods or assets, incur liabilities and
enter into contracts. The legal entity (or unit) always forms, either by itself or sometimes in
combination with other legal units, the basis for the statistical unit52. A statistical unit defined as an
entity about which information is sought and for which statistics are ultimately compiled. It is the unit
that provides the basis for statistical aggregates and to which tabulated data refer.53 These may be
identifiable legal or physical entities or statistical constructs.
5.15. The SNA 2008 provides the standard definitions of the statistical units which are
recommended by the UN Statistical Commission for use in setting up data collection in all economic
areas54 As mentioned before, the MSITS conceptual framework is grounded in SNA 2008. The SNA
2008 definitions of the statistical units should therefore be used for trade in services statistics
purposes as well. Box 5.1 lists and briefly defines main types of statistical units. The statistical units
used in an SBR should be described by at least three sets of variables and characteristics, namely:
i.
Identifier variables: identity number, name, address (including postcode), telephone
and fax numbers, electronic mail address and information to permit electronic collection of data,
value added tax (VAT) registration number or other administrative identity numbers.
ii.
Economic/stratification characteristics: These characteristics are related to activity
classification (principle and secondary), size (e.g. number of persons employed, turnover, value
added) and location variables .e.g. country of global decision centre (UCI), countries where
enterprises or local units are located, information on whether or not the unit is engaged in
international trade.
iii.
Demographic characteristics: recording calendar dates for important events like
commencement of activities, termination of activities, and joining or ceasing to be part of an
enterprise (group) permit an initial demographic analysis of the population of enterprises, local
units and enterprise groups.

Overall quality is not easy to measure, though sev eral of its specific aspects of it can be used as i ndicators,
e.g. coverage, accuracy of the data held, frequency of updates and consistency of processes (see also chapter
20).
51
Guidelines on IES, paragraph 3.29.
52
See International Recommendations for Industrial Statistics 2008 (IRIS 2008), the UN publication
ST/ESA/STAT/SER.M/90 , paragraph 2.15.
53
IRIS 2008, paragraph 2.12
54
In addition to SNA 2008 and IRIS 2008 and ISIC, Rev.4 the compilers may wish to consult the UNSD web –
publication Statistical Units, available at: http://unstats.un.org/unsd/industry/docs/stat_2008_6.pdf.

Box 5.1 Different types of statistical units
The institutional unit is an economic entity that is capable, in its own right, of owning assets, incurring liabilities
and engaging in economic activities and in transactions with other entities. Two main types of units may qualify
as institutional units: persons or groups of persons in the form of households, and legal or social en tities.
An enterprise is an institutional unit in its capacity as a producer of goods and services. An enterprise is an
economic transactor with autonomy in respect of financial and investment decision-making, as well as a uthority
and responsibility for allocating resources for the production of goods and services. Enterprises under the
control of the same owner form an enterprise group to achieve economic advantages such as economies of
scale, control of a wider market and an increase in domestic productivity through more effective busin ess
management.
An establishment is an enterprise or part of an enterprise that is situated in a single location and in which only a
single productive activity is carried out or in which the principal productive activity accounts for most of the
value added. Note that the SNA 2008 also defines in dustries in terms of establishments.
There are two other statistical units that are often referred to in data sources relevant for statistics on the
international supply of services: A kind-of-activity unit is an enterprise or part of an enterprise that engages in
only one kind of productive activity or in which the principal productive activity accounts for most of the value
added; and a local unit which is defined as an enterprise or a part of an enterprise (for example, a workshop,
factory, warehouse, office, mine or depot) which en gages in productive activity at or from one location.
Source: based on the definitions provided in SNA 20 08 and IRIS 2008

B.5.

Creation and maintenance of a statistical business register

5.16. The circumstances and factors, within which SBRs are built, can clearly be vastly different
across countries. The legal frameworks for acquiring data, as well as the access to human, financial
and technical resources will ultimately determine how countries can proceed. There are however some
key themes that recur, including:
i.

The need to build effective partnerships with data suppliers, funding providers and the users
of the SBR by first making sure that the critical role of the SBR in delivering a coherent and
reliable national economic statistics programme is well recognized, and second by putting in
place governance structures and partner engagement mechanisms that are robust;

ii. The need to manage the implementation and operation of the SBR in a manner that allows it
to focus on and achieve its mission-critical purpose, which is to identify the population of
businesses of a country so that they can be surveyed to acquire useful economic data. There
are other secondary yet highly desirable roles that an SBR can fulfil, such as acting as a data
collection management and tracking tool. The original design and implementation plan should
allow for adding in these components, but only once the SBR has fully matured in its role as a
quality statistical frame;
iii. Generally, the approach should be to maintain simplicity to the extent possible. Conceptual
and technical complexities should be added in only when they serve a practical purpose, and
they should never detract the SBR from meeting its larger goals.

5.17. Even though the primary purpose of SBR is to provide a central frame for economic surveys
across statistical programmes the longer-term vision, however, should allow for adding features and
components that will further enhance the SBR’s value-added. For example:

i.
SRB can serve as a source of register-based statistics. This requires seamless
integration of administrative data, adequate quality assurance, raw data treatment and
programming resources;
ii.
A module to track respondents and response burden can be added. This module
requires human IT resources to develop and maintain server and database resources. This
module may or may not be efficient or necessary when using SBR in the early stages, since
population coverage remains limited and resources are directed towards the economic entities
that have the largest impact;
iii.
A receptacle for tracking survey collection outcomes and response rates and a
“survey feedback” mechanism can be established to facilitate analysis of the survey efficiency
and results as well as to update the SBR on a more regular basis.
The IT professionals who will architect the data structures and larger system design will in particular
benefit from having this longer-term vision clearly pre-defined, thereby facilitating the addition of
these modules as the SBR evolves. To re-iterate, the SBR must first and foremost be a solidly reliable
listing of businesses from which statistical surveys can accurately measure the economic trends of a
country. Creating it will be a challenging task, as will keeping it up to date once it is in use. The
challenges entailed by the creation and on-going maintenance will be greatly facilitated by adhering to
the following principles.
5.18. Do not over-extend resources in the early stages by trying to cover all types of business.
While a highly developed SBR may cover a vast amount of the economic population, a new SBR
must focus on covering the population that is both most important and that can be most reliably
captured and reflected. The need to maximize limited human and technological resources, and to use
available funding efficiently, should limit the scope of the initial SBR population. Reflecting the
informal economy, which is highly diversified and for which no administrative data exist, cannot be a
focus of the SBR development project. Typically, for the macroeconomic indicators that are the
driving objective for the statistical programmes to be served, acceptable margins of statistical error
can be obtained by excluding the numerous businesses that are at the smallest end of the size spectrum.
Including the ‘micro’ businesses would add large volumes of records to be maintained, while adding
only very small increments to the statistical aggregates (such as GDP or balance of payments main
components) being produced. This is not to say that having these businesses will not be useful, as they
can inform important policy analysis pertaining to business formation strategies, small-business
financing and other micro-economic issues. This again is a desirable feature that is worth adding, but
only once the core objective of adequately supporting the key indicators produced by the national
accounts and balance of payments has been met.
5.19. Plan for a system that provides both live and snapshot versions of the register. It will likely
be necessary to have two instances of SBR: (1) a ‘live’ version, which can be used to receive updates
so that the updating of records entailed by on-going frame maintenance activities can be instantly
recorded, and (2) a ‘snapshot’ version produced from the live instance on a monthly, quarterly or
annual basis that survey programmes, which can be used to create their particular sample files. The
snapshot also provides a basis for period-to-period comparison of frame quality.
B.6.

Governance of SBR

5.20. Governance and organizational structure is very important — not only for developing the
SBR programme itself, but even more so for the SBR ongoing maintenance and support for its users.
SBR should be, where possible, an independent entity with a dedicated manager. It is a good practice
that the manager’s unit assumes the following responsibilities:
i.
define and document all concepts, in line with international, national, and local
statistical offices standards;
40

ii.

plan and direct the development of SBR system processes and functionalities;

iii.

plan and implement a quality assurance program for the SBR with the goal of
a. assessing the quality and ensuring the frame’s continued integrity
b. defining and producing quality measures for the SBR
c. identifying system improvements, or recommending adjustments to the training
program or procedures, if required;

iv.
profile businesses to delineate those that are larger and more complex, to properly
represent their production output;
v.

ensure that businesses are classified within the proper standard industry classification;

vi.
assign or derive statistical indicators, or create statistical units, from the
administrative registers to create a complete and unduplicated SBR aligned with the needs of
the System of National Accounts and basic economic statistics;
vii.

validate new development strategies, specifications and procedures;

viii.
develop and deliver the courses and material to educate t SBR users such as: profilers,
survey frame specialists, analysts, data coders, staff of survey divisions and collection
units/areas (training could follow a certification process so that those wishing to access the
register must first complete the appropriate level of instruction);
ix.
support the SBR data users e.g., by providing assistance in evaluating their needs for
their survey design;
x.
provide direction and support on legal aspects related to the survey frame such as
access to the SBR and dissemination of the SBR based data;
xi.
maintain a dedicated group tasked with producing files for users and processing all
files related to updating the frame;
5.21. The creation and maintenance of a unified comprehensive statistical business register is a
long-term objective and a challenging task,55 and it is recognized that resources devoted to that
purpose vary between countries. In spite of this, a number of common issues remain that many
national statistical office may encounter in assessing the suitability of a business register for trade in
services statistics purposes.56 For instance, a legal framework should be in place to allow access and
use of these administrative records for the purpose of the business register. In the maintenance process
as many administrative sources as possible should be used to update the register, including, for
example, the administrative company register, register of sole proprietors, register of government
units, register of non-profit units, tax information e.g. corporate tax, value added tax and social
security information.
5.22. Ideally, there should be a unit in the national statistical office responsible for developing and
maintaining the statistical business register. Decentralized systems may want to begin by developing a
system for reconciling the more significant inconsistencies in the data produced by multiple business

Guidelines on IES, paragraph 5.67.
Compilers interested in learning more about the detailed aspects involved in the establishment of an effective
SBR should consult the Eurostat publication Business registers: Recommendations manual. Busines s registers:
Recommendations Manual. http://unstats.un.org/unsd/EconStatKB/Attachment174.aspx.
56

registers, to improve the accuracy of the data through more consistent classification of key enterprises
and the elimination of overlaps and gaps in coverage.
B.7.

Maintenance of SBR

5.23. The statistical business register needs to be continuously updated to reflect changes in the
population of economic entities as accurately as possible. In particular, the number and characteristics
of economic units engaged in trade in services are normally shifting very rapidly, which means that
the SBR data may age fast and soon become useless. The register should be updated at least annually
to record unit creations and deletions, as well as changes in address and stratification variables.
5.24. New enterprises should be recorded in the business register as soon as information about them
is available, preferably before they start trading, so that information about investment in new
buildings and plant can be collected. Changes in the data necessary for the conduct of surveys, such as
addresses of reporting units, should obviously be reflected in the register in the most fast way as
well.57
C.
A satellite register for trade in services and use of business registers to identify potential
mode 4 self-employed service suppliers
C.1.

A Satellite Register for Trade in Services (TIS-R)

5.25. A satellite register of SBR is a register which covers a sub-population of the SBR with
specific characteristics. For instance, a satellite register for trade in services (TIS-R) consists of all
economic units which are engaged in the international supply of services. This sub-population is
identified by linking SBR with data from other sources, which helps the statistician to determine the
target sub-population when indicators are published. Subsequently, survey samples can be drawn
from this satellite register.
5.26. It is a good practice that TIS-R is maintained by a dedicated unit within the agency
responsible for the compilation of trade in services statistics in the country. However, it is strongly
advised that the core information on the economic units in TIS-R is automatically updated from the
central SBR and its content is harmonised with SBR. The harmonization of TIS-R and SBR includes
forwarding the information on the statistical units identified as engaged in the international supply of
services but not yet included in SBR to the SBR managers. For example, TIS-R can include and
provide to SBR information on units, which are identified by the central bank as being active in
resident/non-resident services transactions but not yet registered in SBR. International supply of
services is, in principle, a possible activity for all units in an economy. Therefore, drawing efficient
samples requires including in the sample frame all those units which supply service internationally
and excluding those which are not engaged in this activity. In this context, in practice, it is useful to
have tailored-made satellite register for trade in services.
C.2.
5.27.

Objectives of TIS-R

The objectives of TIS-R can be described as follows:
i.
The TIS-R needs to enumerate all the resident economic units having had in the
recent past international transactions in services. Ideally, the services sub-register should
enable an immediate differentiation of its population by major kinds of services and
distinguish the population of services exporters from that of importers insofar as these
populations may have significantly different features.

Eurostat publication Business registers: Recommendations Manual, paragraph 4.4, page 23.

ii.
The TIS-R needs additional indicators providing information on international
transactions (generally not included in SBR). The SBR is a necessary but not a sufficient
source to determine the TIS population.
iii.
The selection of the population of economic units which are of interest to statistics on
the international supply of services will be an extract of this register. It generally includes all
big players and SMEs selected for participation in a sample survey. So we are distinguishing
three sets of economic units: the overall population of SBR, the subset population of TIS-R
and the sample of TIS-R which is surveyed.
5.28. Variables. Various economic information like turnover, economic activity, number of
employees, balance sheet variables, foreign trade data and data on foreign ownership are necessary
auxiliary variables to carry-out the selection, the stratification of economic units, sampling and
estimation. It is also a good practice to use business registers to identify the self-employed
population that could be potentially supplying their services through mode 4, because merchants
and small manufacturers (including self-employed and one-man companies) should be registered
in them. An indication of cross border trade activities might be also available and should be
reflected.
5.29. Sources. The different sources used to build and update the TIS-R must be considered under
two aspects: the identification of the population and the variables that should be used to identify the
population and to select the enterprises to be surveyed (see the diagram above). Here are some
potential sources to be carefully tested:

The general SBR lists the businesses active in the country, whatever their activity
(domestic and/or international). It may contain relevant information for building and updating
the TIS-R: identification variables, stratification variables relating to the activity and size of
the enterprises, demographic variables and relationships variables (links between units);

ii.

The settlements and international payments databases exist in many countries and
contain the detail that can be derived from an ITRS on the resident units carrying out
international payments through accounts held in resident banks with an item breakdown
(including trade in services);

iii.

The trade registers include a list of resident operators involved in international trade
in goods. Generally a link can be established between trade in goods and trade in services
(e.g., through information on international freight);

iv.

VAT registers are used as a major source for updating other registers existing in the
country. The basic information in the VAT register includes, in general, the variables like
turnover, employment, main activity and total goods and services exports and imports which
may be useful for TIS registers;

Other sources are the register of partial direct reporting companies, FDI register,
balance sheet information, specific registers available from trade associations or regulatory
bodies (insurance companies, postal and telecommunication operators, trusts, securities
dealers, press and other media).
C.3.

Using Business registers to identify potential mode 4 self-employed service suppliers

5.30. In some countries foreign individuals must apply to a supervision office. All new entries and
changes are recorded and could eventually be published nationally. This means that potential mode 4
persons could be observed if this information feeds the business register and subsequently linked to
the trade in services variable. Therefore, demographics of the businesses including in relation to mode
4 can be derived.
43

5.31. In those countries where the compilers have a direct access to the business register or can link
their own BoP register with the business register, the self-employed category could be accessible. For
example, in the EU, very small enterprises should be included in the business registers. Therefore, it is
recommended that this legal type (that is physical persons acting on the basis of small business
authorization) and size class (micro enterprises) be used as a first indication of potential mode 4 selfemployed persons.
5.32. For example, chapter 7 of the Eurostat business registers manual stipulates which legal units
must be recorded in business registers in the EU member states, while chapter 8 provides relevant
information for mode of supply in section 8D (External sourcing of ancillary services) and section 8E
(Activities of workers under exclusive contracts). There are many activities that involve the services
of workers under exclusive contracts: “Commercial representatives, travelling salesmen, insurance
agents. Such workers may or may not be treated as employees of the enterprise, depending on the
nature of the contract binding them to the enterprise”. Therefore, for countries interested in selfemployed mode 4, it is advised to take into account the EU experience and incorporate similar
information in their registers and to enable compilation of supply of services by mode 4 persons.
D.

Country experiences

5.33. This section contains descriptions of three country cases, each of which represents specific
practice adopted by a given country. The experience of Italy shows that a persistent difficulty with the
access to general statistical register and other statistical problems with obtaining a survey frame
relevant for cross-border phenomena may lead to a decision to create a specific, statistical business
register to support the compilation system for the external sector accounts and how it can be done.
The case of Spain demonstrates the complexity of issues which have to be overcome to build a tailormade TIS register when the general statistical business register traditionally used as the statistical
frame for almost all business surveys does not contain variables needed for the direct identification of
exporters and importers of services. The Uruguay experience proves that sustained efforts can result
in developing a register which can provide a basis for compilation of statistics on the international
supply of services under the circumstances the developing countries usually have to face.
D.1.

Country experience: Italy

5.34. In the period 2007-2011, the Bank of Italy modernized its BOP system. The business register
developed by the Bank of Italy plays a central role in the new system. It is used particularly to
facilitate direct reporting via surveys covering different parts of the BOP and the IIP. Although the
use of a single national business register across all domains of economic statistics (not just BOP)
would be ideal, the Bank of Italy had to create its own statistical business because, despite intense
statistical cooperation between the national statistical institute (ISTAT) and the central bank, due to
legislative constraints, the Bank of Italy cannot access the national statistical institute’s (ISTAT)
business register.58
5.35. The Bank of Italy register contains information for the entire corporate sector (about 1.5
million units), with the exception of small one-man-owned firms (individual firms). The coverage of
the register in terms of turnover is near 90 per cent, and even higher for cross-border operations.59
Given the tight deadline for the production of BOP data, timeliness is a particularly critical factor for

At the time of writing of this Guide, the two institutions are working together to overcome the mentioned
legal obstacles.
59
In terms of number of enterprises the coverage of the BoI business register is about 33 per cent. The
apparently low percentage is explained by the fact that the numerous “individual firms” are excluded. In other
words, the true population frame of the register is actually the corporate sector, for which the coverage is
complete (100 per cent). In any case the cross-bord er transactions of individual firms are covered through tax
authority’s data; their cross-border transactions r epresent about 5% of the total in the “other services” aggregate.

the Bank of Italy’s business register. The register is therefore updated daily, with online data
transmissions. This allows to timely account for changes in the population of firms, which in some
cases may significantly affect the external sector statistics (especially FDI). An always up-to-date
business register also contributes to the smooth running of back-office activities (e.g., help desk,
formal communications to enterprises and sanction procedures).
5.36. Enterprises that engaged in cross-border transactions are flagged in the register; based on
information from quarterly bank reports on firms that carried out cross-border settlements. This
integration of sources significantly improves the efficiency of sample selection. The Bank of Italy’s
business register is built by aggregating various source data (mostly received from external data
providers) and converting them into normalised tables. The latter are integrated, forming a relational
database, through a “bridge table” which contains all the identification keys used by the various data
providers. The relational database consists of three tables: a structural data table, which contains
information such as firm’s name and address; a quantitative data table, with balance sheet data and
the flag for the presence of cross-border payments; and an FDI table, concerning enterprise ownership
features.
5.37. The final population list used for direct reporting purposes is extracted from the business
register according to well-defined criteria. The complete population list is usually too large to be used
efficiently in a survey strategy. Therefore, a threshold is applied, according to which only the firms
with total assets or turnover greater than 1 million of euro are included in the subpopulation. In Italy’s
direct reporting system, the business register not only allows for the identification of the target
population, but it also contributes auxiliary information and is used to reduce the population (with a
cut-off strategy), to stratify the population in homogeneous clusters and to derive model-based
estimators directly in the estimation process.
D.2.

Country experience: Spain

5.38. Since 2005, the National Institute of Statistics (INE) has been entrusted to conduct the
International Trade in Services Survey (ITSS), with the aim of helping the Bank of Spain (BE) to
estimate Other Services item of the balance of payments. The ITSS is sent out to enterprises sampled
from the business registers of various institutions because the INE’s Central Business Register does
not contain a variable to identify exporters and importers of services. Populations that are considered
are obtained from different registers, including regular declarants of foreign services transactions
enterprises from the monthly VAT records of Large Companies, and a sample of enterprises (with 10
employees or more) from the Central Business Register (DIRCE); care is taken to exclude double
counting of the same units. More detail on these populations is available in the on-line version of the
guide.
5.39. The results obtained by the ITS survey between 2005 and 2012 showed significant differences
in level (but not in trend) from those obtained by the Bank of Spain settlements system. From 2013
onwards, the populations and statistical frameworks used in the surveys remain very diverse. The
intention is, however, to reduce to the maximum the number of survey respondents who may state in
the questionnaire that they do not carry out such transactions. In the new sampling design, the Central
Business Register (DIRCE) does not form a statistical frame by itself. Instead, its main function is to
characterize the companies of the different populations by main economic activity and size, to enable
to sub-stratify each stratum by these two variables.
5.40. FATS. The set of INWARD FATS statistics is compiled from the INE structural business
surveys, the majority of which use the DIRCE. DIRCE records the "group of enterprises" as a
statistical unit. The system is updated on a yearly basis in a way similar to the other statistical units.
The maintenance processes affect both the basic features of the groups and the units forming them. In
this process access to a complete set of both administrative (corporation tax groups) and private (Dun
& Bradstreet database or INFORMA/IBERINFORM shareholder database) is absolutely crucial.
45

5.41. For each DIRCE company, whether it operates independently or under a group of companies
and the nationality are available variables, which identify the INWARD FATS population. The set of
OUTWARD FATS statistics is obtained directly from an administrative register from the Ministry of
the Economy and Competitiveness called the Foreign Investments Register (RIE).The main variables
include: identification of the company, its legal status, its main business activity, number of
employees, turnover, and other economic variables, % of foreign participation and country, and % of
Spanish participation, among others.
D.3.

Country experience: Uruguay

5.42. The National Statistics Institute (INE Uruguay) is in charge of the creation and maintenance
of the national business register. Information sources used for the maintenance of the business are
divided into administrative sources (such as the tax authority, social security agencies, internal
economic surveys and information from other ministries) and other sources (including phone
directories, web pages, members of commercial chambers, and press releases and other information
on publicly traded companies). The enterprises directory was created in its initial version after the
economic census of 1996. In 2007, a new agreement between INE, the tax authority (DGI) and the
Social Security Agency (BPS) resulted in a new version of the business register with these institutions
being main source data providers. In addition, the Ministry of Tourism provides data on hotel
occupation. Information from companies operating under the free zone regime is obtained through a
census of these companies.
5.43. In this new version of the register, INE includes information on foreign ownership and on
involvement in trade (in services), following the guidelines designed in the Inter-American
Development Bank (IADB) project for Latin American Business Register and the Manual on
Statistics of International Trade in Services 2010. The business register in Uruguay will serve as the
main tool for detecting enterprises engaged in international trade in services. As part of its new
design, information on main and secondary activities, goods and services produced, and the
destination of the sales (national or foreign and country of destination), the origin of capital (national
or foreign) and foreign ownership will also be included. A new survey will subsequently be designed
to better estimate international trade in services and FAT statistics. More information on the definition
of statistical units used in the Uruguay business register is available online.

Chapter 6

Enterprise and Establishment Surveys

Introduction
6.1.
Scope. This chapter describes enterprise and establishment surveys, which are needed to
collect data for the compilation of statistics on resident/non-resident transactions in services and
FATS. The Chapter promotes an integrated approach to the organization and conduct of the surveys,
keeping the response burden as low as possible and linking trade in services and business statistics.
The Chapter consists of the following sections: General description of enterprise or establishment
surveys (Section A), Surveys of resident/non-resident services transactions (Section B), foreign
affiliate statistics and the international supply of services (Section C), and trade in services surveys
with links to FATS and FDI (Section D).
A.

Summary of good practices

6.2.
Enterprise and establishment surveys can be used to provide data for the full range of
resident/non-resident services transactions, as well as for FATS and for additional indicators on the
international supply of services. Such surveys should always be based on clearly defined objectives,
sound sampling framework and collection methodology, and a well-established legal basis. Compilers
are advised to take into account the set of steps for developing and conducting an enterprise survey
described in Box 2.1 of Chapter 2 of the BPM6 Compilation Guide, as well the suggestions for the
creation and updating of the survey frame, and the model survey forms provided in Annex 8 of the
IMF BPM6 Compiler’s Guide. It is a good practice to provide clear and simple
instructions/explanatory notes to respondents on the information to be submitted.
6.3.
Compilers can conduct general enterprise or establishment surveys, which usually cover most
or all services that can potentially be delivered or consumed by enterprises, but it is a good practice to
consider the need for, and use of, focused surveys for specific service categories, such as
manufacturing services, transportation, R&D and finance and insurance. Such focused surveys allow
the compiler to capture a great level of information detail for a specific service category and related
transactions without imposing response burden on those enterprises or establishments that are not
involved in a particular service. In such focused surveys, compilers are suggested to ensure that forms
take into account the specificities of individual service categories (such as mode of transport for
transportation services).
6.4.
For FATS compilers it is suggested to collect data via SBS (inward FATS only), FDI or
designated FATS surveys, weighing their respective costs and benefits as described in chapter 15, and
to include in their survey forms at least the main FATS variables of interest as identified in the
MSITS 2010 (including revenues or turnover, employment and value added). Given that output is a
preferred measure to sales or turnover, it is also important to consider that relevant information is
collected to enable the compilation of this item. Compilers should ensure that the data sources enable
as a minimum the relevant breakdowns of information by (services) activities and country of origin or
destination of control. In addition compilers should consider for revenues (or output) that services
products (at least a total) and destination of sales in the country of establishment of the affiliate can be
separately identified. For those compilers for whom it is difficult to implement the collection of all
these details when starting the collection of FATS, it is important to establish a gradual approach, first
focusing on the information of primary interest to the economy.
6.5.
Compilers should consider that for certain services sectors, it may be logical to use a single
survey to respond to various information needs (e.g. trade in services and FDI/FATS). This could be
applicable in particular for construction activities.

General purpose and description of enterprise or establishment surveys

6.6.
Business surveys can be conducted at the establishment level or the enterprise level and can
provide coverage across the full range of services. They have proven successful for the collection of
trade in services data and FATS in many countries. Detailed descriptions of the types of surveys, their
design, sampling techniques as well as related data editing and data compilation procedures are
described in a number of publications 60 , most notably Chapter 2 and Chapter 3 of the BPM6
Compilation Guide which compilers are advised to consult when developing enterprise or
establishment surveys for trade in services and BOP purposes61 . Compilers also are advised to make
sure that the national definitions of the statistical unit (see chapter 5) comply with the standard
definitions and document any deviations in metadata.
6.7.
One of the first decisions to be made in collecting data is whether to undertake a census or to
compile data from a sample survey that balances data quality with other considerations such as
reporting burden (see Chapter 11 for the comparison of data sources). In determining the reporting
population, various approaches are possible, including a census, a partial coverage collection survey, a
random sample survey and a stratified random sample survey. In practice, compilers in many
countries use a combination of two or three approaches when collecting data from enterprises,
benefiting from their respective different advantages. 62
6.8.
Surveys should always be based on clearly defined objectives, sound collection methodology,
and a well-established legal basis. Properly designed collection forms, full coverage of the population,
well-defined data structures and classifications, and effective data validation and aggregation
procedures are also required.63 The principles of survey design include:
i.
ii.
iii.
iv.
v.
vi.

Specify the objectives and coverage;
Establish the sampling unit and the information to be collected;
Determining the appropriate sample size, if a sample will be used;
Developing a sampling frame, i.e. an exhaustive list from which sampled units are selected;
Developing the sample design, i.e. how the sample is selected from the frame;
Determining the method of collection (paper form, electronic, interview, etc.).

6.9.
Sampling techniques and contacts with respondents. Compilers can choose from a wide
variety of sampling techniques, generally either probabilistic or non-probabilistic. Under probability
sampling, every unit in a population has a calculable probability of being selected in the sample. This
approach is objective and defensible. There is a theoretical basis for the process of extending the
sample results back to the population. Under this approach, estimates of sampling error can be
calculated and inferential statistics can be derived. Non-probability sampling methods should be used
with caution because there is no way to measure their precision. The only way to address the quality
of the survey data produced under non-probability sampling is to compare the results of the survey to
some known information about the population. A frequently used non-probability sampling method is
cut-off sampling (see more detail in box 6.1).

For example in the Eurostat Handbook on the design and implementation of business surveys, available at :
http://epp.eurostat.ec.europa.eu/cache/ITY_OFFPUB/CA-09-97-818/EN/CA-09-97-818-EN.PDF and Eurostat
publication Survey sampling reference guidelines: Introduction to sample design and estimation techniques
(2009), available at: http://ec.europa.eu/eurostat/ramon/statmanuals/files/KS-RA-08-003-EN.pdf. Many
countries published their guidelines as well.
61
Draft BPM6 Compilation Guide (BPM6 CG).
62
See BPM6 CG (p.13-14) for a more detailed explanat ion about survey forms and their advantages and
disadvantages.
63
BPM6 Compilation Guide, Chapter 2, paragraph 2.3.

6.10. Whether selecting a sample using probability or non-probability techniques, compilers must
define the universe (population) from which they will sample, that is, in practical terms, construct a
sampling frame. In most countries it is possible to define the population using various lists of
enterprises (business registers), compiled for administrative purposes. For more information on
sampling techniques, sample frame, sample structure and sample allocation (how to allocate the data
collection between the strata), compilers are suggested to consult the Producer Price Index Manual
(Chapter 5)64. While the Manual focuses on price collection, much of the information can be applied
more generally to enterprise or establishment surveys.65
6.11. The collaboration with respondents to surveys is essential for the production of good quality
statistics, as explained in more detail in the BPM6 Compilation Guide (Chapter 2). Compilers are
suggested for example to hold consultation meetings to make respondents aware of the purpose of the
survey and to help the statistical agency design the survey.66
Box 6.1 Cut-off sampling
Cut-off sampling is a strategy frequently used by c ountries to select samples. In this approach, a pre determined
threshold is established with all units at or above the threshold included in the sample (selected wit h certainty)
and units below the threshold level not included (zero probability of selection). Cut-off sampling is a common
technique in economic surveys in cases when most economic activity is generated by the largest firms . It can be
used to reduce the number of firms required to report in non-benchmark years, thus reducing the burden on
excluded firms and reducing processing costs.
Cut-off sampling generally results in a high degree of coverage among a small number of prospective un its. This
occurs because the distribution of the selection variable (for example, production or sales) is concen trated in a
small number of large establishments. Limiting the target population enables the exclusion of small e nterprises
and/or enterprises that are not primarily engaged in providing services. This method is resource effic ient and
ensures large firms are included. However, there i s no way to determine if firms that fall below the cut-off
behave in similar ways to the firms in the survey.
Sample frame cut-offs are revaluated regularly. If benchmark surveys are conducted, cut-offs are typically left
unchanged between benchmark surveys and are often increased at each benchmark, for a number of reasons .
First, since cut-offs are set in nominal terms, inf lation will eventually cause enterprises that did not grow in real
terms to exceed cut-offs. Therefore, small enterpri ses may face a higher reporting burden due simply t o price
changes. Second, industry consolidation may increase the share of economic activity originating from t he largest
firms. Therefore, it may be possible to increase th resholds without significant loss of coverage.
Compilers must consider resources when using cut-off sampling. Lower cut-offs reduce the amount of data that
needs to be estimated. Since more enterprises are r equired to respond, a greater share of the estimates are actual
data from reports. Estimates with lower cut-offs ar e likely to be more precise. However, lower cut-offs also
impose a number of costs on respondents as well as the compiling agency. Requiring more firms to respond
imposes a burden on respondents since firm employees must devote time and effort to gathering information and
filling out forms (and these costs are proportionally higher at small firms compared to large firms). Once a
report is received, it requires substantial effort by editors to process. Processing more reports is m ore costly and
slows the issuance of data. Cut-off sampling significantly reduces the costs of a survey by reducing the number
of both reports and items filled out and processed.

https://www.imf.org/external/pubs/ft/ppi/2010/manual/ppi.pdf
Likewise, the Eurostat Survey Sampling Reference G uidelines provide more detailed information on this
topic.
66
BPM6 Compilation Guide, paragraph 2.29.
65

Enterprise and establishment surveys for trade in services statistics purposes

6.12. Enterprise or establishment surveys are often used to collect data needed for the compilation
of resident/non-resident trade in services statistics. They may take various forms, such as an extensive
survey covering many or all services (often excluding most of the data needed to compile transport,
travel and government goods services n.i.e.), or a specialised survey which may be developed for
various reasons, e.g. to avoid unnecessary burden on firms that are not likely engaged in trading some
specific services, or for sectors that are important enough to develop a dedicated survey for e.g.
manufacturing services, business process outsourcing services. Existing enterprise surveys such as
structural business surveys or research and development surveys may also be used to collect some
trade in services data.
6.13. Surveys that collect data on trade in services covers the value of exports (credits) and imports
(debits) of services, broken down by type of service (at least for the services sectors of primary
interest to the compiling economy), with an identification of the country of the counterpart in the
transaction (i.e. partner country). Other recommended elements for data collection of the value of
exports and imports of services cover the identification of trade between related parties (intra-firm
trade) and an identification of the way these services were traded (mode of supply).
6.14. Various options exist for collecting data according to these dimensions via surveys.. Given
the complexities surrounding trade in services, surveys should always be accompanied by explanatory
notes to help respondents understand the type of information required. The frequency of the data
collection (and associated levels of details) will depend on the needs identified before the data
collection is established (see chapter 3 on institutional arrangements). While this chapter covers the
aspects more directly related to trade in services data collection, readers can also refer to the BPM6
Compilation Guide for further information. In particular the model survey forms shown in Annex 8
provide useful guidance as to the type of information and questions to be included in trade in services
survey forms. The model forms of interest include: International Trade in Services, Manufacturing
services, Resident Transport Operators, Transactions with Nonresident Transport Operators,
Construction, International Insurance Transactions and International Pension Services.
6.15. In all surveys, questions should be clearly phrased, using wording that businesses understand,
while ensuring that these will fit the EBOPS 2010 definitions, including complementary groupings,
that the data can be used in the subsequent compilation of trade in services statistics. In some
instances, such as for insurance or financial services, the information to be collected cannot be the
value of the service rendered itself but will have to be deducted from data reported in the survey
forms (often derived from some accounting records e.g. insurance premiums and claims). This would
also apply if for some services categories a more detailed list than EBOPS 2010 is included in
questionnaires.67 When elaborating survey forms and notes, this guide strongly encourages compilers
to refer to the CPC to define the service items in terms of levels of details or for the explanatory notes.
A correspondence table between EBOPS 2010 and CPC Version 2 is provided in an on-line annex to
MSITS 2010.68
6.16. This section first presents general guidance on the use of generic trade in services surveys
(B.1), and then describes the use of some specific surveys, namely, transportation surveys (B.2),
manufacturing services surveys (B.3), surveys to insurance and financial corporations (B.4), R&D and
innovation surveys (B.5), Surveys of tourism related activities (B.6), construction surveys (B.7), and
other specific surveys (B.8) that can be used to also derive trade in services information.

67
68

This would be particularly relevant for specific services surveys.
http://unstats.un.org/unsd/tradeserv/TFSITS/msits2010/ebops2cpc.htm.

C.1 Generic trade in services surveys
6.17. A generic trade in services survey is a survey that usually covers most or all services that can
potentially be delivered or consumed by enterprises. It usually does not cover the data on services
transactions that correspond to imports of services by households or individuals (e.g. travel related
transactions), or some specialised services where the number of transactors may be more limited (e.g.
exports of transport services, construction) and a more targeted survey may be more efficient. A
generic trade in services survey is a survey of enterprises with potential relevant balance of payments
transactions with non-resident households, corporations, and governments. Enterprise trade in services
surveys have traditionally been used to capture data used for the compilation of the balance of
payments and IIP,69 but they should to the extent possible also serve the more detailed needs as
identified in MSITS 2010. In particular it can be used to collect data by service type, by counterpart
country as well as by relation between the parties (affiliated and unaffiliated trade). They can also be
useful to identify the mode of supply in resident/non-resident transactions, to develop modes of
supply estimates.
6.18. The organization and conduct of generic trade in services surveys is a complex task which
countries approach depending on their needs and circumstances. Some countries had to develop the
survey systems in view of lack of necessary information from banks’ records on payment transactions
(settlement system). The experiences of France and Austria, which are provided below, provide
valuable lessons for compilers which might decide to develop new or improve existing surveys for
collecting trade in services data. It is a preferred practice for compilers to take into account the set of
steps for developing and conducting an enterprise survey described in Box 2.1 of Chapter 2 of the
BPM6 Compilation Guide, as well the suggestions for the creation and updating of the survey frame.70
A model survey questionnaire is available in Annex 8 of the IMF BPM6 Compiler’s Guide (Form 6).
To respond to the trade in service information needs such a model questionnaire will have to be
amended with some additional breakdowns/questions. Box 6.2 illustrates how such additional
information could be collected. The different dimensions, content and design (as well as formulation
of questions) of the survey forms will have to be chosen according to the needs identified.
Box 6.2 Sample questions for collecting trade in legal services data broken down by partner,
affiliated/non-affiliated trade and mode of supply
Country
A
Legal services
- Of which to related parties
- How were your legal services delivered (% of tota l)?
From your territory to a non-resident overseas
To a non-resident customer temporarily in your territory
Delivered to non-resident customer by a resident
employee, temporarily working abroad

%
%
%

Revenue for year __
Country
Country
B
C

%
%
%

Other
countries

%
%
%

6.19. Compilers looking to incorporate mode-of-supply data in generic trade in services surveys are
suggested to first consider what the costs, burden and priorities are for such a breakdown as compared
to other needs (e.g., partner country, etc.), and if this is needed for all services items or for a selection
of these. Decisions could be made using a step-by-step approach, e.g. in light of policy needs, priority
may be given to breaking down exports of services by mode of supply (for all items, or in first
instance for a selection of aggregates); imports of services by mode might be done in the second stage
of the survey development. Alternatively some simplification rules could be used such as described

69
70

BPM6 Compilation Guide, paragraph 2.2.
Ibid., page 11.

in chapter 5 of MSITS 2010, i.e. a service transaction can be associated to one or two dominant modes.
Only the identification of one (or two) mode(s) may therefore be requested, assuming that the residual
may be derived from the information provided.71 Another option could be to simply ask respondents
to identify one dominant mode in a service transaction (e.g. by ticking a box).
6.20. It is good practice to provide clear and simple instructions/explanatory notes to respondents
on the information to be submitted, in particular on how to determine the modes of supply if relevant
questions are included in the forms. In electronic surveys, prompts and restrictions can be used to help
improve the accuracy of the reported data. For example, asking the respondent to confirm their answer
if they enter a mode of supply that is unlikely to be associated with a particular service type. It should
be considered that respondents are generally more comfortable to estimate proportions of the value
already provided, than giving exact monetary values by mode of supply.
6.21. While processing survey results, imputations should be part of the work. Compilers can use
table 5.2 of MSITS 2010 as the basis (for lower value answers), or by contacting respondents directly
for clarifications (for higher value answers). Experience shows that respondents are very likely to be
able to provide mode of supply information when contacted directly.
6.22. It is also important to note that trade in services enterprise surveys can be the source for mode
4 (contractual service supply) and quantitative indicators (i.e., number of persons or trips). Firms
normally keep staff records (although not necessarily in the same department as the ones that report
data on transactions; i.e., personnel versus accounting departments), which may include information
on the type of work performed and/or whether the staff member received special compensation for
working abroad. Such information could serve the needs for mode 4. Within enterprise surveys, and if
deemed relevant, one could also consider including explicitly mode 4-related questions along the lines
proposed in the trade in services surveys, or other questions designed to identify mode 4 activities.
C.2.

Country experiences 72

C.2.a. Country experience: Austria
6.23. In Austria, the Oesterreichische Nationalbank (OeNB) is responsible for compiling and
disseminating information on trade in services between residents and non-residents. Since 2006, trade
in services statistics have been compiled by taking samples from enterprises, institutional investors
and banks (not including households), instead of the ITRS system that was previously used. To keep
the costs of data collection low and make utmost use of existing data, the OeNB works in close
cooperation with Statistics Austria such that the OeNB concentrates on the financial corporations
while Statistics Austria focuses on the real economic sector.
6.24. On behalf of the OeNB, Statistics Austria compiles quarterly data on service exports and
imports from non-financial corporations. The survey covers information on individual partner
countries and all business activities according EBOPS 2010, except for travel which is measured
separately. The survey is also supplemented with information obtained from public institutions (on
cross-border government services), and from 120 non-profit organizations (on e.g. international aid
services such as education and health services).
6.25. The survey collects data from a stratified sample within the scope of the Structural Business
Survey (SBS). The sample is drawn from the total population of enterprises engaged in trade in
services. To identify this population, Statistics Austria conducts a survey on services exports and
imports in the enterprise sector every five years, and complements this information with former
71

For example a systematic identification of the pro portion of mode 4, assuming that the residual will be either
be mode 1 or mode 2 depending on the type of service considered.
72
Extended country examples are available on the UNSD Compiler’s Guide website.

settlement data (for unit-non-response) and estimates of values below survey thresholds. Since OeNB
decided that the regular survey should cover at least 90% of service exports and imports in every twodigit NACE division, approximately 4,800 non-financial enterprises were selected (reflecting an
initial reporting threshold of EUR 200,000 for both annual services exports and imports). From 2013,
the reporting threshold was raised to EUR 500,000 after administrative data sources have become
available within the European Union (the Value added Tax Information Exchange System).
6.26. The OeNB conducts the surveys in the financial sector. The report on service exports and
imports by Austrian banks as well as imports by insurance companies mirrors the enterprise report run
by Statistics Austria and covers all EBOPS 2010 items. In addition, the OeNB makes use of
administrative data from the Financial Market Authority (FMA) on insurance service exports in
particular. Quarterly data are reported to the OeNB on premiums and claims from insurance service
exports, whereas yearly data include financial claims and liabilities from insurance transactions and
insurance technical reserves. FMA also gathers mirror data from other EU countries on insurance
service imports in Austria. These data become available only with some delay, but they are especially
important for the calculation of life insurance imports.
C.2.b. Country experience: New Zealand - Collecting data on modes of supply
6.27. Most of New Zealand’s trade in services data comes from surveys. Modes of supply data (for
exports) was collected directly through the 2011 Census of International Trade in Services and
Royalties, which collects data on 55 different service types excluding transportation, travel, insurance,
and government services. Respondents were asked to estimate the proportion of their services exports
that were delivered through mode 1, mode 2, or mode 4. A detailed guide accompanied the Census to
provide further help, along with examples.
6.28. Respondents were asked to report their services exports in the upper sections of the survey,
and then asked to estimate the percentage of the value that was delivered through mode 1, mode 2,
and mode 4. It was felt that the respondents would be more comfortable estimating proportions of the
value already provided, rather than giving exact monetary values by mode of supply – fewer
respondents should leave this section blank this way. Testing showed that most respondents
understood the modes of supply concepts, although feedback suggested the guide was helpful as well.
6.29. For policy reasons, New Zealand’s Ministry of Foreign Affairs and Trade was mainly
interested in exports of services, which is why only exports data was collected for modes of supply.
However, modes of supply information could be compiled for imports of services in the same way.
C.3.

Transportation surveys

6.30. Enterprise and establishment surveys of resident and non-resident carriers are important
sources of transport services data. Often the agents of the operators will have to be approached to
collect relevant data, in particular for imports of transport services. When considering data collection
it is important to note that MSITS 2010 follows BPM6 in recommending a cross-classification of
transport services by mode of transport and by kind of service. There are different modes of transport
to consider such as maritime/sea, air, rail, road, or internal waterway transport. The types of services
covered are the transport of passengers, freight and other supporting and auxiliary transport services.
Some modes of transport, given the smaller number of operators or agents, may be easier to survey
(air, maritime, rail or inland waterway transport) than others (road). In addition, given their specific
nature, special forms or enquiries may need to be designed for space or pipeline transport, as well as
for electricity transmission. For these three sectors, the number of entities to which the surveys should
be addressed is also often limited.
6.31. An important point to consider is that operators, which are the providers of the transport
service, in many cases differ from the owner of the transport equipment. In addition, when defining
the residence of the operator, the country of registration of ships and aircraft may also differ from that
53

of the operator (or the owner). Charters or vessels without crew (i.e. operational leasing services) also
need to be differentiated from charters of vessels with crew (i.e. transport services). Such
considerations render the collection of data complex, but clear notes referring to relevant information
can be collected. Specific surveys should be sent to the operators (i.e. as the providers/exporters of the
transport service) of each category of modes of transport and if relevant with a distinction between
transport of passengers and goods. Exports of such services are identified in resident operators’
activities, while imports are identified in non-residents’ operators' activities. Compilers may face
difficulties in getting a representative coverage of the non-resident carriers’ activities. However, when
transport is regulated in a country (air transportation being the best example), non-resident carriers
have to be registered to operate and they typically establish branches or agents in that country. Such
entities in principle know or have access to the information necessary to the compilers. Thus,
compilers could send survey questionnaires to those branches/agents to collect information on imports
of services. Transport surveys often are used to also gather related information such as on auxiliary
and supporting services, leasing services or on fuels and provisions provided to carriers (goods). The
BPM6 Compilation guide proposes model forms 8 and 9 (Appendix 8) that could be used as a guide
for designing survey forms related to transport services. See BPM6 Compilation Guide paragraphs
3.21-3.52 for more information. In designing the survey forms it will also be necessary to factor in the
country of residence of clients (for exports) and of operators (for imports), as well as other dimensions
if identified as necessary.
6.32. Many users, in particular for GATS purposes, would need information on the real value of
international freight transport services contracts (i.e. with no adjustment for either the FOB/FOB
valuation of goods in the BOP, or the FOB/CIF valuation in merchandise trade statistics). This is in
fact the data that are collected in transport surveys. Compilers are encouraged to use the raw data that
may be collected from freight transportation companies (or their agents) to make publicly available
some supplementary estimates of the "real" trade in freight transport services taking place between
residents and non-residents.73
C.3.a. Country experience: Chile
6.33. Data on international transport services are collected by the Central Bank of Chile (CBCH)
through quarterly questionnaires to resident sea and air international transport companies and to the
agencies or representatives of foreign companies in Chile. The transportation questionnaires are
prescribed by the “Compendium of International Exchange Regulations (CIER).” The response to the
transportation questionnaires is mandatory and data are directly reported in US dollars.74
6.34. Both the questionnaires addressed to Chilean resident companies and those addressed to
agencies or representatives of foreign companies in Chile request information on all acts or
conventions that create, modify or extinguish obligations payable in foreign currency. In general,
transport services accompanying settlements in foreign currency are distinguishable in resident
companies’ commercial accounting as it distinguishes revenues and expenses in foreign currency
from those in national currency for income statement purposes companies. In the case of agencies or
representatives of foreign companies, their purchases’ books or expenses’ books distinguish transport
services accompanying settlements in foreign currency.
6.35. The first block of information in the questionnaire includes general backgrounds which are
used to identify the enterprises, the reporting period, and the name of the responsible of the
information (to be able to consult respondents about the reported data). The second block inquires

See Chapter 14 and MSITS 2010, paragraphs 3.107-3. 110.
The instructions to complete the questionnaires an d forms are set out in the Chapter VIII, annex 1.1 and 1.2 of
the CIER and are available on the CBCH website: http://www.bcentral.cl/normativa/cambiointernacional/manual-procedimiento/pdf/capviiianexo1.pdf.
74

about incomes and expenses related to international transport business. Most of the information
collected in this section is used in estimating transport services, but is also used for the estimation of
certain other services categories. For example, rentals of transportation equipment without crew or
major repairs abroad are used to estimate “operational leasing services” in “other business services”
and “repair of goods”, respectively (see details in the elaborated example online).
C.4. Manufacturing services surveys
6.36. Manufacturing services on physical inputs owned by others (or manufacturing services in
short) include activities such as processing, assembly, labelling and packaging, undertaken by
enterprises without ownership of the goods. The BPM6 Compilation Guide model form 7 provides a
sample questionnaire on manufacturing services.
6.37. If imports of manufacturing services are important for the compiling economy, a survey can
be used to collect data from the owners of the goods on the processing fee they paid to non-resident
processors (which may incur cost of materials procured by the processors for use in the production).
In addition, data on the costs and origins of materials supplied by the resident enterprises (owners) to
the non-resident enterprises (processors) for outward processing should also be collected. This covers
both cases where owners’ materials are sent to the processor’s economy directly from a third economy
and from the owners’ home economy. It may be difficult to identify the population involved in such
activities. To identify enterprises possibly involved in processing activities, exports declarations
(when sending goods for processing) or import declarations (when receiving processed goods) from
the customs authorities could be used. The identified enterprises can then be surveyed to collect
various data on the fee and the goods involved in outward processing.
6.38. The enquiries also need to identify the cases where processed goods do not enter or return the
economy of the owner (i.e. they are sold in the economy of the processor or to a resident of a third
economy). If relevant, the survey could go beyond manufacturing services and also cover merchanting
as well as other trading type activities (where there may be some processing component or not). The
sales values and various cost components of the goods for different types of global production
arrangements could therefore be captured for all different cases, including when such transactions
with non-residents are completed offshore.75
6.39. For exports of manufacturing services on goods owned by non-residents, the data collection
has to cover at least the processing fee as well as the identification of the country of the client, as well
as the costs and origins of materials procured by the resident processors. If it is available, information
on the destination of the processed goods would be useful. In principle a survey of processors should
be easier to establish as (i) the number of service providers should be small and (ii) many of these are
often located in special zones such as export processing zones, and are generally subject to special tax
regimes. The survey will have to distinguish the rendering of manufacturing services to non-residents
from manufacturing on the processors’ own account and that for other residents.
C.4.a

Country experience: Hong Kong, China

6.40. Hong Kong’s outward processing activities (processing of goods, which are owned by the
Hong Kong residents, by the non-residents) are substantial and are predominantly carried out in
Mainland China. The goods processed in Mainland China may be sold in Hong Kong but are also
frequently sold directly to other markets. To support the collection of data on manufacturing services
and related activities, the monthly survey on trade involving outward processing in Mainland China
(OPS) and the annual survey on imports and exports of services (ASIES) have been enhanced and a

75
For a more elaborate discussion and guidance, see the UNECE Guide on Global Production and the handbook
on the Impact of Globalization in the National Acco unts.

new quarterly survey on merchanting and other trading activities (QSMTA) has been launched. Two
main cases are identified:
Case I: Outward processing in Mainland China with processed goods returned to Hong Kong
6.41. The OPS survey uses the import/export declarations of Hong Kong’s trade with Mainland
China as the unit of observation. The declarations are stratified by trade types (including domestic
exports and re-exports to Mainland China, and imports from Mainland China) and by commodity
groups. For cases pertaining to imports (from customs-based perspective) from Mainland China which
involve outward processing of goods under “processing and assembling” arrangement, data are
collected on a) the value of raw materials/semi-manufactures sent from Hong Kong to Mainland
China; b) the value and origin of raw materials/semi-manufactures purchased and delivered directly
from places other than Hong Kong to Mainland China; and c) the value of manufacturing services on
goods owned by Hong Kong, distinguishing between the processing fees paid by Hong Kong; and the
value of raw materials/semi-manufactures procured directly by Mainland China processors.
6.42. A significant proportion of OPS respondents are logistics companies (rather than the owners
of imported processed goods) which generally do not have readily available information pertaining to
the additional data required. Therefore, efforts have been made in establishing rapport with major
logistics companies, so that these companies can now either acquire the relevant information from the
goods owners themselves or provide the contact details of the goods owners for subsequent follow-up.
This practice has successfully reduced the non-response rate.
Case II: Offshore trade activities involving outward processing with processed goods sold offshore
6.43. Some of the offshore trade activities involve outward processing in which goods owned by
Hong Kong companies are processed in Mainland China (or elsewhere) before they are sold offshore
directly to non-residents. These offshore trade activities cannot be collected from import/export
declarations and are captured through ASIES and QSMTA.
6.44. In ASIES, offshore trade activities involving outward processing are distinguished from
conventional merchanting activities. For manufacturing services and related activities, data are
collected on similar items as in the OPS survey, as well as on the sales value and cost of processed
goods sold offshore by origin, destination and commodity group.
6.45. The values and destinations of exports of manufacturing services on goods possessed by
companies abroad (inward processing of goods by Hong Kong) is also collected in ASIES. The new
QSMTA survey collects similar data at quarterly intervals to support the timely compilation of
relevant macroeconomic aggregates (namely BOP and GDP).
C.5. Surveys of insurance and financial corporations
6.46. Surveys of insurance corporations. The collection and compilation of data relating to
insurance are described in Annex 2 of the BPM6 CG. In order to obtain source data for insurance
services, in particular for exports of such services, surveys on insurance corporations represent the
preferred source. To ensure an appropriate coverage of the domestic insurance sector, a survey frame
should be available including a list of insurance companies. Insurance agents and brokers are usually
required to register with insurance authorities; therefore, a list of these businesses should be readily
available from official sources. Surveys of insurance corporations could also be used to capture
information of exports and imports of reinsurance services.
6.47. The insurance service provided to residents and nonresidents is derived by determining the
output of the insurance in a way that mimics the accounting practices based on premiums earned and
losses incurred pertaining to the accounting period. An extended model form can be found in BPM6
CG Appendix 8 (form 12). For a survey to insurance companies to be useful for compiling TIS
56

statistics, it should include amongst other things questions about premiums written to, premiums
earned from, and claims paid to, non-residents for different insurance categories (i.e. life insurance),
and to specify also reinsurance ceded to and accepted from non-resident issuers (and vice-versa).
Information on technical reserves dues to non-resident policyholders is also necessary.
6.48. The export of freight insurance can be identified by surveys of insurance corporations, though
small insurers or small lots of export might not be fully captured. It is important to note that premiums
payable on internationally traded goods before they reach the customs frontier of the economy of the
exporter are included in the FOB price of the goods. Freight insurance premiums payable subsequent
to the departure of the goods from the customs frontier of the exporter’s economy are treated as
payable by the importer. This means that relevant detailed questions need to be included in forms so
that the compiler can derive the freight insurance services which should be included in the compiling
economy’s balance of payments to compile relevant estimates of the exports of insurance services.
6.49. The export of freight insurance can be identified by surveys of insurance corporations, though
small insurers or small lots of export might not be fully captured. In contrast, the import of freight
insurance is often estimated from the price of imported goods, in conjunction with the estimation of
transportation services (see Chapter 14. B (iii).
6.50. Surveys of financial corporations. In order to obtain source data for financial services, their
exports in particular, conducting surveys of financial corporations is very useful. In this respect, it is
often financial intermediaries that are engaged in international transactions. Mandatory reports of
assets and liabilities of these corporations may be required for countries’ prudence policy. However,
such call reports may not contain sufficient information on international trade of financial services.76
The BPM6 CG contains a model survey form (form 17 (part E in particular) in appendix 8) with
questions that can be useful to ask financial corporations to obtain information on financial services.
To estimate FISIM, it is useful to ask information on positions and interest payments related to loans
to and deposits from (exports) and loans from and deposits to (imports) non-resident non-banks (see
BPM6 CG page 368).
6.51. To ensure an appropriate coverage, a survey frame should be based on a list of financial
intermediaries. In this respect, depository corporations and securities dealers are usually required to
register with supervisory authorities; therefore, a list of these businesses should be readily available
from official sources. In contrast, a list of other financial intermediaries other than depository
corporations and securities dealers, such as securitisation companies, may not be available in some
countries. In this case, it is important to develop country’s business profile so that such intermediaries
are identified. A full list of financial intermediaries will make it easier to conduct surveys for specific
businesses in the financial sector.
C.5.a. Country experience: the United States collection of insurance data
6.52. The main source of U.S. exports and imports of insurance services is a survey of U.S.
insurance enterprises, "Quarterly Survey of Insurance Transactions by U.S. Insurance Companies with
Foreign Persons (BE-45).77 This survey, conducted by the Bureau of Economic Analysis (BEA),
collects quarterly data on reinsurance premiums sold to and purchased from abroad and annual data
on reinsurance claims paid and received, primary insurance premiums sold and claims paid, and
auxiliary insurance services. For the quarterly survey, there is a reporting threshold. This form
distinguishes transactions with foreign affiliates, transactions with foreign parent(s) and other
members of affiliate foreign group, and transactions with unaffiliated foreign persons.

76
77

See also BPM6 CG.
The entire forms can be retrieved from http://www.bea.gov/surveys/iussurv.htm.

6.53. For the quarterly survey, filing is mandatory if, with respect to these transactions, any of the
following eight items were greater than positive USD 8,000,000 or less than negative USD 8,000,000
for the previous calendar year or can expected to be in the current calendar year, on an accrual basis:
(1) premiums earned, and (2) losses, on reinsurance assumed; (3) premiums incurred, and (4) losses,
on reinsurance ceded; (5) premiums earned, and (6) losses, on primary insurance sold; (7) sales of,
and (8) purchases of, auxiliary insurance services.
6.54. Every 5 years, BEA conducts a benchmark survey of insurance enterprises, “Benchmark
Survey of Insurance Transactions by U.S. Insurance Companies with Foreign Persons (BE-140),” to
collect information on enterprises that fall below the reporting threshold on the quarterly survey.
6.55. To ensure complete coverage of imports of insurance services, BEA asks non-insurance
enterprises in the United States to report premiums paid to and losses recovered from foreign insurers
on another survey, "Survey of Transactions in Selected Services and Intellectual Property with
Foreign Persons." BEA conducts this survey quarterly and also as a benchmark survey every 5 years.
C.5.b. Country experience: France collection of data for financial services
6.56. Over the last three years, Banque de France (BdF) has improved the way it collects data on
financial services. As of 2013, data on financial services are collected according to four channels:
i.

The Relevé de Transactions Économiques (RTE), a monthly reporting concerning a sample of
408 non-financial companies which generate the most important flows with entities abroad. In
2012, 58 of those companies reported that they are involved in financial services flows with
non-residents;

ii. The Enquête Complémentaire sur les Échanges Internationaux de Services (ECEIS), an
annual survey of data from non-financial companies which have generated service flows with
non-resident entities. In 2011, 162 among 3 921 companies reported significant financial
services flows with non-residents;
iii. The Comptes-rendus de Transactions (CRT), a reporting by the financial intermediaries. Data
is collected either on an annual basis – for 1 523 companies – or on monthly basis for the 37
most significant companies; CRT account for over 75% of the overall flows of financial
services with non-residents.
iv. A monthly pro forma reporting concerning the Financial Services of the General Government.
6.57. These reports enable BdF to meet with the breakdowns required in BPM6 and produce an
estimation based on monthly data with a relevant coverage for the monthly financial services flows.
6.58. The national accounts incorporate a measure of imports and exports of FISIM. A joint
workflow with the National Statistical Institute (INSEE) is currently being developed to base the BOP
FISIM on the National Account data for the production process and for back data. This will ensure the
consistency of the BOP FISIM data with those of the national accounts. FISIM geographical
breakdown will be based on the geographical breakdown of the stock of loans and deposits.
6.59. Regarding financial services embedded in margins from buy-sell transactions (b), and asset
management costs deducted from property income (c), methodological work is being carried out,
based on case studies with the industry.
C.6.

R&D and innovation surveys

6.60. R&D and innovation surveys may include variables that are relevant for the estimation of
trade in certain service categories, notably related to computer software and information services;
58

research and development services; Architecture, engineering, scientific and other technical services;
Audio-visual and related services included in personal, cultural, and recreational services; and charges
for the use of intellectual property n.i.e. In particular questions related to the origin of funding of
R&D activities, and payments for the sales of the products and services derived from R&D can be
used to get to better estimates of trade in services in these service categories.
6.61.
When developing and conducting R&D and/or innovation surveys, compilers should consult
the Frascati Manual, which is the internationally accepted standard for such surveys (note that at the
time of writing of this Compiler’s Guide, the Frascati Manual is being revised and is expected to
include a chapter dedicated to the globalisation of R&D). Since 2011, the Manual has included an
online annex designed specifically to address the features and needs of developing countries. This
annex is being integrated into the revised core manual.
6.62. When using questions from R&D and innovation surveys related to payments for R&D,
compilers should be aware that the international funding of R&D may also include donations and
grants and thus do not necessarily represent payments for R&D services. Hence, care is needed to
clearly identify this information for the purposes of compiling trade in services statistics (especially
when compiling import statistics because the coverage of R&D surveys does not necessarily include
firms that buy R&D services but do not perform R&D themselves). Compilers can build on the
prototype questionnaire with a comprehensive list of questions as described in the OECD Handbook
on Deriving Capital Measures of Intellectual Property Products, notably the section III with example
questions on data for international trade in R&D services and R&D output produced in the past.78
6.63. In addition to questions related to payments for R&D services, questions in R&D and related
surveys (e.g., ad hoc commercialisation and licensing surveys) could also give insight into the value
of intra-firm trade in services related to IPPs, notably if no payments are made for those services. Box
X gives some examples of these questions that could be added to derive such information. A
distinction between the types of IPPS could be included.
Box 6.3 Questions for R&D surveys to better capture intra-firm services related to IPPs
i.

What is the estimated value of intellectual propert y licences (whether protected by copyright, patent etc. or
otherwise) that you use in production and receive from your parent or other affiliated party, and who own
and control the underlying asset, but for which no explicit payment was made by you?
ii. What is the estimated value of intellectual propert y assets (whether protected by copyright, patent et c. or
otherwise) that you use in production, whose ownership was transferred to you by your parent or other
affiliated party but for which no explicit payment was made by you?
iii. For dedicated intellectual property producers that receive no revenue through licensing of IPPs: What is
the estimated value of intellectual property assets developed by your firm this year on behalf of your parent
or other affiliated parties, and which may have funded the activity but did not make an explicit payment that
was recorded by you as sales?
iv. For dedicated intellectual property producers that receive some revenue through licensing of IPPs to n onaffiliated parties: What is the estimated value of intellectual property (whether protected by copyright,
patent or otherwise) you provide to your parent of affiliated parties but for which no explicit receipt was
booked as sales.
All these questions can in turn be broken down by type of service category, e.g. R&D, Software, brands or
audio-visual.

78
Two other publications that are relevant for measuring trade in services related to R&D include Globalisation
in the National Accounts, and The Guide to measuring Global Production (forthcom ing).

6.64. It is a good practice to target certain questions to specific groups of respondents. For example,
sellers and buyers of licences to reproduce/distribute are likely to remain within discrete industry
groups.
C.7.

Surveys of tourism activities79

6.65. It is useful to observe the activities which supply products relevant to Travel (credit) and
inbound tourism consumption according to the tourism statistics concept. This necessitates the ability
to identify those products which meet the share-of-supply condition, that is, those in which travellers
and visitors’ consumption represent a relevant share of total supply. Additionally, it will be necessary
to be able to distinguish between the output provided to residents and those to non-residents. In the
following, the supply of accommodation and travel agencies and tour operators services will be
highlighted as possible sources of information on resident to non-resident transactions related to travel
and international tourism consumption.
C.7.a. Supply of accommodation services
6.66. Accommodation services are mostly provided by production units specialized in their
provision, though there are few exceptions. It is generally assumed that the provision of such services
as a secondary output to other activities is limited, meaning that the collection of data will be
restricted to production units producing accommodation services as their main output.
6.67. Providers of accommodation services. There are two broad categories of visitor
accommodation providers: (i) market providers, which receive payment for their services, included
under “Accommodation” (ISIC Division 55) or under “real estate activities” (ISIC Division 68) and
(ii) non-market providers, which accommodate guests free of charge (case of family and friends and
accommodation in one’s own second home or timeshare). Staying with family and friends as a form
of accommodation does not generate any additional production of accommodation services, and
services provided by own second homes or timeshares do not depend on the actual visit to those
homes.80
6.68. Only market providers included in ISIC Division 55 will be considered in the present
compiler's guide, as in most countries, they represent the major providers of accommodation services,
and are those at which specialized surveys are usually directed. For more information on the other
categories of providers, users can refer to the previously mentioned IRTS 2008 Compilation Guide.
Activities under ISIC Division 55 include classes “551 Short term accommodation activities”; “552
Camping grounds”, “Recreational vehicle parks and trailer parks”, and “559 Other accommodation
activities”.
6.69. Most national tourism administrations have their own registers of accommodation
establishments, based on a licensing procedure, which contemplates a more detailed classification (for
instance using a “star” system of qualification of the services provided) but possibly a more restricted
coverage (no inclusion of class 559, or even of class 551) whereas the NSO might have its own
general Business Registry, that classifies generally establishments, without considering these
particularities. This makes inter-institutional coordination all the more so important in order to insure
consistency of the results.
6.70. Survey design. Regarding the statistical design, care should be taken, mostly in the annual
NSO design, to ensure that accommodation establishments located in zones of low general economic
activity (manufacturing, business services) but with a high tourism orientation are not omitted through

This sub-section deals only with the “supply side” of tourism statistics while the demand side is covered in
Chapter 7.
80
See TSA-RMF 2008 paragraph 2.36.

the general selection procedure of statistical units (which is often based on the density of general
economic activity).
6.71. The general sample design should recognize also the existence of small family-owned units. It
may be the case that the licensing system is recording them, but not the general business register.
Such establishments should be included in the annual survey (but might be excluded from the monthly
or quarterly ones that usually focus on relatively big units, following a census type system whose base
should be frequently updated (at least annually) to take into consideration the dynamics of the
activity. Some of these small family-units might also be observed using household surveys or surveys
to associative bodies to which they might belong. Such surveys might inform on the global
performance of their associates in terms of tourism indicators (in particular accommodation units
offered and occupancy).
6.72. Surveys of accommodation units concentrate in measuring occupancy rates, room rates
revenues per night or guest, and the national origin of guests. This type of information can help in
estimating non-resident expenditures on accommodation. In addition, other variables related to
establishments can be compiled, such as: employees per room and their remuneration, total revenue,
revenue derived from sales of accommodation services, value added, remuneration of employees,
details on employment in particular according to the permanency of the work contract (short time
contract for the season are often used), and gross fixed capital formation.
6.73. Frequency of accommodation surveys. Because tourism is a phenomenon which usually
presents high seasonal variations, establishments belonging to these activities are observed monthly,
quarterly and annually. Tourism administrations frequently control the monthly and quarterly surveys,
whereas often the National Statistical Office (NSO) is in charge of the annual, structural survey. The
surveys used for these different observation frequencies should have a harmonized design, to enable
the comparison of their respective results.
C.7.b. Surveys of travel agencies and tour operator services
6.74. The main business of travel agencies and tour operators services consists in trading in
services such as passenger transport, accommodation, package tours (arranged by others or arranged
by themselves). The related fees are paid usually, but in different proportion both by the person
traveling (or his employer in the case of business travel) and by the provider of the service
(international transport, visitor accommodation services, package tours). The person travelling is
usually resident of the same economy as the travel agency, though this is rapidly changing with the
increase of transactions realized through the Internet; in the case of the service provider, it might be
resident of the same economy as the visitors, resident of same economy as the service provider or
resident of any other economy.
6.75. Because their services are used to book or arrange trips, travel agencies and tour operators can
provide indications on the intensity of travel (debit) or outbound tourism consumption. It is
worthwhile mentioning that, in Balance of Payments statistics as well as in statistics of international
trade in services, when services providers (principally international transportation and accommodation
providers) use the services of a travel agency resident in an economy different from their own to make
their supply available to the public, a resident-nonresident service transaction needs to be recorded;
this transaction, however, is not included in travel but in trade-related services.
6.76. As a consequence, and in the case of mainly travel agencies, two important sets of data should
be collected:
i.

Data on the volume of travel agencies’ business in intermediating in the sale of international
transport, international accommodation and package tours to travel abroad (arranged by either
themselves or by specialized tour operators whom, in turn, might be resident in any country).
This might be used as an indicator on the trend of travel (debit) and outbound tourism
61

consumption. The information to be collected may refer to: types of clients
(business/personal), destinations, number of operations and total value of transactions,
treating separately those that correspond to travel on package or without package.
Nevertheless, such a measurement might be fragile as households/persons are increasingly
using the Internet to plan their trips, and thus have direct access to other online travel
arrangers, that might be resident in other economies, and thus escape from this measurement.
ii. Because, within the perspective of International Trade in Services, travel agency services, and
more generally, reservation services paid for by international transport companies,
accommodation service providers, etc are treated as purchase of services, these fees should be
measured separately in order to be included in data corresponding to trade-related services if
they correspond to resident-non-resident transactions.
6.77. In the particular case of international transport, obtaining information is complicated because
of the multiplicity of travel agents and international carriers worldwide, the different modes by which
the payments are collected from the final clients by travel agencies and the way their amount is
assigned among the different stakeholders, the complexity of arrangements among airlines (in
particular code sharing and interlining, by which whoever sells the service might not be the one that
provides it). All of this results in the fact that the value that can be reported by resident travel
agencies about the services sold to nonresident service providers might not be accurate, although it
may represent useful information to be possibly combined with other sources. Improvements in the
measurement of resident-nonresident reservation services could be gained through the possible use of
databases generated by the International Air Transport Association (IATA) regarding the clearing
system between airlines and travel agents (the Billing and Settlement Plan (BSP6)) and from a better
understanding of the roles and modes of remuneration of the Central Reservation Systems (CRS) and
the Global Distribution Systems (GDS), whose residence should be accurately considered.
C.8.

Construction surveys

6.78. Construction is a sector with such special characteristics that compilers should consider to use
a single survey to respond to various information needs, notably those of trade in services between
residents and non-residents and of FDI/FATS. The main reason for this is that international
transactions for the construction sector are sometimes difficult to separate. For example, when a
construction company provides its services through a physical presence over a long term period (more
than 12 months) but without creating a legal entity in the country of the client, and if the operations
are substantial enough, then the compiler may need to consider that a notional institutional unit needs
to be considered from a statistical point of view. In this case, the corresponding international
transactions will be recorded as foreign direct investment financial flows, income and position, and
the data corresponding to its operations should be considered as FATS (see also section D of this
chapter). If the above conditions are not met, the international transactions will be considered trade in
services between residents and non-residents.
6.79. By reviewing all the information obtained through a single survey, the compiler can
subsequently (and coherently across surveyed firms) make decisions as how to use the data, i.e.
classify transactions as trade in services between residents and non-residents or FDI and use some of
this information for FATS (i.e. sales/turnover, see also section D). Such a unified survey could also be
used to collect additional operational data on the activities of these service providers.
6.80. In order to correctly compile construction it is particularly important to precisely identify the
residence of the enterprise realizing the construction work. Indeed, a construction enterprise
established in one economy may undertake construction of projects in another economy through a
subsidiary or a branch; i.e., via a direct investment relationship. In other cases, it may decide to
conduct services directly from its home base. In the former case, the construction activities are
regarded as the activities of foreign affiliates and the corresponding international transactions are
62

considered direct investment, whereas in the latter case they are considered export of services. For a
number of cases (i.e., for larger projects that take more than one year), the delineation between trade
in services between residents and non-residents and activities of foreign affiliates may be difficult to
establish. A number of criteria to consider are suggested in BPM6 and MSITS 2010 and are described
further in Chapter 14. Given the difficulties in establishing a clear delineation, it may be useful, in
particular for countries where (many) big contractors have projects abroad, to collect all information
on their activity through a single survey, thereby providing the necessary information for compilers to
then distinguish between exports/imports of construction services or FDI/FATS transactions.
6.81. When classified as trade in services between residents and non-residents, construction
transactions may be considered as mode 3 supply, i.e. commercial presence, or mode 4, that is
presence of natural persons, to account for cases involving the presence of resident workers
(employed by the resident construction enterprise or self-employed) in the non-resident/host
economy. It should be noted that, although in general mode 3 concerns mostly domestic sales of
foreign affiliates, the construction services for works carried out in less than one year may also be
regarded as commercial presence according to the GATS..81
6.82. Although business surveys are designed to collect information on both the transactions of
construction companies and those of their clients, the coverage of the latter could be more difficult in
the case of construction. The services provided by the contractor are likely to be closely related to the
sector of activity of the enterprise (i.e. construction). This is less true for the clients, i.e. any
household or companies of any industry may buy construction services. In addition information is
needed on both the value of the construction project and on the inputs, in particular those bought in
the country where the project is taking place. Hence, for construction a specific survey may be more
appropriate, in particular for those economies which are specialised in the exports of construction.
C.8.a. Country experience: Italy
6.83. In Italy construction data are compiled by the central bank (Bank of Italy), in the wider
framework of BoP statistics. Construction-related information is collected through a Quarterly nonfinancial transactions questionnaire (TTN), consisting of regular reports on other services (i.e.
services other than travel and transport), intangible assets, unilateral transfers and compensation of
employees. Therefore, the construction item compilation is based on a general purpose sample, both
for credits and debits (see Chapter 14 – Section B.3.c., paragraphs 14.118 – 14.124 for data
compilation of construction).
6.84. Among the around 3,000 firms sampled with the TTN survey, about 450 enterprises are
mainly engaged in construction activities, according to the reported NACE code. They cover about 23
per cent and 30 per cent, respectively, of the population total turnover (around 180 billion euro) and
foreign turnover (65 billion euro).82
6.85. The TTN questionnaire begins with a general part. Construction specific information has to
be reported in a subsequent dedicated section of the questionnaire, where constructions abroad and
constructions in Italy are distinguished. Despite the fact that the TTN survey is meant to mainly
address non-financial items, the construction section aims also at contributing to the compilation of
FDI, including the related flows of income.

MSITS 2010, paragraphs 5.53-54.
The foreign turnover is estimated on the basis of a survey carried out by the National Builders Association
(ANCE) in 2012 on 40 Italian construction enterprises (ANCE, Rapporto 2012 sulla presenza delle impres e di
costruzione italiane nel mondo, http://www.ance.it/ docs/docDownload.aspx?id=8546).
82

6.86. Firms are required to list the construction projects in which they are involved either as
contractors or clients and the counterpart is a non-resident entity, and to provide detailed information
for each project to according to the guidelines of MSITS 2010.83
6.87. In Italy’s system, the attribution of the status of “branch resident in the host country” is based
on a simplification criterion: the entity carrying out the works is considered a resident branch of the
host country if the project lasts one year or more. Consequently, depending on the estimated duration
of the works, the construction activity is either regarded as an FDI-related operation, if the
construction work extends over a period of at least one year, or as a service transaction in the opposite
case.
C.9.

Other specific surveys (e.g. health, legal, IT enabled)

6.88. Given the high interest in specific sectors, in particular for exports, it may be useful to
consider establishing a specialized survey to collect more detailed data on a particular service (e.g. a
detailed breakdown of the services products or activities, detailed information on partner countries, on
modes of supply used in these services international transactions). Often, the collection of statistics is
motivated by a strong policy demand for more detailed information on one or more service category,
when for example this category accounts for a substantial share of the economy or employment.
6.89. Although such a survey may be specially tailored to respond to the specific (national) needs,
it is important that the information collected is compatible with the international definitions as
outlined in MSITS 2010 for example for residence, detailed services items and modes of supply. The
MSITS 2010 in particular indicates that for further breakdowns by type of service the CPC is used as
a guide (see EBOPS 2010-CPC Version 2 correspondence table84).
6.90. Finally, it should be noted that if a specific sector and its international transactions are very
important for a country, compilers may wish to consider collecting all the required information
through a single survey (i.e. not only trade in services between residents and non-residents, but also
structural business statistics and/or FATS), and then make decisions as to how to use the data (i.e.
classify transactions as trade in services between residents and non-residents or FDI and use some of
this information for FATS (i.e. sales/turnover). Such a survey could also be used to collect additional
operational data on the activities of these service providers
C.9.a

Country experience: India’s compilation of education services

6.91. In order to better analyse trends in trade in higher education in India, the Directorate General
of Commercial Intelligence and Statistic of India (DGCIS) initiated a survey on international trade in
education services, under the overall guidance of the Central Statistics Office, Ministry of Statistics &
Programme Implementation. The scope of the survey was restricted to Higher Education services, i.e.,
technical education, management education, medical education and general discipline (University
level courses). The survey period was year 2010-11, as all financial information for that year had
already been audited and was accessible. The fieldwork of the survey as well as processing and
compilation of data has been outsourced to the Indian Institute of Foreign Trade under the overall
guidance and supervision of DGCIS.
6.92. A stratified sampling method was used to select the institutes/universities providing education
services under each of the categories of education, and by mode of supply, across eight cities in India.
Further, all the important Universities and Technical and Management Institutes in India enrolling
foreign students were within the coverage. The questionnaire was developed in consultation with

83
84

MSITS 2010 paragraph 3.133.
http://unstats.un.org/unsd/tradeserv/TFSITS/msits2010/ebops2cpc.htm.

relevant agencies to collect detailed information on trade (both export and import) in higher education
from Indian Universities / Institutions. The questionnaire has 16 subsections, covering among others
(a) identification of the service provider, (b) transactions with non-residents, (c) employment by type
of staff, (d) details of revenues / receipts from foreign universities / institutes (including foreign
students studying in India and in-house Indian faculties visiting abroad), (e) details of expenses /
payments to foreign universities / institutes (including Indian students studying abroad and foreign
faculties visiting India), and (f) identification of trade barriers in higher education services.
C.9.b. Country experience: India’s compilation of software and IT exports
6.93. In view of the growing importance of exports of Information Technology and Information
Technology Enabled Services (IT & ITES), the Reserve Bank of India (RBI) has conducted annual
surveys on software and IT exports services on regular basis from September 2008 onwards. It covers
elements of the EBOPS 2010 categories computer and information services as well as a broad range
of other business services, which have the particularity that they may have a strong IT component
(either to produce or to deliver the service). The survey is designed to collect detailed disaggregated
level information on export of software services broken down by activity, type of services (onsite/off-site) and country of destination along with the four modes of supply. For determining the
proportion of each mode of supply within the services transactions, survey respondents are asked to
report the estimated value of services provided through modes 1, 2 and 4. This information is sought
both on a fiscal year basis as well as on a quarterly basis. No treatment different to what is applied to
other results of this very detailed survey (i.e. for non-response etc.) is applied for modes of supply.
The survey also collects information on the software business of foreign subsidiaries/associates of
Indian companies (foreign affiliates), distinguishing between software business done in host country,
with sales to either India or to other countries. The collected statistics are cross-checked and validated
with the information collected independently by two other sources, namely, the National Association
of Software and Services Companies (NASSCOM) and the non-physical (off-site) software exports
figures collected through a different format called Softex.
D.

Foreign affiliate statistics and the international supply of services

6.94.
This section deals with the collection through surveys of enterprises of data necessary for
FATS. For most countries that are compiling FATS, business surveys are usually the most appropriate
way to obtain the information needed to provide meaningful FATS in a reliable and cost-effective
fashion. The elements for data collection detailed in this section focus on the specific aspects of
services. The main definitions of the FATS conceptual framework are briefly described in Chapter 1,
whereas more details as to the definitions of some of the variables are provided in Chapter IV of
MSITS 2010. Needs in terms of compilation are described in chapter 15. As most of the variables are
directly drawn from the data collected, it is important that these definitions are considered when
designing the survey form as well as the accompanying explanatory notes.
6.95. There are two basic approaches to collect FATS through enterprise surveys. These two
approaches are not necessarily mutually exclusive. The first approach, which can be used only for
inward FATS, identifies from existing data on resident enterprises the subset that is foreigncontrolled. The second approach entails conducting surveys that request information directly on the
operations of resident affiliates of foreign enterprises and foreign affiliates of domestic enterprises
(either through a dedicated survey, that can be specially tailored to respond to needs, or through an
existing FDI survey which offers more limited options as to the information that can be inquired).
MSITS 2010 recognizes the advantages and disadvantages of each approach and the need for
compilers to demonstrate flexibility in adapting the FATS data collection to each country’s statistical
infrastructure while maximizing the use of existing data.
6.96. A usual precondition for the establishment of a foreign affiliate is the development of
investment flows leading to an FDI relationship. An understanding of the FDI universe provides a
65

strong basis for developing an understanding of the FATS universe and, therefore, ensuring a sound
data collection mechanism for FATS. The FDI universe of enterprises has a broader coverage than the
FATS universe, but a data collection program for FDI statistics can provide a base for a data
collection program for FATS, either directly or indirectly. The direct use of this data collection is that
key FATS variables might be incorporated into existing FDI surveys. Alternatively, if the information
is not available from the FATS data source itself, links to existing data on FDI could be used (i) for
inward FATS to identify the data of interest in existing sources of information on resident enterprises
(i.e. to identify the foreign controlled subset of firms) or (ii) for both inward and outward FATS to
identify foreign-controlled affiliates for which FATS variables could be collected on a separate
survey. Finally it is important to note that in the case of regional headquarters which report data on
behalf of affiliates further down in the chain and located in various countries, the respondent should
be required to report the data on a non-consolidated basis, i.e. allocating the variables describing the
operations of the affiliates to the affiliates' actual country of location.
6.97. Three types of surveys are particularly useful to compile FATS: Structural Business Surveys,
FDI surveys, and specific FATS surveys. These are discussed in more detail below. It should be noted
that the collection of FATS does not need to be a stand-alone exercise and that compilers can choose
to collect FATS data and other information in a single survey, as described in B7 and B8.
D.1.

Structural business surveys

6.98. Structural business statistics (SBS) surveys represent a class of surveys designed to provide
detailed economic and financial data about the activities of domestic resident business enterprises.
They may be encompassing all types of activities or could be specifically tailored for firms primarily
engaged in specific sectors (e.g. wholesale and retail trade activities). Reporting and statistical units
for these surveys can be identified at several levels of consolidation, ranging from establishments
(single physical locations or local activity units) to fully-consolidated enterprises that may include
several subsidiaries and operate in several different sectors of the economy.
6.99. SBS surveys can be used to provide FATS measures only for inward FATS, by identifying
the subset of resident domestic corporations that meet the relevant criteria for statistical units and for
which the recommended data items are either already collected for other purposes or can be added to
the surveys to meet the FATS requirements. As described in Chapter 15 this approach requires the
existence of indicators of foreign ownership in the domestic business register, a sample frame that
meet the FATS requirements, or the ability to link SBS data with FDI survey or an FDI register
through the use of common identifiers. An example of this approach was implemented in the United
States; i.e. the linking of inward FDI enterprise data with the more comprehensive and detailed data
collected every five years at the establishment level in the economic censuses. Where existing
domestic statistics are used as the source of information, links to FDI data can be used to supplement
the SBS data with information, such as by identifying the country of the controlling entity.
6.100. In general SBS surveys already cover most of the variables of interest identified in MSITS
2010, or include the relevant elements to compile them. However in many cases some of the elements
or breakdowns are absent. When envisaging the use of SBS surveys for FATS, compilers are advised
to coordinate with statisticians in charge of the structural business statistics data collection program,
to identify what is already available, and what would need to be added in the survey(s). This would
need to be established according to the priorities identified, keeping the burden for respondents to the
minimum as SBS surveys are generally already quite long. The FATS variables often covered by
existing surveys are the following: sales, employment, purchases of goods and services for
intermediation consumption, as well as those for resale (often only covering goods), assets,
compensation of employees, taxes on income, investment. Some breakdowns of these variables are
often included and compilers will need to investigate how these could be used in the context of
MSITS 2010. Some variables, for example on R&D, may be included or covered by supplementary
66

surveys. Finally the forms may include some additional variables of interest to the compiling
economy.
6.101. The additional characteristics which are often absent from existing structural business surveys
but are of interest in the context of services are the following:
i.

For sales/turnover: a breakdown into sales of goods and sales of services (in first
instance), and if possible a further breakdown into main categories of services (a longerterm goal), at least those of interest to the compiling economy, and on a basis compatible
with EBOPS 2010.

ii. Identify separately (i) sales (of services) made in the host economy, (ii) the exports to the
economy of the UCI and (iii) to third economies.
iii. Other geographical breakdowns of sales/exports as well as for purchases/imports (of
services) would also be useful. If possible a distinction should be made between
affiliated and unaffiliated parties. However given the higher burden this may entail, the
possibility offered by structural business surveys may be limited (see also Chapter 15
sections E.1 and E.3).
iv. For employment, identify non-resident employment, of which intra-corporate transferees.
6.102. Using SBS surveys to collect FATS data has a few advantages. The FATS variables will be
compatible with measures of the domestic economy. Moreover, it allows FATS variables to be
compiled with no (or little) added burden to reporters and with fewer resources than a separate data
collection program. However, it may be difficult to include the full range of recommended FATS data
items due to limitations in the domestic statistics.
D.2.

Foreign direct investment surveys in the context of FATS

6.103. Where there are existing FDI surveys, registers used for FDI would typically be used to
identify foreign-controlled affiliates that qualify for FATS treatment. Alternatively, key FATS
variables could be incorporated into existing inward or outward FDI surveys. Many countries have
inward and outward FDI surveys, and the use of these vehicles to collect some basic FATS can be
considered as a quick and cost-effective solution. In particular for outward FATS. However,
compilers need to keep in mind that there are a number of drawbacks, some of which are listed below.
6.104. The priority would be to cover the main FATS variables, namely sales/turnover (if possible
with an estimated breakdown into total goods and total services, excluding income elements) and
employment (if possible distinguishing employees recruited from overseas and intra-corporate
transferees). If not available from other sources, the survey should enable the identification of (i)
subsidiaries and branches (possibly including further qualitative information on the subsidiary), (ii)
the principal activity of the affiliate, and (iii) the country of establishment of the affiliate for outward
and of the UCI for inward. All affiliates should be covered whether directly or indirectly controlled.
However given that FDI surveys often only gather information on immediate relationships, in the first
stages of implementation only data on direct control could be collected, but a medium to long-term
objective would be to also gather information on affiliates indirectly controlled.
6.105. Compilers should note that incorporating FATS-related questions in FDI surveys may
increase the response burden imposed on enterprises that are not part of the FATS population. In
addition FDI surveys may need to be conducted more frequently (for example, quarterly), whereas
FATS may be needed less frequently (for example, annually, at least in the initial stages of
development of the FATS framework at the national level). If a country conducts an annual survey of
direct investment as well as monthly or quarterly surveys, it may be appropriate to only incorporate
FATS variables in the annual one. Concerning the possible extension of FATS variables and
67

breakdowns to be covered in FDI surveys, this should also be carefully assessed given the main
reasoning behind FDI surveys, and the broader population covered by those. This is why compilers
are advised not to go beyond the list of items identified above, and that if there is a need to go beyond,
then that a survey complementary to the regular (annual) FDI survey is developed, which could also
be used to collect more detailed FDI data.
6.106. Extended business registers might also be a suitable means of maintaining such information.
This approach has already been adopted in some countries which use them to maintain data on foreign
ownership.
6.107. While FDI statistics have several differences from FATS variables. FDI statistics may provide
useful indicators relating to commercial presence for those countries that have not yet begun to
compile FATS. In particular, FDI positions can serve as an indicator of a country’s interests in using
commercial presence to supply services internationally. In addition, FDI statistics can be used in
conjunction with FATS to indicate the extent to which the operations of affiliates were financed with
funds from direct investors, as well as the extent to which the income generated by affiliates accrued
to direct investors.
D.3.

Foreign Affiliates Surveys

6.108. Some special surveys could be designed to collect FATS which would complement existing
data collection systems such as SBS or FDI. These surveys would be limited to the subset of the FDI
population that qualify for FATS and may be conducted as an extension of the existing surveys or less
frequently than the FDI surveys. Naturally information from the business register or the FDI register
would be necessary to identify the target population. Such surveys could be conducted separately for
both inward FATS and outward FATS, or would enable more possibilities in terms of collecting
detailed data, in particular for outward FATS. This approach may also be appropriate when an
existing FDI data collection program is not already in place or cannot be easily adapted for the
requirements of compiling FATS. In general separate surveys would be required for inward FATS and
outward FATS due to the distinctly different populations, although in certain circumstances this
would not be the case, such as when for outward FATS many of the mother companies of the
compiling economy are themselves controlled from abroad. It is also important to note that the ability
to conduct separate surveys may be limited by legal and institutional conditions in the compiling
economy.
6.109. One of the main drawbacks of conducting a dedicated FATS survey is that it may be costly
and burdensome to develop additional surveys. However, such a solution has also a number of
advantages as it can be specially targeted towards the relevant inward and/or outward population.
More importantly, it can be targeted to the needs of users and its design can more closely follow the
international recommendations, including the development of detailed guidelines for survey
respondents. Such a survey could have all or most of the various dimensions as listed in MSITS 2010.
As a general consideration, one should note that again it may be easier to approach the inward FATS
population for detailed data, than it is for the outward. As an alternative approach, the existing SBS
could be extended with questions for the inward foreign affiliates, which could be overall more cost
effective.
6.110. The information that can potentially be covered in a FATS survey includes all the
requirements of international manuals (MSITS 2010) and guidelines (BD4 and HEGI). In the context
of the international supply of services, these are also listed in Chapter 15 of this guide. The
information to be covered will also depend on the national needs identified. At least in the first stages
of development of a FATS survey, the collection could cover the following aspects:
i.

If not already covered by the register(s) or related FDI surveys, information on the type of
affiliate (subsidiary, branch, associate), its main activity, the country of the UCI (for
inward) and the country of establishment of the affiliate (for outward).
68

ii. Variables: sales/turnover (if possible with a breakdown into total sales of goods and total
sales of services), employment (if possible distinguishing employees recruited from
overseas and intra-corporate transferees), purchases of goods and services for
intermediation consumption, as well as those for resale.
iii. For sales identify separately (i) sales (of services) made in the host economy, (ii) the
exports to the economy of the UCI and (iii) to third economies.
iv. Other details and breakdowns could then be sought as experience in the use of the FATS
survey develops, and depending on the results obtained and the use that can be made of
these.
D.4. Country experiences
D.4.a. Country experience: the United States
6.111. In 1976, the International Investment Act authorized the U.S. BEA to collect data on the
finances and operations of foreign-owned enterprises in the United States and of U.S. parent
enterprises and their foreign affiliates.85 The Act was expanded to include trade in services in 1984.
The Act made responding to BEA’s surveys mandatory and required the BEA to maintain the
confidentiality of the data collected. Another act passed in 1980, the Paperwork Reduction Act,
governs the collection of data from the public by any government agency. The Paperwork Reduction
Act requires agencies to minimize the burden they place on private businesses and citizens in
collecting information.
6.112. BEA’s surveys of the operations of MNCs collect data on balance sheets; income statements;
sales; employment and employee compensation; research and development expenditures; property,
plant, and equipment; taxes; trade in goods; and the components necessary to estimate value added.
While the surveys cover both majority-owned and minority-owned affiliates, more data are collected
for majority-owned affiliates.
6.113. BEA uses a system of benchmark surveys (conducted every 5 years), and annual surveys to
collect data on FATS. Benchmark surveys cover the universe of multinational enterprises, but data
collection is much less detailed for smaller reporters and for minority-owned reporters.
6.114. Annual surveys are conducted between benchmark surveys. The smallest reporters are
exempt from reporting on the annual surveys, and BEA uses statistical sampling for the medium-sized
reporters. Large reporters are required to report annually and provide the most-detailed information.
The reporting thresholds are based on the reporter’s assets, sales, or net income. Estimates are made
for reporters not required or failing to report in a given year, so the published statistics cover the
universe. Smaller majority-owned foreign affiliates report less detail than larger majority-owned
foreign affiliates. To present statistics on the operating data of all majority-owned foreign affiliates,
BEA estimates items that are only collected for large majority-owned foreign affiliates for the smaller
majority-owned affiliates. These estimates are based on relationships among the data items for a
panel of comparable larger majority-owned foreign affiliates.
6.115. Both for inward FATS and outward FATS (regarding the U.S. parents), data are collected on
an enterprise group basis and cover the fully consolidated domestic entity. For outward FATS, the
U.S. parent is required to report on the operations of its foreign affiliates but this tends to be less
consolidated. First, affiliates can never be consolidated across countries. Second, affiliates cannot be
consolidated across industries unless they were part of an integrated production process. Finally,
affiliates cannot be consolidated if they do not have the same ownership structure.

86
Housing censuses may also be of interest. A housing census collects information pertaining to all liv ing
quarters and occupants thereof in a country or in a well delimited part of a country.

6.116. Data are collected on an accrual basis and generally follow U.S. Generally Accepted
Accounting Principles (U.S. GAAP). Data are reported based on the enterprises’s fiscal year. For
outward FATS, items recorded in foreign currency are translated into U.S. dollars following U.S.
GAAP, which calls for assets and liabilities to be translated using the exchange rate on the date of the
balance sheet and for revenues and expenses to be translated using weighted-average exchange rates
for the period.
6.117. For inward FATS, operations are classified by the country of the ultimate beneficial owner
(UBO), which is equivalent to the UCI. However, a few data items are classified by the country of
the foreign parent. For outward FATS, the statistics are classified by the country where the affiliate is
located. That is, where the affiliate’s physical assets are located and where its primary activities are
carried out. In most cases, the country of location and the country of incorporation are the same.
However, in some cases, a business enterprise is incorporated in one country but part or all of its
physical assets are carried out in a second country. If all of its operations are in a single country
outside of its country of incorporation, then the affiliate is treated as a single affiliate in the country of
its physical presence. If the affiliate has physical assets in each country, it is treated as two affiliates.
6.118. For classification by industry, BEA uses industry codes derived from the North American
Industry Classification System (NAICS) generally at 4-digit level. For inward FATS, each affiliate
reports up to 10 industries in which it has sales. These are used to assign a primary industry code to
the affiliate in three steps. First, a given affiliate is classified in the NAICS sector that accounted for
the largest percentage of its sales; NAICS sectors are at the two-digit level. Next, the affiliate is
classified into the 3-digit subsector within that sector. Third, within that 3-digit subsector, the affiliate
is classified in the 4-digit industry for which its sales were largest. For outward FATS, the same
process is followed to assign U.S. parents and each foreign affiliate to a primary industry.
6.119. BEA asks reporters to break out their sales into goods, services and investment income. For
inward FATS, enterprises are asked to further break out their sales of services between sales to U.S.
persons and sales to foreign persons. For outward FATS, BEA asks that sales of goods, services, and
investment income each be broken out by destination: to the United States, to the host country, and to
third countries. For each of these, BEA asks that the reporter further distinguish between sales to
affiliated parties and sales to unaffiliated parties.
6.120. The data reported have to pass a large number of computerized edit checks and reviews for
consistency with data such as quarterly FDI surveys. As a result of this edit and review process, a
number of changes are made to the reported data, usually after consulting with the reporter. In some
cases, usually involving smaller affiliates, estimates based on industry averages or other information
are substituted for missing or erroneously reported data.
D.4.b. Japan
6.121. The Survey on Overseas Business Activities of Japan (outward survey) and the Survey of
Trends in Business Activities of Foreign Affiliates (inward survey), conducted by the Ministry of
Economy, Trade and Industry (METI) serve as Japan’s FATS. They are conducted on an annual basis
and data correspond to those recommended by MSITS 2010, with however some deviations i.e. :
i.

In case the parent company is mainly involved in financial, insurance and real estate
industries, it and its affiliates are excluded from the outward survey;

ii.

Inward data are based on ownership ratio exceeding one-third, while outward data are
based on 10 percent standard, while the majority standard is adopted by MSITS 2010;

iii.

Ultimate investors are not identified in inward surveys.

6.122. The Survey on Overseas Business Activities of Japan (Outward survey) aims at presenting the
actual conditions concerning overseas business activities of Japanese corporations that will serve as a
basis for future industrial and trade policies. Survey targets are parent companies: Japanese
corporations which, as of the end of March, own overseas affiliates, excluding those in the financial
and insurance industry or real estate industry (hereinafter referred to as "Parent Companies"), as well
as overseas affiliates (“Subsidiaries" and "Sub-subsidiaries" are collectively referred to as "overseas
affiliates"). The following overseas affiliates are surveyed:
i.

A foreign affiliate in which a Japanese corporation has invested capital of 10% or
more;

ii.

A foreign affiliate in which a "subsidiary," funded more than 50% by a Japanese
corporation, has invested capital of more than 50%;

iii.

A foreign affiliate in which a Japanese corporation and a subsidiary funded more than
50% by a Japanese corporation has invested capital of more than 50% cumulatively.

6.123. This outward survey is based on Japan’s Statistics Law. For 2013 survey, 8,662 parent
companies comprised the population and questionnaires were sent to the entire population. 76.4% of
those enterprises replied. Survey forms have to be filled out by parent companies. In the case that
multiple Japanese companies are parents, the company with the biggest share in the equity (leading
company in the case that the shares of two companies are the same) of a foreign affiliate should fill
out the form.
6.124. Outward FATS information collected include overseas affiliate profile; investment; operation
status; Decline in control share; Employment on a full time equivalent (FTE) basis employment; sales
turnover; purchase and related costs; profits; appropriation of earnings; R&D expenses; capital
investment; major products broken down in intermediate and final goods.
6.125. The Survey of Trends in Business Activities of Foreign Affiliates (Inward FATS) is conducted
based on Japan’s Statistics Law. The 2013 survey covered 5,463 companies of which 60.3 percent
replied. The following enterprises were covered:
i.

A company in which more than one third of shares or holdings is owned by foreign
investor;

ii.

A company funded by a domestic company (in Japan) in which more than one third of
shares or holdings is owned by foreign investors, in which the total ratio of the
foreign investor’s direct and indirect investment is more than one third of the shares
or holdings of the company concerned;

iii.

Companies that fall under (1) or (2) above, in which the principal foreign investor’s
direct investment ratio is more than 10%.

6.126. Inward FATS Information collected include the company outline; operation status;
employment on a FTE basis; type and function of the establishment (i.e. is the affiliate in Japan a
regional headquarter); sales turnover; purchases of which imports; payments to the principal foreign
investor (dividends, interest on loans; royalties); R&D expenses; plant and equipment investments;
profit and assets.

Chapter 7

Surveys of Persons and Households, and Population Censuses

7.1.

Scope. This chapter discusses the issues related to the use of population censuses, surveys of
persons and households and their organization in the context of data collection for measuring the
international supply of services. Section A provides an introduction and a summary of good practices,
The chapter promotes an integrated approach and cooperation between related statistical domains
(e.g., BOP and tourism statistics) by describing these sources and their use in compiling statistics on
the international supply of services (Section B). Section C covers population censuses. Section D
covers household surveys, construction of the target population and sampling frame; conducting a
household survey; measuring travellers’ expenditure using a household survey; seasonality issues; and
reference period. Section E describes labour force surveys, particularly the development of a labour
force survey module for mode 4. Section F deals with border surveys and provides details on
organizing surveys at different border points, including airports, road borders, railways, and cruise
ships. Section G provides an overview of complementary surveys of persons traveling.
A.

Introduction and a summary of good practices

7.2.
This chapter mainly refers to population censuses, household surveys (including labour force
surveys) and border surveys. Surveys of persons and households can be used to compile international
transactions in services in which individuals (or households), as major purchaser or supplier of
services, can be relatively easily identified for survey sampling purposes. For trade in services this
would typically refer to travel-tourism, transportation, and mode 4 related transactions, but with the
growing use of e-commerce by households, it may be relevant to consider other types of services or
modes. Such sources as well as population censuses can also be used to collect non-monetary
information, again in relation to mode 4 as well as mode 2. With respect to collecting balance of
payments services data for households, compilers should consider if the information is not already
covered by another data source such as described in this guide. However in some cases it may be
necessary to turn to surveys of persons or households, where these are important for trade in certain
services.
7.3.
Population censuses, person and household surveys are generally used as major data sources
for other statistical frameworks, such as social and demographic statistics, tourism statistics etc. They
can also be used to collect information on economic transactions. Persons and more generally
households are in many economies important economic agents in the context of the international
supply of services. The usefulness of such sources will need to be carefully assessed by compilers for
each type of information that could be used. Therefore, an integrated approach is imperative to the use
of such sources for the purposes identified in MSITS 2010 and this guide. Following the basic
principle of efficient statistical work – collect once, use many times (multiple-purpose use) – it is
good practice to design a survey such that its results may be used for compilation in several statistical
domains to ensure the highest possible efficiency. Given that the organization and conduct of these
surveys will be the responsibility of different units in a statistical system, it implies that if a need is
identified compilers actively seek cooperation in all stages of the statistical process with these units.
For example, such a close cooperation is absolutely essential in border surveys which have to provide
data for both tourism statistics and statistics on the international supply of services. Such cooperation
will need to be formalised through an institutional arrangement (see Chapter 3). Cooperation with
demographic statistics is important as well, for example, to ensure the proper use of population
censuses in the context of the organization of the household surveys and as a benchmark in grossingup their results.
7.4.
Population censuses, surveys of households and of persons can be used for various purposes
to collect the required information, but none can serve all needs. This guide suggests the following:

Population censuses could be used to gather benchmark information (in particular to compile
outgoing mode 4 data, but not only limited to this) or used for household sampling. This
could be done by adding questions directly in the census, or by developing a tailored module
to the population census.

ii. Household (limited to outbound travel) and border surveys are particularly relevant for
collecting data related to international travel (including transport), such as on the
expenditure/consumption of households or persons while outside their country of residence.
Such sources could also be used to collect other information on characteristics of those
travelling. A breakdown by purpose of travel and by type of product consumed should be
collected to the extent possible (as a first priority according to the recommendations of
MSITS2010/BPM6 and IRTS 2008, and with further items if relevant to the compiling
economy).
iii. For border surveys it is important to ensure that no category of persons are excluded from the
survey (e.g. cover persons beyond the definition of visitors used in tourism statistics), but a
clear distinction is made for each category so the compiler can select the population of
interest to his/her statistical domain.
iv. Labour force surveys can be used to collect various types of information in relation to the
needs of MSITS2010, for example, for the mode 4 variable “number of persons or trips”,
either by adding directly questions in the labour force survey, or by developing a specific
module (see suggestions in Box 7.3). A labour force survey could be used for collecting more
targeted information on outbound business travel (mode 2 consumption for personal purposes,
mode 2 consumption for business purposes).
v. Household surveys could also be used to collect data on
o consumption of other services (e.g internet purchase/consumption of services, mode
4),
o International provision of services of members of households, in particular for selfemployed persons (modes 1 or 4)
vi. Complementary surveys of persons traveling (e.g. at places of accommodation or in sites of
tourism interest) could be used, but it is not a general recommended practice (could be done
under certain circumstances). Indeed there are a number of challenges in using these sources,
i.e. identification of non-residents, subsets of the population, stay not terminated at the time of
surveying, etc
vii. There could be interest in having some surveys targeting categories of persons given their
specific (spending) characteristics (e.g. students, patients).
viii. The use of household surveys or surveys of persons should be carefully looked at given the
cost of them and the likely prevalence or sparsity of the activity within the broader
population. Increased response burden on the households/persons and the burden for the data
collection as well as the reliability and relevance of data obtained should also be considered.
ix. A particularly useful source for mode 4 (and related mode 2) would be business travel
surveys. Synergies could also be found in this context with labour force statistical sources
(e.g. comparing/joining samples).
x. Given that there are strong synergies with related statistical frameworks (e.g. tourism
statistics, labour statistics) or that surveys of persons or households or population censuses are
primarily designed to serve other information needs and given that it is considered good
practice that one source serves multiple needs, compilers are strongly encouraged to find
73

synergies and discuss the possibilities of using these sources to respond to the MSITS 2010
data needs. Chapters 2 and 3 provide more information on legal and institutional frameworks.
xi. When designing the data collection system using surveys of households or persons, special
attention should be brought to the sampling/selection of households or persons, selection of
informants, cost, response burden, sparsity of responses obtained etc. Compilers should
weight how these sources could be used for the various data needs identified. In particular
increasing the size of samples may significantly increase costs and alternatives may involve
using auxiliary information to better target the population to be covered.
B.

General Purpose and Description of Population Censuses and Persons and Households
Surveys

7.5.
A population census, hereinafter referred to as a census, provides at a specified time reliable
data on social, demographic and economic characteristics, of all persons in a country or in well
delimited parts of a country, especially for small geographical units. This includes information on the
size, composition and spatial distribution of the population in addition to socio‑ economic and
demographic characteristics. It is a major source of social statistics, with the obvious advantage of
providing reliable data—that is to say, data unaffected by sampling error. In general the census
collects information for each individual in a household, usually for the whole country or for well‑
defined parts of the country. Censuses are conducted periodically in the majority of the countries in
the world. The international recommendation is that a national census be taken at least every 10 years.
Some countries carry out censuses more frequently because of the rapidity of major changes in their
population and/or its housing circumstances. For more information on population censuses, compilers
are encouraged to refer to the Principles and Recommendations for Population and Housing
Censuses, Revision 2 (UN, 2008).86
7.6.
SNA 2008 defines a household as a group of persons who share the same living
accommodation, who pool some, or all, of their income and wealth and who consume certain types of
goods and services collectively, mainly housing and food.87 In other words they are groups of persons
with common economic interests.
7.7.
The internationally endorsed recommendations for the organization of household sample
surveys are contained in the UN publication Designing Household Survey Samples: Practical
Guidelines.88 The publication provides recommendations and good practices in such key topics as the
planning and execution of surveys; sampling strategies; sampling frames and master samples;
documentation and evaluation of sample designs; construction and use of sample weights; estimation
of sampling errors for survey data; non‑ sampling errors in household surveys; data processing for
household surveys. Many countries have in place household survey programmes that include both
periodic and ad hoc surveys.
7.8.
The compilers should familiarize themselves with the recommendations contained in the
publication referred to above in order to be in a better position to participate in the planning and

Housing censuses may also be of interest. A housing census collects information pertaining to all liv ing
quarters and occupants thereof in a country or in a well delimited part of a country.
87
SNA 2008, paragraph 24.12. SNA 2008 recognizes the importance of the household surveys, but, at the s ame
time it is noted that the conventions adopted by su rvey statisticians and those of national accountant s are not
always the same. A household expenditure survey for example may not include estimates of imputed rental of
owner-occupied dwellings or own account production. It may measure income after tax and measure
expenditure on a cash and not on an accrual basis. Compiler should be aware of these limitations when using
these sources.
88
Available at: http://unstats.un.org/unsd/publication/seriesf/Seriesf_98e.pdf.

conducting of the household surveys in their countries for purposes identified in this guide. Household
surveys are a widely used statistical tool89 . Many countries have therefore developed their own
detailed methodological guidelines on their design and conduct90. Such guidelines are periodically
reviewed and amended as necessary. It is good practice that compilers actively participate in this
process and develop amendments which would ensure that relevant needs identified in MSITS 2010
are taken into account.
7.9.
Compilers are also advised to review the IMF publication International Transactions in
Remittances: Guide for Compilers and Users which suggests that household surveys or labor force
surveys could be used to collect information relevant to mode 4 by including a number of specialized
modules or questions in existing surveys, or that specialized surveys could be conducted through
which relevant households would be identified. This would help analysts understand the relations
among the supply of services, employment status, etc. This should be done by identifying ways of
specifically targeting the population of interest in order to avoid increases in the cost of using such
sources, the lack of results obtained, as well as the possible increase of response burden.
7.10. As any sample survey, entities (enterprises or individuals) surveyed must be representative of
the population. Since there is no register for those traveling, compilers should look at administrative
records or other sources that could play a role of register. For some countries, even measuring the
number of people crossing the borders represents a challenge; for example, cases in which there are
border unions (as is the case for EU countries).
7.11. A survey of persons is a data collection pertaining to (specific groups of) individuals.
Household surveys provide access to persons by first selecting households. Compilers could transform
household surveys into a sample of persons by the following procedure: at the beginning of the
interview, the interviewer asks how many persons above a certain age are permanent members of the
household. From this range of persons the target person (e.g. the oldest, the second-oldest, etc.) is
chosen by a random procedure. The interview is then conducted with this randomly chosen target
person only.
7.12. Due to the unique nature of tourism (as described later in this chapter), sample surveys of
households and of persons (e.g., border surveys) are often used to collect information for travel
imports (debits) and tourism expenditures. Thus, given the importance of tourism and travel, some
parts of this chapter will frequently refer to the surveys conducted for tourism statistics purposes.
Compilers should therefore have knowledge of the basic tourism statistics concepts in order to make
appropriate use of such surveys in the context of MSITS 2010. Basic concepts of tourism statistics are
provided in box 14.2 of chapter 14 of this Guide. For more detailed information, also see
International Recommendations for Tourism Statistics 2008 (IRTS 2008) and its accompanying
compilation recommendations.91
C.

Population census

7.13. Population censuses can be used to benchmark some of the data collected through other
means or estimated.92 It could also be used as a sampling frame for surveys. This could be relevant for

See also the International Household Survey Network at www.surveynetwork.org.

An example, household surveys are extensively treated in the Statistics Canada publication “Survey Methods
and Practices”, available at: http://www5.statcan.gc.ca/bsolc/olc-cel/olc-cel?lan g=eng&catno=12-587-X.
91
See IRTS 2008, paragraphs 2.4 – 2.12. Add referenc e to ITRS 2008 compilation guide.
92
It should however be kept in mind that an increasing number of countries are moving from traditional
questionnaires to census a population to a register-based census, which may reduce the potential use o f this
source in the context of this Guide.

persons moving temporarily abroad under a mode 4 regime and to obtain a sense of medium to longterm developments. This includes information on the importance of outbound mode 4 for the country.
More specific surveys or modules may then gather additional detail focusing on employment at home
and abroad and additional questions relevant to mode 4. For example, to identify more specifically
persons who went temporarily abroad in the context of mode 4, one could add the following
questions:
i. Did you travel abroad for work in the past [x] 93 months? Yes/No
ii. If Yes, for what purposes?
- fulfil a contract.
- negotiate a contract or establishment of affiliate.
To avoid any confusion with the work related movements not covered by mode 4 work, questions and
response alternatives are needed to identify the country of residence of the employer at the time.94
7.14. If this is identified as a potential source of information, compilers are encouraged to approach
the staff in charge of the census. The contents of the questions respond to various needs, and it is once
again necessary to analyse if and how such a source could be used (i.e. either existing information or
by adding a module/questions to the census). Given the long preparation time necessary to develop
and conduct it, compilers need to contact the staff in charge of the census at the early stages of its
design.
C.1

Country experience – Canada

7.15. At the time of writing Canada did not collect mode 4 data. However, some of the questions
could be useful. For example, the Canadian census questionnaire95 identifies if the person is a selfemployed or an employee, his/her occupation, the industry of the firm, and if this person was
generally working in or outside Canada during the previous week. Some countries which often have
part of their population travelling abroad for work purposes may be interested in knowing more on the
purpose of such travel to determine whether they are mode 4 relevant. When information is collected
on the duration of residence, information could also be gathered to find out if the employer is a
multinational or not, or if the person is in the country for mode 4 relevant purposes or not (i.e. with
employer based in country of origin or not).
D.

Household surveys

7.16. As indicated above, household surveys are primarily used to collect socio and demographic
information, but can also be useful sources for collecting information as identified in MSITS 2010, in
particular for the following purposes:
i. For households as consumers of services (debits/imports):
-

Collect information on expenditure patterns of households, or some of its members,
which may have travelled abroad. This information will mainly be used to compile the
payments for international passenger transport or travel (as well as on outbound tourism
expenditure). Socio-economic characteristics of these households as well as the
characteristics of their travel abroad may also be a useful source of information in
relation to the variable number of trips/persons for mode 2.

Number of months to be determined by the frequency of and period covered by the census.
See Chapter 1 and Box V.2 of MSITS 2010 for more i nformation on the identification of the employeremployee relationship.
95
http://unstats.un.org/unsd/demographic/sources/census/quest/CAN2006enL.pdf.
94

Collect information on expenditure of households on payments/imports for (other)
services. This may particularly be relevant for Internet-based transactions or services
provided through telecommunication networks, in particular expenditure related to
intellectual property products (e.g. software or audiovisual products), as well as when the
supplier is physically present to provide his services (e.g. health services providers,
cleaning personnel).96

ii. For households or its members as suppliers of services (credits/exports):
-

Collect information on the supply of services to non-residents by members of the
household:
1. Members of the household supplying services to non-residents by traveling
abroad (i.e. mode 4), either as self-employed (value of contracts and information
on characteristics of business travel abroad, e.g. business-mode 4- trips) or on
behalf of their employer (only information on characteristics of business visits
abroad). A labour force survey may be particularly relevant in this context (see
section E).
2. Members of the household supplying services to non-residents without traveling
abroad, e .g. through the Internet (mode 1) or through the presence of these nonresidents in the premises of the household (mode 2), most likely to be accounted
for under travel credits (or inbound tourism expenditure or consumption).

7.17. Some specific household surveys, such as business travel surveys could be a timely and
relevant source on outbound temporary labour mobility, which could be useful to collect (detailed)
information on mode 4 (see the example of Netherlands below), as well as mode 2 for this particular
category of business travellers. There may also be scope to think further into the direction of
extending the usefulness of both business travel data and labour force data by joining samples.97
7.18. Section B provided a general introduction to household surveys. In principle, all resident
persons are attached to one resident household and only one. Therefore, in order to survey residents,
resident households can be used as the sampling frame, as it will ensure total coverage, if the
sampling frame is up-to-date. 98 The information can be collected by interview, through the visit of an
interviewer to the household. This is the most common procedure, as it facilitates the control of the
survey.
D.1

Household surveys and services items of EBOPS 2010: travel, transportation and
other services

7.19. Using a household survey to collect information in relation to residents traveling abroad (i.e.
outbound travel) will serve the needs of the balance of payments (travel and passenger transport
items) as well as the needs of tourism statistics in particular with respect to international tourism
consumption (including international passenger transport). Residents as members of households will
be asked (among other questions) about their travel abroad, the circumstances and characteristics of

Compilers will have to ensure that only service contractual relationships will be taken into consideration, see
Chapter 1, Chapter 14 and Chapter 16.
97
For a discussion on the principle of modular and integrative data collection refer to
https://circabc.europa.eu/sd/a/45f2b51e-9ba4-4ad1-a b9a-d09a5bb02af4/DSS-2014-Sep07.6%20VIP%20streamlining%20and%20integration%20of%20the%20European%20social%20surveys.pdf.
98
Note that some special dwellings may be more diffi cult to enumerate - for example, retirement and pal liative
care homes, trailer parks and long-stay hostels, re sidential colleges, holiday homes, etc.

the travel, and the expenditure attached to them (products and values). It should be underlined that
travel (and tourism) refers to individuals, not to households. Nevertheless, when persons travel
together and belong to the same household, data on (shared) expenditure by the travel party can be
collected. Households are used exclusively for selection purposes, but not for providing results:
households as such do not travel.
7.20. The fact that household surveys are generally conducted at the home of the household will
facilitate the possibility for the persons been interviewed to check some information, for instance
credit card invoices, etc. The interviewer can collect the information on paper questionnaires, or using
some electronic device which makes it possible to have a first control of the consistency of the
information (see Chapter 21). Such procedures can be costly; as a consequence some countries use
rather telephone interviews: the procedure is cheaper, but with less possibility of asking more
sophisticated questions. The telephone interview could also be used as a first step to approach
respondents through some simple questions, and then the respondents could be asked if it is possible
to send a questionnaire with deepening questions.
7.21. The travel item of resident/non-resident trade in services statistics (as well as tourism
consumption) is expenditure oriented. Compilers will need an estimate of the value of expenditure per
person in a given period, and this value will vary depending on numerous factors. In addition the
information on characteristics of those traveling and their travel itself may be of interest to compile
other variables. Compilers will therefore need information as shown in Box 7.1. The information on
characteristics of travel which has to be collected might differ from person to person within the same
household, even though they have travelled in the same party for this trip (e.g. their purpose of travel
might be different, and also their expenditure).

Box 7.1 Information needed for international passenger transport, travel and their
breakdowns, as well as for mode 2 number of persons/trips
The points below summarize the information to be collected from surveys of persons or households to
compile statistics on the international supply of services (e.g. international passenger transport, tr avel and its
breakdown). As shown below this is strongly related to information needs on tourism expenditure.
Information is needed on:
i.
The characteristics of travel (such as, main purpose identifying clearly the different tourism and other
travel purposes, duration, origin and destination, modes or transport, forms of accommodation used,
organization -i.e. with or without package, traveling alone or in a party, with persons of the household or
with others, etc.). This usually does not present major difficulties;
ii. The characteristics of expenditure: Expenditure to be included refers not only to that part directly paid
for by the persons traveling themselves, but also t hat paid by others for their benefit. This is true, both for
travel and for tourism expenditure. The consequence is that, for each trip, and each category of expen diture,
it is necessary to ask: (i) whether there has been an expenditure, and the value of expenditure, (ii) who has
paid for it (oneself, a business, the government, other person of the travel party, another person, et c.), (iii)
the type of payment used (such as credit cards, cas h, travelers checks, etc.), (iv) the number of pers ons to
which the reported expenditure corresponded to, as an individual might pay for him/herself and for any
other persons within a travel party; and
iii. The breakdown of expenditure. Finally, it is important to take into consideration both the breakdowns
suggested in tourism statistics and in resident/non- resident services transactions.
Tourism expenditure should be broken down into the following functional categories. These
categories have been defined so as to facilitate re sponse.
1)
2)
3)
4)
5)
6)
7)
8)

Package travel, package holidays and package tours
Accommodation
Food and drink
Local transport
International transport
Recreation, culture and sporting activities
Shopping
Others

In balance of payments and in international trade in services statistics, the recommended breakdown
is:
1) International passenger transport
2) Travel, broken down into the following categories:
a) Goods
b) Local transport services
c) Accommodation services
d) Food-serving services
e) Other services
Of which,
Education services
Health services

Box 7.1 (continued)
It is important to note that beyond the breakdowns recommended above, compilers may also identify other
breakdowns of importance at the level of their econ omy (see CARICOM proposal for creative/cultural
services). In addition the breakdowns of expenditures in tourism and trade in services statistics are not totally
equivalent, as the orientation of the tourism statistics classification is by purpose, whereas in trad e in services
statistics it is by types of products; meaning, for example, that expenditure in fuel for the car woul d be
classified as “transport” in tourism statistics, an d in “goods” in balance of payments statistics. To compile
recommended breakdowns it will therefore be necessary to collect the necessary (detailed) information to
respond to all needs. Due to these circumstances, d ifferent designs of forms are possible: to observe
separately and with different frequencies travel and average expenditure per day; or to observe both,

flows and expenditure, simultaneously using a unique questionnaire. The same source could also be
used for associating information related to mode 2 trips/persons.
Challenges in collecting information on expenditures/consumption by type of product consumed. Regarding
the collection of data on the expenditure by persons traveling, there are a number of challenges in ob taining
data by type of product consumed. Visitors frequently share expenditure within a group of persons traveling
together. Persons traveling do not always know the amount of expenditure attached to their trip as the y do not
pay directly for all the expenses (accommodation, transportation, etc.): there might be agreements in
receiving some implicit or explicit payment in kind (transport, meals, accommodation); this happens fo r
residents traveling for personal purposes but this case might be even more frequent for those traveling for
business purposes. Persons traveling usually perceive their expenditure by "family" of products, not
necessarily being able to single each (detailed) type of products they are actually consuming. For instance,
separating goods from services might not be as stra ightforward as it seems such as for medical expense s
which may include drugs and services, education expenses which include cost for lectures and include books
and other services, transport might include goods (like gas), etc.). Persons traveling for business pu rposes
(mode 4 and others) , or those that move frequently from their country of residence to their country of
employment might not have a clear perception of the expenditure associated with each of their travel (in
particular for expenditure such as accommodation, e tc.). These aspects need to be taken into consideration
when devising questions in the survey form to collect such information.
Seasonality of travel and the survey organization. Travel is often highly seasonal; it cannot be obser ved over
a reduced period of time, and then be extrapolated to the whole year as travel, in particular that for personal
reason, will be influenced by climate conditions in the country of origin and of destination, periods of
vacation, etc. As a consequence, frequent observati ons will be required. It is a good practice to collect
information on a continuous basis, though results might be produced with a different frequency: quarterly, or
with other types of groupings, for instance, the peak season, and the low season, though seasonality might be
different depending on the purpose of the trip (business, work, study, personal, etc.).

7.22. Reference period. Additionally, because of memory effects, the period of reference to be
used, both regarding the travel/trips and the expenditure associated to them, has to be very short. This
Guide considers as a good practice if a month is used as a reference period in order to reduce
telescopic errors (improper date assigned to trips) and recalling errors (improper characteristics and
expenditure), see box 7.2. As a consequence, because most persons might not have made any trip
during such a short period of reference, the selected sample should be sufficiently large so as to
collect enough valid information.99 A solution could be to ask about travel in the last X months and
then deal with the memory effect by weighting up information relating to older travel.

99
However cost, burden and sparsity of responses con siderations are also needed. Alternatives to increa sing the
size of the sample may involve using auxiliary information to better target travellers in household surveys.

Box 7.2 Research on the effect of expanding the period of reference for reporting
tourism trips in Spain100
Research carried out by the Spanish Instituto de Estudios Turísticos (IET) confirms how
relying on a respondent’s memory for information covered by a domestic and outbound
tourism survey can cause two distinct types of errors entirely unrelated to sampling, that
have often opposite signs:
 Telescopic error: the actual date of an event (an expense, a trip) is unconsciously and
erroneously moved up to a more recent date.
 Pure memory effect: an event that in fact occurred during the observation period is
simply omitted, because, subjectively, by the time of the interview, it seems to have
occurred long before.
The IET developed a complex observation methodology using three overlapping samples, in
which respondents were asked about trips taken three months, two months and one month
before being contacted. As a result of this study, it appears very clearly that the longer the
delay in contacting respondents after the observation period, the fewer the trips they report.
For example, three statistically validated measurements for the number of trips taken in June
2006 by residents of Spain are as follows:
 12,991,044 if observed in July
 12,745,211 if observed in August
 12,247,920 if observed in September
Source: Guardia, T. and Garcia, S. (2008)
7.23. Issues to consider. As to the process of selecting households, individuals and travel/trips,
various possibilities are conceivable and each of them has their pros and their cons. Regarding the
selection of households:
i.
All households might be selected with equal probability; or with different probabilities,
taking into consideration differences in propensity to travel (case of the rural population
for instance; or when countries have a stratified universe according to level of income
as higher income is often positively correlated with propensity to travel); or
ii.
A moving panel of households can be used, that is renewed periodically; this allows to
design profiles of behaviour overtime; nevertheless, persons who frequently travel and
thus are absent when requested to answer the survey will tend to be substituted by more
sedentary persons that tend to travel less, thus skewing the results of the survey.
7.24.

Regarding the selection of informants:
i.
Select randomly one person within the household that will inform on his/her travel and
expenditure; the drawback is that much of the cost of visiting a household will be lost,
particularly taking into consideration that traveling over a short period is not so frequent
so that the randomly selected person might not have travelled but other persons of the
household might have;
ii.
Include only persons over a certain threshold (10 years, 15 years,…); this looks
attractive, but small children usually travel with other persons of the household, and
though the children usually do not pay themselves for their expenditure, their

100

Memory Effect in the Spanish Domestic and Outbound Tourism Survey (FAMILITUR) A paper presented
by Teresa Guardia and Sandra Garcia from the Spanis h Instituto de Estudios Turísticos to the OECD 9th
International Forum on Tourism Statistics, Paris, N ovember 2008.

iii.
7.25.

participation in a travel party decreases the average expenditure per person per day of
all the members of the party; or
Include everybody: this is often the method followed.

Regarding the selection of travels:
i.
All travels can be selected both for description of the trips taken and the attached
expenditure; this is often the method followed but when some persons travel frequently,
it might be a drawback with too large a questionnaire (but this might not be so frequent
if restricting to outbound trips); or
ii.
All trips can be counted (outbound and domestic, i.e. in relation to tourism information
needs) but the characteristics of the travel and the expenditure are only collected for one
of them (usually, the most recent one); this is usually the principle mostly followed by
compilers.

7.26. Using a household survey to collect the above-mentioned information could be done in
various ways. It is possible to attach a “travel/tourism” module to an existing household survey
(usually, a labour force survey or an income and expenditure survey), or to a design specific
observation procedures to observe the required variables. Because of the particularities of tourism and
travel mentioned above, the first alternative, though usually considered initially by countries, does not
result in obtaining all necessary data and should be complemented by a specific survey to observe
tourism and travel. Further information is provided in the ITRS 2008 compilation guide.
7.27. Data for other BOP services items could be collected through a household survey, such as for
example on the services consumed by households on-line, downloaded or received by e-mail or
telephone (e.g. legal services, medical advice, audiovisual or software downloads, gambling).
Although households may easily identify the type of online purchases they make, it may be difficult
for households to identify whether they are buying their service locally or if they are importing a
service and from which country. An additional difficulty is that the country of billing or marketing
may not necessarily be the one which is actually supplying the service to the household. Services
provided through mode 4 could be easier to identify, i.e. when the supplier of the service is physically
present to render the service. This identification of mode 4 transactions for relevant services could be
particularly interesting to serve the needs identified in this guide. However it will be necessary to
ensure that respondents understand the difference between an employment relationship and a service
contract. This information could be collected through existing household surveys (e.g. expenditure
and income survey, specialized survey on the use of the Internet or telecommunications network by
households), or could be collected by developing a specific module or survey. For all these other
services sampling will be important to consider, as any household could potentially be a buyer of
services. In addition many of the payments will most probably be low, and may be below thresholds
that have been established in the data collection system. Many of the indications provided above on
the selection of households and informants to collect information on travel/tourism and transport is
also relevant for other services. To summarize, although it may be appealing to approach households
to collect data for payments for services other than travel/tourism expenditure or related information
(including international passenger transport), compilers need to consider if such an approach would
provide meaningful and exploitable results for compiling other balance of payments services items.
D.2

Household surveys and (mode 4) receipts for exports of services

7.28. If the population of self-employed persons can be clearly established within households, it
may be relevant to collect data on the value of the contracts for services provided abroad by these
persons. Separately identifying those who travelled abroad to provide their services would be
particularly relevant to obtain a breakdown by modes of supply for self-employed persons. However
the issues identified above on identification of the population, informants, sampling (i.e. relevance of
this data source) are even more prominent in this context as the number of self-employed exporting
their services abroad may be relatively small and difficult to identify through the use of a (specialized)
82

household survey. Some specialized survey could help collect some information, such as the business
travel survey which is described below.
7.29. To collect information concerning mode 4 it can be considered to add a specific question in a
household survey about the reason for the trip (“business trip”). In combination with the sociodemographic characteristic “self-employed” it can be used as a basis. If such a survey is conducted as
a telephone sample compilers have to be aware that self-employed persons are difficult to reach and
therefore consider using a specific sample of mobile phone numbers in addition, or use a web-based
survey.
D.3

Household surveys and mode 2 and mode 4 number of persons/trips

7.30. Data on the characteristics of mode 2 and mode 4 trips (i.e. contractual service supply), or the
persons who are themselves traveling, could naturally be collected at the same time as the
expenditure/receipt data described above. The type of the information that could be collected through
household surveys is the following:
i.

on mode 2 for outbound travel (i.e. those traveling for personal purposes and those for mode 4
or other work purposes);

ii. on mode 4 for outbound travel, i.e. contractual service supply, intra-corporate movements and
services selling/establishment of commercial presence; and
iii. on mode 4 for inbound travel, i.e. import of services by households through mode 4contractual service supply/inbound; or inbound intra-corporate transferees if these become/are
residents of the compiling economy. For the latter the issue of identification of the relevant
population is particularly difficult.
7.31. Some compilers may use some household surveys to collect specifically information on
outbound business travel. As indicated above if such a survey is conducted as a telephone sample
compilers have to be aware that frequent business travellers are difficult to reach and therefore
consider using a specific sample of mobile phone numbers in addition or the development of a webbased survey. In general when using a household survey one would focus on a specifically targeted set
of the population, i.e. those travelling for business reasons, and may serve multiple needs such as for
national accounts, balance of payments or trade in services statistics. As it is targeted survey, some
specifically tailored questions could be included to help the compiler gather some information that
may be of interest for mode 4 in particular. Of course, as for other data collection described in this
chapter, the level of detail one can expect from such a data source is relatively limited. However by
targeting some of the questions, there are some possibilities to obtain more details on those categories
of specific interest to the compiling economy. Considering confidentiality issues and the sample size,
mode 4 data could be presented or aggregated to broad economic or geographical categories.
D.4

Country experiences – Austria

7.32. In Austria sample surveys of the Austrian population proved to be the most efficient solution
to gather information on domestic and outbound tourism expenditures. Every quarter a representative
sample of the Austrian population is asked about their travel behaviour with the help of computer
assisted telephone interviews (CATI). In the Austrian questionnaire travel expenses include “all
expenses related to the trip”. During the interview it is clearly stated that goods and services, bought
in the name of the traveller and made available without compensation (i.e., as a gift or invitation)
should also be included (for example invitations to dinner, expenditures for goods and services).
During the interview it is mentioned that imaginary rent for free accommodation should not be
included.
83

7.33. The advantage of CATI is that the spoken text can be standardized and better monitoring of
the interviews is possible. Supervisors, interviewer training, pre-tests and interviewing instructions
ensure the quality of the interviews. Plausibility checks are integrated to prevent typos and outliers.
Warnings are programmed for example to pop up during the interview if the stated expenses are
below specified expenditure minimums. This gives the interviewers the opportunity to mention again
that expenditures made on behalf of the respondent have to be included. These consistency checks
make sure that typos and outliers are identified (signals) and replaced (hard errors) immediately.
D.5

Netherlands - Survey on outbound business travel

7.34. The CBS of Netherlands has purchased a new survey on business travel to measure their
expenditure. It also extended the list of purposes of the trip, which could be (somehow) related to
mode 4. The Continuous Business Travel survey (CBS) is a web-based survey for outbound business
travel using the TNS-NIPO-base101 which consists of more than 150.000 people and is held every
second year whereas one measurement (retrospective) every quarter in the respective year is executed.
The CBS is measuring business trips (overnight stays), same day trips and number of travellers. The
sample consists of Dutch residents aged 18 years and older, identified as working population and who
did at least one business trip abroad in the period of review. Every quarter of a year 10.000 panel
members from the TNS-Nipo-base receive a screening question to determine if they belong to the
sample. Out of them 1.000 Dutch residents receive electronically a complete questionnaire (to be
returned within two weeks). The survey is a computer assisted web interview. Travel item is grossed
up to the stratification remarks of the traveller going abroad (gender, age, region, municipality and
education) to the total of the Dutch working population.
7.35. The questionnaire was extended regarding the purpose of the trip which can be related to
mode 4. It suggests 15 types of activity conducted abroad (of which "marketing and sales",
"installation and repair", "research, teaching, consultancy", "visit to clients", "visit to head office,
branches" etc.) as well as indications about the nature of employment (self-employed, employee and
civil servant). Additionally, the survey contains variables on the professional group e.g. farmer, higher
professional educated, owner of an enterprise etc. as well as on the level of education.
7.36. Therefore, the group of business travellers providing services abroad can be narrowed down
in the sample. This information could give first indications on the number of persons and could be
linked to the value of service in the respective service category. However, to derive statistics on mode
4 transactions, one would need to make assumptions, in particular on the size of the activity. The
business travel survey gives some fist views about the number of persons carrying out activities
abroad and also their allocation to some service categories. For analytical and trade negotiation
purposes the partner country (destination country of the trip) might be interesting as well (the
questionnaire contains a separate question on which countries were visited during the business trip
abroad). Considering confidentiality issues and the sample size mode of supply data could be
presented or aggregated to broader economic or geographical categories: e.g., EU-28, other country in
Europe, Asia, Africa, NAFTA, and ASEAN.
E.

Labour force surveys

7.37. As indicated above, a labour force survey (LFS) is a standard household-based survey of
work-related statistics. In the context of statistics on the international supply of services, LFS could be
used for the following purposes:
i. The most relevant use of this source is to collect information on mode 4 outbound
movements, i.e. contractual service suppliers, intra-corporate transferees/employees directly
recruited by foreign affiliates and services sellers/responsible for setting up commercial
101

TNS-NIPO is a Dutch survey agency.

presence (and to a lesser extent inbound for the two former, of course if the persons become
resident).
ii. Collect information on consumption/expenditure of those travelling abroad for business
purposes (mode 2, relevant for travel debits and outbound tourism expenditure), i.e. goods
and services actually consumed by the person in the context of his travel, not on behalf of his
employer.
iii. Collect information on acquisition of services on behalf of their employer, or more generally
on a business to business basis (mode 2).
iv. Collect information on sales/exports of services on a business to business or business to
consumer basis, which will mainly be relevant for self-employed persons (modes 1, 2 or 4).
7.38. Labour force surveys are widespread and reasonably standardized. To these surveys, modules
might be added for multiple-purposes. For example, such a module could be designed to get
information on persons moving in the context of mode 4. Another possibility is to use the survey
system but change the subject-matter in each round, i.e. from questions on the labour force to
questions related to mode 4 movements.
7.39. Labour force surveys can identify the individuals in the household that are self-employed
(employers or not) or employees, as well as the occupation of their (main) job and the industry or the
type of production of the employer if the individual is an employee (and in some instances the size of
the company), or that of the self-employed persons. This is of interest in the context of mode 4.
However the important factor for identifying mode 4 is whether the individual went abroad in the
context of his/her work, but remains based in the home country where his place of employment is.102
7.40. To use labour force surveys with the purpose of separating mode 4-related movements from
international labour mobility, requires including a limited number of questions on (recent) visits
abroad by household members for the purpose of work. For example, questions about the contracting
parties, the duration and forms of payment, etc. This possibility may be mainly relevant for sending
countries in respect of mode 4 which covers contractual services suppliers and those travelling for
negotiating purposes. In the case of intra-corporate movements as well as migration of self-employed
persons, these surveys could also be relevant for receiving countries, but with a different set of
questions.
7.41. For those countries/regions where mode 4 is potentially important and/or the population is
relatively well covered by the sample, it may prove useful to add appropriate questions in the surveys.
A specific module could be developed similar to the one proposed by ILO for labour migration.103 In
order to identify such movements, some questions as illustrated in box 7.3 could be added to the
questionnaire (or developed in a specific labour force survey module on mode 4). If deemed relevant
other questions could be added to gather more information as outlined at the beginning of this section,
however it is once again important to note that compilers need to analyse the eventual benefits and
mainly the costs associated with the use of such a source, in particular if they want to go further than
the purely mode 4 information needs.
7.42. It is recognised that such an approach cannot be generally adopted without a thorough
analysis of the importance of the different mode 4 categories for a country. Some are more concerned
with the fact that their workers are temporarily sent abroad to fulfil services contracts, whereas others
are more in the position of receiving many self-employed persons, or intra-corporate movements of
personnel (i.e. incoming) who operate from a base in the host economy (as per relevant modes 3 and 4

102
103

See chapter 1 and MSITS 2010 box V.2.
At the time of writing the module for labour migration had already been applied in Thailand and Armenia

commitments, often combined). In addition, this may be limited to a specific region within a country;
and such an additional module may then only be envisaged for respondents of that particular region.
7.43. Potential benefits could exist if countries with an LFS would be willing to cooperate at the
national and/or regional level on producing such labour migration or mode 4 statistics for mutual
benefit. The employment status in combination with demographic indicators has the potential to be
useful for the compilation of mode 4 statistics. Another potential benefit of cooperation or
coordination could be to ensure consistency with enterprise surveys that have similar subject matters.
7.44. It is therefore good practice to cooperate with other agencies which have common or close
interest in collecting such information, e.g. national accounts, balance of payments, migration or
tourism statisticians. Integrated enterprise and social surveys such as the Structure of Earnings Survey
or the Canada Workplace and Employee Survey, USA National Compensation Survey, could also be
built to have information on duration at workplaces abroad collected and they could be potentially
valuable sources covering information on wages, industry and other relevant characteristics. It is
therefore important that compilers contact the units who are in charge of the respective surveys or
subject matters. Chapters 2 and 3 provide more information on legal frameworks and institutional
arrangements.
E.1

Country experiences – European Union member states

7.45. The European Labour Force Survey (ELFS) programme is an example of the (at the time of
writing unexploited) possible use and benefits of cooperation at the regional level on producing such
labour migration or mode 4 statistics for mutual benefit, given that all member countries of the
European Union and EFTA have (reasonably well) coordinated their LFSs. 104 An additional
illustration of the possibility offered by related sources is the European Working Conditions Survey105
which provides an overview of working conditions in Europe using different indicators for working
conditions of both employees and the self-employed. The employment status, which is included in
this survey, as well as demographic indicators, once combined, have the potential to gather
information useful to compile mode 4 statistics.

104

For more detailed information on the results, com mon practices, methods and definitions of the European
Labour Force Survey, see the EU LFS dedicated section on Eurostat's website at:
http://epp.eurostat.ec.europa.eu/portal/page/portal/employment_unemployment_lfs/introduction.
105
Source: http://www.eurofound.europa.eu/surveys/ewcs/index.h tm.

Box 7.3 Sample questions for inclusion in a labour force survey module on mode 4
The following questions could be included for sending countries:
A.
Did you work in a country other than your country of residence in the past period (e.g. year,
months)?
B.
In the past year were you employed in a country outside your country of residence?
C.
If so, in which country is your employer?
D.
Were you employed by an employment agency?
The following questions would then be necessary to capture further information on contractual service
suppliers (i.e. those who are self-employed, or tho se who are resident and employed in the compiling
country, i.e. those who replied "No" to question B):
E.
In the context of your work did you travel abroad in the past year?
F.
How long did you stay abroad?
G.
How many trips did this entail?
H.
Did you travel abroad to fulfil a service contract in the destination country:
i. with a third-part client?
- was it a firm?
- was it a household?
ii. with an affiliated firm?
It may be of interest to obtain further data on ser vice sellers/persons negotiating the establishment of
commercial presence:
I.
Did you travel abroad to negotiate a contract?
- Was it a service contract?
J.
Did you travel abroad to negotiate the establishment of an affiliate/branch?
Some questions could be added to identify if an employee travelled abroad on his/her own, or with
colleagues,:
K.
If you are an employee, did you travel with some of your colleagues? If so how many?
Compilers should pay attention to the issues of double counting, interpretation and extrapolation of
information obtained for this question. A solution could be to design processing algorithms to avoid
double- counting.
For a self-employed/employer the following question could be asked:
L.
Did you send some of your employees to fulfil a contract with a client abroad? If so, how many?
For receiving countries it may be useful to gather information for intra-corporate transferees and selfemployed migrants. The identification of the country of residence (origin or in last period) may be a first
indication. If the individual comes from a foreign country, then the following questions would be useful:
M.
Did you come to work in an enterprise affiliated wi th your previous employer?
N.
Did you enter this country to develop a business?

Border surveys

7.46. Border surveys are a type of survey of persons. Border surveys apply to persons who are
traveling. They are counted and/or surveyed when they enter or leave the country. Such sources can
be used to collect information on characteristics of trips/travel, and are useful for deriving estimates of
travel and tourism expenditures as outlined in Box 7.1, however, they could also be used to collect
information to compile data on the number of mode 2 and mode 4 trips/persons. In 2005, UNWTO
carried out a study “Tourism as an International Traded Services” in which 26 out of the 34
responding countries (in a sample of 48 countries considered to be representative of a wider group of
countries) specified using border surveys. Given the strong link with tourism information needs, it is
good practice to identify how stakeholders could cooperate to collect the details necessary to respond
to all information needs. Chapter 2 and 3provide more information on the legal and institutional
arrangements.
87

7.47. When to survey? The circumstances in which information is collected are relevant for the
quality of the data. In the case of inbound flows, this information has to be collected from the person
as he/she is leaving the country in which he travelled. In the case of outbound flows, this information
can be collected, either at the moment of re-entering the country of residence, or after the travel,
within the usual environment of the person. This allows observing actual expenditure because they are
interviewed once and this expenditure has already taken place. If they would be asked about their
expenditure before their travel ends (e.g. at entrance for inbound travel and at exit for outbound
travel), then we would be observing their expected expenditure. The particular case of inbound travel
makes the observation challenging, as time is short upon departure, and the person might not be in the
appropriate mood to answer a complicated questionnaire. Additionally, if the flows of certain
categories of persons are small, it would possibly require a sample of a relatively bigger size or a
deliberate over-sampling with suitable correction methods. It is strongly recommended that a specific
set of questions is designed to enhance the accuracy of estimates and break them down into
categories.
7.48. UNWTO developed a model border survey included in the IRTS 2008 Compilation Guide.106
The model questionnaire covers 5 parts: A. - Travellers, B.- Means of transport, C. - About your stay,
D. - Acquisition of services in your country or other country before arriving in our country, E. Acquisition of goods and services in our country booked or paid either before, during or after the trip.
The UNWTO Compendium of Tourism Statistics also presents a broad range of data and indicators
that are being collected by a significant number of countries in line with the IRTS 2008 which could
give an indication of what data should be collected to compile a basic set of data.
7.49. Organization of border surveys. Border surveys need to have very different organizations
depending on whether they refer to airports, land borders, sea ports, ports on rivers, or to the case of
cruises. The more extended practice refers to airports. The type of organization of the surveys, the
sample design, and the questions that can be asked differ also extensively, because of the different
circumstances in which the surveys can be held, in particular the time that can be assigned to the
interview or to filling in the questionnaire.
7.50. Means of conducting border surveys. As other surveys the border survey can be conducted
through different means such as (i) distributing the questionnaires on paper to pre-selected persons
and having them answer (provided the questionnaires are available in different languages, with clear
instructions and easy to understand); (ii) using a paper questionnaire, but filled by an interviewer; or
finally, (iii) an electronic questionnaire with an interviewer using a handheld device and collecting the
information and feeding it directly into a validation program. In comparison to a household survey,
border surveys have the characteristic that it is impossible or difficult to get back to the informant if
some information turns out to be inconsistent. 107 Consequently the questionnaires should be clear
enough to minimize errors in the provision of the required information. For example, a well-designed
electronic questionnaire to be filled directly on a tablet computer by a person as he is leaving the
country visited - at some strategic location where he/she has to wait anyway - could facilitate the
conduct of face to face border/travel surveys.108

106

International Recommendations for Tourism Statistics 2008 Compilation Guide, annex 1.
Some countries have tried returnable questionnaires, but this has not proved being very useful, as those that
actually return it are very few, and possibly do not have the same profile as those that neglect to answer.
However, New Zealand developed an on-line questionnaire where random sampling is done after airport
security, and respondent provide information at a later stage.
108
The Computer-Assisted Personal Interviewing system (CAPI) technique was used in 2013 in Morocco for the
first time in the survey on travel conducted by the Ministry of Tourism at borders. The CAPI technique ensures
conducting interviews by means of tablet computers in which questionnaires are installed. Answers are entered
in computer and data is sent in online to the database after predefined checks and control processes. This sort of
tool helps improve quality, reduce costs and save time.
107

7.51. Treatment of traveling parties. Because visitors often travel in parties, in which the most
important expenditures are shared, some countries have tried to use different questionnaires for
persons traveling alone and for those traveling in parties so as to facilitate the collection of
information (since common expenditure would only be collected once for the whole party, while
personal characteristics of party members need to be collected for each of them). This might be a
good practice, especially if dealing with travel for recreational purposes.
7.52. Working with specific characteristics. For borders in which the flow of persons is almost
permanent (land borders, airports), and in which it is difficult to maintain the periods of observation
overnight, it may be useful to work with specific characteristics (for instance, in air, given origins or
destinations often group flights in certain time brackets).
7.53. Finally, an observation should be made on the survey procedures themselves: when observing
the flows of persons over the border (with the exception of cruise ships), it is not possible to
determine a priori whether the person is a resident (leaving or returning) or a non-resident (arriving or
leaving) yet only those having terminated a trip should be interviewed on their actual trip. The
imbalance in the questionnaire responses corresponding to the different situations might result in false
information.
7.54. Compilers should refer to Box 7.1 for information on the characteristics and breakdowns
needed. Box 7.1 in particular outlines the difficulties regarding the breakdown of expenditure by
product and on issues of seasonality and the conduct of the survey. Care should be taken nevertheless,
that the extension and details that can be collected will depend to a large extent on the duration that
can be assigned to the interview. For example, in the surveys at a land border, the time that can be
dedicated to the interview is very short so that it might not be possible to enter in many details and the
questionnaire should contain only the most essential types of questions.
7.55. It is important to ensure that no category of persons are excluded from the survey (e.g. cover
persons beyond the definition of visitors used in tourism statistics), but a clear distinction is made for
each category so the compiler can select the population of interest to his/her statistical domain.
F.1.

Details on the organization of the surveys at different border points

7.56. Airports. The observation of travel flows through airports and their corresponding
characteristics and expenditure should be relatively straightforward as the flows are clearly defined
(various types of administrative records should be available from airlines, customs control, airport
administration, etc.) and persons traveling have time to answer (either when they leave: after checking
in and before boarding; or when they arrive: after disembarking and before leaving the terminal). This
makes it possible to interview them at the end of their travel and observe the corresponding
expenditure. In some cases, it might also be possible to distribute questionnaires on board the aircraft,
before arriving or just after take-off, provided that all airlines cooperate.
7.57. In the case of airports, the statistical design should take into account the information provided
by the airlines on anticipated flights, their origin or destination, their capacity and expected number of
passengers. On this basis, and making some assumptions on the proportions of passengers residing
abroad or in the country of reference, it is possible to define a sample. For a given flight, some
countries might decide to survey all passengers, whereas others prefer selecting respondents (on a
quota basis or systematically), or even allow the interviewers to select the persons they interview. For
the sake of completeness and accuracy, it is preferable to interview all persons of a selected flight.
Quotas, usually based on country of residence, provide skewed results, as the country of residence is
not a good explanatory variable of expenditure, if not associated with others, such as duration of stay
or purpose of trip. In the case of interviewers selecting informants, the skew can be all the greater. An
additional difficulty may reside in interviewers who do not have access to airline lounges for the
conduct of surveys and can only access potential respondents at departure gates. It is suspected that
89

those that use the services of airline lounges have significantly different expenditure patterns from
those who do not have access to the lounges, which can be a serious bias for the estimates.109
7.58. Land borders (road). The situation at land borders (road) is very different from that
previously described for airports as, frequently, the flows of persons over the border is not well
known, either because there is no control at all (such as the case of the European Schengen countries),
or because the control is limited to a certain subset of flows (as is the case with the existence of
bilateral agreements that often facilitate the movements of persons living in the vicinity of the border
post), or even because of the physical impossibility of the border control authority to control all the
(legal or illegal) border-crossings. Therefore, it is a good practice to measure the flows of persons
over the border and to qualify them subsequently as either visitors or other categories of persons
traveling. This might be done automatically with counting devices that capture vehicles (private cars,
buses, trucks), to which an average number of passengers can be imputed, and also capture the license
plate, from which the country of residence of passengers in private cars can be derived. In the case of
buses, in most countries, a list of passengers and their nationality should usually be available.
7.59. Land borders (railways) In the case of land borders crossed on international railways, the case
is relatively similar to that of airports. The railway companies should be able to provide the number of
passengers (and in some instances on some rough characteristics, e.g. 2nd class versus 1st class
passengers). A sample could be designed, on the basis for example of the persons seated in a given
place, which is the kind of design that railway companies use for their satisfaction surveys.
7.60. Cruise ships. For some countries, such as those in the Caribbean, persons arriving aboard a
cruise ship represent a very significant share of total arrivals and travel expenditure. The number of
passengers aboard and their characteristics in terms of residence are known, as well as the
characteristics of the crew. In the case of cruise ships, ferries, yachts and all types of recreational
vessels, the captain is usually requested to provide the port authority with a list of passengers and
crew on board (a manifest), indicating name, surname, nationality, passport number, and any
additional information that authorities might decide to request. Expenditure in the country visited
occurs when cruise passengers disembark, although they might also purchase packages on board to
visit places of interest. Cruise ships necessarily use specific moorings, and embarking and
disembarking are controlled by customs officers. It is a good practice to apply a simplified
questionnaire, either to all passengers or to a sample, in order to collect information on their
expenditure.
F.2

Border surveys and number of mode 4 trips or persons

7.61. Border surveys could also be used for collecting further information on mode 2 and 4 trips or
persons (or details) if there was already an initial indication of the purpose of travel (e.g. obtained
through the counting mechanism identified above or through administrative records, e.g. E/D cards or
immigration counts). . The relevant questions in existing questionnaires should relate to travel for
personal, business or professional purposes, the two latter being more closely related to mode 4.
Information on characteristics of mode 2 and 4 border crossers could be obtained by adding
supplementary detail to the existing border surveys/passenger surveys questionnaires. However, it is
important to note that the information collected in such sources will most likely relate to number of
trips rather than number of persons. When it comes to mode 2, the former would be the preferred
option, whereas for mode 4 there would be an interest in both.
7.62. If the flows of certain categories of persons are small (and this might happen for some
detailed categories of service suppliers) it would possibly require a larger (and more costly) sample or
a more targeted one than when fewer categories are considered, in order to obtain sufficient
109

Note that many mode 4 persons may transit through airport lounges.

information. For mode 4, a mechanism to target more specifically business travellers would be useful
(e.g. airport lounges).
7.63. IRTS 2008 already suggest the collection of information that would be very useful for
approaching the number of mode 4 trips/persons. Recommended elements include a breakdown of
purpose of stay into personal and business and professional activities. Among the professional
activities, self-employed are explicitly mentioned as well as other types of persons travelling for
professional reasons (businessmen, investors). Information on attendance of meeting, conferences or
congresses, trade fairs and exhibitions is also recommended110. The compilers should further consult
the IRTS 2008 Compilers Guide and closely cooperate with the compilers of tourism statistics in
order make the most effective use of such a data source.
7.64. A starting point for identifying mode 4 would be to identify people travelling internationally
for business or professional purposes. As previously described such information could potentially be
available from arrival and departure card data which could be used for a first stratification/selection
for identifying business travellers (also see chapter 9). However, not all of the people in this group
will represent mode 4. For example, people travelling overseas to receive a service (like attending a
conference or workshop) are also likely to select business or professional as their purpose of travel in
an arrival or departure card unless these cards include a separate option for this category.
7.65. Some small modifications can be included in existing surveys to capture mode 4 movements
of natural persons. Adding such questions to border surveys will require close cooperation between
those in charge of tourism and trade in services statistics. The questions suggested in IRTS 2008 on
the purpose of the business trip could, for example, be extended as shown below. This would allow
the differentiation between persons attending meetings, conferences, trade fairs and exhibitions and
those who are of interest for mode 4.111 The suggestions below are built from the perspective of
incoming persons. Similar questions could be developed from the perspective of outgoing persons.
7.66.

If business or professional visit/trip
Are you here for
Attending meetings, conferences, trade fairs and exhibitions
Other business and professional purposes

The next question in order to identify mode 4, combined with the information on the length of stay,
could then be formulated in different ways. Two options are given below. :

110

The scope of this business and professional category is described in paragraph 3.17 of IRTS 2008: “Business
and professional. This category includes the activities of the self-employed and employees as long as they do
not correspond to an implicit or explicit employer-employee relationship with a resident producer in t he country
or place visited, those of investors, businessmen, etc. It also includes, for example, attending meeti ngs,
conferences or congresses, trade fairs and exhibitions; giving lectures, concerts, shows and plays; promoting,
purchasing, selling or buying goods or services on behalf of non-resident producers (of the country or place
visited); participating in foreign Government missions as diplomatic, military or international organization
personnel, except when stationed on duty in the country visited; participating in non-governmental organization
missions; participating in scientific or academic research; programming tourism travel, contracting
accommodation and transport services, working as gu ides or other tourism professionals for non-resident
agencies (of the country or place visited); participating in professional sports activities; attending formal or
informal on-the-job training courses; being part of crews on a private mode of transport (corporate jet, yacht,
etc.), etc.
111
Care should be taken about the use of the non-specific term "meeting", meetings can be with a specifi c client
to deliver a particular service and would be of int erest under mode 4.

First option:
Are you employed
from domestic economy?
from abroad?
or are you self employed?
Second option, set directly more focus on the mode 4 supply of service:
Are you here to fulfil a service contract?
as a self-employed
as an employee sent by your employer
(alternative: Did your employer send you abroad?)
Are you coming to work in an affiliate of your employing company?
as a services sales person?
to negotiate the establishment of a (services) affiliate?
It is important that these questions be formulated to be understandable by respondents, for instance
using explanatory notes.
7.67. Information on the occupation of the person could be collected, possibly presenting a short
list relevant for the compiling economy. For example, specific information could be sought for those
supplying maintenance and repair services as this is a category of particular interest for many
economies. Alternatively a choice of activities of the service supplier (i.e. the employer) could be
proposed, e.g. specify the service activity of your employer: "Agriculture, forestry and fishing",
"Water supply; sewerage; waste management activities", "Construction", "Information and
communication" etc. The activities of the employer (preferably consistent with ISIC) could also be
listed in the notes attached to the questionnaire, but this would be considered as a second best
solution.
7.68. Some persons moving internationally under mode 4 (especially intra-corporate transferees)
will be staying abroad for longer than 12 months, so could be excluded from (tourism/travel) border
surveys if respondents are told not to respond if this is the case. However, it would be a good practice
to adopt an integrated approach – to design and conduct the border survey to collect also information
on all respondents, and then the compiler would select the information that is needed for different
statistical domains (e.g., for tourism statistics only select visitors with trips of less than 12 months,
while other persons, either visitors or those traveling for other purposes, with stays of various lengths
could be of interest in the context of MSITS 2010.)
7.69. In addition compilers need to take into consideration the national treatment with regards the
duration determining whether or not an international traveller will be required to register as resident
rather than be regarded as a visitor. This may also imply to adapt the data collection method.
7.70. As indicated above some rough estimates may be derived with no or little additional cost in
the collection system. Unless it is possible to develop some specific modules towards well targeted
mode 4-related samples or identifying ways to better exploit the existing data collection, estimates for
mode 4 derived from border/passenger surveys may only be done with a certain degree of
approximation.112
7.71. At least in some countries using border surveys to collect additional mode 4 information may
involve certain challenges, such as costs, the size of the sample which may need to be enlarged to
ensure the representativeness of small sets of the population, the interview duration (and consequently
112

See chapter 16 for more information. The UNSD webs ite provides information on the existing border surveys
and the compilers are advised to familiarise themse lves with the questions relevant for mode 4.

the survey total cost) may increase or the survey form may be become more complex or lengthy.
Compilers must not forget that border survey managers are already pressed by other users, in
particular the tourism sector users, who are willing to expand the questionnaire for tourism-related
aspects. This is why it is important that there is a strong co-operation between the (potential) users
(i.e. tourism, balance of payments, trade in services, etc.) of such data to identify the synergies and
priorities according to the specific information needs of the economy.
F.3

Italy – border sample survey

7.72. International tourist flows are very important in Italy. Bank of Italy uses an extensive border
sample survey. The size of the survey and its sample design enables the production of high-quality
analytical statistics on many aspects of international tourism in the country. The intended objectives
of the data collection system are essentially: (i) Quality of the statistics of the “travel” item of the
balance of payments and a better adherence to statistical standards established at the international
level and (ii) the provision of disaggregated data on a large number of characteristics of the tourism
market, for use by central and local government bodies, the tourism industry and researchers. The
technique used for the collection of the data consists in interviewing a representative sample of
persons crossing borders, both residents and non-residents, in transit at the Italian borders, while at the
same time estimating the number and nationality of those in transit. The sampling is carried out
independently at each type of border (roads, railways, international airports and ports), which are
selected as representative. The general logic of the survey provides estimates of expenditure on
international tourism in Italy through the application of two distinct operations at border crossing
points chosen: the counting operation and the interview.
7.73. The counting operation is mostly based on the technique of systematic sampling, that is, with
the observation of one unit each “n”, with “n” predetermined. The counts provide the number of
persons traveling at each border point disaggregated by country of residence. The counting operation
is necessitated by the lack of administrative information on physical flows of persons with the
required level of coverage, detail and timeliness. The face-to-face interviews provide the estimate of
the expense and a set of attributes that allow its disaggregation and qualification. The interviews are
conducted using a structured questionnaire submitted to a random sample of persons traveling at the
end of the stay abroad (i.e. when residents are re-entering Italy and non-residents are leaving it). This
technique involves fewer errors in the respondents’ memory of the expenses incurred than, for
example, does a telephone survey conducted some time after the travel is complete.
7.74. The questionnaire is the same for all border points. The main information - with varying
levels of detail – that is requested include: Sex, age and occupation, Location, Means of transport used
(with possible detail of the airline or ship used), Reason for the trip (if "vacation", the type of
holiday), Place visited (foreign country for residents of Italy, Italian province for residents abroad),
Number of nights spent during the trip, Type of accommodation used, Travel arrangements (inclusive
or not inclusive), Total expenditure, broken down by type of product (transportation, lodging,
restaurants, shopping and other services), Method of payment, Level of satisfaction about various
aspects of the place visited and of the stay .
7.75. In 2011, 145,000 interviews, or about 1.1 per thousand of total Italians and foreigners
crossing the borders of the country, and about 1,550,000 counts of persons traveling, have been
carried out. The sample is stratified by different variables for each type of border. The stratification
variable "direction", with the two modalities "to Italy" and "to abroad" and the variable "type of
carrier", with four modalities (road, rail, air and sea), are recorded exhaustively; i.e., respondents are
Italian and foreigners crossing all types of borders. The survey covers 82 border points (42 for roads,
5 for rail, 24 for airports and 11 for ports), i.e. the most important in terms of annual flows, although a
limited number of small border points were selected to capture origin-destination routes that were
otherwise poorly represented in the survey. The selection was based on ISTAT (Italian statistics
office) data when the survey was begun, and thereafter updated through the evidence of the border
93

survey itself, which monitors some border points on a rotating basis. The border points considered
cover the percentages of the total, according to ISTAT and ENAC (National Civil Aviation Authority)
data. The choice of number of border points sampled was reduced over time to optimize costs. Some
specific procedures for conducting the counting operation and the interviews were established for
each type of border.
F.4

Philippines – Measuring Inbound Tourism

7.76. The main source of inbound statistics in the Philippines is the arrival card which is filled up
by all travellers entering the Philippines. The Philippine Department of Tourism (DOT) and the
Bureau of Immigration (BI) are jointly responsible for encoding, processing, and generating reports
on the volume of international visitor arrivals, as sourced from the arrival card, which has been
identified as a “Designated Statistics” for tourism under Executive Order in 1996. Though a proposal
to remove the arrival card has been raised, the DOT and BI are continually working together to
improve the appearance and dynamics of the arrival card to ensure that it remains a vital and critical
data source for decision-making by the government and the private sector with regard to tourism
development and promotion. The critical data in the arrival card relevant to tourism are the country of
residence and purpose of visit.
7.77. A Visitor Sample Survey (VSS) is also being administered by the DOT on a regular basis in
all international airports of the country to generate statistics on visitors’ demographic or profile, their
travel characteristics, including psychographic information. The most critical data in the survey is the
determination of length of stay and average expenditure of visitors, which are important parameters
for estimating visitor receipts. The VSS complements the data gathered from the arrival card as a
major source of information for inbound tourism statistics and the tourism satellite account (TSA)
F.5

France – characteristics of international visitors and tourism trips

7.78. In France, a border survey is conducted every quarter (20,000 questionnaires) by a private
subcontractor. Entitled “Enquête auprès des Visiteurs de l’Etranger” (EVE), the survey permits the
collection of quantitative data on trips (same-day visits included) in France by non-resident visitors,
and is conducted just before they leave French territory. One major purpose is to collect data on
physical flows of non-resident visitors (arrivals, tourist nights, same-day visitors, etc., broken down
by country of usual residence). Another major purpose is to provide data requested by the statistical
service of the Central Bank (Banque de France) for use in estimating the travel receipts item for the
French Balance of Payments. The survey also collects data requested for French tourism market
analysis (categorical data on non-resident visitors, main purposes of trip, places visited in France,
activities during their stay, types of accommodation, modes of transport used, etc.). The General
Directorate for Competitiveness, Industry and Services is interested in both monetary and nonmonetary variables. The sample is stratified in order to calculate results for 22 countries or group of
countries. Manual headcounts are conducted and EVE questionnaires are used at the same locations,
except in the case of questionnaires relating to travel by road, which are used at car parks along
motorways near the French border. Dates of arrival and departure are indicated by the respondent.
G.

Complementary surveys of persons

7.79. It is possible to survey non-residents traveling in the compiling economy either in places of
accommodation mainly (hotels and other kinds of accommodation) or in sites of tourism interest.
Although this type of procedures presents some limitations many countries use it as an
alternative/complement in cases where border control systems are incomplete or not reliable. The
information to be ideally collected is outlined in Box 7.1. It could also be used to collect quantitative
and qualitative information on mode 2 or mode 4 as deemed relevant.
94

7.80. The first challenge is to identify the non-residents among the guests of the means of
accommodation. In the case of the observation of these persons at the collective accommodation, two
major limitations usually exist. First of all, not all stay at collective accommodation. Studies in many
countries have shown that the form of accommodation is a determinant of average expenditure per
day, so that such variable observed for those staying at collective accommodation cannot be
extrapolated to others. Another limitation is associated with the fact that when surveying at places of
accommodation, they still have not terminated their stay, so that their average expenditure per day
cannot be extrapolated to their expected length of stay as expenditure is not uniformly spread over the
whole duration. Additionally, a person on a unique trip might use more than one hotel, a situation that
alters the probabilities of being selected.
7.81. A similar situation occurs in the case of surveys at visitor attraction where the drawbacks are
as follows: not all persons traveling visit those attractions, the probability of visiting an attraction is
not known, and a given person might visit more than one attraction. As a consequence, information
based on surveys at visitor attraction could be biased. Any derived information for the whole
population should take into account this bias.
7.82. Finally complementary data could be collected on specific sets of the population traveling.
This could be particularly relevant for health patients or students. Indeed given the particular
characteristics and expenditure patterns of some categories of persons, it may be relevant to capture
information through a dedicated survey, e.g. expenditure survey of foreign students (other than
education fees if these can be collected through other more relevant means).
G.1

Country experience – Austria and below threshold establishments

7.83. Austrian accommodation statistics provide information about the number of arrivals and
nights spent in rented accommodation establishments in Austria. The data is based on a cut-off
sample. About 1600 municipalities (two third of the total number of municipalities) are selected
according to the number of nights spent (at present the threshold is 1000 nights spent per year) and
report the data to Statistics Austria. The data includes all rented accommodation establishments, i.e.
no threshold concerning the number of beds per establishment is applied. The results of the
accommodation statistics show the importance of rented private accommodation in Austria. Based on
Regulation 692/2011, Austria reported to Eurostat 77.16 million nights spent by non-residents in
2012, but this figure only represents 81% of the total number of nights registered in rented
accommodation establishments, because in addition to the 77.16 million nights, 17.89 million nights
were registered by non-residents in rented private accommodation establishments. So in 2012, 19% of
the nights registered by non-residents in rented accommodation establishments in Austria were spent
in rented private accommodation establishments, which are out of the scope of Regulation 692/2011.
G.2

Country experience – Australia and survey of students

7.84. Education related travel services, one of the largest contributors to service exports in
Australia, are derived using a modelled approach. This approach combines data on the number of
foreign students studying in Australia with estimates of expenditure by foreign students on course fees
and on other goods and services.
7.85. Estimates on the number of foreign students studying in Australia are based on quarterly
stock data for a range of study related visa sub-classes provided by The Australian Government
Department of Immigration and Border Protection. Estimates of student numbers for the interquartile months are compiled using information from the ABS's Overseas Arrivals and Departures
data with a small adjustment made to account for permanent migration.

7.86. Monthly estimates of course fees per student registration are derived from International
Student Fees data supplied by Australian Education International. This data set is supplied by
educational sector on a per-semester basis which is then split into months according to the number of
students studying in Australia for each month in that semester (as estimated above).
7.87. Monthly estimates of per-capita expenditure on other goods and services by foreign students
are derived from data collected in the Survey of International Students Studying in Australia
conducted by Universities Australia. This survey provides estimated average weekly expenditure on
goods and services for selected educational sectors, which is then converted to a monthly
estimate. As this survey is conducted irregularly, most recently in respect of the 2010 calendar year,
estimates are interpolated between survey years and extrapolated to the most recent period with
reference to the ABS's All Groups Consumer Price Index.

Chapter 8 International Transaction Reporting Systems
8.1.
Scope. One of the sources used for statistics on international trade in services is the
international transactions reporting system (ITRS). This chapter describes how ITRS can be used and
discusses its advantages and disadvantages as a data source. It should be noted that the MSITS 2010
and this Compilers Guide deal with the use of the ITRS for the compilation of statistics on
international trade in services. In doing so, they keep consistency with IMF’s Balance of Payments
and International Investment Position Compilation Guide (BPM6 CG) that discusses the ITRS113 as a
source for compiling the Balance of Payments and International Investment Position statistics.
A.

Summary of good practices

8.2.
It is advised that compilers using ITRS as a source for statistics of international trade in
services review the advantages and limitations of this data source, as summarized in table 8.1, and
assess their relevance and comparative importance in the context of their countries. In particular, it is
a good practice that compilers explore to what extent it is possible to mitigate the limitations of this
source (e.g., under-coverage of small transactions in the case that reporting thresholds are in place,
absence of data on services that cannot be measured or identified via ITRS, problems with the
recording of partner countries, as well as lack of information on breakdowns by modes of supply) by
using additional data sources such as surveys.
Table 8.1 Summary of advantages and limitations of ITRS
Advantages
-

Timeliness and high frequency
facilitating data compilation
Comprehensiveness (new reporting
entities can easily be covered)
Limited number and stability of
respondents (banks can report on
behalf of their clients)
Burden of reporting and compiling is
not heavy once it is implemented
Reporting is often mandatory

Limitations
-

Use of thresholds (simplification or exemption)
has effects on data accuracy
Transactions approximated by settlements
Possible misclassifications of services items, in
particular for some of the detailed EBOPS 2010
items
Geographical allocation may be biased (country
of settlements versus country of transactions)
Difficult to manage with an increased number
of transactions and growing share of intercompany transactions with netting practices

B.
General description and purpose of ITRS
8.3.
The ITRS is a system of collecting data of individual international settlements and/or
transactions as reported by banks, enterprises and/or households. 114 In this Compilers Guide, the ITRS
system in which reporters are resident banks that report to the authority (the central bank in most
cases) on behalf of transactors (enterprises or households) is called a bank ITRS, while an ITRS based
on reports to the authority from transactors themselves is called a direct reporting ITRS.
8.4.
In most countries that maintain ITRS, the reporting is mandatory. Because the ITRS is often a
by-product of present or past foreign exchange controls, the data collected involve settlements (as

113

See BPM6 CG for more information on the use of ITR S for other parts of the Balance of Payments at
http://www.imf.org/external/pubs/ft/bop/2007/bop6comp.htm
114
See also BPM6 Compilation Guide, paragraph 4.2.

opposed to transactions115 ). The most comprehensive ITRS collects data on: (1) settlements with nonresidents through resident banks, (2) settlements with non-residents through non-resident banks, (3)
settlements with non-residents through intercompany accounts and (4) transactions without cash
settlements.116
8.5.
Historically, the ITRS started with settlement reports by resident banks in most countries.
However, as resident transactors increasingly started using non-resident banks for international
settlements, direct reporting from these transactors (i.e., not through resident banks) became necessary
to maintain the completeness of data collection. In addition, requests of resident banks to reduce
reporting burdens have made data collection depend more on reporting from transactors.
8.6.
When the bank ITRS is complemented with direct reporting, the information included in the
ITRS can be extremely useful to identify the major players involved in international trade in services,
especially if the reports not only include settlement amounts, but also the name of transactors.117 In
some countries, the framework of ITRS is maintained even after the introduction of surveys exactly in
order to capture international settlements with large amounts. By accumulating such information, a
balance of payments enterprise register can be established, which can be used when conducting
periodical surveys to measure small value transactions or adopting survey systems in place of the
ITRS (see the chapter 5 on business registers and surveys frames). For a proper maintenance of such a
register, the international payments database would need to include, at least once a year, the total sum
of transactions in services conducted by individual enterprises.
8.7.
Finally, in addition to its use for statistics on the international trade in services, the ITRS can
also be used for statistics on other financial transactions, as the ITRS could cover a wider range of
financial operations, including currency exchanges in foreign exchange companies and banks’
information on settlements, as well as position and flow data regarding foreign assets and liabilities.
C.

Using the ITRS for collecting data on international trade in services

8.8.
There are many reasons why ITRS can be useful to collect data on international trade in
services. First of all, the data are collected in a timely and frequent manner.118 Data can be reported as
soon as international settlements are done, in an exhaustive way, at least when there is no reporting
threshold or it is very small. The transaction codes included in ITRS that identify the purpose of
settlement generally facilitate the classification in conformity with the recommendations of MSITS
2010, except in some cases where it is difficult to obtain the appropriate information (such as FISIM
and construction), and for certain detailed EBOPS 2010 classifications, as discussed below. Indeed,
classification of international settlements according to the type of transaction (although with
precautions against misclassification) is the most important condition for ITRS to be used in
compiling statistics of international trade in services.
8.9.
An additional benefit of ITRS is that it can be maintained with relatively light burdens of
reporting and processing data119 . In the bank ITRS the reporting burden is concentrated in banks
where the procedures of reporting are relatively mechanical (once ITRS is implemented) and can be

115

In countries where the ITRS was not maintained after the abolishment of foreign exchange controls, surveys
have become main sources for statistics on trade in services and the focus has already shifted from se ttlement to
transaction data.
116
Ibid., paragraph 4.3.
117
This point is especially important for countries c hanging their collection system from banks ITRS to direct
reporting and surveys.
118
BPM6 Compilation Guide, paragraph 4.69.
119
Although setting up or restructuring ITRS may be costly and may require a period of more intense contact
with respondents to inform and train them on the reporting forms, procedures, and coding systems.

computerised. Data compilation is generally also efficient because its procedure is also mechanical
and compilers do not have to depend on complex statistical techniques that are often used for grossing
up survey results. In addition, new reporting entities can easily be covered if they have significant
transactions. This makes data reporting more comprehensive and stable and reduces compilers’
burdens significantly.
8.10. When assessing the 12 BPM6 main services components individually, data from ITRS may
be useful in compiling statistics on international trade in services for 1) manufacturing services120; 2)
repairs and maintenance; 3) insurance 4) charges for the use of intellectual property n.i.e.; 5)
telecommunications121, computer and information services; 6) other business services; and 7) personal,
cultural and recreational services. ITRS is also useful for financial services with the exception of
FISIM, whose amounts are often estimated based on accounting figures of financial corporations, as
discussed in Chapter 14. ITRS captures construction but it has certain limitations, as discussed in
paragraph 8.18. ITRS can be used partially for transport and travel. Since the compensation for
transport services is often included in the total price of the goods that are transported, they are
difficult to identify corresponding international settlements. In addition, in the case of travel, an ITRS
may not ensure appropriate coverage in countries belonging to monetary unions (as the euro area), as
the residents of these countries who are traveling will spend using the same currency as their own
during intra-area journeys. Consequently, this mode of payment cannot be tracked by banks and other
financial intermediaries.
8.11. Although ITRS has benefits for collecting data on international trade in services, it also has
some major drawbacks that compilers should be aware of and should aim to mitigate when using this
system.
8.12. First, resident banks reporting under ITRS may have difficulties in correctly classifying each
transaction to their relevant transaction codes, especially for complex cases (for example, when
services are bundled with other (financial) transactions, in the case of interest payments including
fees). Misclassifications may occur if the information on the purpose of international settlements is
not correctly transmitted by transactors to reporting banks. A breakdown of trade in services by
EBOPS2010 categories (and supplementary items and complementary groupings), as recommended
by MSITS 2010, requires more detailed transaction codes which exacerbates this problem due to the
number of codes used.
8.13. Secondly, the transaction partner country may be incorrectly classified. ITRS generally
includes country codes of counterparties of international settlements. In some cases, however, the
country of the settlements counterparty and the country of transaction counterparty differ from each
other. This occurs when settlement vehicles are used by transacting enterprises, for example.
8.14. Thirdly, small value services transactions may not be captured due to the existence of
reporting thresholds. For example this may be particularly problematic for small value transactions
made by households, e.g., for telecommunications, computer and information services as well as
personal, cultural and recreational services, which can be small taken individually but may represent
large amounts summed up at the level of the total economy. The bank ITRS generally has thresholds
to reduce reporting burdens. The threshold can be either an exemption threshold (no reporting
required under the threshold) or a simplification threshold (a lump sum amount is reported without
indication of the nature of the transactions). The general trend is for reporting thresholds to augment,

120

While ITRS may provide information about the fee charged, it may not provide the data necessary to adjust
the corresponding goods transaction in the trade in goods account
121
However, depending on its set up, ITRS may also su ffer from under-coverage as payments for some
telecommunication services are made on a settlement (net) basis). Care should be taken to ensure the gross
reporting of transactions (see BPM6 CG: para 3.11 and page 287).

as the regulations related to foreign exchange controls are gradually lifted. For example, in EU
countries, since the reporting threshold was raised from EUR 12 500 to EUR 50 000, a considerable
loss of information has been observed. Box 8.1 explains options how the lack of information due to
thresholds may be mitigated.
8.15. A fourth caveat to address is that ITRS-based data may not accurately reflect the timing of the
transaction, as the transaction and the settlement time may not always match122. Under ITRS, the time
of settlements is the only choice for recording the time of transaction. However, there are cases where
settlements are deferred or advanced for some reasons, e.g. in construction and in insurance services.
8.16. Regarding construction, as noted in Chapter 6C(i), the use of ITRS may give rise to several
biases. First, it may prove difficult to correctly identify the relevant settlements corresponding to
services transactions, in particular, in defining the residence of units conducting construction projects
(for example, from a balance of payments perspective a site office may be regarded as a resident if
substantive (long term) construction work is undertaken, even if legally they are not necessarily
treated as residents in the host country). Second, difficulties may arise in collecting data with the
required level of detail (e.g. for the identification of goods and services sourced from the economy
where the construction is taking place, and used as an input in a construction project). Third, the ITRS
thresholds set in some countries may be too high to properly cover some of the payments related to
construction projects (especially for smaller-scale work or especially in the case of fractionated
settlements)..
8.17. In some countries, the authorities oblige ITRS reporters to use foreign exchange rates fixed by
the government, rather than prevailing market foreign exchange rates. If large discrepancies exist
between fixed rates and market rates, due to fluctuations of market rates and/or the time lag between
transactions and settlements, reported data do not reflect the economic reality of international trade in
services.
8.18. Netting contracts can make the settlements amounts smaller than transaction amounts. Netting
is a common practice for telecommunication, postal services etc. It is also common for MNEs using
treasury centres. Ideally, reporting under ITRS should be made based on the data on a gross basis, i.e.,
before netting. In that case, a direct reporting ITRS is more appropriate compared to a bank ITRS with
indirect reporting. If transaction data, rather than settlement data, are reported on a gross basis using
the prevailing market exchange rates, most of technical limitations of the ITRS could be overcome.
Although the timeliness of data reporting might be sacrificed to some extent, such an evolution of the
ITRS could be beneficial for the improvement of the accuracy and completeness of reported data.
8.19. Finally, ITRS is not a complete source of information. It does not provide adequate detail for
all service categories recommended by MSITS 2010 (as mentioned above in 8.10), nor for preparing
the recommended breakdowns (e.g., for government goods and services n.i.e.); in such cases
administrative data may provide useful information especially if government’s international
settlements do not go through private banks.
8.20. As regards statistics on the international supply of services by mode, MSITS recommends
that each EBOPS components be allocated either to one dominant mode, or broken down by mode. In
particular, it is often difficult to identify modes of supply (e.g. modes 1 and 4) in the ITRS
transactions, as well as partner countries, although those transactions may be reflected in transfer of
funds or exchanges of foreign currency with a certain time lag.

122
Also see chapter 11 Comparing Data Sources for a more in-depth discussion of the pros and cons of using
the ITRS as a data source for selected service cate gories.

100

Country experiences123
D.1.

Country experience: South Africa

8.21. The Research Department of the SARB, responsible for the compilation and dissemination of
quarterly BoP statistics, relies mainly on ITRS (managed by the Financial Surveillance Department),
to compile statistics on international trade in services. The electronic ITRS was introduced in 2001,
replacing the previous paper driven system, as a compulsory reporting system to be used by all
Reporting Entities (banks and bureaux de changes) using the UN EDIFACT international electronic
reporting standard. In the transition (which coincided with the liberalisation of exchange controls),
special care was taken to ensure that the data needs for compiling balance of payments were covered.
In ITRS, the Reporting Entities provide all details of payments made to foreign parties by South
African residents as well as payments received by South African residents from foreign parties,
irrespective of the value.
8.22. SARB considers that ITRS is a cost effective reporting system that provides accurate, timely,
and reliable information which is validated via the SARB Data Exchange Architecture (SARBDEX).
The information contains comprehensive details of all transactions including the names, surnames,
addresses, ID numbers, company registration numbers, telephone numbers, and physical addresses of
transactors. The confidentiality of the data is guaranteed by using Internet encryption, SARBDEX
senders’ validation and SWIFTNET.
8.23. The South African ITRS has been enhanced several times since its introduction. The most
recent improvement (end 2013) has also ensured a better alignment with the requirements of BPM6,
although this required an additional substantial investment by reporters. The ITRS is used to assist the
compilation of balance-of-payments statistics in general, but from the point of view of the SARB has
special benefits for collecting trade in services data. Given that the South African ITRS has no
threshold, it can identify potential new services traders well, and can serve as a sampling framework
for surveys aimed at better understanding specific transactions in international trade in services.
8.24. The ITRS is seen as a primary tool to obtain statistical information for statistics on
international trade in services. To mitigate any potential downsides related to inaccurate reporting
(e.g., misclassification) the Financial Surveillance Department conducts regular onsite and offsite
inspections. In addition, the systems of the Reporting Entities are certified annually to ensure
compliance.
D.2.

Country experience: Japan

8.25. The core data source for compiling statistics on international trade in services in Japan is
ITRS, which is regulated by the Foreign Exchange and Foreign Trade Act. Under Japan’s ITRS, all
residents who made payments to or received payments from non-residents under a certain condition124
are obliged to report their underlying transactions. Resident transactors must report their cross border
transactions to the Minister of Finance through the Bank of Japan (BOJ). Transactions via nonresident banks are reported directly.

123

A more elaborate description of the French and South African country examples in this chapter is available
online. Also the Japanese example is described more elaborately online, and includes coding schemes, example
forms, and more detail on the below-threshold statistical estimation methods.
124
There are some exemptions from reporting requireme nt, including transactions below the reporting threshold
(current reporting threshold is 30 million JPY or the equivalent in foreign currency per transaction, in principle)
and payments/receipts related to export/import of goods which crossed Japan’s custom. Other exemptions are
also specified in the ministerial ordinance.

101

8.26. Information reported includes information on the transactor, the counterparty, the date, type
of payment (payment or receipt), currency used, value, and the classification and description of the
purpose of the transaction. For this last variable, reporters use a BOP classification code to report their
purpose of transaction, choosing from a list of 186 BOP codes as provided by ministerial decree (61
of which relate to trade in services).
8.27. Under Japan’s ITRS, transactions that do not involve cash payments are required to report on
gross settlement basis. For example, if a resident imports and exports services under offsetting or
netting arrangements, not only the amount settled, but also the underlying transactions on gross basis
must be reported.
8.28. The Japanese ITRS reporting threshold has gradually increased to a relatively high level (30
million JPY). Therefore, a considerable amount of transactions may be missing, especially small
value service transactions. In order to estimate trade in services (other than transport and travel125)
below the threshold, Japan will employ statistical methods starting with the implementation of BPM6.
Empirical researches suggest that the frequency of transactions increases exponentially as a value of
transaction decreases. Statistically, a Pareto distribution fits the data well. Assuming that transactions
below the threshold also follow Pareto distribution, data below the threshold can be estimated. These
estimations are made on annual basis, for each service item.
8.29. Compared to using information on transactions from before a threshold increase, the Japanese
approach to estimating below-threshold transactions has the advantage as the estimations can be
updated periodically with recent data. This means that structural changes and price changes over time
can be captured. Further, since the fit of the Pareto distribution will be assessed on a regular basis,
statisticians are able to choose another statistical approach if the implemented method does not fit the
data well.
D.3.

Country experience: France

8.30. France used ITRS as a data source for statistics on international trade in services up to 2010.
Under the French ITRS, banks reported all payments between residents and non-residents registered
in their books, whether the transaction was on their own account or on the account of their clients.
8.31. Since 1990, the data from Bank was complemented by direct reporting by firms with annual
cross-border transactions greater than EUR 30 million. This system still exists and covers
approximately 500 firms. These firms are required to report all transactions (on a monthly basis) and
positions (on a quarterly basis) with non-residents conducted through accounts with domestic banks or
banks abroad or through intercompany accounts.
8.32. The French ITRS had several advantages. It delivered information in a very timely and
frequent manner and it was highly detailed, allowing in principle for an almost exhaustive knowledge
of international services transactions. In addition, compiler’s data access was facilitated and costefficient because data processing and compilation occurred at Banque de France DG Statistics, and
Banque de France is also the institution overseeing payment systems and operations of resident banks
in payment systems.
8.33. However, the French ITRS also displayed serious limitations common to other national ITRS
such as misclassifications, stemming from the fact that the largest part of transactions was classified
by intermediaries (banks) on behalf of their clients. Potential remedies would involve either additional
resources or increase the risk of substantial data omission.

125

Transport is captured mainly by direct reports. Travel is estimated by using boarder survey, household survey,
and official statistics.

102

8.34. Also, transactions are registered on a cash basis, while BPM6 recommends the accrual basis
of recording. Furthermore, the system did not capture all transactions that did not have a payment
counterpart, such as intra-group transactions and was able to register transactions made via nonresident banks. As the evolving patterns in international trade show increased settlements through
non-resident bank accounts and in intra-group flows, these limitations become progressively
prominent.
8.35. Finally, ITRS is vulnerable to administrative changes imposed for other motives, but that
may have negative consequences for collecting and producing statistics. For example, in Europe, the
creation of the Single European Payments Area (SEPA) that aims to improve the smoothness of crossborder payments and to lower their cost, implies that banks’ reporting obligations are adjusted and
reporting thresholds are increased (up to 50.000 EUR as of 2008).
8.36. In order to assess the magnitude of the ITRS limitations, Banque de France has organized a
parallel run between the “old” ITRS-based system and the new system based on the ‘Enquête
Complémentaire sur les Echanges Internationaux de Services’ (ECEIS), in order to perform all the
appropriate testing on the ECEIS survey (see Chapter 6F(i)). The results indicated that ITRS could be
replaced by a direct reporting system. It should however be noted that some traces of the old system
are kept in the new gathering system: ITRS still provides data from all transactions higher than EUR
50 000 that intermediaries complete for their clients vis-à-vis counterparties located outside the
SEPA. This survey is particularly useful to identify firms that have to be integrated into the direct
reporting system.

103

Chapter 9

Administrative Records

9.1.
Scope. This chapter describes main advantages and limitations of administrative records as a
source of data needed for measuring the international supply of services. Advantages include
reduction of costs, reduction in reporting burden by companies, filling of data gaps, and improvement
in statistical business registers. Typical limitations are restrictions on access to information due to
confidentiality, methodological differences with the statistical standards, consistency issues,
timeliness, and other quality concerns. The chapter also presents good practices in the use of
administrative records, which demonstrate that when their limitations are well understood and
effectively addressed, such records can be a valuable data source in statistical compilation. The
chapter consists of the following sections: a summary of good practices (section A); general
description of administrative records (section B); examples of administrative records for use in
statistics on the international supply of services (section C); and country practices (section D).
A.

A summary of good practices

9.2.
It is a good practice that administrative records are used in the compilation of statistics on the
international supply of services, as well as in the maintenance of the SBR and the trade in services
satellite register, following the policies adopted in the national statistical systems, that national
legislation and/or regulations grant the compiling agency adequate access to the administrative data
sources (see Chapter 2), and that appropriate institutional arrangements are in place between the
compiling agency and the administrative authorities (see Chapter 3).
9.3.
Compilers should fully understand and document relevant aspects of the administrative data
sources, including (a) information on how these data are processed, registered, and stored, (b)
methodologies and definitions, (c) coverage, (d) periodicity and timeliness, (e) changes in regulations
pertaining to the administrative data being collected and (f) budgetary constraints at the administrative
authority. In particular, compilers should be well aware of how the concepts and definitions used in
administrative sources match with those necessary for the compilation of data on the international
supply of services. The use of unique identifiers of economic units by all administrative authorities
(as is done in increasingly done in many countries) would greatly facilitate the integration of the
administrative records into the statistical system and help to eliminate duplication in coverage.
9.4.
In particular, this Guide advises to use customs records as a source for trade related transport
and insurance services while immigration records and entry/departure cards (E/D cards) - as a source
for measuring the flows of inbound and outbound visitors. Compilers are encouraged to consult
UNWTO’s list of proposed items for inclusion in E/D cards.
9.5.
Compilers are encouraged to use immigration records for estimation of the movements,
numbers, and presence of foreign natural persons under mode 4 of supply of services and for
compiling some FATS variables, if the immigration authority collects work permits containing
information on foreign entrants applying for work.
9.6.
Finally, it is advised to use records of tax authorities, especially value added tax (VAT)
declarations, as a source on values of trade in services, foreign affiliate relationships, and the
movement of natural persons. Moreover, tax records often include information on the ownership
relationship between businesses and employment, and on income from foreign sources separately
from income earned from domestic operations, which can also complement other sources in order to
identify enterprises and individuals with investments abroad.

104

General purpose and description of administrative records

9.7.
Administrative records are compiled for regulatory purposes or to support and document the
administration of various government programmes, such as immigration regulations, social security
benefits, education, and public health services. Important potential sources of data for the compilation
of statistics on international trade in services include customs records, records of immigration
authorities, and records of tax authorities.
9.8.
The use of administrative data has become a cornerstone of economic statistics programmes
in many countries 126 . Administrative data are a valuable source for updating statistical business
registers. However, other sources with similar information should also be considered, such as income
statements made by public enterprises and public quasi-corporations.
9.9.
In some cases, statistical outputs are produced at the national level from information obtained
from different administrations and/or from various levels of the same administration. The
characteristics of each source are important, as are the differences between the sources and how these
differences affect statistical output. For example, different regional offices of the same administrative
agency may experience different changes to their budget, capacity and workload, which may
influence the timeliness and consistency of the data are processed and registered.
9.10. It is important, therefore, to collect, understand and document information (metadata) about
the administrative data sources which is critical to the design of the data compilation system. Also,
such documentation will improve coherency of the resulting statistics by helping users to assess the
advantages and limitations of data obtained from different administrative sources. The information
which compilers should document include: details of how the source data are processed (e.g.,
collected and stored), concepts and definitions of data variables, changes in data collection
methodologies, available metadata on the coverage and number of data reporters, changes in sample
size and/or in sample composition, breaks in series for any reason, time and date when the
administrative data were processed and revised, as well as features of the environment in which the
system of administrative recordings and which can influence quality of such recording (e.g., such as
changes in regulations pertaining to the scope and composition of administrative data being collected,
level of permanency of the mandate to collect such data, resource requirements and possible
interruptions to staffing and technological needs, and budgetary constraints, among others).
9.11. Ideally, access to the administrative sources should be guaranteed to compiling agencies by
national legislation. However, availability of legislation is not a sufficient condition for the efficient
use of administrative records. A cooperative approach to the development and utilisation of
administrative records for statistical purposes is likely to be far more effective in obtaining access to
administrative records than an approach based on legal arguments. Managers in organisations
gathering administrative information should be made sensitive to the importance of the data and
contribution of each data provider to the overall statistical system.
9.12. With the increasing role of administrative data in the overall statistical process, clear
arrangements between statistical offices and administrative authorities must be in place and reviewed
regularly to ensure continuity of the use of these important data sources in the statistical system.127
The agreements should contain clauses about confidentiality, coding, data transfers and their
frequency and the content of the administrative database. Strict measures should be taken within the
statistical agencies to ensure the administrative data records remain confidential, are treated as survey
micro data and used only for statistical purposes.

126
127

See Guidelines on IES, paragraph 5.98.
See chapter 3 for more information on effective institutional arrangements.

105

B.1.

Advantages and limitations of administrative data

9.13. If administered and maintained properly, administrative records can offer strategic and
statistically important advantages over direct collection of corresponding data from respondents. The
advantages and limitations of using administrative data must be considered, including the following:

Methodological soundness. As administrative recording must comply with relevant
legal and other administrative concepts and requirements, the resulting records typically
adhere to the adopted administrative standards of methodological soundness and consistency
in terminology. Therefore, the use of administrative records for statistical purposes normally
implies their certain transformation to better approximate statistical concepts. When the
adequate transformation procedures are developed and systematically applied, administrative
records can be a reliable and valuable primary and complementary source of statistical
information (see paragraphs 9.11 – 9.13 below);

ii.

Cost. Administrative records are a relatively inexpensive source of information
compared to surveys and censuses, which is an important factor for statistical agencies faced
with tightening budgets. However, the cost to statistical agencies may be higher than
anticipated if the administrative data require complex transformation and/or processing to
meet statistical requirements;

iii.

Coverage. Due to their nature, administrative records usually have the advantage of
covering a large segment of the economy, if not the entire economy. Moreover, due to the
administrative character of the data, non-response is normally negligible and data are subject
to substantial scrutiny, which should generally ensure their accuracy. However, statistical
compilers should be aware that the coverage and content of administrative records can be
subject to discontinuities resulting from changes in regulations or administrative practices.
Also, not all variables in administrative records may receive the same level of attention; for
example, revenues may be examined closely, whereas less effort may be devoted to ensuring
that industry codes are correct. Nonetheless, with the increasing demand to produce statistics
for small domains, where sample surveys may be difficult to implement, administrative
records represent a valuable alternative or complement to sample surveys, when the
limitations above are considered and addressed;

iv.

Periodicity and timeliness. The availability and timeliness of administrative records
may not align well with data release deadlines of statistical agencies. For example, individuals
and/or entities may not be required to report to administrative agencies at common intervals,
resulting in some data being reported monthly, while others are reported quarterly or
annually. Moreover, annual data may be presented in administrative records on a calendar
year basis or on a fiscal year basis. There may also be delays before the administrative data
can be used and procedures established for allocating the records to the proper period;

Response burden. Administrative records allow for the reduction of the burden of
statistical inquiries made by compiling agencies. Administrative records can be adapted and
compared against data collected by other methods (e.g., surveys) for validation purposes.
Linking administrative records to other administrative data sets, survey data or census data is
increasingly being performed to produce richer datasets for statistical utilization;

vi.

Data compilation. Administrative records can be used to complement survey data and
in the absence of information, such as non-response to surveys, administrative data can be
very useful in the imputation process.

106

9.14. In order to make administrative records useful to statistical compilation, it is necessary to
show how the concepts, definitions and classifications of administrative data can approximate the
economic variables required by the statistical system. For example, a correspondence table can be
established that describes how business accounting type data are linked to the economic variables of
the statistical agencies. Methodologies should be developed on how the administrative data can be
used as extrapolators for economic data collected from official surveys and censuses using economic
accounting concepts and definitions.
9.15. Additional work is needed to link the legal entities used by administrative authorities to the
enterprises and establishments used by statistical agencies. One of the prerequisites in using
administrative data is to establish rigorous mappings between the various structures of entities
maintained by administrative departments and the statistical agencies. This is necessary to ensure that
there is no duplication in coverage and to match up information from various sources. Some countries
have a unique identifier of economic units which is used by all administrative authorities. This clearly
greatly facilitates the integration of the administrative records into the statistical system.
9.16. When using administrative records, it is good practice that the file(s) used for statistics are
separate from the system in which the administrative records and transactions are stored for
operational use. This separation enables compilers to manipulate the data without interfering with the
administrative processes. It also helps to maintain the integrity and consistency of the data and
preserves the confidentiality of the outcomes of statistical editing and adjustment.
C.

Examples of administrative records
C.1.

Tax records

9.17. Records of tax authorities can be a very important source of information on values of trade in
services, foreign affiliate relationships, and the movement of natural persons. For example, tax
records, especially value added tax (VAT) declarations, may include the value of services transactions
between residents and non-residents, as well as the location of the service transactors. Moreover, tax
records often include information on the ownership relationship between businesses, which may be
useful for compiling FATS. Additionally, tax records on businesses may include employment
information. This information can be combined with individual tax return records, which identify
independent service suppliers or employees of service providers who go abroad to supply a service, to
compile data on the movement of natural persons. Tax records often also identify income from
foreign sources separately from income earned from domestic operations. These records can be useful
especially for conducting surveys or for checking information obtained from other sources128 and to
identify enterprises and individuals with investments abroad.129
9.18. Use of tax records for statistical business registers purposes. Tax records are one of the
sources for business registers and survey frames, as they contain unique identifiers, names, and
incomes of businesses. The systematic and persistent updating and maintenance of business registers
based on tax records will normally lead to significant quality improvements in the business register
(coverage, timeliness and accuracy), reduction of operational costs and business compliance costs.
9.19. Where a VAT has been introduced and includes services, the VAT declarations also belong to
the potentially very promising administrative data for statistics to identify the mode of the
international supply of services (especially modes 2 and 4). The information is often reported
monthly, covers most business units in the economy and, since the main principles of a VAT system
128

As used by tax authorities, definitions of the terms foreign and income may differ significantly from those
used in the compilation of statistics on the international supply services. The compiler should take care, in using
tax data directly in compilation of such statistics , that any such differences are accounted for.
129
IMF, Balance of Payments Compilation Guide, paragr aph 6.46.

107

are similar in different countries, there are common features, items and even details that could be used
for statistical purposes in many countries.
C.2.
Customs records for estimating value of resident/non-resident trade related transport
and insurance services
9.20. Customs records are one example of how administrative sources can be used by compilers of
trade in services statistics. In general, customs declarations include the freight charges and insurance
for shipments of merchandise and can be used to determine the value of trade related transport and
insurance services.
9.21. If foreign merchandise trade statistics provide both the FOB and CIF values of imports then
the values of freight costs and insurance premiums can be obtained from such statistics. However, a
method also needed to be developed to separate freight costs from insurance premiums. When both
valuations are not available on a regular basis, it may be possible to analyze the supporting import
documentation supplied to customs to obtain freight costs and insurance premiums. Such analysis
could be achieved by means of a properly designed sample survey of the customs records. In some
countries, import documentation may also provide the name or registration of the vessel carrying the
imported goods. The compiler could match this information against the lists of vessels operated by
residents; if no match is found, it could be assumed that the transport service was provided by a nonresident operator.130 If customs data can also be used to measure freight on imports—for example, by
taking the difference between imports CIF and imports FOB and deducting an estimate for insurance
premiums—these data could be matched with information on the vessel to determine transport
services provided by vessels operated by non-residents.131
C.3.

Immigration records and entry/departure cards

9.22. In most countries, immigration records and entry/departure cards (E/D cards) are a valuable
source of information on the movement of persons across borders, even as the increasing mobility of
persons and facilitation of the free movement of persons across borders, in some parts of the world or
under certain circumstances (e.g., in the case of the European Schengen area) is changing this picture.
9.23. Migration statistics usually distinguish between visitors, other short-term individuals traveling
abroad and migrants, where migrants could – for instance – be defined as persons staying in the
country for a period longer than 12 months.132 The number and characteristics of migrants and shortterm visitors are usually part of international migration statistics,133 guidelines for which may be
found in Recommendations on Statistics on International Migration, Revision 1 (1998).134
9.24. Both short-term visitors and migrants are of interest for the compilation of statistics on the
international supply of services by mode of supply. For short-term visitors, information could be used
to measure earnings and expenditure of resident short-term visitors abroad and non-resident shortterm visitors in host countries.135 This information could also be of interest in the context of gathering
information on the number of mode 2 and mode 4 persons, which is further described in chapter 16.
9.25. Some common examples of the categories covered by migration statistics and their relation to
travel and tourism statistics are presented in box 9.1. However, it should be highlighted that
immigration records or E/D cards have been put in place for purposes other than to measure trade in
services. Records of these administrative procedures should, therefore, be carefully evaluated on a

130

IMF, Balance of Payments Compilation Guide, chapter 12, box 12.2.
IMF, Balance of Payments Compilation Guide, chapter 12, paragraph 12.58.
132
IMF, Balance of Payments Compilation Guide, paragr aph 304.
133
IMF, Balance of Payments Compilation Guide, paragr aph 306.
134
Available at: https://unstats.un.org/unsd/pubs/gesgrid.asp?ID=116.
135
IMF, Balance of Payments Compilation Guide, paragr aph 305.
131

108

country-specific basis to understand the degree to which they could be useful. This exercise might
even lead to adoption of the administrative records for improved measurement of trade in services (as
well as tourism). Such adaptations are more likely if it can be demonstrated that they lead to
improvements in the procedures used for registration and/or monitoring of the movement of persons
across borders being done by the responsible agency.
9.26. Immigration records contain information provided by travellers when applying for visa as
well as passport information recorded at the time of border crossing. Entry/departure records contain
information provided by a traveller to immigration authorities when crossing the country border. The
Box 9.1. Common categories of visitors, other short-term individuals traveling abroad and
migrants captured in immigration records and their relation to travel and tourism statistics
Tourism
Travel
Business
purposes
Short term
workers, etc.
Other
business
purposes
Personal
purposes
Education
related
Health
related
Other
Mode 2 number
of persons
Mode 4 number
of persons

Holiday,
leisure
X

Religion,
pilgrimage
X

Education
purposes*
X

Health
purposes*
X

Employment**

Business
visit ***
X

X
X
X

Only to
work in
foreign
affiliate

* For tourism would only qualify if for less than one year, for travel no time limit if residence (i.e ., centre of
predominant economic interest) does not change.
** For the purpose of being employed in host economy; i.e., employer-employee relationship in host cou ntry.
*** Employer-employee relationship remains in home country.

information available from these two sources and reconciled by the immigration authorities are often
the basic source for establishing the flows of inbound and outbound visitors. The E/D cards usually
collect information136 on a census basis on name, sex, age, nationality, current address, date of arrival
(and/or departure), purpose of trip, main destination visited and length of stay.
9.27. Usually, immigration authorities provide data based on arrivals, in which case, for inbound
travellers, the data that are collected refer to the expected length of stay. Some countries reconcile
entry and exit cards by matching their identification number in order to establish the actual length of
stay.137 Some difficulties might arise in this operation owing to the existence of unmatched cards as a

136

See the UNWTO publication Tourism as an international trade service (2005), page 40 on the results of
survey of country practices in the use of E/D cards, available at: http://www.unwto.org/statistics/border.pdf.
137
For example, the Australian Bureau of Statistics compares the actual length of stay to the expected length of
stay on the entry card, which allows for calculatio n of a "realisation ratios" so that numbers of visitors expecting

109

consequence of errors in the process (e.g., lost cards or errors in the capture of the data), lack of
coordination in recording authorizations of change in status, or change in expected stays (illegal
immigrants who entered as tourists being also a possible source of discrepancy). However,
notwithstanding the existing challenges and limitations, the compilers should make most use of them
for compilation of statistics on the international supply of services.
9.28. Immigration records and E/D cards can be combined with information on patterns of
expenditure and compensation of employees to form the basis for a data model to estimate travel
services by multiplying the actual number of trips of short-term travellers and the estimate of per
capita expenditure from surveys of actual expenditure. Preliminary estimates for a period can make
use of data on the expected length of stay and anticipated expenditure.
9.29. Immigration records and E/D cards can be an important source of information on mode 2
movements of persons as such persons may constitute a large proportion of those recorded in these
sources. However, E/D cards are not seen as a sufficient source for mode 4 movement of people, as
E/D cards are meant to collect data for immigration control purposes, not for the identification of
natural persons crossing the border to supply a service. Mode 4 types of movements only represent a
very small proportion of total entries/ departures as registered by E/D cards. Moreover, the
identification of mode 4 persons is not an issue for compilation of tourism statistics, as such persons
are part of the visitor population, regardless of the fact they are travelling in order to provide a
service.
9.30. The use of E/D cards requires that statistical compilers have a clear understanding of their
coverage and content, most importantly:
i.

The geographic coverage of E/D data needs to be clearly stated to assess how representative
the cards are for the full population of persons crossing borders. It is, for example, important
to know if E/D cards are used at all border crossing points or their use is restricted in some
way (e.g., covers air passengers only, or applies to other types of arrival such as by sea, by
land, by river, etc.). This is particularly relevant for countries with long and open international
land borders or borders delimited by rivers, where geography makes crossing the border easy
or where border controls are absent at some crossings. Border control authorities usually will
have an estimate of what is beyond their present control procedures, but this estimation might
need to be permanently monitored to detect changes over time.

ii. What categories of persons are covered? Are there specific conditions that exclude certain
persons from border controls, in addition to those represented by the non-controlled crossing
points (e.g. nomads, refugees, border workers, etc.)? In many countries, nationals are often
exempted from border controls or detailed reporting requirements. Frequent border crossers
may have special permits, may not be registered for each crossing, may be excluded from the
controls altogether, or may be covered only by a global estimate. Finally, certain types of
border crossing might be subject to less cumbersome procedures (for instance private airports,
or land borders used by nationals of neighbouring countries).
iii. What is the actual content of the data? Access to detailed micro-data should be ensured to
allow for corrections and controls. In general, compilers should not expect border control
operations to provide all the information needed to measure movement of persons across
borders and to provide all the required variables and the type of distinctions that would be
needed for description and analysis. A recurring example is that many E/D cards do not
request the residence of the respondent, only the nationality. Additionally, not all controls in a

to stay more (or less) than a year can be adjusted at the time of arrival to allow for an estimate of those who will
actually stay for the amount of time they expected.

110

given country will be the same at all border points, nor will contain the same questions (e.g.,
questions at land borders might be kept to a basic minimum, because of the time constraint).
iv. Quality of the data collected has to be assessed. There are various repeated inconsistencies in
the information taken from administrative sources that stem from their specific functions. The
main interest of border authorities, for instance, is controlling the flows of non-nationals and,
as a consequence, other data (e.g., a national’s country of residence, origin or destination –
often different from the origin or destination of the flight, and detailed purpose of trip) are of
less interest to them and are not always well-collected or stored. The concern of border
control authorities is that the declared purpose be consistent with the type of traveller’s visa or
resident permit presented, which may induce travellers to declare a purpose in line with their
visa (e.g., recreation instead of convention/conference, or seeking business opportunities).
Revisions, checks and controls are needed to make E/D card information usable for purposes
other than migration.
9.31. UNWTO proposed that E/D cards should include a number of variables for the measurement
of travellers’ flows useful for tourism statistics as well as for modes 2 and 4,138 such as date of
entry/departure, the gender, age, place of birth, nationality, country of residence, port of entry, mode
of transport, length of stay and purpose of the visit. In particular, if a further differentiation of the
business purpose could be recorded on the E/D card (e.g., the difference between intra-corporate
transferee or employment by a local business) then this information can be used to identify mode 4
movements. The full list of proposed variables on E/D cards can be found in the on-line version of the
guide.
9.32. In countries that require a temporary work visa for mode 4 (at least for citizens of some
countries) administrative records from the visa issuing department (usually the Ministry of
Immigration) may be a more reliable source of information than the E/D card. If there is no
requirement for a specific visa (for instance for the duration of the stay or its repetition), the proposed
E/D card might not always enable identifying mode 4, unless a “purpose of trip/visit” is requested. In
should be noted, however, that the declared purpose of trip may be influenced by the type of
visa/permit that has been issued and not accurately reflect the reality. In the case when business visas
are used, they can be directly identified using immigration data. Moreover, it is usually recommended
that E/D cards identify the different subsets of travellers (such as visitors as defined in tourism
statistics) through indirect observation of derived characteristics.139
9.33. Also, it should be noted regarding mode 4, that the interest might not be on the total number
of trips, but on the combination of number of persons, number of trips taken by these persons and
duration of each trip and total per person. The E/D card could provide information on repetition of
trips, only if the persons are uniquely identified in the E/D card and the registrations at different
border crossing points are stored in the same database.
C.4.

Work permits

9.34. In many developed countries where migration regulations are complex, each legal foreign
entrant applying for a work permit is asked to register his/her nationality, occupation, purpose and
place of visit, length of stay, etc. The information on inflow and stock of such persons could be
collected from the immigration authority, and with the help of such information it may be possible to

138

The MSITS 2010 sets out the framework to define mode 4 in the context of the existing tourism and
migration statistical recommendations. Statistical compilers should review this framework and in parti cular table
V.3 in MSITS 2010, which presents the Links between the coverage of the Recommendations o n Statistics of
International Migration (RSIM), Rev.1 and that of IRTS 2008 in terms of purpose of trip or migration and
length of stay, and identifies categories of interest for GATS Mode 4.
139
See “Tourism as an International Traded Service” ( 2005) paragraph 2.45.

111

estimate the movements, numbers, and presence of foreign natural persons under mode 4 of supply. In
some other countries, registrations of work permits may be available from the immigration or labour
authority and may include additional information on the foreign resident, such as date of entry into the
country and detailed occupation type, as well as information regarding the resident employer, such as
name, location, type of business, or even gross annual income and number of employees. Such data
could serve as a major source to compile the number of foreign employees of service suppliers or the
number of foreign nationals providing services as an independent supplier. Moreover, such
information could be a complementary source of information for compiling FATS statistics (at least
for those foreign affiliates in the services industry that may be hiring foreign nationals) – both in
terms of identifying such foreign affiliates as well as estimating their sales and/or output and number
of employees. Data derived from documents issued in the context of social security coordination and
employment services could also be useful sources of information.
9.35.
Work permits and visa application records. Only those permits relating specifically to mode
4, 140 or those that are clearly identified as a sub-group of a mode 4 category, can be used for
compilation of mode 4 persons. While rarely possible, 141 immigration registers could, in theory,
include the relevant information.142 A solution could be to add questions to visa/permit application
forms to get more detail on the reasons for migration, visit or permit request, and, to consequently
obtain information more directly related to mode 4 as defined in MSITS 2010, and, therefore, be
comparable between countries. For example, current visa information may indicate the length of stay
or whether a migrant is moving under entrepreneur or skilled migrant categories, 143 but may not
indicate whether a person is moving as an intra-corporate transferee or is being directly recruited by
an overseas service producing company. In general, statistics on visas granted for work (or education)
reasons are available; however, further breakdown by reason of stay (employment contract, service
contract, or official residence permits categories) and for short-term visas of less than 3 months are
typically not available.
C.5.

Population Records

9.36. Registers of individuals (e.g., population registers, registers used to assist in the
implementation of legislation, or for monitoring specific or overall activities) could be used to
measure the number of mode 4 persons, in particular when such registers relate to a specific relevant
part of the population. Once again, it is important to analyse to what extent such a source can be used
for statistical purposes (i.e., the type of information included, reasons for inclusion in a register, how
the data are stored in the register, etc.). Although the use of such registers is clearly possible from the
perspective of receiving countries, the use of a register source could be particularly relevant in some

140

As defined in MSITS 2010.
It is important to note that in certain cases coun tries have established commitments based on existing specific
types of visas or working permits, which may or may not comply with the definition of mode 4 as defined in
MSITS 2010 (i.e. it may have been designed to serve complementary or additional administrative needs). When
this information is used to analyse mode 4, it shou ld be clearly specified that such information relates to the
commitment made, but does not necessarily refer to mode 4 as defined statistically.
142
Actually such information is often used by analysts and policy makers to support decisions, including
concerning labour migration. By ensuring that relevant mode 4 information is also included and reliabl y
recorded in such a source would greatly improve the understanding of the different types and characteristics of
international movements, and consequently help making the distinction between labour migration and mode 4
trade in services.
143
See also: http://epp.eurostat.ec.europa.eu/portal/page/portal/population/introduction. The technical guidelines
for data collection under Art.6 of Regulation 862/2007 – Statistics on residence permits specify under those
called grants of permission to stay to third-country nationals for reasons related to remunerated acti vities,
categories for those activities encompass (e.g., hi ghly skilled workers, EU Blue Card, Researchers, Se asonal
workers and other remunerated activities). The last category could be expanded for mode 4 purposes.
141

112

"sending" countries where mode 4 movements are important and in the case such information is
recorded in a register (e.g., by the labour ministry, or by a specific sectorial ministry such as health,
transport, etc.) or any other official source.
9.37. The statistical compiler must ensure that it is possible to distinguish in the register data
between those going abroad for employment purposes and those travelling for business purposes as
defined in MSITS 2010 chapter 5 (i.e., most likely the registers would include those going abroad on
the basis of services contracts, but compilers must ensure that the register does not also cover those
travelling abroad for negotiating purposes). Registers used to assist in the implementation of a
legislation (e.g., for tax purposes) may by particularly useful for sending countries in the context of
mode 4 self-employed persons (i.e., those who remain residents of their economy of origin). The use
of such a source would be especially relevant if it were possible to link information on individuals
from this source with information on the same individuals from other sources, such as immigration
records. Also, it is important to identify if this linking is legally (and statistically) feasible.
D.

Country experiences
D.1.

China - compilation of mode 4 person numbers

9.38. According to the Foreign Trade Law of China, qualified enterprises that seek to supply trade
in services supplied through the presence of natural persons or to enter into contractual services with
natural persons need to apply and get approval from the government. The governmental agency
responsible for approving these applications of enterprises or service suppliers also collects and
aggregates data on outflows and stocks of mode 4 persons in the category of contractual services. A
statistical institution has been jointly established by the Ministry of Commerce and the National
Bureau of Statistics in China, to which relevant enterprises are obligated to report required
information, such as service project, outflow and stock of persons, occupation or overseas work,
length of stay, and destination of country, etc. The national authority is responsible for aggregation
and annual publishing of the data.
9.39. Sometimes basic information, such as occupation, purpose of visit, visit destination, etc., is
being compiled separately by different authorities (such as immigration, tourism or labour), each with
their own statistical purpose and measurement scope, resulting in statistics that do not always match
the requirements for measuring the number of mode 4 persons. As a result, it is difficult to extract
useful information from the existing statistical systems. For instance, occupations may be classified
by first category (i.e. manager, technician, professional, etc.), and it may be impossible to further
break this down by industry details. One potential solution to this problem is for compilers to
cooperate more closely with the relevant sectors and ask for additional useful information.
Alternatively, statistical systems of the authorities (other than the compiling authority) could be
redesigned in order to meet the data needs for compiling number of mode 4 persons. Another solution
is that compilers can establish a new and specific statistical system for the compilation of the number
of mode 4 persons. Both solutions require close cooperation among the different agencies involved
(e.g., immigration, labour, tourism and other sectors). Redesigning the existing system requires more
communication and understanding, and increases daily workload for the agencies involved.
Establishment of a new system may improve the coverage and accuracy of the data, but requires a
large capital input, more coordination, and could raise the risk of leaking personal and business
information.
D.2.

Philippines and Customs Data for compiling freight and insurance services

9.40. The compilation of statistics on the resident/non-resident trade in services in the Philippines is
heavily based on administrative records of the Bureau of Customs (BOC). These records are
especially useful for estimating freight and insurance services. The cost of cargo insurance is stated in
the attached invoice for Import Entry Internal Revenue Declarations (IEIRD). If the insurance value is
not declared, then the value of insurance is estimated.
113

9.41. In 2007, a special survey was conducted for the purpose of input-output tables, which also
enabled a verification of insurance and freight values generated from administrative sources. (See the
on-line version of the guide for the 2007 Survey on Selected Imported Commodities (SSIC)
questionnaire). Freight and insurance costs are also validated by collecting and comparing data
collected from various government and private entities, including the Insurance Commission, the
Bureau of Shippers, Asian Terminal Incorporated, Cargo House, and FedEx-Philippines. For IEIRD,
the freight cost is stated in the customs declaration or in the attached invoice. If the freight value is not
declared then it is estimated. The details of the estimation procedure are given in the on-line version
of this guide.
9.42. Computation/Estimation: exports, insurance and freight cost. For Export Declarations,
insurance and freight costs are not reported on the export document but rather on the attached invoice.
For multiple commodities, freight and insurance costs are computed by prorating, using Total
Insurance/Total FOB or Total Freight/Total FOB as the multiplier and the individual FOB value as the
basis in the computation. Insurance and freight costs are not imputed if not reported.
D.3.

New Zealand and tax records

9.43. Statistics New Zealand (Statistics NZ) is in the process of establishing a new business register
(BR), which will also include micro businesses that are not currently recorded. Since the 1920s,
Statistics NZ has used tax data from Inland Revenue (IRD) 144 on incomes of individuals, selfemployed workers and companies, which has been integrated with tax data from the Goods and
Services Tax (GST)145 since the mid-1980s. This integration was made possible by linking the BR
enterprise statistical unit to the business tax number of GST-registered businesses
9.44. Increasingly over the past 20 years, as the use of tax and other administrative data to maintain
BR has been extended, Statistics NZ has observed that BR quality has improved in terms of coverage,
timeliness and accuracy; BR operational costs have been reduced; and business compliance costs
resulting from BR update surveys have been reduced. Challenges in transforming administrative data
for statistical purposes that Statistics New Zealand has addressed include: better understanding of the
rules and processes that define the administrative data, including coverage, timing, quality, and
completeness of the administrative data as well as the differences between the statistical units on BR
versus units defined in the tax system; and developing appropriate methodology and processes to
transform the administrative data to the statistical model. Addressing these issues has involved:
i.
Reliance on tax data for small businesses with a simple structure where the statistical
unit directly matches the tax unit structure and supplementing the tax data for large and
complex businesses with data collected by Statistics NZ;
ii.
Using models to derive the required statistical outputs from tax data. For example,
modelling two and six monthly GST tax returns to produce quarterly data;
iii.
Using statistical techniques, such as estimation/forecasting, to address timeliness
issues;
iv.
Using tax data that are correlated to a required statistical variable (not available in the
tax system) to model the required variable;
v.
Making available to users a very clear definition of the statistical outputs produced
and providing overlaps between series on the old and new definitions;

144
145

See chapter 3, section D.4, for a description of the MoU between IRD and Statistics NZ.
The Goods and Services Tax is a Value-Added tax th at covers almost all business activity.

114

vi.
Requesting changes to administrative rules so that the administrative data better fit
with the statistical requirements.
9.45. Maintenance of the BR is conducted continually, based primarily on tax data, as well as on
Companies Office data, survey feedback, media reports, and company reports, among other sources.
Units on the BR are stratified into three tiers on the basis of their structure, GST activity and
employment numbers, thereby allowing Statistics NZ to focus its resources on the largest businesses.
All units are updated using primarily tax data, as well as an annual update survey and other
information to supplement and verify the tax information for the larger enterprises. The BR has a
monthly quality monitoring programme which also identifies opportunities for quality improvement
and extending the use of administrative data.

115

Chapter 10 Other Data Sources
10.1. Scope. This chapter describes sources of information which were not intended for statistical
use and are not covered in Chapters 5-9 but which, nevertheless, can be used for collecting data
needed for measuring the international supply of services. In particular, the chapter presents
recommended practices (section A); describes other data sources and Big Data (section B); and the
use of specific data sources, namely payment cards (section C), mobile phone records (section D),
external party sources (section E), and linked micro-data (section F). Each section presents some
related country experiences.
A.

Summary of compilation guidance

10.2. It is a good practice that compilers explore the possibilities of improving quality of statistics
on the international supply of services (e.g., coverage and timeliness) offered by the use of additional
data sources and develop action plans to gradually incorporate them into the data collection activities
taking into account both their strengths and shortcomings. . Compilers are encouraged to use of
additional data sources as necessary including payment cards records, mobile phone records, external
party sources, such as financial statements of companies, reports of chambers of commerce, records of
investment promotion agencies, records of business associations, surveys conducted by other
organizations (e.g., various line ministries may conduct special studies on industries they supervise
and may conduct surveys for this purpose), private databases, on-line search engines, on-line travel
booking sites, data compiled by trading partners, and linked microdata.
10.3. It is further advised that compilers consider incorporation of various additional data sources in
the context of Big Data initiatives and undertake pilot projects in selected areas and to weigh the
benefits of using Big Data with the objective to improve timeliness, fill data gaps and reduce cost,
while maintaining a high-level of quality of official statistics on the international supply of services.
Many of the aforementioned other data sources are often placed under the umbrella term “Big Data.”
For the purposes of this Guide, Big Data are sources that can be described as high volume, high
velocity and wide variety of data that demand cost-effective, innovative forms of processing for
enhanced insight and decision making.146
B.

Other data sources and Big Data
B.1.

Good practices in using other data sources and Big Data

10.4. Compilation of data needed for measuring the international supply of services is a complex
process and requires the use of both traditional and other data sources. Compilers are encouraged to
consider using other data sources in the context of national initiatives to use Big Data as a new
potential data source. This Guide advises that compilers undertake pilot projects in selected areas and
to weigh the benefits of using Big Data in combination with the existing traditional data sources in
order to improve timeliness, fill data gaps and reduce cost, while maintaining a high-level of quality
resulting official statistics on the international supply of services. For instance, Big Data may be
useful in compiling travel and tourism expenditures. However, compilers should be aware of, and be
prepared to address, the challenges in using Big Data, including legislative issues surrounding the
access and use of data; privacy concerns; financial costs in sourcing data relative to its benefits;
management and protection of the data; methodological (in terms of data quality); and technological.
10.5. It is a good practice that any use of Big Data be part of an organized and efficient data
integration process and be subject to strict standards commonly applied in quality control, data

146

See “What Does “Big Data” mean for Official Stati stics,” by UNECE (2012), paragraph 6.

116

modelling, estimation and imputation. All of these modifications to the statistical process should be
done in a transparent way, be part of the quality assurance programme, and be properly reflected in
metadata.
B.2.

Purpose and description

10.6. In our modern world, more and more data are automatically generated through a variety of
devices, including mobile phones and sensors, and via many computer applications. The amount of
data and the frequency at which they are produced have led to the concept of Big Data. Examples of
different classes of Big data are:
i.
Commercial or transactional data (arising from transactions between two entities);
e.g., credit card transactions, bank transactions, on-line transactions (including from mobile
devices), retailers’ sales records, etc.;
ii.

From sensors; e.g., satellite imaging, road sensors, climate sensors, etc.;

iii.
From tracking devices; e.g., tracking data from mobile telephones, global positioning
systems (GPS), etc.;
iv.
Behavioural; e.g., online searches (about a product, a service or any other type of
information), online page views, etc.;
v.

Opinion; e.g., comments on social media, etc.

10.7. Even though administrative data (arising from the administration of a programme involving
collection of certain information, be it governmental programme or not) are usually seen as standard
data source, they do have some of the characteristics associated with the Big data sources, and could
therefore also be included here, if the volume of data and the velocity, in which these data become
available, are high; e.g., electronic records on medical procedures, hospital visits, insurance
transactions, education programmes, value added tax records, etc.
10.8. Major complaints about official statistics usually include lack of timeliness and high costs.
Big Data is often automatically generated and are accessible in real-time. Therefore, it could certainly
be envisioned that Big Data should complement official statistics in order to improve timeliness and
cut costs.
10.9. However, many challenges need to be dealt with in order to effectively use Big Data in
official statistics, such as:
i.

Legislative challenges with respect to the access and use of data;

ii.
Privacy issues to comply with confidentiality rules, to gain and maintain
public trust and challenges to obtain acceptance of data re-use and link to other
sources;
iii.
Financial challenges regarding the potential continuous costs of acquiring,
hosting and processing the large data sources;
iv.
Management issues regarding policies and directives about the rules, roles
and regulations to adequately protect and secure the sensitive data sources;

117

v.
Methodological challenges, specifically with respect to representativeness of
the data, the volatility of data sources over time and the need for adequate estimation
and model techniques in making the data useful and comply with quality standards;
vi.
Technological issues related to hosting and accessing the data sources, the
data processing, the system maintenance and the over-time storage of huge amounts
of data.
B.3.
Using Big Data for purposes of official statistics on the international supply of
services
10.10. Examples of Big Data which can be used for the compilation of statistics on the international
supply of services include: electronic medical records of hospital visits (may aid in compiling health
services), mobile phone records (are helpful in tracking movement of international visitors); credit
card records (may be useful in tracking a wide range of tourism expenditures and in compiling other
trade in services transactions). Such data can be used provided that involved operators and merchants
are identified in those data as belonging to service industry. More information on such applications is
included in next section of this chapter.
10.11. Compilers are, therefore, encouraged to treat Big Data as a new potential data source and to
undertake pilot projects in selected areas. It should be noted that seizing the potential of Big Data
would require attention to the organization of an efficient data integration process and review of
existing methods on data modelling, estimation and imputation. All of these modifications to the
statistical process should be done in a transparent way, be part of the quality assurance programme
and be properly reflected in the metadata.
Box 10.1 Country example: Project of Eurostat on using Big Data
Eurostat conducted a 15-month feasibility study on the use of mobile positioning data for tourism statistics.
The project will explore the usefulness of using these data for tourism statistics (and related domain s) and
will assess their strengths and weaknesses. Issues to be studied include access (and continuity of access),
trust (of producers and users of statistics), costs , concepts (in translating the existing tourism statistics
concept to a new data source) and other methodological topics (e.g., representativeness and sampling
within a very large number of observations). The ab ility of handling large data files held by mobile
operators is considered a critical obstacle to overcome if the project is to be successful. The inclus ion of
this project in the work programme is, among other reasons, based on promising research results in a
number of countries.

B.4.

Good practices in using payment cards data

10.12. Payment cards records are considered by this Guide as a potential source for compiling
statistics on international trade in services in countries that have a favourable institutional
environment concerning payment cards, especially if there is a well-established and extended network
of automated teller machines (ATM) and point-of-sales (POS) terminals, along with a massive use of
payment cards in national and international transactions.
10.13. A sound understanding of the processes of payment card transactions and the actors involved
in this process, as well as what to measure, is crucial in order to find and communicate the
specification of data to be collected from the payment cards institutions. Efforts should be made to
obtain standardised nomenclatures, compatible with the statistical classification. Over time the
specification also needs to be reviewed and eventually modified in order to reflect changes in crossborder payments infrastructures and patterns.
118

10.14. It is a good practice to compare the expenditures obtained from the payment card data with
other traditional sources. For instance, if payment card data are being considered for use to in the
compilation of travel statistics, these data should be compared with surveys on average daily tourist
expenditure conducted by statistical agencies or publicly available from the tourism industry.
10.15. It is advised that an indicator of whether the card was present at the point of sale be collected
from the payment card institution, as this information is useful in isolating cross-border transactions
or for identifying the type of service involved in the transaction, particularly those related to ecommerce.
10.16. Purpose and description. Payment and bank cards (such as credit and debit cards) are
important payment instruments in national and international transactions in both goods and services.
Records of card-issuing institutions can provide data on international transactions of the card-holders,
be they individuals or companies. A comprehensive payment cards database can provide detailed and
accurate information concerning operations performed with payment cards, with a number of
variables that can be used to characterize the operations, namely related to the geographical
breakdown and the types of goods and services.
10.17. Payment cards data offer an advantage in collection of information on such transactions on
surveys, as it would be difficult to identify those individuals or entities involved in international
transactions by means of surveys. Moreover, individuals often have imperfect recall of specific
transactions, whereas payment card records are generally accurate and complete.
10.18. Using payment card records for purposes of statistics of the international supply of services.
Payment card records can be a valuable source for the compilation of travel statistics. Using the
information reported by payments’ institutions, it is possible to obtain the number and value of
operations performed in resident ATM and POS with cards issued abroad, and operations performed
abroad with cards issued by resident entities, as well as the characteristics associated with the cards
and the type of channel used. Such data provide, on a monthly basis, a significant measure of travel
expenditure, both in terms of credits and debits.
10.19. Payment cards data can also be used as a good proxy for the geographical breakdown of
travel, if it is assumed that the country where the issuer bank is located is the country of residence of
the traveller. However, compilers should be aware that this assumption may not be appropriate for
some smaller countries, where a considerable number of individuals use payment cards issued in
countries other than their country of residence.
10.20. Regarding the purpose of the trip, payment cards data can also be a helpful data source. The
bank identification number (BIN code147), which identifies the cardholder’s designated account as an
individual or a company, allows to distinguish business cards from other types of cards. This
information can be used as a proxy to estimate expenditure made by business travellers and other
travellers. However, an important issue that needs to be considered is the regular use of personal cards
for business travel, which leads to the need for complementary data sources to estimate this
breakdown. A separate alternative breakdown of travel into types of goods and services is required
according to BPM6 and EBOPS 2010, which can be integrated with additional requirements of other
statistical domains, namely the tourism statistics and tourism satellite account. Payment cards
databases can provide important information to meet this new breakdown, using the activity

147

Recently designed as IIN (Issuer Identification Nu mber).

119

classification of the goods and services providers. These variables can be used as proxies to identify
the goods and services acquired by travellers. 148
Box 10.2 Country example: Description of U.S. Merchant Category Codes1
The U.S.-based payment card industry uses merchant category codes (MCCs) to classify each merchant
and its credit, debit, and cash withdrawal transact ions into industry groups. One or more MCCs are
assigned to each U.S. or foreign merchant by the me rchant’s acquiring financial institution, based on the
merchant’s primary type of business. Each transacti on is then assigned the MCC used by the merchant to
process the transaction. Most members of the payment card industry collect detailed data on transactions
by country and industry group, for which the industry classification is determined by the company’s
existing internal MCC groupings.

In total, there are approximately 600 MCCs on the universal list used by the U.S. payment card industr y.
Although the payment card merchant categorization was not specifically designed to conform to the BPM6
main services components, many of the MCCs directly correspond to particular services categories.
However, MCCs are applied broadly in that a merchan t that sells both goods and services may do so under
the same MCC. If transactions data are being used to measure travel purchases, this distinction is not
important, but it may be if it is intended to be us ed for other purposes.

10.21. The main advantages of using payment cards data as a source for the compilation of the travel
item are, among others, the wide-ranging coverage of travel-related transactions; the existence of a
limited number of respondents; the timeliness and frequency of the information, as data are available
with a short delay and on a monthly basis; the detailed information on the characteristics of both
travellers and goods and services providers; and, finally, a reduced cost in terms of compilation, since
payments institutions need to process this information for their own use, imposing a reduced statistical
burden. However, coverage with payment cards data will likely change over time.
10.22. Moreover, purchases made with a payment card from a foreign business identified as a
service provider (by an industrial or activity code) in which the card is not present may be an
indication of a mode 1 service provision. Identification of such information may be particularly
relevant for Internet transactions in services related to IPP products (e.g., music, video, software).
10.23. There are some challenges that have to be considered when using payment cards data.
Primarily, there exists the challenge of excluding non-tourist related transactions, such as imports or
exports of goods or services other than travel. For this purpose, the use of the transaction amount
along with the economic activity of the goods and services providers can be used to exclude
transactions that should be classified under other items in the balance of payments. Also, an indicator
of whether the card was present at the point of sale is useful in isolating some of these transactions,
particularly those related to e-commerce.
10.24. Moreover, payment cards data are not comprehensive for travel expenditures, due to the fact
that other means of payment can be used. In this case, it may be useful to compare the travel

148

For this purpose, correspondence tables between nat ional industrial or activity classifications (e.g., between
NACE or MCC) and travel expenditures on goods and different types of services should be developed.

120

expenditures obtained from the payment card data with other traditional sources, such as surveys on
average daily tourist expenditure or statistics from the tourism industry.
10.25. Another challenge is the fact that classifications used by payment card processors are not
necessarily the same as those used by statistical compilers. In fact, efforts should be made to obtain
standardised nomenclatures, compatible with the statistical classifications.
10.26. One more task to deal with when using payment cards data is to account for time lags
between the moment of the payment and the time of the trip. An additional potential difficulty is the
development of the global financial and payments systems that may lead to emergence of more
international brands and processors, which brings further complexity to the collecting system.
10.27. Using diverse data sources is crucial to have a comprehensive system that facilitates the
collection of travel-related transactions made by residents abroad and by non-residents in the
compiling economy. One of the main challenges in designing a compilation system for travel item is
integration of different data sources, which have different degrees of coverage, different periodicities
and that may contain overlapping data. There is a trade-off between reducing the impact of possible
double counting and covering the various types of expenditures of travellers as much as possible. The
extensive detail of the payment cards database could allow the compiler to mitigate this risk.
Country experience: Iceland
10.28. Statistics Iceland has been using information on payment card data since 2009 as
supplementary information to the trade in services survey for the compilation of the travel item.
Information is received quarterly from all three payment card companies in the country, of which two
issue credit cards.
10.29. Data on the use of foreign payment cards in Iceland links the ID number of the services
provider receiving payment to the business register (or national registry for those not registered in the
business register) and NACE and EBOPS classifications. For some NACE numbers, like hotels and
restaurants, all transactions are automatically included in travel, whereas all other new ID numbers are
examined manually. In order to avoid double counting with the trade in services survey, businesses
are asked not to report transactions on the survey if the payment was made by payment cards. The
data also includes the country code of the country where the card was issued; the amount of the
transaction; and the date of the transaction.
10.30. Some potential challenges with this data are misclassifications of the NACE category; ID
numbers for which no NACE category is available (however, transactions with such service providers
are typically of small values); missing or incorrect data; double-counting with the trade in services
survey; misclassification of country that issued the payment card (i.e., residents of country A with
payment card from country B); large amounts of data that are ATM withdrawals and cannot be
identified. All ATM withdrawal are therefore assumed to be travel transactions.
10.31. For data on transactions made with domestic cards abroad, only credit cards are included
because the information on debit cards is not sufficiently detailed. Statistics Iceland receives total
figures for the usage of debit cards abroad, which confirms that the use of credit cards abroad is much
more common than the use of debit cards. Given that most of the debit card transactions are ATM
withdrawals, Statistics Iceland assumes that the entire amount of the debit card transaction data are
travel expenditures.
10.32. The data from the credit card companies includes the merchant category code (MCC) ID
name and number, which is linked to the EBOPS classification. While most MCC numbers are
manually examined, some, like hotels and restaurants, are automatically included in travel. Ecommerce data is excluded. The data also includes identification of the card type (i.e., individual or
121

enterprise), which may be useful in identifying travel transactions (e.g., if an individual card is used to
purchase goods, it is assumed that these transactions are likely travel-related). The data also The data
can also supplement information on transportation and other business services collected on the trade
in services survey.
10.33. The challenges in using data on credit card usage abroad are the fact that MCC categorization
is imperfect and of unknown reliability; e-commerce transactions are often difficult to recognize; the
volume of different sales/service providers makes individual investigation difficult; transactions are
recorded in the payment card data at the time of purchase, not when the service is delivered,149 and
large amounts of data that are ATM withdrawals and cannot be identified. All ATM withdrawal are
therefore assumed to be travel transactions.
B.5.

Mobile phone records

10.34. This Guide encourages compilers to consider employing the use of mobile phone applications
(“apps”) to collect information about mobile phone users’ movements and behaviors. Such apps can
be easily downloaded to users’ devices and can retrieve information to enable data analysts to conduct
very detailed studies of users’ movements and behaviors. It is advised that compilers clearly request
the consent of the mobile phone users before using such apps and describe how the data will be used
for official statistical purposes.
10.35. For all types of mobile phone data, compilers must take precautions to make the data
anonymous and check that confidentiality rules are being applied appropriately. Privacy concerns are
of critical importance and compilers should reassure users that the data will only be used for official
purposes and will be aggregated in order to maintain confidentiality. Again, as recommended in
chapter 2 of this Guide, legal acts on confidentiality of official statistics should be well-established.
10.36. Compilers will likely need to adjust the mobile phone data to correspond with the definitions
of official statistics. It is recommended that algorithms be developed to automatically organize the
data according to statistical definitions as much as possible.
10.37. It is further advised that the residency of a mobile phone user is determined by the residency
of the mobile operator associated with the user’s account.
10.38. Purpose and description. The use of mobile phones is widespread in many countries, which
enables statistical compilers to obtain data on location data (geographical coordinates in time) and
other important indicators about a mobile phone user and his or her activities.
10.39. Data gathered with the help of mobile phones can be divided into active and passive
positioning data. In the case of active positioning, the collector of the statistics contacts phone owners
and asks information about their location, themselves and their behavior(e.g. about their travel). In the
case of passive mobile positioning, data are automatically stored in the memory files of phone
operators or other recording systems.
10.40. Using mobile phone records for purposes of statistics on the international supply of services.
Both active and passive positioning data obtained from mobile phone records can be important
sources for statistics on the international supply of services. In particular, mobile phone records are
advantageous in the compilation of travel item and mode 4 services, as information on the country of

149

Timeliness is particularly an issue for cases in w hich accommodations are prepaid and when Icelanders
purchase fares online from foreign airline companie s using payment cards. Timeliness is less a concern with
regard to other travel expenditures.

122

residence and current location of the mobile phone user facilitates the identification of international
travellers, and, possibly, providers of mode 4 services.
10.41. Active mobile positioning. In the case of active mobile positioning, information about the
location of a phone, the user, the travel behavior of the user, and/or provision of services, is found by
making special inquiries, which generally requires the consent of the individuals chosen to participate
in the study. Active positioning is related to surveys and software downloadable on smartphones.150
As a result, it is possible to obtain very accurate information about the movement, means of transport,
expenses, provision of services, and motivation of the chosen respondents. Active positioning data are
geographically accurate.151 Such detailed data enable data collectors to analyze mobility within the
destination, and to conduct market research of a particular region. More study is necessary on how to
engage with mobile phone users; the involvement with users could be limited to turning on GPS with
consent to providing answers actively on a regular basis. Those studies should cover the technical
possibilities of phone applications, their user-friendliness and the incentives for users to participate.
Of course, the treatment of the data should be transparent for those users providing the active mobile
phone data. Recruitment of mobile phone users for outbound travel could be more difficult and more
costly due to the different standards of foreign operators. Still, this area is a very important source of
gathering more detailed statistics for performing various studies on travel/tourism and the provision of
mode 4 services.
10.42. Passive mobile positioning. In the case of passive mobile positioning, statistical data are
obtained from secondary sources of mobile phone use, which are most often the phone use
information automatically recorded in the systems of operators, such as Call Detail Record (CDR),
Erlang, and Anonymous Bulk Location Data (ABLD), among others. The advantages of passive
positioning are the huge masses of data involving all phone users and the relatively cost-effective data
collection method. The shortcomings, however, are protection of the privacy of persons, the
difficulties in obtaining the data from operators and the lack of characteristics included in the data.
The main convention for defining the residence of persons traveling is by the residence of the mobile
operator related with him/her. Other conventions for any other statistical purposes can be introduced
as well. The use of passive positioning data in the area of travel/tourism is rapidly growing, because it
is difficult to get an adequate overview of the movements and mobility of people in the increasingly
mobile world with open borders.
10.43. Call Detail Record (CDR) is one of the most widely used sources among the passive
positioning data suitable for compilation of travel item as well as for tourism statistics. The data are
obtained from Data Warehouse or from the billing record of the mobile network operator’s system;
i.e., from the places where information is gathered about phone users from issuing invoices. The time
of a call, the location of a call, the duration of a call, the cost and other characteristics are recorded as
such information. Production of statistics from such data requires making the data anonymous and
standardized, as well as checking the data, because the privacy of people and business confidentiality
of operators need to be protected. CDRs are usually issued as impersonalized data, either aggregated
for certain types of user groups or pseudonymously with randomly generated IDs. Spatially, the CDR
data are usually issued with the accuracy of a network cell (the cell with its location is called the Cell
Global Identity - CGI). This level of accuracy suits well, for example, for generating the main
variables of tourism statistics and several data collection systems using CDRs were developed for this
purpose. The CGI is, however, not accurate enough for preparing detailed analyses of movements of
persons. For example, CGI can be used for identification of transit visitors in travel (airports and
seaports, main transit lines thru the country, etc.). Moreover, there is noise in the roaming data that
needs to be acknowledged, and travelers in border areas of some countries may pick up cells of
neighboring countries without physically entering the country.

150
151

Examples of such software are MySense and Posques.
Smartphone-based studies in this area commonly utili ze GPS resolution of the accuracy of 1-10 m

123

10.44. A caveat on the use of mobile phone records is that some travelers who are in a country for a
period of time (e.g., less than a year) may purchase local subscriber identity module (SIM) cards for
their phones and thus not be covered through mobile phone records.
10.45. The methodology of preparing passive mobile positioning data requires adjusting the mobile
data to correspond with the definitions of official statistics. For example, the duration of a visit, the
number of nights spent and transit tourism are assessed on the basis of ordering single call activities,
with the uneven distribution of call activities in time and space posing a methodological problem. For
this reason, it is necessary to develop algorithms for organizing data, segmenting visitors and the
visits, for example.
Country experience: Estonia
10.46. The Bank of Estonia relies on mobile phone records for estimating travel and the Tourism
Satellite Account (see chapter 14.B.IV and chapter 16.A for mode 2). Estonia compiles the statistics
based on data from Positium Data Mediator, software developed by the Department of Geography of
the University of Tartu.
10.47. In Estonia, algorithms are developed for segmentation and cleaning of data available in the
Positium Data Mediator. For example, special methods are required for eliminating cross-border and
accidental roaming noise. In the preparation of the base models of inbound, outbound and domestic
tourism, Positium Data Mediator uses the data from Statistics Estonia and the results of the surveys
regularly ordered from TNS Emor for calibration. These results are used to prepare a base model of
data, which must be representative of various visitor segments and geographical areas and also take
into consideration the market shares of various operators and the ranges of the radio coverages of
various mobile networks.
10.48. Inbound tourism is determined on the basis of the log files of the roaming service of
operators. The travelers (phone-owners) that have arrived to the destination from abroad are divided
into visitors from different countries of origin on the basis of the registrations of the phones (so-called
resident countries of the tourist or nationality); the duration of a visit is measured by each phone on
the basis of the number of days at the destination when call activities were performed. Outbound
tourism is determined on the basis of the roaming log files of the phones originating from the
countries of origin. Visiting statistics by countries are found on the basis of the temporal and spatial
distribution of the call activities performed abroad; foreign visits are segmented according to the
requirements of statistics, if necessary. Comparison of call activities obtained on the basis in passive
mobile positioning data and data on accommodation nights in Estonia shows high correlation between
these two data sources (see Figure 10.1)

124

Figure 10.1Comparison of call activities in passive mobile positioning data and accommodation
nights in Estonia (Positium LBS)

External party sources
C.1. Good practices in using external party sources

10.49. External party sources have some inherent limitations. While they can provide insight and a
different perspective than are available from other more traditional sources, compilers should be
aware of the unique challenges with using statistics compiled by other non-statistical entities.
10.50. Primarily, external party sources vary in terms of data quality and timeliness, which
ultimately depend on the quality of the statistical processes in place at the external party. It is
recommended that the compiling agency become as familiar as possible with the data sources,
collection methods, methodology, and statistical rigor of the external institution and document their
strengths and weaknesses. It is strongly recommended that the compiling agency contact the external
source directly to discuss these aspects in greater detail; to alert the data provider that its data may be
used for official purposes; and to open the lines of communication so that any changes in
methodology or timeliness can be transmitted to the compiling agency in a timely manner. Some
external parties may be willing to work more closely with the compiling agency in order to better
meet the official statistical requirements; however, even in this instance, the NSO should continually
monitor the quality of the data and be aware of the limitations and weaknesses.
10.51. The compiling agency should assess whether the data available from the external party is
helping to measure the trade of the service category of interest, or other types of statistics related to
the international supply of services. In many cases, the definition of the industry or other aspects
related to international supply of services will not match the official statistical definitions. Such
misalignments may lead to misinterpretations and misuse of the data. Again, it is strongly
recommended that the compiling agency learns all such definitions and methodological aspects of the
external party source.
10.52. Timeliness and frequency are often weaknesses of external party sources. The compiling
agency should be prepared for breaks in series, changes in methodologies, changes in survey sample
sizes, and complete extinction of the source. In particular, it is recommended that the compiling
agency develops plans for cases where there is a reasonable expectation that the external party
125

source might substantially change or expire. It may be advisable to use an infrequent data source as
an input to a data model that estimates the value of the service transactions, or to interpolate missing
years or to fill other data gaps, rather than relying on the data source solely on its own.
10.53. Access to the data may also pose a challenge, especially if the external party typically
charges a fee for access. It is a good practice that the compiling agency contact the external party
source directly to request the best possible terms of use.
10.54. It is further advised that compiling agencies appropriately document all data sources in the
metadata, including in which years the data source was used, and to keep documentation of the
external party source, including historic data, revisions. methodological documents, communication,
and any user agreements.
C.2.

Purpose and description

10.55. Compilers could make use of other external party sources, such as financial statements of
companies; reports of chambers of commerce; records of investment promotion agencies; records of
business associations; surveys conducted by other organizations; private databases; on-line search
engines; on-line travel booking sites; and data compiled by trading partners. In particular, studies and
research publications of public or private institutions that include relevant statistical information can
be resource- and cost-effective sources. Ministries also may conduct special studies on industries they
supervise and may conduct surveys for this purpose. Moreover, industry trade associations may
publish research studies on performance of the industry, particularly in international markets, as a
means of promotion. Such data sources could be useful for compilers as a means of identifying
enterprises involved in international transactions, updating the names and characteristics of businesses
in the SBR, validating survey results, and filling in data gaps.
C.3.
of services

Using external party sources for the purposes of statistics on the international supply

10.56. National and international organizations that represent services industries often compile
registers of the enterprise members, including names, locations, and contact information. Such
information can serve as a source for updating and maintaining the service providers included in
business registers.
10.57. Such organizations may also survey their members for quantitative and/or qualitative
information on their operations, including domestic and international sales or market share, and
conduct market research to compile estimates or forecasts of sales. Quantitative information from
such studies can be used to validate or complement statistical survey results or to serve as a basis for
estimates when no other data are available.
10.58. National ministries associated with specific industries are also important sources of data and
can range from ministries of tourism and investment promotion to regulatory bodies on insurance,
health, and education. Such agencies may already be providing to the NSO more traditional
administrative data based on regulatory requirements. However, these agencies may also conduct
special research reports or one-time studies that can supply greater detail or cover a different industry
than its routine scope of work. Such data can be used to validate or complement other data sources.
10.59. Financial statements, annual reports, and profit and loss statements of enterprises can also be
a wealth of information for statistical compilers. Often such reports include information and data on
the firm’s international operations, sales, and market share by country or region. The level of detail
that is reported varies widely, depending on the country and type of enterprise. However, this source
should not be overlooked as a means of identifying enterprises involved in international transactions,
updating the BR, validating survey results, and filling in data gaps.
126

10.60. Private database companies that maintain databases of financial, statistical and market
information on global companies and private consultant groups are also commonly used by statistical
institutions, either to complement or substitute other data sources. Such private databases are usually
accessible for a fee. Examples of such providers are Compustat, Dunn & Bradstreet, Amadeus
(Bureau van Dijk), Business Monitor, Deloitte and Insight Research Corporation, among others.
These databases can be good sources for updating and maintaining SBRs, as they typically provide
company names, addresses, telephone numbers, names of contact persons, enterprise size by sales
value or employee count, and other basic characteristics, including industry or market sector, market
share, or number of establishments. These private sources often conduct market research studies that
may include analysis of domestic and international sales, market share, actual values of sales, and/or
estimates or forecasts based on historic trends.
10.61. Finally, internet searches of individual company names, industry sectors, or trade groups are
an extremely efficient and cost-effective way of validating information in SBRs and survey results.
C.3.a. Country experience: Australia - legal services statistics
10.62. The Australian Bureau of Statistics (ABS) collects data for legal services my means of a
quarterly survey of resident Australian businesses (the International Survey of Trade in Services). The
ABS captures data from a broad range of respondents, but not a complete range as resource
constraints and collection difficulties restrict data coverage.
10.63. The data from the ABS survey is collected on a country/dollar value basis asking only for the
value of receipts and payments of total legal services. There is evidence that a significant percentage
of international trade in legal services is provided by consultants and sole traders. It is currently
beyond the resources of the ABS to cover and process such a potentially large number of additional
respondents. The true volume in resident/non-resident legal trade, therefore, can only be estimated or
modelled based upon this anecdotal evidence.
10.64. The legal services data currently obtained by the ABS on a quarterly basis is compiled solely
from the Survey of International Trade in Services. The final trade in legal services statistics are
published on a quarterly and annual basis, as part of the International Trade in Services series of
publications, and posted for general access on the ABS website.
10.65. The Australian Attorney-General’s Department in 1990 established the International Legal
Services Advisory Council (ILSAC). Amongst other tasks, ILSAC produces an annual Statistics
Survey on International legal trade. Unlike the ABS survey, the ILSAC survey is sourced from data
representing all four Modes of Supply. ILSAC also noted that when compiling data for international
legal trade, the ABS identifies earnings of overseas branches of Australian law firms as “returns on
investment” rather than “exports”, as the ILSAC survey does.
10.66. Initial results from the ILSAC Statistical Surveys showed that the ABS data were understating
the extent of international trade in legal services. Significant efforts have been made in recent years to
improve the quality of ABS data and as such, the ABS is in contact with ILSAC to better align the
data provided by both organisations. It is now possible to reconcile the differences which continue to
appear on an annual basis between data published by both the ABS and ILSAC. As the two surveys
become better aligned, the size of trade in legal services can be more accurately measured.
C.4.

Bilateral or “Mirror” data

10.67. Exchange of statistics between countries (especially main partners) is seen as a particularly
relevant exercise for compilers of statistics on the international supply of services. However,
bilateral (or “mirror”) statistics should not be seen as a major data source on their own, but should
127

rather serve as a reference for checking and adjusting collected data. In this connection, it should
be noted that due to differences in applied concepts and data collection/compilation procedures,
such “mirror” data should be consulted with caution and after necessary reconciliation. It is also a
well-known phenomenon that mirror statistics on the international supply of services often contain
asymmetries, due at least in part to coverage, sampling and methodological differences. An
investigation into bilateral asymmetries between the United States and Canada is presented below.
10.68. Mirror data could potentially be useful to collect information on Mode 4 number of
persons as "sending" (i.e., exporting) countries may be able to more easily collect information for
some categories (especially contractual-service suppliers and those travelling for negotiation
purposes) than for others (e.g., intra-corporate movements, self-employed migrants). The
"receiving" (i.e., importing) countries may have such information as well.152
C.4.a. Country example: Reconciliation of the U.S.-Canadian Current Account
10.69. The Canadian-U.S. current account reconciliation, which explains the differences between
the official bilateral statistics published by Statistics Canada and those published by the U.S. Bureau
of Economic Analysis (BEA),153 is undertaken because of the extensive economic links between
Canada and the United States. Reconciliation of the U.S.-Canadian current account has been
undertaken each year between 1970 and 2008, and periodically since 2008. The reconciled estimates
are intended to assist analysts who use both countries' statistics and to show how the current account
estimates would appear if both countries used common definitions, methodologies, and data
sources.154, 155
10.70. Differences occur in the official statistics of the U.S. and Canadian current accounts because
of variations in the definitions, methodologies, and sources used by each country. Some of the
differences are sometimes in components of the current account for which data are preliminary and
subject to revision; these differences may be reduced or eliminated when final data for these
components are incorporated.
10.71. The longstanding Canadian-U.S. current account reconciliation is among the leading
examples of the benefits of international data exchanges. As a part of the reconciliation process,
Canada and the United States have evaluated the accuracy of each other's statistics, and as a result,
each country now includes in its official statistics data that are provided by the other country. The
exchange of data between Canada and the United States for transactions such as trade in goods,
travel, passenger fares, Canadian and U.S. Government transactions, and some large transportation
transactions covers a substantial portion of the value of the Canadian and U.S. current account and
has eliminated some of the differences in the Canadian and U.S. official statistics. In addition, the
reconciliation process has identified areas where errors and omissions may exist helping each
country to target data improvement efforts.
152

More information on potential sources for mode 4, and by direction of movement (ingoing/outgoing), is
provided in table 11.8.
153
U.S. Bureau of Economic Analysis and Statistics Canada, Reconciliation of the United States-Canadian
Current Account, 2010 and 2011.
http://www.bea.gov/scb/pdf/2013/01%20January/0113_unitedstatescanada.pdf and http://www.statcan.gc.ca/pub/13-605-x/2013001/article/11772-eng.htm#n1.
154
A detailed article on the methodology was published by BEA in "Reconciliation of the U.S.-Canadian
Current Account" in the November 1992 Survey and by Statistics Canada in Reconciliation of the CanadianUnited States Current Account, 1990-1991. Statistics Canada also published a shortened version in the
December 1992 Canadian Economic Observer and in Can ada's Balance of International Payments, Third
Quarter 1992.
155
To better understand the differences in investment income, each country’s positions on bilateral foreign
direct investment and on some portfolio investment accounts are also compared during the exercise of t he
Canadian-U.S. current account reconciliation.

128

10.72. Although the U.S. and Canadian official statistics are reconciled and extensive exchange of
data take place between Canada and the United States, differences in the official statistics remain.
Complete substitution of the reconciled statistics for official statistics and complete exchange of data
are not feasible for several reasons. For some accounts, the protection of the confidentiality of the
source data bars the exchange of data. A few differences are attributable to different requirements
for integrating the international and national (domestic) accounts in each country.
10.73. To reconcile the official Canadian and U.S. bilateral current account statistics, the official
statistics are first restated to a common basis, that is, they are adjusted for definitional and
methodological differences; and then statistical adjustments are applied to reach the reconciled
values. The framework for restating the statistics to a common basis mainly follows the international
guidelines published in the International Monetary Fund's Balance of Payments Manual. The official
Canadian and U.S. statistics now largely conform to the international guidelines, but some
differences from the international guidelines, and between the Canadian and U.S. statistics, remain
because of data limitations, difficulties in determining country attribution, and differences in
classification. In addition the international guidelines can sometimes provide for more than one
acceptable treatment.
10.74. Definitional adjustments mainly reflect data limitations and differences in country
attribution. Methodological adjustments mainly reflect differences in classification.
10.75. Statistical differences reflect the use of different source data in the United States and
Canada, the difficulty in determining country attribution because of insufficient data, the
preliminary nature of some data (particularly for the most recent year), and the use of sample data
between benchmarks. For both the northbound (United States credit transactions and Canada debit
transactions) and the southbound (United States debit transactions and Canada credit transactions)
statistics, most of the statistical differences are in the categories of services other than transport and
travel and in the investment income accounts.
C.4.b. ECB/Eurostat - Bilateral data exchange – FDI network
10.76. The FDI Network is a joint ECB/Eurostat initiative, which has been developed in close
cooperation with FDI compilers from all Member States. Its aim is to tackle asymmetries in FDI and,
as a consequence, to improve data quality. To reach this purpose, the FDI Network facilitates a secure
exchange of information between national compilers on specific FDI transactions and positions.
10.77. The transmission and exchange of confidential data in the context of the FDI Network is
foreseen exclusively for statistical purposes and in particular for the purpose of increasing the quality
of the EU/euro area balance of payments statistics.
10.78. Transactions taken into account for a specific reconciliation round are related to transmissions
exchanged in a certain period. These transactions were introduced into the template and sent out by
Eurostat asking the FDI Network users to amend the information of those transactions where they
acted as an initiator or – in a second sheet - as a counterpart. The compiler Member State completed
the template for their transactions. A similar procedure is used for the reconciliation of positions.
10.79. In accordance with the provisions of the FDI Network Manual, Eurostat regularly monitors
the results of the reconciliation process of the transactions exchanged through Network. FDI
asymmetries can be analysed from two different points: at EU-28 156 aggregate level using data
available from quarterly BoP, as well as annual FDI, and on a bilateral level using mirror FDI data for
156

Croatia is not included in the tables above (as be fore 1st July 2013).

129

those countries more actively using the FDI network. The analysis of the counterpart countries are
based on net figures reported by countries, while the transactions and positions exchanged true the
network are expressed normally in gross values. The asymmetries are calculated separately for inward
and outward flows/stocks. The tables indicate the decrease of asymmetries and the status of several
transactions within a period.
10.80. The bilateral exchange of data (flows/stocks) could improve the quality of transmitted data.
Reconciliation in the FDI network leads to a number of closed cases. When reconciliation cannot be
made, it is typically due to incorrect information received information from the company involved or
some methodological differences in Member States, particularly due to the use of different valuation
methods.
D.

Linked microdata
D.1.

Good practices in using linked microdata

10.81. In addition to the various data sources that have been discussed in this chapter and the
preceding chapters, it is important to note that the integration of two or more existing data sources can
also provide a lot of additional information relevant for the compilation of the international supply of
services, as is described in more detail in chapter 13. In order to facilitate such microdatalinking, it is
first advised that all economic entities are included in a SBR that serves as the central sample frame of
the statistical system.
10.82. Microdata are the observation data that may be collected on an individual statistical unit,
including an individual person, household, business or other entity.157 Examples of microdata may be
the name and location of a survey respondent; enterprise identification number from the BR;
information on the number and characteristics of employees of an enterprise; etc. Such microdata may
already be available from existing data warehouses within the NSO, or may come from other
administrative data sources or the other sources cited in this chapter. If microdata are being linked
across agencies or across different data providers, it is recommended that an aggregate list of the
entire population of individuals or enterprises from all sources are marked with an indicator of
whether it has been linked or not. A manual review of the entities that have not been linked may
reveal duplicates or other inconsistencies that can be resolved.
10.83. This Guide advises that compilers conduct a thorough review of enterprises with similar
names and multiple establishments, in order to identify the appropriate establishments to include in
the linking exercise and to avoid double-counting.
D.2.

Purpose and description

10.84. Compilers may consider using microdata to link characteristics of reporting individuals and
enterprises in order to get a different perspective on a target population, to identify inconsistencies or
overlaps in the data, or sometimes to even fill data gaps.
Often microdata are not publicly released other than for officially-sanctioned research studies, which
typically involves a rigorous approval process.158

157
158

OECD glossary of statistical terms. http://stats.oecd.org/glossary/detail.asp?ID=1656.
See chapter 20 for a more detailed discussion of t he release of microdata.

130

D.3.

Using linked microdata for purposes of statistics on the international supply of

services
10.85. Linked microdata on enterprises may prove most valuable in compiling foreign affiliate
statistics (FATS) and data on modes of supply. Namely, linking microdata on foreign ownership
indicators (available from the BR, company financial statements, or other industry sources) will help
to identify the target population for FATS surveys and service transactions through Mode 3.
Moreover, microdata on the characteristics of employees of enterprises that may indicate foreign
residence could also help in identifying these target populations in relation to mode 4.
10.86. Microdata on labor statistics and/or individuals’ residence will also help compilers in
identifying presence of natural persons for compiling supply of services through mode 4.
10.87. Linking microdata can also provide opportunities for the NSO to conduct special research
studies on different aspects of international supply of services which may be of particular interest to
the country, such as the characteristics of domestic firms engaged in trade in services; determinants of
imports of services; and effect of activities related to foreign affiliates and foreign direct investment
on the domestic economy, among many others.
D.4.

Country experience: the United States

10.88. The U.S. Bureau of Economic Analysis (BEA) jointly engaged with the U.S. Bureau of Labor
Statistics (BLS) in a microdata-linking study of multinational enterprises based in the United States in
2011.159 The aim of the project was to uncover more about the geographic, occupation, and wage
distributions of employment by U.S.-based multinational enterprises s. While this linking study does
not directly relate to statistics on international trade in services, the methodology employed
demonstrates how such linking is possible and can open the door to future extensions and applications
in the area of FATs and statistics of trade in services by modes of supply.
10.89. The project combined firm identifiers from the BEA 2004 Benchmark Survey of U.S. Direct
Investment Abroad with BLS microdata on employment in establishments of these firms, for a pilot
group of the largest U.S.-based multinational manufacturing enterprises. BEA data on multinational
enterprises based in the United States were used to match a pilot group of U.S. parent firms of
multinational enterprises with their establishments appearing in BLS data. The pilot group consisted
of the largest 500 U.S.-based multinational manufacturers (by primary industry of the U.S. parent) in
the BEA’s firm-level data from the 2004 Benchmark Survey of U.S. Direct Investment Abroad. The
efforts at matching were based primarily on the names, locations, and employer identification
numbers (EINS)160 provided in this survey.
10.90. Identification of the establishments of firms in the pilot group were based on the BLS
Quarterly Census of Employment and Wages (QCEW), which collects information on total employment by month and total wage bills for all U.S. establishments covered in the Unemployment
Insurance program, as well as detailed information on the industry of main activity (at the six-digit
North American Industry Classification System (NAICS) level) and geographic location (at the
Census block level) for each establishment. These establishments were then matched with those

159

Elizabeth Weber Handwerker, Mina M. Kim, and Lowel l Mason, “Domestic employment in U.S.-based
multinational
companies,”
Monthly
Labor
Review,
October
2011.
http://www.bls.gov/opub/mlr/2011/10/art1full.pdf.
160
The Internal Revenue Service assigns the EIN, or Federal tax ID number, to identify a business entity. Most
large companies with many establishments report mor e than one EIN to the BLS Quarterly Census of
Employment and Wages. The one or two EINs that comp anies report to BEA in the Benchmark Survey of U.S.
Direct Investment Abroad generally match only a fra ction of these large companies’ establishments.

131

sampled in the BLS Occupational Employment Statistics (OES) survey, which provides data on the
distribution of their employees’ occupations and hourly wages.
10.91. The automatic matching efforts entailed matching EINs between BEA firm-level data and
BLS establishment-level data. BEA firm-level data contain only one or two EINs per firm, while BLS
establishment-level data contain one EIN per establishment, and the establishments of each firm may
report many different EINs in the BLS data. Thus, additional EINs for each firm were found by
matching firm names and addresses with the establishment names and addresses in the QCEW, as
well as by using enterprise family lists (lists of employers that operate under different names but are
part of the same enterprise) from other BLS programs, enterprise information in the Compustat
database, and other sources of data on firms.
10.92. However, such automated matching procedures are imperfect: some firms are matched with
unrelated establishments, while other firms appear to be matched with only a fraction of their
establishments in the QCEW. Accordingly, the lists of all establishments found through automated
matching were reviewed manually, and the establishments matched in error were removed. Then, the
QCEW was searched for additional establishments identified from enterprise Web sites, filings from
the Securities and Exchange Commission (SEC), and enterprise annual reports.
10.93. The firms identified in BEA surveys were considered to be “adequately matched” with BLS
establishment data if the total employment of all matched BLS establishments for a particular firm
was within 20 percent of the total employment reported in the BEA survey. The establishments of the
adequately matched firms were then linked with establishments in the OES survey data. Although a
large portion of U.S. employment of U.S.-based multinational manufacturing enterprises was found in
the QCEW and OES data, the missing employment was not random. The multinational enterprises
that were found to match with the BLS establishment data were different from the firms that remained
unmatched. For example, it was more difficult to match privately owned firms (which generally
disclose less information than publicly owned firms and, in particular, do not file annual reports with
the SEC) and firms that have undergone liquidation or reorganization since the survey date.
10.94. Furthermore, the sample design of the OES survey is intended to produce estimates at the
State and industry levels, not to provide estimates for the unusual subsample of multinational firms
examined in this study. The OES survey collects information from a sample of establishments rotating
in 3-year panels, with sample probabilities that vary by establishment size. The probability that a
larger establishment is included in the OES sample over the course of 3 years is greater than the
probability that a small establishment is included. This difference could affect the distribution of
occupations and wages in the subsample of establishments that are matched with multinational firms.
Consequently, the sample and nonsample variance of these estimates may be large.
10.95. Future research could be based on other information collected by BEA, such as the
magnitude and scope of foreign direct investment, the amount of intrafirm trade, the destination
countries for foreign direct investment, the enterprises’ degree of “global engagement,” and their trade
in services.

132

Chapter 11 Comparing Data Sources
11.1. Scope. This Guide discussed, in Chapters 6 to 10, a variety of data sources including
enterprise and establishment surveys, surveys on persons and households, international transactions
reporting system (ITRS), administrative records and other data sources. This chapter provides a
comparison of the advantages and disadvantages of these sources in compiling the various EBOPS
components, as well as FATS and modes 2 and 4 number of persons and trips. The advantages,
identified with a plus (“+”) sign, and disadvantages, identified with a (“-”) minus sign, are analyzed in
comparison tables by systematically addressing the coverage of transactions (C), the accuracy of
reporting (A), the timeliness and frequency (T), the relevance (R) and the burdens of reporting and
processing data (B) involved in each data source. In doing so, this chapter forms the basis for Chapter
13 where the integration of data from different sources is discussed.
A.

Summary of good practices

11.2. Given the fact that the whole chapter is about comparing data sources and good practices for
the collection of five main services groupings (manufacturing services on input owned by others and
repair services, transport, travel, other services, government goods and services), FATS and the
modes of supply, the user is advised to read sections B (resident/non-resident trade in services), C
(FATS) and D (Modes of supply) for specific discussions of good practices in these area.
11.3. It is however generally considered a good practice that compilers, guided by the comparison
tables available in this chapter explore and compare the possibilities offered by the sources readily
available nationally and uses those that present an advantage (identified with a plus (“+”) sign),
relatively to other sources, for collecting the information for the supply of services element under
consideration. If no source is available yet for the topic under consideration, it is considered good
practice that the compiler carefully compares and assesses, which type of data collection or
combination of data collections would be most useful to set up in his national context, using the
guidance of the comparison tables in this chapter,
11.4. To assess the advantages and disadvantages of different sources with respect to their coverage
of the specific service category, the good some collection practices are marked with “++”in order to
illustrate that they, relatively, present most advantages. However, the reader should always be aware
that depending on national circumstances, the suggested source might not always be the best option
for all countries and in practice other existing sources or a combination of available sources can be
more suitable to collect the necessary information.
B.

Comparison of data sources for resident/non-resident transactions

B.1.
Preferred collection practices and comparison of data sources for manufacturing
services on physical inputs owned by others and maintenance and repair services n.i.e.
11.5. This guide suggests data on manufacturing services on physical inputs owned by others (and
corresponding goods), short, manufacturing services, could be collected most preferably through
surveys that are sent to the manufacturing companies which organise international production (for
imports of manufacturing services) and to processing companies/sole-proprietorships (for exports of
manufacturing services). It is important to explain and verify that respondents understand the scheme
of production networks in which they are involved. Surveys would also be an efficient source for
collecting information on maintenance and repairs n.i.e. (paying special attention to identify
separately maintenance and repairs related to construction and computer services given the definition
of these items).

133

11.6. ITRS is an alternative data source for these items. For manufacturing services it should be
noted that it only captures fees paid to processors, meaning that transactions without settlements (e.g.
between affiliated enterprises), may not be captured. Although customs data may also be used to
derive estimates on manufacturing services, it is widely recognised that the difference between the
value of imports and that of exports of processed goods in general does not represent the processing
fee. Customs data could be useful in identifying companies engaged in processing, could be used for
reconciliation exercises when processing fees are obtained from ad-hoc surveys or could also be
useful to complement other methods or alternatively, as input in a data model (by major type of
manufacturing). Some (regular) benchmark surveys may complement those to validate the model
outcomes. Customs data are also potentially useful for identifying goods being sent abroad for repairs
and maintenance. Although difficult information on fees could be requested in customs forms (see
BPM6). Given the tax concessions provided to some of the firms that may be involved in processing,
eventual statements to tax authorities may provide relevant information.
Table 11.1 Comparison of data sources for compiling manufacturing services on physical inputs
owned by others and maintenance and repair services n.i.e. (import and export)
Enterprise and Establishment
Surveys
(EES)

ITRS

Administrative Records
(AR); customs data; Tax
concessions

Coverage
(C)

(++) cover enterprises engaged
in such activities
(-) may not fully cover full
population if threshold, in
particular for resident SMEs

(+) cover settlements related to
transactions
(-) does not cover all
settlements if threshold is
applied

Accuracy in
Reporting
(A)

(+) may require explanatory
notes/follow up when
respondents do not fully
understand the scheme of the
production network

Timeliness,
Frequency
(T)
Relevance
(R)

(-) often lag of more than one
month; quarterly frequency

(+) if settlements of processing
fees occur between unrelated
parties
(-) misclassifications; may
include other (goods or
services) transactions between
processing company and its
client
(+) lag of a only few days;
monthly frequency

(+) customs data could be
useful for identifying
companies involved in
manufacturing services or
processing and maintenance and
repair, but declaration forms
may need to be adjusted
(-) Most probably not possible
to ask for fee charged.
(+) information from Tax
authorities.
(+) customs data: if gross flows
related to goods for processing
and repairs and maintenance are
properly identified (mainly to
adjust goods). Possible to ask
for fee

Burdens of
Reporting and
Processing Data
(B)

(-)

(-) in case of difficulty of
grossing up

(+) can collect detailed
information on manufacturing
and maintenance and repair
activities

(+) if different processingrelated transactions are
identified by separate
transaction/codes
(-) additional information
needed on corresponding
movement of goods to adjust
general merchandise

(-) especially for banks
reporting on behalf of
transactors

(+) once implemented

(+) implies advantages and (-) implies disadvantages in terms of each element.

134

(+) lag of a few weeks; monthly
frequency
(-) difference between value of
imports and exports often does
not reflect manufacturing (or
repair) fee (e.g. holding gains
and losses)

(+) if no additional work

(+) in terms of coordination
with data collecting units

B.2.

Good collection practices and comparison of data sources for transport

11.7. Enterprise and establishment surveys of resident and non-resident carriers are the main
sources to collect transport services data in many cases. Compilers are advised to make sure that the
survey is elaborate enough to cover most of the EBOPS categories related to transport services. An
additional difficulty - which applies to all sources for transport - lies with the information necessary
for compiling data based on balance of payments recording rules. 161 Compilers should also be aware
of potential difficulties in obtaining data reflecting the activities of non-resident carriers’ for freight
related services (i.e. imports of freight transport services- debits).However, in many cases transport
operators establish branches or agents in client countries who may then respond to surveys. For
economies with regulated modes of transport (air transportation being the best example) this may
even be more relevant. Alternatively countries can survey resident importers/exporters about their
transport expenditures paid both to resident and non-resident carriers.
11.8. Information on passenger transport services can also be obtained from traveller surveys
including questions on passenger fares of international transportation, although the same limitations as
encountered in travel (the capacity of the respondent to provide the right information, and the
compiler’s ability to know the population) would apply.

161

The reason for this is that, in line with BPM6 recommendations (para 10.78) all freight costs up to t he
customs frontier of the economy of the exporter are shown as incurred by the exporter and all freight costs
beyond the customs frontier of the exporter are shown as incurred by the importer (see chapter 14 and BPM6
CG 12.35).

135

Table 11.2 Comparison of data sources for compiling transport services (including postal and
courier services)
C

EES

(++) general enterprise
surveys cover transactions
by residents reporting on
their transactions involving
non-residents (less relevant
for passenger transport
imports)
(++) surveys on
agents/branches of nonresident carriers cover
imports of residents (for
passenger transport debit
could be targeted to ticket
selling agents)

(++) specific surveys may
cover major resident
carriers providing services
to non-residents (for both
passenger and freight
transport credit)
(-) may not fully cover
resident small carriers

(-) often lag of more than
one month, quarterly
frequency
(+) can be designed so that
relevance is ensured

(-)

(-) in case of the difficulty
of grossing up

PHS

(partly +) household
surveys or border surveys
can cover transport service
(freight and passenger
fares) imports by residents

(-) border surveys can
provide information on
passenger fares incurred
by non-residents.

ITRS

(+) cover settlements
related to transactions
(-) does not cover
settlements under
thresholds
(-) may be difficult to
capture freight export and
import included in goods

AR; customs data

(+) customs data could be
useful for identifying
freight costs
(-) do not cover transport
occurring between two
foreign countries

(-) households/individuals
may not identify cost
components of package
tour; depends on sample
representativeness for
international transactions
(-) lag of more than one
month, quarterly frequency

(+)but compilers need to
pay attention to partner
country attribution.

(-) difference by types of
merchandise cannot be
fully reflected

(+) lag of only a few days,
monthly frequency

(+) lag of a few weeks,
monthly frequency

(+) can be designed so that
relevance is ensured
(-) recollection of
information may be
approximate

(-)

(-) in case of the difficulty
of grossing up

(-) cannot easily identify
complex transactions

(-) typically cannot
identify freight costs

(-) especially for banks
reporting on behalf of
transactors
(-) in terms of aggregation
of micro data

(+) once implemented

(+) if no additional work

(-) in terms of aggregation
of micro data or
coordination with data
collecting units

(+) implies advantages and (-) implies disadvantages in terms of each element.

11.8. ITRS may cover most exports and imports of transport services categories, but may not

properly distinguish freight exports under fob contracts where the transporter is a resident, and
imports of freight transport under c.i.f. contracts, as the transport value is usually included in the value
of merchandise. In these cases, compilers need additional information to estimate the imports of
freight transport services. In addition, when using an ITRS, compilers should pay attention that the
country of operations may be different from country of registration or residence of the owners
(especially important in marine shipping).

11.9. Administrative records such as customs documents could provide useful information for
modelling freight costs, as long as they include information about the commodity traded, weight, the
origin and destination, and the mode of transport. Obviously, this also depends on the information
136

available about the carrier involved. For example, administrative records may not have information on
transport costs by type of merchandise. They also will not be helpful for transport of cargo or
passengers occurring between one foreign country to another foreign country by a resident
transporter, and administrative records might also be a weak source to measure transactions related to
auxiliary transport services (such as loading containers, storage or air traffic control).
B.3.

Good collection practices and comparison of data sources for travel

11.10. Travel related transactions are unique in comparison with other services in that individuals
are involved systematically in the consumption of the related products. The most efficient way to
collect data from those traveling is therefore through surveys of persons or households where samples
are asked to report information on their spending, while outside their home country. However, there
are certain limitations to such surveys. The first issue comes from the capacity to reach relevant
households/persons, notably non-residents at the time they leave the country. Reaching residents
when they return to their home country is also a difficult task but they can be inquired later (i.e.
through household surveys).

11.11. Secondly, respondents may not have a perfect idea of all their expenses made during their
travel especially if they have to fill the survey while they are on their way to leave the country. Some
expenses could have been paid well in advance such as the hotels or some transport. Expenses
engaged in the context of business travel could be paid or reimbursed by its enterprise (in some cases
he/she may even not know the exact value of the service transactions paid by his employer).

11.12. ITRS could cover payments by resident and non-resident travel agencies as well as large
payments for travel. Payments by travel agencies may include expenses on hotels, domestic transport
and meals during the travel. Transaction data from foreign exchange bureaus could also approximate
travel expenditures, although such data could contain non-travel transactions. Another potential
weakness which can be important in the context of travel is that settlements and service delivery may
not occur in the same time period.

11.13. Other data sources, such as payment and bank card (credit and debit) transactions, could also
be explored. These sources are not perfect, as purchases with credit card or withdrawals with debit
cards could have other purposes than travel spending. Therefore, the use of further information
relating to the credit and debit card transactions should be explored, such as the merchant code and
the information concerning the point of sale, in order to identify the relevant travel transactions.
Compilers should be aware that persons travelling do not always use a card as a way of payment but
may use cash (or transactions may be paid by a third party). However, payment and bank card
information might be very helpful to establish trends and geographical breakdowns and it offers a
much larger sample than any survey on persons or households likely can provide. In the context of
personal and household surveys, administrative records (including information from entry/exit cards)
as well as mobile phone records, often the information is combined in a model to derive travel
information.

137

Table 11.3 Comparison of data sources for compiling travel
EES

(+) cover major
resident suppliers,
e.g., travel agencies,
hotels, hospitals,
education facilities

(+) potential source
for business travel
(-) do not cover nonresident suppliers

PHS

(++) border surveys
capture expenditures
by non-residents

(++) border or
household surveys
capture expenditures
by residents

(+) if respondents
fully understand the
residence of the
customers

(-) lag of more than
one month, quarterly

(+) can incorporate
breakdown by
products partially
(-) recollection of
information may be
approximate

(-)

(-) in case of the
difficulty of grossing
up

(-) possibly depend
on unrepresentative
samples and
frequency of survey
(-) members of
households may not
identify cost
components of
package tours
(-) recalling errors
might exists for
household survey
(-) uncovered
periods may exist
when the survey is
not frequently
conducted
(-) lag of more than
one month (border
surveys more
timely), quarterly or
annually
(+) can incorporate
detailed questions,
e.g. expenditure by
products

(-) response-rate of
household surveys

(-) for border
surveys in operation
costs
(-) if grossing up
survey results is
difficult

ITRS

(+) cover settlements
by resident and nonresident travel
agencies
(+) capture large
payments
corresponding to
travel expenditure
(+) data of foreign
exchange companies
cover most foreign
exchange
transactions by
individuals if not
paid in advance
(-) does not capture
travel payments
below threshold (as
travel payments are
small)
(-) does not cover
settlements under
thresholds
(-) payments and
services may not be
in the same period
(-)foreign exchange
data do not represent
entire expenditure
and can include nontravel related
payments

registration of
immigration office
(including E/D
cards, border
counts), can be used
for identifying
number of persons
crossing borders;
official records may
be used for students,
medical patients
(-) possibility of
abrupt suppression
of sources

ODS

(+) payment and
bank card data cover
most payments

(-) depend on
immigration control,
but some countries
may not identify
relevant population

(-) credit/debit card
data can include
non-travel related
payments

(+) lag of a few days,
monthly

(+) lag of a few
weeks, monthly

(-) cannot be
designed easily so
that details of travel
expenditures are
correctly collected

(-) cannot be
designed easily to
identify relevant
population

(+) if non-travel
amounts can be
excluded

(-) especially for
banks reporting on
behalf of transactors

(+) once
implemented

(+) if no additional
work

(+) in terms of
coordination

(+) if no additional
work

(+) in terms of
coordination

(+) implies advantages and (-) implies disadvantages in terms of each element.

138

(+) partner country
data may be used as
supplementary
source.

(+) mobile phone
records can be used
for identifying both
inbound and
outbound travellers

B.4.

Good practices and comparison of data sources for other services162

11.14. For most of the other services, the guide suggests enterprise/establishment surveys could
provide a large share of sources for estimates. The advantage of these surveys is that they could be
designed to collect specific information corresponding to the compilers’ needs.
11.15. Surveys are however costly to process, and are also costly for the respondents as they have to
train employees to complete questionnaires. This is particularly true for transactions in services which
are difficult to extract from accounting documents. Increasing the response burden to small entities
could be counterproductive as it could affect the interest of these small entities to respond to other
surveys for which the presence of these entities is more important.
11.16. To improve the efficiency in data collection, enterprise/establishment surveys could be
designed to target a specific activity or a specific industry in particular for exports. In certain cases
this may also be relevant for imports of services (e.g. affiliated companies, activities for which
outsourcing/offshoring of activities may be important). Respondents could then be asked to provide
more details for services activities in which they are likely involved and less details for other services
where they expect to have negligible impact. In addition it is important to note that collecting
representative data for imports of services may be more difficult than for exports as any resident
entity (large or small) can import services. However, as indicated above, a number of factors could be
used to define those enterprises/establishments which are more likely inclined to be big importers of
services.

11.17. As for any other survey, the degree to which a sample is representative of the population is a

key indicator in the quality of the results. Compilers should take advantage of all other available
sources of data to supplement and validate survey results, and to reduce processing costs and
reporting burdens.

11.18. ITRS has the great advantage of providing a larger coverage than what enterprise or
establishment surveys could do. ITRS comes with a strong degree of obligation for the respondents to
provide the information requested. ITRS data may require low compiling costs, as the compilers can
concentrate their efforts and resources on items that are not sufficiently covered. However, in some
cases, it might be difficult to incorporate details, e.g. EBOPS, in reporting forms.

11.19. Administrative records cover a large part of transactors but may not be sufficient for all
services categories. A promising source in this respect are the tax records which include the value of
services sold or purchased to or from non-residents (Value Added Tax for services) as well as the
location of the service transaction. This information can be particularly useful to identify service
traders. The advantages of these records are their low costs and their timeliness, as well as
completeness (Tax or Finance Departments may have greater powers to convince respondents). The
drawback is that since the main purpose of this source is not to produce statistics, the validation of the
results might not be adequate for the compilers’ needs, and also the compiler has no control over
changes in (or even abolishment of) the data source by the Tax authorities.

162

Excluding manufacturing, government goods and services n.i.e., transport services, and travel.

139

Table 11.4 Comparison of data sources for compiling other services (excluding manufacturing
services on physical inputs owned by others, maintenance and repair services n.i.e., and
government goods and services, n.i.e.)
C

EES

(++) surveys cover
major resident
enterprises
(-) may not fully
cover resident
small enterprises;
may be difficult to
ensure
representativeness
of imports data

(+)

(-) lag of more than
one month,
quarterly
(+) can incorporate
detailed questions
relatively easily,
e.g. complex
transactions, such
as construction.
Could also be
tailored to target
specific activities,

SPH

(+)can cover internet
transactions (e.g.
consumption of
software, audiovisual products)
(+) Can collect
information on
persons physically
present to provide
services (health,
cleaning)

(+) (targeted) labour
force survey can
cover mode 4
persons supplying
services abroad
(+) Household
survey can cover
self-employed
residents supplying
services through
mode 1
(-) recollection of
services consumed
through internet may
be approximate
(-) lag of more than
one month, quarterly

(+) if transactions can be
identified for some specific
sectors
(+) Tax records can help
identify service traders and
may include the value of
services sold or purchased
to or from non-residents.

ODS

(+) data on financial
market and institutions are
useful complementary
sources for estimating
insurance and financial
services

(+)if settlements are
occurring between
unrelated parties
(-) misclassifications
(+) lag of a few
days, monthly

(+)

(+)if resident/non-resident
transactions separately
identified

(+) lag of a few weeks,
monthly

(+) data are provided
timely and frequently

(+) all the resident
(-) mapping may be
non-resident
difficult as not designed
transactions can be
for statistical purposes
captured once the
system is designed
properly
(-) for long term
construction projects
can be difficult to
establish the
location from which
work is conducted

B
(-) in particular for
(-) large samples to
(-)especially for
(+) if no additional work
imports
ensure
banks reporting

representativeness
indirectly on behalf
(+) in terms of
(-) in case of the

of transactors
coordination with data
difficulty of
(-) in case of the

collecting units
grossing up
difficulty of
(+)once
grossing up
implemented
(+) implies advantages and (-) implies disadvantages in terms of each element.
R

(+) Can be designed
so that relevance is
ensured.

ITRS

(+) cover
settlements related
to transactions
(-) does not cover
settlements under
thresholds

(-) mapping may be
difficult as not designed for
statistical purposes

(+)

(-) if statistical techniques
need to be used

11.20. Surveys and the ITRS cannot be used to directly compile some services such as FISIM or
insurance services which need supplementary information and require models to be generated.
However they could provide important partial information for the measurement of these service items
(e.g. insurance payments). Commercial databases on financial institutions could also be useful for
these items if they contain sufficient detail. For construction, surveys may provide more detailed and
relevant data than the ITRS (see also Chapter 8).
11.21. For some specific detailed service categories, surveys on persons and households, which are
primarily used to collect socio and demographic information, could be useful. For example, although
normally small in comparison with enterprises or establishments transactions, purchases of products
downloaded online by individual consumers (software or audiovisual products) or commissions paid
140

directly to non-resident brokers by individuals directly buying or selling securities abroad would most
likely be covered only through such surveys. (see chapter 7)
B.5.

Good practices and comparison of data sources for government goods and services n.i.e.

11.22. Similarly to travel, government goods and services n.i.e. are transactor-based (i.e. government
units supply services or purchase goods or services). Compilers should use administrative records for
the majority of services supplied by government units. Administrative data could also be used to
capture data on expenditures by government units abroad, as well of staff (e.g. by obtaining data on
their wages to derive an estimation of their expenditure). For some components, such as goods and
services acquired by government units, by staff and their dependants working in foreign embassies
and military enclaves (for example diplomats or military personnel), these transactions could be
obtained through surveys of the units. For staff and their dependants surveys could be targeted
towards embassies. For example, these could include questions on the number of diplomats and other
government employees, and their earnings, to estimate expenses. ITRS if well designed could capture
expenditures by government units; however it could be difficult to capture transactions with foreign
government enclaves located in the reporter’s economy. In addition, it cannot be used to capture
expenditure by diplomats and their dependents.
Table 11.5 Comparison of data sources for compiling government goods and services n.i.e.
C

A
T
R

Surveys

(+) surveys (less likely) on
own country’s
governmental agencies
cover goods and services
acquired from nonresidents

(-) surveys on foreign
embassies cover goods and
services consumed by nonresident
(+) if sufficient responses
(-) lag of more than one
month, quarterly
(+) can incorporate
detailed questions (e.g.
earnings)

(-)foreign embassies may
not report

(-) if estimation is needed

ITRS

(+) cover settlements
related to transactions
(-) does not cover
settlements under
thresholds

(++) the Central Bank
dealing with international
settlements of the central
government may provide
comprehensive data or
Administrative records
from Ministry of Foreign
Affairs
(-) transactions in kind are
not covered

ODS

(+) budget data and
statements of government
accounts provide
information on costs of
government services

(-) can include payment s of
other goods and services
(+) lag of a few days,
monthly
(+) all the resident nonresident transactions can be
captured once the system is
designed properly (-)
difficult to capture
transactions with foreign
governments’ enclaves
located in the reporters
economy
(-) Cannot capture
expenditure by diplomats
and their dependants.

(-)especially for banks
reporting on behalf of
transactors

(+) once implemented

(+)

(-) budget data are mere
assumption
(-) publication of
statements not timely
(-) cannot be designed so
that necessary data can be
identified

(+) lag of a few weeks,
monthly
(-) cannot be designed so
that necessary data can be
identified

(+) very low additional
costs

(+) in terms of
coordination

(+) implies advantages and (-) implies disadvantages in terms of each element.

141

(+) very low additional
cost

(-) if estimation is needed

Good collection practices and comparison of data sources for FATS

11.23. For FATS, notably the information on services provided via mode 3 (commercial presence),
enterprise/establishment surveys are the most useful source of information. As noted in chapter 6, an
understanding of the FDI framework provides a strong basis for the establishment of the FATS
universe; key FATS variables could even be incorporated into existing FDI surveys. Administrative
records or commercial database on foreign affiliates are also useful but their forms and questionnaires
cannot be easily designed to provide data useful for statistical and analytical purposes. Thus, they
could rather be used as supplements to the survey on foreign affiliates.

11.24. ITRS, which provides strong sources for resident/non-resident transactions, is not necessarily
useful for compiling FATS, because ITRS only captures settlements between foreign affiliates and
residents, and such information does not measure the activities of foreign affiliates (including the
provision of services).
Table 11.6 Comparison of data sources for compiling inward and outward FATS
C

EES
Inward FATS

ODS

(++) survey used to collect
information on a compiling
economy's business structure,(
identifying the foreign-controlled
population of firms, as well as the
country of the UCI)
(++) (key) FATS variables can be
incorporated in FDI inward surveys.
(++) dedicated inward FATS surveys
(-) may be difficult to fully cover
resident SMEs

(+) business registers, tax returns or
regulatory reports cover (most)
resident enterprises.

(+) industry associations, commercial
databases, e.g. financial statements of
major enterprises, could be useful
(+) surveys conducted outside the
main statistical program such as by
sector specific ministries (e.g.
research and development surveys)
(-) Partner country data. However
level of detail likely small for use,
and care should be taken over the
interpretation of concepts

(+) customs data for the trade
variable, but through linking
exercises

Outward FATS
(++) key FATS variables can be
incorporated in FDI outward surveys.
(++) dedicated Outward FATS
surveys
(-) may not fully cover foreign
affiliates of resident SMEs

(+) business registers, tax returns or
regulatory reports could be used to
identify if residents have affiliates
abroad, and use relevant information
if included in records.

(+) industry associations, commercial
databases, e.g. financial statements of
major enterprises
(+) surveys conducted by specific
ministries (e.g. research and
development surveys)
(-) partner country data, However
level of detail likely small for use,
and care should be taken over the
interpretation of concepts

A
T

Inward and outward FATS
(+)
(-) lag of more than one month,
quarterly

(+)
(+) lag of a few months, annually

(+) can incorporate detailed questions

(-) FDI surveys: portion of FDI firms
are not part of FATS universe and
FDI surveys may be conducted on a
quick turnaround whereas FATS data
are usually collected on an annual
basis.
(-) in case of difficulty in grossing up

(-) cannot be easily designed for
statistical purpose

(+) if no additional work

(+) in case of the need of
coordination with data collecting
units

(-) cannot be designed for statistical
purpose

(+) no additional costs

(-) in case of the need of statistical
techniques to use data

(+) implies advantages and (-) implies disadvantages in terms of each element.

142

D.
Preferred collection practices and comparison of data sources for mode 2 and 4
numbers of persons
11.28. Table 11.7 and 11.8 provide an overview of the different sources from which the number of
modes 2 and 4 trips and/or persons can be derived, as well as some of their characteristics both from
the perspective of the sending (outgoing individuals) and receiving (incoming individuals) country.
Compilers should carefully evaluate how residence is defined in their country, be aware of the laws
and regulations in place for migration and trade in services, and assess how this can be used in a
statistical context.
11.29. Table 11.7 on mode 2 movements of persons show that the sources of interest would most
likely be the same as some of those used for the collection of travel information. ITRS cannot capture
such information, but household surveys and border surveys can capture information on outgoing
individuals travelling for personal or business reasons. Often these surveys will need to be elaborated
to identify the travel motive (business or personal reasons). Labour force surveys and enterprise
surveys constitute appropriate sources to capture the number of outgoing persons travelling for
business reasons. For incoming mode 2 persons, specific surveys targeting students, medical
personnel and tourists could be used as well as border surveys which can be conducted as persons
leave the country. Administrative sources like border counts or entry and departure cards can be used
to get a first and timely estimate of the number of individuals leaving or entering the country to
consume service; but compilers should be aware that it might be difficult to add supplementary
questions to these sources. Often the data will need to be combined, in particular for compiling
breakdowns, to obtain relevant data on the number of mode 2 movements/persons, e.g. through the
use of data models using numbers from border counts E/D cards and information from border or
household surveys.
11.30. Table 11.8 breaks down the categories of mode 4 persons into four categories: employees
and self-employed contractual services suppliers, intra-corporate transferees/foreign employees of
foreign affiliates and service sellers.
11.31. Regarding mode 4 types of persons, enterprise surveys (i.e. trade in services, structural
business or general enterprise surveys) constitute an important source both for incoming and
outgoing employees and of self-employed providing services under a contract (i.e. contractual
service suppliers). Surveys are particularly relevant given that they can be specially tailored, in
particular in the case of trade in services surveys. Inward FATS surveys, if they include variables
measuring exports and import of services of the foreign affiliate could possibly also include
information on the persons (and their numbers) who provide these services as employed
contractual service suppliers. For outgoing persons, outward FATS surveys may provide
information on employees on service contract working in the affiliate located abroad. The number
of intra-corporate transferees/ foreign persons directly recruited by affiliates and possibly service
sellers could also be captured through inward and outward FATS surveys. In particular, outward
FATS surveys could allow the identification of employees of the parent company who work in the
affiliate as intra-corporate employees.
11.32. As in the case of mode 2 persons, border and household surveys can also constitute an
interesting source both for all types of incoming and outgoing employees and self-employed mode
4 persons. Information regarding residence and identification of the employer-employee
relationship of the individuals will be crucial as household surveys only cover resident persons.
Finally, as for mode 2, the data will most probably need to be combined, in particular for
compiling breakdowns, to obtain relevant data on the number of mode 4 movements/persons, e.g.
through the use of data on characteristics sourced from surveys of persons or households,
combined with counts of those crossing borders. Use of data from partner countries may also be
relevant, in particular that it may be easier to capture the relevant information for sending countries.
143

Table 11.7 Comparison of data sources for compiling mode 2 movements of persons

(+) for incoming if
respondent fully
understands the residence
of the customer
(-) time lag of more than
one month, quarterly

Sources of data
PHS
< outgoing>
(++) household surveys
can capture information on
individuals travelling
(++) labour force surveys
can capture data on
individuals travelling on
behalf of their employers
(i.e. in the context of
business trips)
(++) border surveys
provide information
(mainly on characteristics)
of persons travelling.
(-) some adjustments to fit
both tourism and trade in
services concepts needed
(-) sample size might need
to be increased
< incoming>
(-) most likely not relevant
in case of household,
persons and labour force
surveys
(++)border surveys provide
information (mainly on
characteristics) of persons
travelling (often as they are
leaving)
(-) some adjustments to fit
both tourism and trade in
services concepts will be
needed
(-) sample size might need
to be increased
(-) depend on
unrepresentativeness of
samples and frequency of
survey
(-) time lag of more than
one month, quarterly

(+) can be designed so that
relevance is ensured

(-)

(-) in case of the difficulty
of grossing up

(-)

(-) in case of the difficulty
of grossing up

EES

(+) enterprise surveys can
include questions on
individuals travelling on
behalf of their employer.

(+) surveys targeting
tourism, education or
health related
establishments can include
questions to non-residents
to whom services were
supplied.

AR
< outgoing>
(+)border counts, E/D
cards or other types of
immigration records can be
used as first estimates of
number of mode 2 persons
leaving the country to
consume services

(+) border counts, E/D
cards or other types of
immigration records can be
used as first estimates of
number of mode 2 persons
entering the country to
consume services

(+) time lag of a few
weeks, monthly
(-) cannot be easily
designed/modified to fit
statistical needs
(-) possibility of abrupt
suppression of sources

(+) low costs

(+) in terms of aggregation
of micro data and re-use of
existing data
(-) coordination with data
collecting units

(+) implies advantages and (-) implies disadvantages in terms of each element.

144

ODS

(+) mobile phone records
can be used for identifying
both inbound and outbound
number of persons
travelling

(+) partner country data
may be used as
supplementary source.

(+) lag of a few weeks,
months
(-) for partner data
(-) for partner data need to
ensure that residence etc. is
compatible with national
definitions

(+) if no additional work

(+) in terms of coordination

Table 11.8
mode 4
EES

Comparison of data sources for compiling mode 4 persons/trips by category of
SPH

AR
ODS
Coverage (C)
"Contractual service suppliers (Employees, Self-emp loyed) and Intra corporate transferees"

(++) trade in services, structural
(+) border surveys
(+) E/D cards for
persons>
business, or general enterprise surveys whether when
rough estimates
(++) surveys covering
(++) inward FATS surveys (for
entering or leaving the (+) work
statistical needs of other
imports of services of foreign
country, to capture
permits/visas
domains, such as surveys
affiliates, i.e. incoming persons to
information on
/immigration records of a sectoral ministry (legal
provide services to foreign affiliates),
characteristics of
(+) legislation
services, R&D, education,
Can also identify intra-corporate
those crossing borders implementation
health)
transfers where employment
(-) household/labour
register (e.g. tax)
relationship is transferred to resident
force survey or
(-) registers of
(+) partner country data,
affiliate
population census to
persons do not cover
in particular for receiving
capture information
non-residents.
countries
on characteristics of
those crossing

borders: in general
(+) registers of
(++) trade in services, structural
non-residents are not
persons if remain
business, or general enterprise surveys covered.
residents.
(++) inward FATS surveys (for
(+) commercial
exports of services of foreign
registers(+)
affiliates) Can also identify intra
legislation
corporate transfers where employment (+) border surveys
implementation
relationship is transferred to nonwhether when
register (e.g. tax)
resident affiliated unit
entering or leaving the
(+) outward FATS surveys: for
country, to capture
(-) for incoming and
outgoing persons going to supply
information on
outgoing only if
services to non-resident affiliates.
characteristics of
coverage correspond
Can also identify intra-corporate
those crossing borders to MSITS2010
transfers where employment
(+) household/labour
definitions.
relationship is transferred to nonforce survey or
resident affiliated unit.
population census to
capture information
on characteristics of
those crossing
borders, if remain
residents.
Services sellers/commercial presence negotiations

(++) trade in services, Structural
outgoing persons>
outgoing persons>
business, or General enterprise
(+) Border surveys
(+) E/D cards
surveys
can cover information
(+)visas/immigration
(++) possibly inward FATS surveys
on services
records
sellers/commercial
(-) only if coverage
presence negotiations
correspond to
whether when
MSITS2010
entering or leaving the definitions.
country
Accuracy in Reporting (A)
(+) for
(+)
(+)
A (+) in particular trade in services and
FATS surveys; but requires clear
border/passenger
explanatory notes (in particular to
surveys (but with
clearly define mode 4 categories).
limited details)
(+)for outgoing for
household/labour
force surveys and
population census
(-) for incoming for
household/labour
force surveys and
population census

145

Timeliness, Frequency (T)
(-) uncovered periods
(+) time lag of a few
may exist when the
weeks, monthly
survey or census is
frequency.
not conducted on
regular basis
(-) time lag of more
than one month,
quarterly
Relevance (R)
(+) can collect detailed information on (+) all sources can be
(-) cannot be easily
contractual service suppliers
designed so that
designed/modified
(employees, self-employed) and intra
relevance is ensured.
(-) for incoming and
corporate transferees, in particular for
outgoing, for all
trade in services survey which can
sources only if
collect detailed information on
coverage of
number of persons in addition to
categories close to
values. Also useful for intra-corporate
MSITS2010
transferees in particular FATS surveys
definitions
(-) for self-employed outgoing, only if
(-) possibility of
covered (i.e. threshold)
abrupt suppression
of sources
Burdens of Reporting and Processing Data (B)

(-) in particular need to keep track of
(-) may need to
(+) if no additional
additional information which may be
increase samples
work
stored in other departments (e.g.
substantially

human resources)

(-) in terms of

in case of difficulty of coordination
(-) in case of difficulty of grossing up.
grossing up.
(-) lag of more than one month,
quarterly

(-) for partner data

(+) can be designed so that
relevance is ensured

(-) keep track of additional
information stored in other
departments (e.g. human
resources)

(-) in case of difficulty of
grossing up

(+) implies advantages and (-) implies disadvantages in terms of each element.

Country experiences

11.33. For compiling statistics on the international supply of services, most countries face the
challenge of having to rely on several sources with different coverage. The data collection is often
done in a context of scarce resources and having in mind the necessity to limit response burden.
Therefore the compilers need to make best possible use of available sources. The following
examples illustrate some country practices in dealing with the challenge of having to use and
compare characteristics of different data sources in order to collect relevant information.
E.1.

Country experience: Canada

11.34. For the majority of the statistics on services transactions, apart from travel and transport
services, Canada uses a specific survey on trade in services and several industry-oriented surveys.
These enterprise/establishment surveys are supplemented by administrative data recording non arm’s
length transactions with non-residents. Each source has its own coverage and its own collection of
service statistics which are not necessarily equivalent to EBOPS. However, as resources are limited
and as the reduction of response burden is an important objective for the Canadian compilers, there is
a necessity to use several sources to produce the best possible estimates.
11.35. Canada’s Industry-oriented surveys collect different information on revenues and expenses by
enterprises/establishments for specific industries. These surveys offer a broad coverage of the service
industries but the level of detail of the collected information is limited. These surveys are therefore
complemented by a specific survey on International Transactions on Commercial Services (BP21S)
targeting 3500 enterprises and collecting information for 32 categories of services. Transactions are
146

reported by country of trade as well as with the relation with the trade partners (either related or not).
The accompanying explanatory notes should in principle improve the quality of the responses.
11.36. Compared with enterprise/establishment surveys, administrative data provide, in principle, a
better coverage as all Canadian legal entities are required to fill the tax form if their total transactions
(not only in services) with their foreign related entities are above C$1 million. For example, for the
year 2010, more than 12,000 entities have filled that tax form and half of them have reported service
transactions. The tax form supplies transactions by name and country of the non-resident entities.
11.37. However, this tax form is limited to transactions with foreign related parties. That limitation
virtually eliminates Statistics Canada’s capacity to use this administrative information as a “datareplacement” source. In other words, the administrative data could not completely replace survey
data. Furthermore, several service categories are aggregated on the form to simplify the work of the
respondents. Very few definitions come with the tax form which may lead to errors or
misinterpretations by the respondents.
11.38. The files provided with annual administrative data are updated twice a year. As it is an
external source, less quality control could be applied on the information provided. For example, if no
transaction is reported for a certain entity, the reason could be absence of transaction but the
transaction could also not be processed yet. The main purpose of this tax form is not to monitor
transactions in services. Thus, verifications by tax authorities are usually done at a more aggregated
level so transactions for the same unit could be reported under one activity in one year and under
another activity the following year without further verification.
11.39. More information on Canada’s Industry oriented surveys and the more specific International
Transaction in Commercial services survey is provided online.
E.2.

Country experience: Japan

11.40.
Japan uses ITRS as a primary data source for compiling trade in services. However, as
some transactions cannot be captured through ITRS, various data sources have been implemented to
supplement ITRS. Other data sources include enterprise surveys for specific business based on the
Foreign Exchange and Foreign Trade Act, administrative source, and data provided by private
institutions. Uses of data sources other than ITRS for compilation of trade in services are summarized
in the Table below. More detail of Japan’s experience is given in the online version of the Compiler’s
Guide.

147

Table 11.11 Japan’s sources* of trade in service items
Source data
debit
ITS, ITRS, trade statistics, other
sources
ITS, ITRS
ITS, ITRS, trade statistics, other
sources
ITRS
ITRS
Border surveys, AS, ITRS, other
Household surveys, AS, ITRS,
sources
other sources
ITRS
ITRS, survey ordered by written
notification
ITRS, ITS
ITRS, ITS
ITRS
ITRS
ITRS, survey on freight
ITRS, survey on freight
insurance, other sources
insurance, trade statistics, other
sources
ITRS
ITRS
ITRS
ITRS
ITRS
ITRS
ITRS
ITRS
ITRS, AS, ITS
ITRS, AS, ITS
credit
ITS, ITRS

Sea transport
Air transport
Other transport
Travel
Manufacturing services
Maintenance and repair services
Construction
Insurance and pension services
Financial services
Charges for use of intellectual property
Telecom., computer & info. services
Other business services
Government goods and services n.i.e.

* note: ITS = International transportation survey; AS = Administrative sources

E.3.

Country experience: Philippines

11.41.
The Bangko Sentral ng Pilipinas (BSP) is the lead agency in the compilation of statistics on
the international supply of services in the Philippines. The BSP is able to compile all the main
services accounts, mostly from the bank reports in the International Transactions Reporting System
(ITRS), supported by the company responses to the Cross Border Transactions Surveys (CBTS) and
administrative reports from other agencies and industry associations.
11.42.
The main challenge of the BSP in the compilation of statistics on the international supply of
services is anchored in maximizing the benefits of the ITRS and CBTS. On the one hand, the
challenge of the BSP in the use of ITRS, is addressing possible misclassification of transactions by the
banks. On the other hand, the BSP is confronted with the challenge of improving the generally low
response rate and the coverage of its CBTS to include other industries, to better capture relevant
international services transactions. The low response rate of surveys stems from the absence of a
specific provision in the charter of the BSP that would allow it to enforce compliance by non-financial
corporations to provide data or information.
11.43.
Reports submitted to the BSP by commercial, thrift and foreign banks under the ITRS
contain information on bank positions in foreign currency-denominated assets and liabilities and all
payments and receipts which bring about changes in the banks’ position. This information is being
utilized in the compilation of some of the services sub-items such as transport, travel, construction,
insurance and pension, financial (including FISIM), charges for the use of intellectual property, n.i.e.,
telecommunications, computer, and information services, other business services, personal, cultural
and recreational services.
11.44.
In addition to the monthly CBTS, the BSP in coordination with the Department of Trade
and Industry (DTI) is also conducting the annual Information Technology-Business Process
Outsourcing (IT-BPO) Services Survey that covers companies engaged in Business Process
Outsourcing (BPO) activities.

148

11.45.
The BSP also utilizes administrative-based data provided by regulatory and administrative
government agencies (as customs data) to compile some of the services sub-accounts (Manufacturing
services of the physical inputs owned by others, freight import and exports, Insurance etc.). The
Department of Foreign Affairs (DFA) data on annual budget of Philippine embassies/consular offices
abroad are used as approximations for payments on government goods and services n.i.e.
11.46.
For non-resident expenditures in the Philippines (travel receipts), data are derived from the
Visitors Sample Survey (VSS) conducted by the Department of Tourism (DOT), which provides
information on the average expenditure of foreign tourists and their average length of stay in the
Philippines. For tourist-related travel expenditures abroad by residents, data are sourced from the
ITRS and CBTS. Beginning with the 1999 report, travel receipts include non-resident overseas
Filipinos (OF) expenditures in the Philippines during home visits. Travel debits cover expenditures of
resident OFs in the host countries where they are deployed. Information on resident’s expenditures
abroad via credit card payments are obtained via reports of the Credit Cards Association of the
Philippines.

149

Part III

Data Compilation

Scope. Data compilation is understood as a set of statistical procedures performed on the collected
data to derive new information according to a given set of rules and resulting in intermediate data and
final statistical outputs. Data compilation includes, among other things, integrating data from different
sources, the use of weighting schemes, methods for imputing missing values or source data, statistical
adjustment, balancing/cross-checking techniques and relevant characteristics of the specific methods
applied. Part III includes an introduction and overview of data compilation within the modes of
services supply statistical framework (Chapter 12), followed by discussion of integration of data from
different sources (Chapter 13), compilation of resident/non-resident trade in services statistics
(Chapter 14), compilation of foreign affiliates statistics (Chapter 15), compilation of other indicators
for modes of supply (Chapter 16) and concluding with a description of estimation and modelling of
missing data, and forecasting or back-casting (Chapter 17).
Chapter 12
Introduction and overview of data compilation within the modes of services
supply statistical framework
12.1. Scope. This chapter provides an introduction and an overview of data compilation with the
modes of supply framework. It first covers the information needs of data users, then provides an
overview of using an integrated approach to data compilation, reviews the need for phased
implementation of suggested data compilation procedures, and finally provides an introduction to
using models and estimates in data compilation, which will be further covered in chapter 17.
A.
Data compilation within the modes of services supply statistical framework: issues to
consider
A.1.

Information needs

12.2. The information needs include the value of trade in services or sales/output of services by
foreign controlled foreign affiliates and other monetary statistics (foreign direct investment flows and
stocks, research and development expenditure in affiliates, income flows relating to foreign affiliates
active in services sectors etc.). There is also interest in non-monetary data, in particular when it comes
to assessing commitments made in trade in services agreements (e.g. number of mode 4 persons,
number of foreign affiliates established abroad in the context of mode 3 etc.).
12.3. To provide a better guidance on how to satisfy the information needs of data users, this Part
of the Guide contains three chapters specially focused on particular areas:
i.
Compilation of statistics on resident/non-resident trade in services is covered in
Chapter 14. This chapter complements the guidance provided in the BPM6 Compilation
Guide Chapter 12. It elaborates conceptual issues which are directly related to the data
compilation; describes challenges and good practices in the compilation of particular EBOPS
services categories in total and their detailing by trading partner, as well as their allocation to
modes of supply; and provides guidance on services transactions between related enterprises;
ii.
Compilation of FATS is covered in Chapter 15, which contains a general description
of compilation of foreign affiliate statistics, elaborates a set of FATS variables and the related
data compilation issues and describes relevant country experiences. The chapter has a special
focus on the data needs with respect to the international supply of services. It in particular
identifies the variables and breakdowns of interest for assessing mode 3 commercial presence,
such as the need to have more detailed statistics for services activities, that output is a
preferred measure to sales/turnover and that these variables need to separately identify
150

sales/output of services products for all activities as well as separately identify those that are
destined to consumers of the country of establishment of affiliates;
iii.
Compilation of other monetary and non-monetary) indicators, in particular for mode
2 and mode 4 (i.e., number of mode 2 persons travelling abroad and purchasing services and
number of 4 persons crossing borders and temporarily abroad in the context of services
contracts) is an important part of the information needs. Challenges and good practices
relevant to this topic are described in Chapter 16. In particular, Chapter 16 provides an
overview of data variables on the movements of natural persons under mode 4 of services
supply, describes the uses of various data sources in the compilation of the variables and
provides a comparison of different sources and guidance on the organization of the data
compilation process. The chapter also provides information on the possibilities offered by the
linking of trade and business statistics, and a list of services sectoral statistics that would be
useful for a full analysis of the international supply of services.
B.

Focusing on an integrated approach

12.4. Part III gives a special attention to the promotion of an integrated approach to compilation. As
described in Part II there are several groups of data sources which can be used in the compilation.
This raises the issue of how to properly integrate data from various sources. Chapter 13 provides
guidance on this topic by discussing issues and good practices in consolidation and merging of data
and exploring possibilities of compilation using data generated in other statistical domains. It also
proposes a number of different approaches, taking into account the suitability for economies and
considering differences in compilation systems.
12.5. Chapter 14 proposes to integrate data sources for compiling resident/non-resident trade in
services statistics. For example for travel, it is proposed to link with tourism statistics, in particular to
develop coherent sets of data. Such integration focuses in particular on the breakdown by purpose of
travel, by type of product consumed or for compiling the supplementary item tourism related services
in travel and passenger transport. In the case of data for modes of supply, such an approach would
enable compilers to arrive at more detailed estimates of mode 2 supply of services, in particular travel
by type of product consumed and linking to tourism statistics (i.e., tourism-related
consumption/expenditure as suggested in ITRS 2008 or MSITS 2010 chapter 14, section C), and
using customs data or merchant category codes from credit card data. The scope of different statistical
domains has to be taken into account when integrating data from different sources (e.g., chapter 14,
section B, iv).
12.6. Chapter 15, while focusing mainly on compilation issues related to FATS, also gives a special
attention to the possibility to use or re-use existing data by linking with micro-level information; e.g.,
through business registers (with a common identifier) for the compilation of these statistics. It should
also be considered to use data from the Structural Business Statistics (SBS) for further development
on international services data by mode of supply accordingly for mode 1 and mode 4 as well as to
combine the information on FATS to gather more insight details for mode 3.
C.

A phased implementation of the recommendations on data compilation

12.7. This Guide advises a phased implementation approach for compilation of data within the
statistical framework for measuring the international supply of services. The proposed approach can,
in particular for modes of supply, serve as a reference for producing approximations of certain
statistics. Part III therefore identifies the main conceptual and statistical challenges in measuring the
international supply of services, in particular for the modes dimension, also beyond balance of
payments services and foreign affiliates statistics. This part evaluates to what extent existing statistics
can match the information needs of trade negotiators as well the needs of economic analysts, in
particular in the context of the GATS; it also identifies what would need to be developed in data
151

compilation systems to respond to information needs when it comes to other indicators useful for
assessing the international supply of services.
12.8. Data on trade in services have been compiled for a number of years in the context of the
BPM5, but compiling more detailed statistics by type of service and by partner remains a challenge
for many compilers. Not much information is currently compiled on activities of foreign affiliates, as
FATS is a relatively new statistical domain. Moreover, modes of supply are not always easy to
observe and consequently many data compilers have not yet compiled these data. On the one hand,
the concept of modes of supply is not yet embedded in the existing national compilation systems. On
the other hand, a complete new compilation of statistics on international supply of services by mode
will necessarily cause a higher burden and additional costs for compiling economies, which is why a
phased approach should be adopted based on what is proposed in MSITS 2010. Statisticians, both
national and international, also need to have a more profound knowledge of how different services
sectors operate in various economies. Compilers can choose between different approaches which are
elaborated in Part III, in particular chapter 14. The respective size of an economy, the structure of the
service sector or the importance of specific services could be indicators that could be used for
focusing on one approach or the other, or adopting a combination of several approaches.
12.9. A simplified allocation and direct compilation of statistics on the international supply of
services by modes. For the development of modes of supply information, a simplified allocation of
FATS and balance of payments data to modes of supply is proposed in MSITS 2010.163 The so-called
mechanical allocation of services categories either to one dominant mode or the indication of a
distribution to several modes in the BoP, as well the simple allocation of FATS to mode 3, is
considered as a very first starting point. This approach makes use of existing services data within the
BoP/FATS framework to compile or to estimate statistics by mode of supply and is further explained
in Chapter 14, section C and chapter 15.
12.10. But this simplified allocation is a reasonable very first step due to relative low costs and a
minimum burden for compilers. Chapter 14 then further elaborates on the development of estimates
on a broader basis taking economic, political-economic and socio-economic issues into account.
Chapter 14 then focuses on the actual compilation of data, in particular by suggesting the
identification of modes in trade in services surveys (mainly mode 1 and 4, but also mode 2 under
certain conditions). Focussing on different user interests, specific sector studies would already serve
several purposes e.g. policy makers could also be interested in specific studies such as for services
incidental to agriculture or environment services. This approach could be also combined for further
elaboration and adjustments of the simplified allocation.
12.11. Finally, as outlined above, beyond the monetary value of the international supply of services,
users also need additional monetary and non-monetary data, which are mainly described in chapters
16 and, for mode 3, chapter 15 (e.g., employment, number of enterprises). Although these additional
needs are considered as longer term objectives for compilation, when developing the system,
compilers should keep in mind these additional needs in order to be able to factor them in easily for
future stages of development for a complete information system to serve user needs.
D.

Use of models and estimates

12.12. This Guide recognizes that available data sources have their limitations and, therefore,
compilation has to include the use of models and estimates because part of the information cannot be
collected or can be gathered only at an unsustainable cost, the available sources cannot provide the
required coverage, detail, frequency and/or timeliness foreseen by the international standards, and the

163

MSITS 2010, Chapter V and Table V.2, as well as ch apter 1, paragraph 1.5.

152

various sources to be combined in the collection system are partially overlapping. Chapter 17 is
specifically devoted to these issues and covers such topics as imputation for filling data gaps and for
data editing purposes, forecasting and compensating for lack of timeliness of data sources, , backcasting and revising time series, allocating all services transactions to the relevant individual EBOPS
services categories and appropriate trading partners, and model-based estimates.

153

Chapter 13 Integration of data from different sources
13.1. Scope. As described in Part II, several sources can be used in the compilation of data within
the modes of services supply statistical framework. In practice, information from various sources will
often have to be combined to obtain the required level of detail and quality, while at the same time,
reducing excessive burden for respondents. This chapter reviews the potential challenges compilers
face when integrating data sources that cover the same activities or the same entities and provides
guidelines on how to resolve these issues. More specifically, this chapter presents a summary of good
practices and defines the relevant terminology (section A), and describes the integrated business
survey program (section B), integration, consolidation and merging of data from different sources
(section C). A country experience is presented in section D.
A.

Summary of good practices

13.2. Integrating data from different sources is advised as the principle way to ensure the
production of more detailed and more comprehensive statistics as well as to reduce the burden on
survey respondents. Compilers are encouraged to determine which data sources are most appropriate
on a case-by-case basis carefully considering the strengths and weaknesses of each data source. More
specifically, compilers should be aware of the statistical units used in each source, the entities
covered, the services categories identified, the variables compiled (e.g., value of services
exports/imports by EBOPS categories for compilation of trade in services between residents and nonresidents, output or turnover/sales and employment for compilation of FATS etc.), the availability of
geographic breakdown, the time period of reference, the presence of thresholds, and the survey
frequency. Ideally, compilers should identify the source(s) in which they have more confidence, to be
used as a benchmark for the other data sources.
13.3. To optimally exploit the advantages of data integration it is advised that the potential of using
linked data is considered from an early stage onwards and that decisions on survey design and
sampling frameworks, for example, are made while keeping in mind that the data will be linked later
on. This forward thinking requires ‘positive coordination’ across surveys, that is compilers should
ensure that the entities covered by different surveys and registers have sufficient overlap in order to
avoid large shares of non-matched records (which make data integration more difficult and may create
statistical biases). It is recognized that such coordination may be seen as a long-term strategy by
countries where the capacity to compile statistics on the international supply of services needs
significant strengthening.
13.4. This Guide encourages national statistical agencies to adopt an integrated business survey
programme, as outlined in the Guidelines on Integrated Economic Statistics.164 Specifically, under
these guidelines, it is recommended that the collection of statistical information move from a
stovepipe approach, in which each statistical program collects information for its own purposes,
towards an integrated approach, which aims to integrate survey design and implementation across all
statistical programme areas. This recommendation is particularly relevant in the context of statistics
on the international supply of services, given that information needs cut across various statistical
domains. Such integration of the data collection procedures will reduce the burden on respondents. It
will also create a common standardized statistical framework for presenting more coherent statistics
for the entire economy (covering business statistics, short term statistics, national accounts and
international statistics) for evidence-based policy making.

164

Guidelines on Integrated Economic Statistics http://unstats.un.org/unsd/nationalaccount/docs/IESGuidelines-e.pdf.

154

13.5. It is a good practice that SBR is used to identify the common statistical unit and to provide the
sampling frameworks. In addition it is advised that the ITRS and the trade register (which includes
merchandise trade statistics) are linked to the SBR (with the enterprise as common statistical unit).
This is particularly relevant for those services categories where the link between trade in goods and
trade in services is strong, such as manufacturing services.
13.6. Compilers should take note that for purposes of this chapter the term “data integration” is
used to refer to bringing together information from two or more sources of data, with the object of
better understanding and presenting the nature of a transaction that would otherwise not be possible
based on any one source alone. “Data consolidation” is understood as the summation of data from
multiple sources, in which the sources generally provide information on non-overlapping parts of the
total. “Merging data” refers to the process of combining data (especially data covering the same
activities or the same entities) from different sources via common denominators, such as a unique
identifier in the BR, when incomplete information is available from individual sources. Merging data
is the final step in the process and creates a unified view of an entity or activity. Before merging data,
compilers must first be aware of the definitions and methodologies used across the various data
sources to ensure internal consistency of the merged data and to prevent duplication or overlap of
records.
B.

Integrated business survey programme: an overview

13.7. The idea of developing and achieving an integrated system of business and international
statistics has a long history among national and international governing bodies of economic statistics.
The motivating source of this work has been the policy and statistical benefits that can be obtained
from an integrated statistical system for national and international coordinated economic policy
planning in an interconnected global economy.
13.8. The benefits to data users, producers and providers, include, but are not limited to, the
following:
i.

A common standardized information management framework that governs the statistical
production process over time and across countries;
ii. Statistical production processes that help reducing response burden and costs while
improving transparency for data users and providers;
iii. Collaboration in the development and application of common methods and IT tools for
data processing, documentation and exchange through the adoption of standardized
information management model;165
iv. Addressing the demand for evidence-based mutually reinforcing stories and descriptions
of the sequence of events that are obtained from coherent statistics, covering business
statistics, short term statistics, national accounts and international statistics.
13.9. The starting point for integrated economic statistics is the use of the internationally accepted
standard for macroeconomic accounts, the SNA (preferably its latest edition, the 2008 SNA) as the
conceptual organizing framework. As a result of years of work in harmonizing macroeconomic
statistical standards, the use of the 2008 SNA as the conceptual framework provides consistency and
coherence with other internationally accepted standards and international recommendations, such as

165

Concept note for the earning centre on explaining the integrated business statistics programme for st atistical
information management, presented at the Forty-fift h session of the Statistical Commission.
http://unstats.un.org/unsd/statcom/statcom_2014/seminars/LC_Integrated_Business/Concept%20Note%2012%2
0Feb%202014.pdf.

155

the BPM6 and other specialized manuals, such as the Manual on Statistics of International Trade in
Services 2010.166
13.10. Standardized surveys across statistical programmes. The most effective means of ensuring
integration and consistency is through a holistic redesign of data collections (surveys and
administrative data) that minimizes inconsistencies through the use of common standards and
methods, integrated survey design and a central business register. It is a good practice that the
collection of statistical information move from a stovepipe approach towards an integrated approach.
In a stovepipe model, where each programme collects information on its own and for its own
purposes, an entity surveyed may be asked the same questions with different definitions. A stovepipe
approach, therefore, not only increases survey burden, but is also likely to produce statistics that are
difficult to align. In contrast, integrated questionnaire design and integrated surveys help to resolve
these data-collection issues. 167 The integrated approach is particularly relevant for statistics on the
international supply of services, which are compiled with information from various statistical
domains.
13.11. Under an integrated approach, all economic data collections gradually change their objectives
and statistical designs and enhance the coherence of statistical outputs. When designing a collection
for various sets of economic statistics, compilers will need to think beyond their current work to how
those sets will integrate with other statistical outputs. Likewise, questionnaires have to apply concepts
and definitions that are consistent with those used in other surveys and administrative collections.168
13.12. A national statistics act may provide a national statistical office with the authority to approve
the design of all statistical surveys operated by government agencies.169 One practical approach to
achieving integrated surveys is a reorganization of the statistical agency towards a more functional
structure. A separate division with responsibility for developing and maintaining classifications and
standards can lead a programme to apply consistent definitions and classifications to all statistical
outputs. Creating a division that has responsibility for developing SBR to be used for the production
of all economic statistics is an effective way to ensure that units are selected and classified
consistently across all statistical outputs. Establishing an independent statistical methods division can
help to ensure that all surveys are designed in a coherent way.170
13.13. It is important to note here that the subject matter expert or compiler of statistics on the
international supply of services may not always have full control over the design of such integrated
surveys, if the design is managed by an independent division. It is important for compilers to
recognize that success in achieving integration across many programmes requires extensive
collaboration and communication across the organization, support from senior leaders in the
organization, and adaptability and flexibility.171
C.

Integration, consolidation, and merging of data

13.14. The compilation of data within the statistical framework for describing the international
supply of services requires the use of a wide range of data sources, as some sources are

166

Guidelines on Integrated Economic Statistics 2013, p. 17.
http://unstats.un.org/unsd/nationalaccount/docs/IES-Guidelines-e.pdf.
167
Ibid. pp. 49-50.
168
Ibid.
169
Ibid. pp. 28-29.
170
Ibid. p. 39.
171
For a country example of the Integrated Business Statistics program at Statistics Canada, please see:
http://unstats.un.org/unsd/nationalaccount/workshops/2014/St_Lucia/IBSP.ppt. Samples of integrated surveys
from Statistics Canada are available on the UNSD we bsite.

156

complementary and their combined use can result in the production of more detailed and more
comprehensive statistics.
13.15. Integration of different sources and re-using surveys for several purposes also has the
potential to reduce the burden for compilers, in particular in the context of compiling modes of supply
data. Furthermore, linking data on a micro-level could allow for broader types of analysis of the
resulting data. The potential of the approach of using information from other statistical frameworks
therefore deserves further exploration by the compiler. For example, in the context of the compilation
of inward FATS, there is the possibility of integrating information from FDI surveys and domestic
enterprise surveys. Other examples include using information from sources on migration, tourism,
household expenditures, population, or taxes (see chapter 15). This information could be used in the
compilation of resident/non-resident trade in services data (see Chapter 14) or to compile quantitative
indicators on mode 2 or mode 4 (see chapter 16).
13.16. Consolidation and merging of data sources provides multiple advantages for the compilers.
First, it improves coverage and the diversity of information. Second, it usually reduces the resources
required to collect the statistics. Third, it leads to higher quality of resulting statistics. However,
compilers must be aware of the risk that some of the data sets resulting from integration of various
data sources may be internally inconsistent.
13.17. Consolidation is rather a straightforward summation of data sources from multiple sources in
which the sources provide information on non-overlapping parts of the total. Meanwhile, merging
data sources is usually not a simple task, as compilers must find common denominators between data
from multiple sources that cover the same entity or activity. In this process the role of the business
register is of vital importance in linking microdata.
13.18. Compilers should take into account all relevant dimensions of data sources before merging;
for example, the content of each source could be more or less compatible with a certain services
category definition. For such reasons, compilers should have an in-depth knowledge of the
methodologies and definitions used by the other data sources to ensure consistency and comparability
of the resulting merged data. Various dimensions of the different sources must be analysed by the
compilers during merging, most notably: entities covered, services categories or activities identified,
availability of geographic breakdown, and period of reference.
13.19. Entities covered. Sources could collect statistics at different structural levels of the entity.
Some sources may target the legal entity, while others could survey entities responsible for the
production. Companies could merge, be acquired, disappear, or simply modify their organisational
structure so that entities might end up being different from source to source if they have been
surveyed at different points in their structural composition. There could also be differences between
how the business register defines the structural organization of an entity and how that entity defines
its own structural organization. This last point might result in the statistical compiler unwittingly
incorrectly comparing entities from different sources.
13.20. It is a good practice that compilers analyse the coverage, possible overlap, and potentially
differing definitions of variables across different sources available, and make determinations of which
data sources are most appropriate to use on a case-by-case basis.
13.21. Services categories. Compilers should compare the services categories available from each
source to be merged. Details and aggregations provided could be different from source to source. In
the absence of clear definitions and guidance, respondents may interpret survey questions incorrectly
and include invalid transactions or exclude pertinent transactions.

157

13.22. Geographic breakdown. Data sources could have different levels of geographical breakdown,
or even none at all. For example, the country of transaction identified in the International Transaction
Reporting System (ITRS) could be different from the country that actually purchased or sold the
service.
13.23. Period of reference. Values reported on a monthly basis from one source could differ from
the annual value reported by another source (e.g., where there is an annual reconciliation or balancing
process). Moreover, difficulties may arise when integrating data from administrative sources on a
fiscal year basis with information collected on a calendar year basis. Similarly, there may be
challenges in bringing together data reported on a monthly or quarterly basis with data reported on a
through-the-year basis in another source.
13.24. MSITS 2010 recommends the use of the accrual basis for determining the time of recording of
transactions. The accrual basis provides the most comprehensive information because all flows are
recorded (including nonmonetary transactions, imputed transactions and other flows). The change of
economic ownership is central in determining the time of recording on an accrual basis.172
13.25. Other factors, such as threshold for certain sources or the frequency of the survey, may also
affect the comparison between sources. If a source is available only once every two years, compilers
should develop a strategy to preserve the comparability for the year that source is not available. All
these possible differences between sources may complicate the comparability and integration of data
sources, not only at a global level but also at a micro level.
13.26. Possible approaches and solutions. There is no quick fix to resolve problems when merging
data. Compilers must have in-depth knowledge of the data with which they work and need to
recognize the strengths and weaknesses of each source. Data sources are rarely equivalent; one source
may provide reliable information for some transactions but be weak for others. Ideally, if compilers
could identify a source in which they have more confidence, such a source could then serve as a
benchmark against which the other sources will be compared.
13.27. Linking various data sources. It is vital that the compilers avoid approaching businesses
multiple times with different surveys covering the same or similar information. Above all, statistical
surveys should not request information that the business has already supplied in another data
gathering mechanism. The central statistical business register should be linked to the trade register to
enable the analysis of the effects of the international supply of services on production, employment
and enterprise performance.173 The enterprise is the suggested statistical unit to be linked between
trade and business statistics; thus, data collected and registered at the level of the declaring unit of
trade operators can be aggregated to the level of the whole enterprise via characteristics available in
the BR. The linking of trade and SBR allows generating relevant information on the structure of the
international supply of services without collecting additional data from businesses.
D.

Country experience
D.1.

Canada

13.28. Three different sources are used in Canada to generate most trade in services statistics
(excluding travel and transport). First, the annual survey on trade in services (International

172

BPM6 Compilation Guide, paragraph 1.21.

173

See also chapter 16.

158

Transactions in Commercial Services – BP21S) is sent to Canadian enterprises of varying size and
economic sector. 174 The second data source is the set of approximately fifteen annual industry
surveys covering a large sample of establishments in service industries. The third source is
administrative data on non-arm’s length annual transactions between residents and non-residents.
13.29. Statistics Canada integrates these three sources and is thereby able to improve coverage, make
better use of limited resources, and reduce the burden on respondents. The key to integrating these
sources is that each entity covered by the three sources is linked to the Canadian Business Register
(CBR). The CBR covers and contains information on ownership, industrial activity, organizational
structure, and the business identification number175 of businesses engaged in the production of goods
and services in Canada.
13.30. The quality of the CBR directly influences the compiler’s ability to integrate the different
sources. A team at Statistics Canada is constantly working to ensure the CBR reflects the structure of
Canadian economy as best as possible.
13.31. The three sources cited above are not considered to be equal in terms of quality and details
provided. Data on the industry surveys are indirectly measured in a process by which respondents
indicate the percentage of their total revenues related to total trade in services. No breakdown is
available for transactions conducted between related partners and transactions between unrelated
entities. Furthermore, the geographical breakdown is only available for total trade in goods and
services. The administrative records cited only present trade with affiliated parties and aggregated
groupings of services.
13.32. Both the industry surveys and the administrative data have limited details in term of EBOPS
categories. However, the annual survey on trade in services (BP21S) is very detailed, with 32
categories of services collected, broken down by partner country and by affiliation.
13.33. For the majority of the responding entities, data are available from only one of the three
sources cited, which case, reported values are kept by default. When there are data available on an
entity from more than one source, the data from the BP21S survey are used if those values are larger,
equal to or even slightly lower than the other sources. Other situations are reviewed on a case-by-case
basis. Other analysis is also done to insure the validity of the reported transactions, for example,
compilers compare current reported transactions with data from the previous year.

174

To supplement the BP-21S, more than 200 entities f rom the insurance sector are also surveyed with two
surveys of Canadian incorporated insurance companies and Canadian branches of foreign insurance companies.
Both surveys collect financial transactions as well as service transactions.
175

This identification number is assigned at the time of registration with the Canada Revenue Agency. For more
information about the Statistics Canada Business Register, please consult “Statistics Canada Business Register –
A Brief Guide, http://www23.statcan.gc.ca/imdb-bmdi/document/1105_D2_T1_V3-eng.pdf.

159

Chapter 14 Compilation of resident/non-resident trade in services statistics
14.1. Scope. This chapter discusses the compilation of resident and non-resident trade in services
statistics. It is important to stress that this chapter does not aim to be exhaustive, but rather that it aims
to complement the existing guidance provided in BPM6 CG on specific aspects, and focuses on the
needs outlined in MSITS2010. Compilers are also strongly suggested to consult the online version of
this Compilation Guide, where more compilation guidance will be made available, e.g., on certain
services items and complementary groupings, as well as the EBOPS-CPC conversion table that may
facilitate the classification of certain transactions in the EBOPS classification.
14.2. After the summary of good practices (section A), section B describes the compilation of the
EBOPS 2010 service categories by bundling them into several groups that are often similar in the
nature of their compilation and use of underlying source data:
i.
ii.
iii.
iv.
v.
vi.

Goods-related services (including manufacturing services on physical inputs owned
by others, maintenance and repair services, as well as freight transport and insurance);
Passenger transport and travel;
Construction;
Insurance, pension and financial services;
Services related to intellectual property products (IPP), telecommunication,
computer, and information services, other business services, as well as personal,
cultural, and recreational services);
Government goods and services n.i.e.

14.3. Note that in the case of transport, the treatment is split between freight transport and
insurance (goods-related) and passenger transport (travel-related). For each of these subsections, the
compilation of providing partner country detail is separately discussed. Such a breakdown is
recommended in the MSITS 2010, at least at the level of the 12 major components of the BPM6
classification of services, and, where possible, at the more detailed EBOPS 2010 level.176
14.4. Subsequently, section C discusses the compilation of trade in services by modes of supply
insofar as it is related to resident/non-resident transactions. Section D elaborates the statistical
treatment relevant to services transactions between related enterprises.
A. Summary of good practices
14.5. When compiling resident/non-resident trade in services statistics, it is good practice to follow
the BPM6 recommendations on the principles of recording residence, valuation, time of recording,
unit of account and currency conversion. The highest priority should be given to the compilation of
data at the level of BPM6 main EBOPS categories, followed by the introduction of the EBOPS 2010
level of detail, the compilation of related supplementary items and the compilation of complementary
groupings. Also, compilers are also strongly advised to consult the online version of this compilation
guide where more compilation guidance will be made available on e.g. certain services items and
complementary groupings, as well as the EBOPS-CPC conversion table that may facilitate the
classification of certain transactions in the EBOPS classification.
14.6. When compiling goods-related services statistics, this Guide advises to note that enterprise
surveys generally represent the most efficient method to collect the required information. ITRS and
customs declarations, as well as other data sources, may also be used although these are generally not

176

MSITS 2010, paragraphs 3.57-3.58.

160

a sufficient source of data and may need to be supplemented with enterprise survey data. For
manufacturing services on inputs owned by others, compilers should be aware that the difference
between the value of the goods before processing and the value of the goods after processing may
differ from the value of the manufacturing fee. Compilers should also evaluate if and how received
data has to be adjusted, especially in the absence of complete information or misreporting. Compilers
also need to carefully distinguish between manufacturing services and manufacturing on own account,
using the economic ownership of the inputs as the distinguishing criterion. Also a clear difference
needs to be made between transactions related to manufacturing services on inputs owned by others,
and those related to merchanting, based on whether or not the nature of the good changes.
14.7. For freight transport and freight insurance services, it is not always possible to measure the
value of services directly. However, the information may be derived from the CIF/FOB adjustments
that are made to the merchandise trade statistics for Balance of Payment purposes. Compilers should
note however that when calculating the freight transport and insurance costs for BOP purposes, the
statistical value does not necessarily reflect the sum of the actual invoice values, but instead follows
the principle adopted in the accounting framework.
14.8. Regarding passenger transport and travel, compilers are advised that transport should be
recorded on a gross basis. This means that the value of passenger transport includes fares and other
expenditures related to the carriages of passenger. When passengers fares are a component of package
tour payments, the compiler should, separate passenger transport and other travel components.
14.9. The compilation of travel item is closely related to the compilation of tourism statistics. Thus,
it is useful for compilers to understand the conceptual framework of tourism statistics. The BPM6 and
EBOPS 2010 recommendations should be followed and compilers are advised to provide further
breakdowns of travel, in order to better assess the scope of travel activities, gauge their possible
impact in terms of economic activity, and ensure consistency between travel and other related
statistics, such as the tourism statistics, tourism satellite account or supply and use table.
14.10. When compiling statistics on construction, compilers should take care to value transactions
on a gross basis, i.e. inclusive of all goods and services used as inputs to the work (whether
subcontract or not), other costs of production and the operating surplus that accrues to the owners of
the construction enterprise. It is particularly important to precisely identify the residence of the
enterprise realizing the construction work in order to differentiate between exports of services and
direct investment operations, using criteria such as the length of a project. Given the complexity
involved in cross-border construction projects and the level of detail required means that surveys are
the preferred method for collecting data on construction. Such surveys could possibly also be used to
collect information on FDI and other resident-non-resident transactions.
14.11. Regarding insurance, pension and financial services, it should be noted that some services
are provided without explicit fees, which means that their compilation needs a certain degree of
estimation and imputation. It is good practice that insurance and pension services are based on the
earned premiums, which are consistent with the accrual recording. To compile the exports of
insurance services, compilers can generally obtain comprehensive data from surveying resident
insurance enterprises. If comprehensive surveys are not feasible, compilers may be able to obtain
information directly from domestic insurance corporations, which may also be used for estimating
imports of insurance services.
14.12. In order to avoid the volatility and negative figures possibly generated by catastrophic events,
an adjustment in claims due is required to reflect a more long-term view the functioning of the
insurance sector. The compiler is invited to analyse the three possible adjustment methods for
smoothing the amounts of claims by policyholders on insurance corporations, i.e. the expectation
approach, the accounting approach and the sum of cost plus “normal profit” approach and use the
most efficient depending on national circumstances.
161

14.13. For financial services, transactions without explicit fees are included in FISIM. FISIM is
produced only by certain financial corporations and only on the loan and deposit instruments on their
balance sheets. Compilers need to take special care to properly calculate the reference rate. As
recommended by the Advisory Expert Group (AEG) on National Accounts, the calculation of the
reference rate should be determined according to national circumstances, using preferably either a
reference rate based on a single observable exogenous rate for a specific instrument, such as interbank
lending rates; a reference rate based on a weighted average of observable exogenous rates of
maturities with different terms (weighted by the stock of loans and deposits in each maturity) or a
weighted average of the endogenous interest rates on loans and deposits. The most comprehensive
source data for exports and imports of FISIM come from (i) resident financial corporations (surveys
or administrative data collections or financial supervision authorities), which can identify the deposits
and loans with non-residents, as well as (ii) surveys of selected nonfinancial corporations,
households, and NPISH, which could provide data on residents’ accounts with financial corporations
abroad to support compilation of FISIM import
14.14. Services related to intellectual property products may be particularly difficult to measure,
because the distinction between the categories from a reporter’s perspective may not always be clear,
and because intra-firm transactions, which are frequent in these services categories, may be affected
by transfer pricing phenomena or more general tax planning issues. When recording transactions on
the use of intellectual property n.i.e., compilers should note that franchise and trademark licensing
fees related to charges for using non-produced assets and should hence, theoretically according to the
SNA, not be recorded. However, often, the payments are bundled with additional service items that
make it difficult to disentangle the pure payments for the use of the underlying brand and, as such,
both the 2008SNA and BPM6 recommend that such payments are recorded as payments for services.
For payments related to licences for the use of outcomes of R&D, care should be taken to differentiate
between payments for entire originals. Similar differences exist for computer services, where
compilers should distinguish between licences to reproduce (of more than one year) and software
originals. In contrast, compilers should note that R&D services primarily relate to ‘new’ research and
development, such as newly produced customized software, or transactions in ‘originals’, where
ownership and the concomitant rights are transferred to the purchaser. Care is needed in this context
to differentiate this from the licences to reproduce identified above. Compilers are strongly suggested
to consider using the Frascati 177 based survey approach provides perhaps the best mechanism to
improve their measurement of trade in R&D services.
14.15. For the compilation of data on resident/non-resident transactions in education or health
services, primarily the collection of data is often problematic. Compilers are suggested to consider
that sources of information from specialized entities may be necessary (e.g., the Ministry of Education
and the Ministry of Health) and also to explore other types of sources, including higher education
institutions, surveys on health insurance companies in combination with travel surveys (e.g. with
reference to the purpose of stay); data available at embassies and consulates; or administrative data
from health and social insurance.
14.16. Government goods and services n.i.e. includes only those services related to government
functions that cannot be classified to another specific service category. The majority of government
services transactions are most commonly compiled using administrative records. Data on government
expenditure abroad should also be available from an ITRS, although it may be more difficult to
capture expenditures by foreign governments and international institutions located in the compiling

177

OECD Frascati Manual: Proposed Standard Practice f or Surveys on Research and Experimental
Development, 6th edition. Note that at the time of writing the Frascati Manual is being revised, in part to
provide better alignment with the National Accounts.

162

economy using an ITRS, in which case an enterprise survey of nonresident bank accounts or a survey
of foreign embassies and international institutions could be used.
14.17. When allocating resident/non-resident trade by mode of supply, compilers may start with a
simplified allocation, following table V.2 in MSITS 2010 and applying a three-step procedure of
allocate, evaluate and refine. However, compilers should treat such an allocation only as a first rough
approximation of resident/non-resident services transactions by mode of supply as this technique has
important limitations. This Guide strongly encourages compilers to gather additional information on
modes of supply that may be collected for all services sectors using (added questions to) generic
business surveys. Alternatively, compilers may choose to obtain information only for a selection of
specific service sector(s) of interest to the compiling economy using focused sources of information,
possibly using a sector specific survey that also integrates various other statistical domains (such as
FATS, SBS, innovation or others). It may be useful to consider collaboration with other agencies and
institutions that have a special interest (and/or expertise) in obtaining modes of supply data. In these
circumstances, it is important the agency in charge of the collection and compilation of official trade
in services statistics (generally the NSO or central bank) works in coordination with such specialised
agencies so that the international recommendations for the collection and compilation of trade in
services statistics are followed and to ensure that the information that is collected can be used in a
broader context. It is suggested that the compiler investigates how to collect/compile more details for
those items where mode 4 or mode 2 (both involving natural persons) is deemed to be important for
the compiling economy.
14.18. Finally, compilers are advised to develop appropriate procedures to separately identify service
transactions between related (affiliated) enterprises, as recommended by MSITS2010. It is a good
practice to carefully examine the valuation of services transactions between related enterprises, since
the recorded transactions could be under or overestimated and misrepresent the real flows of the
economy. Compilers should be concerned that intra-firm trade might not reflect the real trade, as
measured by market prices, to benefit from fiscal and taxation regulations, since transfer pricing is
used for the valuation of services transactions.
B.

Compilation of individual service categories

B.1.

Goods-related services

B.1.a. Introduction
14.19. This sub-section deals with goods-related services, which include manufacturing services on
physical inputs owned by others; maintenance and repair services n.i.e.; and freight transport and
insurance (as sub-component of transport). These service categories are strongly related to flows of
imported and exported goods. However, the value of services cannot necessarily be directly derived
from the gross value of merchandise trade statistics. Additional information, from surveys, for
example, is often needed to measure the exact value of the services provided.
14.20. Manufacturing services on physical inputs owned by others include activities such as
processing,assembly, labelling and packing that are undertaken by enterprises that do not own the
goods (see also paragraphs 3.66-3.77 of MSITS 2010). Such services may also be referred to as
contract manufacturing, toll manufacturing, or toll services and are also described in BPM6178 and the
BPM6 CG.179. In the case of manufacturing services imports (these services can, for example, be

178
179

BPM6, para. 10.62-10.71 and boxes 10.1. and 10.2 a nd table 10.2.
BPM6 Compilation Guide, para. 12.5-12.25 and para. 3.15-3.17.

163

rendered under a special customs procedure - outward processing of goods), the principal180 (goods
owner) has goods it owns processed by another unit (processor) which is resident of another economy.
The principal pays a fee to the processor for the services provided. In the case of manufacturing
services exports (these services can, for example, be rendered under a special customs procedure inward processing of goods), a processor transforms goods belonging to the principal who is resident
of another economy. In return for a processing fee, the processor transforms the goods using its own
labor and capital. Over the course of the transformation process, the economic ownership of the goods
remains with the principal. The manufacturing service equals the processing fee; i.e., the value of the
contract between the owner of the goods and the manufacturer. Gross values of goods associated with
manufacturing services should be identified as EBOPS supplementary items in economies where such
an activity is significant.181
14.21. Maintenance and repair services n.i.e. includes maintenance and repair work by residents on
goods that are owned by non-residents (and vice versa) The value recorded for maintenance and
repairs is the value of the repair work done, not the gross value of the goods before and after repairs
(see also paragraph 3.78-3.79 of MSITS 2010).182
14.22. Freight transport and related freight insurance services are part of EBOPS categories
transport and insurance services, respectively (see 3.97-3.103 in MSITS for freight transport and
3.178-3.180 for freight insurance). It refers to freight charges by residents of the compiling economy
with non-residents on goods with change of ownership, as well as on freight where there is no change
of ownership.
B.1.b. Compiling statistics for goods-related services
14.23. It is suggested that compilers consider that enterprise surveys generally represent the most
efficient method to collect information on goods-related services. The surveys could include questions
on the value of the service fee, as well as the value of the goods sent and received for processing or
for repair. This information can be used to adjust the goods account to measure merchandise trade on
a change of ownership basis.

180
In this chapter, the following terminology is applied: The principal is a unit that enters in a contractual
relationship with another unit (here called processor) to carry out some part (or all) of the producti on process;
the processor is a unit that carries out a specific production process based on a contractual relationship with a
principal. The activities performed by the processo r are denominated 'on a fee or contract basis'. Processing is a
contractual agreement according to which the principal requires the processor to carry out a specific production
process. The term 'sub-contracting' is sometimes used as well. In this context, the production process also
includes supporting activities. Manufacturing services on physical inputs owned by others can be performed on
goods leaving the county of the principle and entering the country of the processor under various arra ngements
depending on the customs laws of the involved countries including (but not limited to) specific customs
procedures called outward and inward processing. Outward processing customs procedures means the Customs
procedure under which goods which are in free circulation in a Customs territory may be temporarily exported
for manufacturing, processing or repair abroad and then reimported with total or partial exemption from import
duties and taxes. Inward processing customs procedures means the Customs procedures un der which certain
goods can be brought into a Customs territory conditionally relieved from payment of import duties and taxes,
on the basis that such goods are intended for manufacturing, processing or repair and subsequent exportation.
See IMTS 2010, paragraphs 1.19 – 1.21 and annex B, para. B.12-B.1
181
BPM6, paragraph 12.5.
182
No distinction is made between those repairs that maintain the item in working order and are consider ed as
intermediate consumption, as opposed to those that extend the efficiency or capacity of the good or ex tend its
life and thus are included in capital formation.

164

14.24. ITRS and customs declarations may also be used for the compilation of goods-related
services, although these sources are generally not sufficient and may need to be supplemented with
enterprise survey data. When using ITRS data for information on fees for manufacturing services or
maintenance and repair services, the compiler should ensure that such amounts do not include
payments for other goods and services. Administrative sources, including tax records, can also
provide useful information (see BPM6CG 12.20-12.25 for manufacturing services and 12.35-12.37
(including tables 12.1 and 12.2 for freight transport).
14.25. Compilers may consider combining data from customs declarations with the results of
enterprise surveys, when both datasets are available. While such merged data would not necessarily
provide compilers with advantages in developing aggregate measures of goods-related services, such
data could be very useful, in combination with appropriate screening questions on enterprise surveys,
for determining partner country attribution, and for determining the detailed product composition of
exports and imports for those firms that indicate they either receive or send goods abroad for
processing or repair.
14.26. Compilers may also consider comparing the trend in the value of goods under inward and
outward customs procedures (and the use of the nature of transactions codes) with the trend in the
magnitude of manufacturing services over time. While the actual monetary amounts will be different,
compilers could expect the growth trend between the series to be similar, which could provide a
useful quality check for compiling statistics on the international trade of manufacturing services.
B.1.b.1 Compilation of manufacturing services
14.27. Compilers should be aware that the difference between the value of the goods before
processing and the value of the goods after processing may differ from the value of the manufacturing
fee for various reasons, including the sale of processed goods in the economy of the processor or in
third countries, holding gains or losses, the inclusion of brand names after processing, and others (see
BPM6 CG 12.14).
14.28. The compilation of data on manufacturing services on inputs owned by others may be
facilitated by following the change of ownership principle in general merchandise trade statistics.
However, the recording of the movements of goods across borders and statistical surveys on trade in
services are usually independent and therefore not mutually consistent.
14.29. As to export and import of manufacturing services, Box 14.1 provides an example of a
possible data adjustment in the compilation of resident/non-resident trade in services statistics when
not all necessary information is available to the compiler and misreporting of goods after processing
can be expected to be the main cause of the difference between the processing services and the value
of goods before and after processing. It also provides an example on other adjustments of data on
import and export of goods related to import of processing services that may be conducted in order to
compile balance of payment statistics. Table 14.2 presents possible combinations of related
movements of goods and services across the borders for processing, after processing and also without
any cross-border movements. It also explains the difference in recording of exports and imports of
goods between the IMTS2010 (cross-border concept) and BPM6 (ownership concept). It is obvious
that in some cases there is no cross-border movements of goods related to the handover of goods
between processor and principal (either before the processing or after the services are provided).

165

Box 14.1 Numerical example of possible adjustments to import and export of goods for the
balance of payments purposes in relation to import of processing services in the absence of full
information
Assume that country A imports processing services f rom B in value of 7 (5 is related to processed goods
returned to the economy of principal A, 1 - to goods that were sent for processing without returning t o country
A and 1 - to goods sourced abroad which were processed in B and imported (brought) to country A after
processing). A’s IMTS records: export of goods for processing (returning to country A after processing ) is 10;
but import of goods after processing is only 5 (therefore it has to be corrected for misreporting).
The margin between goods sent for processing (10) and returning after processing (5) is -5. The value of these
goods cannot be accounted for as general merchandise in the balance of payments and their value is therefore
subtracted to obtain the gross flows of exports and imports according to the BPM6/MSITS2010
recommendations. In addition, the import of services related to the goods returning back to A is 5 so the
adjustment of import after processing has to be +10 (=5+5), i.e. the adjustment of general merchandise
according to the BPM6/MSITS2010 is -10.
Apart from the adjustment for misreporting, there may be also other adjustments in goods due to insuff icient
data sources and imported processing services. Assume that country A exports goods for processing that is not
returning back after processing in value of 10. Strictly speaking, the goods should be also subtracted from
general merchandise since no change of ownership occurs when crossing the borders and exports of goods
should be recorded only when sale abroad after processing occurs (e.g. in value of 11, 10 + 1 for the
processing services). However, there is usually no data source for these sales abroad and thus the goo ds
exported for processing without returning back may not be subtracted from general merchandise (10). In
addition, the import of services for 1 (corresponding to the value of imported services processed on goods
sold abroad) is supposed to be imputed to the export of goods. Analogously, assume import of goods which
were purchased abroad and processed in B (in value of 11, purchase for 10, services for 1). The value of
import according to BPM6/MSITS2010 should include only the value of goods brought from abroad
(excluding the 1 corresponding to the value of imported services processed on goods purchased abroad and
imported to country A).

166

Table 14.2(a)
Export of processing services; resident = processor (contractor)
Recording in BOP, NA, and IMTS
Residence
of the
principal
(owner of
the goods
to be
processed)

Residence of
Origin of
entity
Residence
the good
acquiring
of
before
transformed
processor
processing goods from
principal

Export

Debit

Credit

1 Goods received by resident processor for
Import of goods
processing. Transformed goods leave
for processing
processor’s economy once processed.

Export of
processed
goods

Service:
Processing fee.

Resident

Goods received by resident processor.
Transformed goods are purchased by Import of goods
2
residents of processor’s economy once for processing
processed.

Goods: Purchase
of processed
goods

Service:
Processing fee.

Resident

Goods purchased by non-resident principal
in the resident processor’s economy.
3 Transformed goods are purchased by
residents of processor’s economy once
processed.

Goods: Purchase
of processed
goods

Service:
Processing fee
Goods: Sales of
goods (input) to
non-resident
principal.

Resident

Compiling
economy

Import

Goods and services that should be
recorded in BOP and NA of
processor’s economy
(change of ownership principle)

Non-resident
Abroad

Nonresident

Export of processing services
resident = processor (contractor)

Goods that should be captured
in the IMTS of processor’s
territory
(physical movement of goods)

Goods purchased by non-resident principal
in the resident processor’s economy.
Non-resident 4
Transformed goods leave processor’s
economy once processed.

167

Export of
processed
goods

Service:
Processing fee
Goods: Sales of
goods (input) to
non-resident
principal.

Table 14.2(b)
Import of processing services; resident = owner (principal)
Recording in BOP, NA and IMTS
Residency
of the
principal
(owner of
the goods
to be
processed)

Residence
Origin of
of entity
Residence physical
acquiring
of the
inputs
transforme
processor
before
d goods
processing
from
principal
Resident

Import of processing services
Resident = client/owner (principal)

Export

Resident

Nonresident
Nonresident

Nonresident

Goods and services that should be
recorded in BOP and NA of
principal’s economy
(change of ownership principle)

Import

Credit

Debit

Resident principal sends goods (inputs) to
non-resident processor.
Export of goods
Transformed goods are returned to the for processing
economy of principal.

Import of
processed
goods

Service:
Processing fee

Resident principal sends goods to nonresident processor.
Export of goods
Transformed goods are sold abroad by for processing
resident principal to non-residents.

Goods: Sales of
goods (includes
processing fee)

Service:
Processing fee

Resident principal purchases goods from
non-resident.
Transformed goods are sold by resident
principal to non-residents.

Goods: Sales of
processed goods
(includes
processing fee)

Service:
Processing fee
Goods: Purchases
of goods (input)

Resident principal purchases goods (input)
from non-resident.
Transformed goods are sent to economy of
principal.

Import of
processed
goods

Compiling
economy
Nonresident

Goods that should be captured
in the IMTS of principal’s
territory (physical movement of
goods)

Abroad
Resident

168

Service;
Processing fee
Goods: Purchases
of goods (input)

B.1.b.2

Manufacturing services: borderline cases

14.30. It is important to separate manufacturing services on inputs owned by others from transactions
that may have similar aspects, but that should be recorded differently according to BPM6.
14.31. First, compilers should carefully distinguish between manufacturing services and manufacturing
on own account, using the economic ownership of the inputs as the distinguishing criterion. Even if the
processor not only transforms the goods owned by others (i.e., the principals), but also provides some
physical inputs in the process, their exported output will be recorded under manufacturing services (i.e.,
including the value of physical inputs provided by the processor). This distinction is particularly relevant
for countries with locations or zones with special taxation arrangements, such as export processing zones
or free zones. Compilers should be aware that enterprises that operate in these zones are not necessarily
by definition engaged in manufacturing services. If such enterprises are the economic owners of the goods
that they process (whether purchased domestically or from the rest of the world), their exported output
would not qualify as manufacturing services but as general merchandise. Similarly, many enterprises may
be engaged in manufacturing services, without necessarily being in special locations (see also (see also
12.15-12.16 of the BPM6CG).
14.32. Some enterprises may be affiliates from direct investors abroad and may use technology
transferred to them by this direct investor.183 Compilers should note that the mere existence of a direct
investment relationship does not imply the provision of manufacturing services, even when the processed
products are sold to the direct investor. Since the classification is solely determined by the economic
ownership of the inputs, this would still be recorded by the affiliate as manufacturing on own account
(and hence not a manufacturing service).
14.33. A second area where compilers need to make a careful distinction is the difference between
transactions related to manufacturing services on inputs owned by others, and those related to
merchanting (see also 12.19 in BPM6CG). When in the case of merchanting, the nature of the goods does
not change; the gross values of the associated goods must be recorded as trade in goods (negative and
positive exports).
14.34. In the case that goods under a merchant’s ownership may be subjected to certain manufacturing
services that change the condition of these goods, relevant purchases and sales of the goods should be
recorded under general merchandise instead of merchanting, whereas the minor processing or packaging
fee would still be recorded under manufacturing services.
14.35. Compilers are suggested to also consult the Guide on Measuring Global Production for more
suggestions on how to deal with the different cases that compilers may be confronted with, as well as
possible solutions to compile the related items in the goods and services account.184

183

Technology transfer includes patenting, licenses, know-how, technical assistance and the provision of research
and development services. See OECD, Measuring Globalization: OECD Handbook on Economic Globalization
Indicators, 2005.
184
UNECE (2014) Guide to Measuring Global Production.
http://www.unece.org/fileadmin/DAM/stats/documents/ece/ces/bur/2014/Guide_to_Measuring_Global_Production_
-_CES.pdf.

169

Country experience: Czech Republic
14.36. In the movements of goods within the EU the processing services cannot be estimated by the
difference between goods crossing borders before and after the processing (e.g. a long-term negative
margin on outward processing with other EU countries has been recorded). There are at least three main
reasons for this:
i.
The non-resident principal has the goods after processing shipped directly from the
economy of the processor to a third economy. The resident processor had reported receiving the
goods for processing into Intrastat (system for collecting information and producing statistics on
the trade in goods between EU countries) but the non-resident does not report them as goods sent
after processing but as a simple merchandise export.
ii.
Goods after processing are sold by the non-resident principal to a resident of the
processor’s economy (i.e. without the goods leaving the processor’s economy). Non-resident
principal must be registered in the processor’s economy for VAT purposes. The processor had
only reported acquiring goods for processing into the Intrastat but there is no cross-border
movement of goods after processing between the principals’ and processors’ economies.
iii.
Goods sent for processing and especially received after processing may be significantly
affected by misreporting or asymmetric reporting. Goods sent for processing within the EU are
not under custom control in the Czech Republic. For this reason it is difficult to keep reporting
consistently on both sides in manufacturing chains. For example processor A reports physical
entry of goods into the country (for processing) in certain value, eventually processes it and
transfers it to processor B, which after its own processing reports the exit from the country with a
completely different value.185
14.37. The reasons mentioned under (i) and (ii) lead to differences between processing services and
margins resulting from gross cross-border flows in goods for processing. Since the compilation of export
and import of goods is based on change-of ownership principle186 (not on cross-border principle), there is
no need for any additional adjustments. In case of (i), the goods received for processing by resident
processor is excluded from general merchandise as well as all cross-border exports by non-residents, even
though they are reported as general merchandise. In case of (ii), the goods received for processing by
resident processor is also excluded from general merchandise and only sales by non-residents within the
Czech market is recorded as import of goods. The sales of non-residents are known from VAT statements
submitted by non-residents.
14.38. However, the difference caused by point (iii) must be eliminated by additional adjustment of
goods excluded from general merchandise in order to keep the balance on trade of goods and services
unaffected. In fact, the first adjustment is done mostly at the level of Intrastat data-collection by

185

Holding gain/losses or overheads are not likely to be among them as the entity obliged to report the movements
of goods across the borders in Intrastat is the processor (not the owner of goods).
186
National concept of foreign trade in goods, which represents change- of ownership approach, takes into account
sales and purchases by non-residents within the Czech market, and eliminates all cross-border movements of goods
reported by non-residents since these cross-border flows cannot be considered as transactions related to the Czech
economy.

170

comparing the processing services (declared by exporters or importers of services) and declared crossborder movements of goods (reported into Intrastat by the same reporting units, as these units are usually
the same – exempt for chain-linked processing). The adjustment requires sufficient data comparison of
both data sources on the individual level of reporting units. Nonetheless, other adjustment may be needed
at the aggregate level, mostly because of the misreporting caused by chain-linked processing.
14.39. For the purpose of these adjustments ad-hoc annual survey on a voluntary basis along the regular
survey on international trade in services is conducted among the processors (exporters of services) and the
principals of goods to be processed abroad (importers of services). Detailed information is collected on
the relationship between services exports and imports and the related movements of goods across the
border and the handover of transformed goods to the principal within the processor’s economy. The
additional adjustment of goods due to misreporting may be carried out in order to keep the margin
between goods after processing and goods before processing in accordance with the particular processing
services related to the movements of the goods (especially when goods is returning back after processing
– see Box 14.1).
Hong Kong imports of manufacturing services and related goods transactions
Case I: Outward processing to Mainland China with processed goods returned to Hong Kong
14.40. In the processing activities between Hong Kong (principal in the reporting economy) and
Mainland China (non-resident processor), the import value of the manufacturing services is derived from
the total of the processing fee charged by the non-resident processor and the amount reimbursable by the
principal for the materials procured by the processor.
14.41. The cost of materials supplied by the principal from or via Hong Kong for processing without
change of ownership is not included in “Exports of goods” in the balance of payments statistics. By the
same token, the value of the processed goods returned to Hong Kong should not be treated as “Imports of
goods”. The cost of input materials supplied by the principal to the processor from a third economy
directly is included in “Imports of goods” of the economy of the principal.
Case II: Offshore trade activities involving outward processing with processed goods sold offshore
14.42. Similar to Case I, the value of manufacturing services is reported under “Imports of services”.
The import value of the manufacturing services is derived from the total of the processing fee charged by
the non-resident processor and the amount reimbursable by the principal for the materials procured by the
processor. Margin earned from the offshore sale of the processed goods are not included in “Exports of
services”.
14.43. As for trade in goods, the cost of goods for processing supplied by the principal from or via Hong
Kong to the processor without the change of ownership is not included in “Exports of general
merchandise”. The value of the processed goods sold offshore directly without returning to Hong Kong is
treated as “Exports of general merchandise”. The cost of materials supplied by the principal to the
processor from a third economy directly is included in “Imports of general merchandise” in the economy
of the principal.

171

Hong Kong exports of manufacturing services and related goods transactions
14.44. For illustration purposes, it is assumed a Hong Kong company provides manufacturing services
on the materials owned and supplied by an Australian company (non-resident principal) and the processed
goods are returned to Australia. The value of manufacturing services is reported under “Exports of
services”. The respondent can provide the export value of manufacturing services based on the total of
the processing fee charged and the amount reimbursable by the non-resident principal for the materials
the processor procured. For trade in goods, the value of the processed goods is not treated as “Exports of
general merchandise” because the ownership belongs to the Australian company. Besides, the value of
input materials supplied to the processor by the principal from Australia or a non-resident are not included
in “Imports of general merchandise”. However, the value of materials procured by the Hong Kong
company from a non-resident to be used in the processing activity are already recorded under “Imports of
general merchandise” and would be part of the manufacturing fee.
B.1.b.3 Freight transport and freight insurance services: CIF and FOB valuations
14.45. It is not always possible to measure the value of the freight transport and freight insurance
services directly. However, freight cost estimates (including both transport and insurance premiums) can
be derived from the CIF/FOB adjustments that are made to the merchandise trade statistics for Balance of
Payment purposes. The difference between CIF and FOB values are considered as freight costs (transport
and insurance premiums). These costs are generally deemed payable by the importer of the goods. Where
they are provided by residents to non-residents, or by non-residents to residents, there will be recordings
in the balance of payments. Table 14.1 displays these differences in recording.
Table 14.1 Valuation principles of merchandise statistics and Balance of Payments
Merchandise trade
(IMTS statistics)
Goods
(BOP statistics)

Imports
Exports
Import and
exports

Statistical values CIF (cost insurance freight) and statistical
values FOB (free on board)
Statistical values FOB (free on board)
Statistical values FOB (free on board)

14.46. As table 14.1 shows, the recommendation for merchandise exports is that they are valued on a
FOB basis, which follows the BPM6 principle of recording general merchandise credits FOB. Imports in
international merchandise trade statistics are recommended to be compiled on a FOB basis (also,
countries are encouraged to compile FOB-type value of imported goods as supplementary
information187 )whereas in the BOP, the imports of goods should be valued on an FOB basis; i.e., the value
of transportation cost and insurance premiums from the frontier of the exporter to the border of the
importer is recorded as imports of these services (if the services are provided by non-residents) and the
remaining value is recorded as imports of goods.
14.47. Compilers should note that when calculating the freight transport and insurance costs for BOP
purposes, the statistical value does not necessarily reflect the sum of the actual invoice values,188 but
instead follows the principle adopted in the accounting framework (see the numerical example in box III.4

187

See: IMTS 2010, paragraph 4.8
The report from Statistics Denmark provides an overview on differences between the statistical CIF and FOB
value. EuroGrant 2009 Balance of Payments, Grant agreement number 20821.2009.001-2009.715, final report May
2010; the Glaab report recommends collecting invoice values instead of statistical value; the import and export of
freight services should be collected.
188

172

in MSITS2010). However, many users are also strongly interested in obtaining the data on the actual
costs, as this reflects the total value of freight related services that are provided, which is why MSITS
2010 suggests making these data available to users on a supplementary basis (i.e. beyond the
BPM6/EBOPS2010 recommendations). The valuation of freight transport services on a transaction basis
indicates market price transactions excluding any kind of adjustments or estimations. Such additional
information189 can be used as supplementary information to the FOB/FOB valuation principles in balance
of payments statistics or the FOB/CIF principle in international merchandise trade statistics.
14.48. A variety of methods for estimating freight and insurance on imports are described in BPM6 CG
table 12.2. A common approach of CIF/FOB adjustments is to calculate conversion factors, or ratios of
CIF imports over FOB imports, that should take into consideration the mode of transport, the
geographical distance and the type of commodity transported. International merchandise trade data from
customs can be used to estimate CIF/FOB ratios, provided that these data include information on invoice
and delivery terms. Dedicated surveys are also used in some countries.
14.49. To compile the ratios between the statistical value (CIF-type) and the invoiced value (of the
transactions with FOB-type delivery terms190), compilers are suggested to use those transactions for
which FOB type delivery terms is available, and to calculate the ratio between their FOB-type invoiced
value and the corresponding CIF-type statistical value. The ratio thus compiled can subsequently be
applied to the total statistical value (CIF-type) to obtain a FOB type value. Such ratio should be calculated
for freight transport and insurance, respectively. When there is no sufficient information available from
customs merchandise trade statistics, a combination of the above techniques and direct data collection or
surveys can be used to allow a full coverage of all the cases of imports and exports, types of contracts,
and residence of the carrier.
14.50. Freight insurance services are usually estimated by compilers on the basis of the insurance
premiums, the former being often calculated as a ratio to the non-life insurance premiums. Depending on
the terms of the freight, additional adjustments may be needed based on the value of the goods being
traded (i.e. if the terms are not on an FOB basis) and the terms of contract (i.e., whether the shipment
between the exporter and the importer’s borders is undertaken by a resident or a non-resident).
14.51. It is advised to compile the value of total freight insurance, whereas the freight transport
component has to be broken down by mode, as follows: sea, air, other, of which, space, rail, road, inland
waterway, pipeline transport and electricity transmission.
Country experience: Germany
14.52. In Gernamy, the following formula is applied for the calculation of the transportation costs:
Transportation Costs = weight x freight rate (subject to mode of transport, product group (in the case of
sea transport), and distance). In order to apply this formula the information matrix of table 14.5 is used.

189
190

MSITS 2010, paragraphs 3.107-3.110.
The delivery-terms-FOB method has been developed by Statistics Netherlands.

173

Table 14.5 Information matrix to calculate transportation costs
Weight

Product
characteristic
x

Distance

Unit costs

Nationality

(x)

Sea Cargo

Air Cargo

(x)

Inland Waterways Cargo
Road Cargo

X
X

x
x

Rail Cargo

Pipelines

14.53. In order not to increase the burden for the reporters the information needed for the estimation are
taken from the following public data sources.
14.54. The weights for imported and exported goods are taken from foreign trade statistics and are hence
available on a monthly basis.
14.55. Freight rates. As a second best solution, we use information extracted from publications
specialized in transport an internet research, which focus on more than 30 major carriers and data from
transport services associations. Based on this information the transport costs for a standard container191
were calculated for the various modes of transport. With regard to sea transport, owing to the availability
of additional information, a further breakdown of the transport costs by containers, bulk goods and crude
oil is made.
14.56. The transport prices for each mode of transport, which fluctuate considerably over the course of a
year, are updated on the basis of the transport price indices calculated by the Federal Statistical Office.
14.57. Mode of transport. As an expedient, a “main mode of transport” is used as a basis for the
estimation. The longest part of the total transport route (and hence as a rule also the stretch with the
highest value) is covered using this mode of transport. Information on the mode of transport is derived
from the foreign trade statistics, in which an obligatory enquiry is made about the active mode of
transport used to cross the EU’s external border. This is mainly sea or air and a good approximation of the
actual main mode of transport. If another mode of transport is stated in the trade statistics, it is assumed
that in the case of non-European countries, the goods were transported by sea to the large ports in the
Netherlands and Belgium192 and then transported onwards to Germany on the stated mode of transport.
Table 14.6 shows an example of calculating transportation costs.

191
192

20ft equivalent unit (TEU) with an average loading weight of 14 tons.
For mineral oil imports, Italy (Trieste) is assumed to be the unloading port for the Middle East countries.

174

Table 14.6 Example of calculation of transportation cost for imports from China
Imports from
China: 1,109,537t
Sea cargo
+ chain sea->inland
waterways
+ chain sea->road
+ chain sea->rail
Air cargo
Inland waterways
cargo
Road cargo
Rail cargo

Mode of
transport
85.6%

Distance

Product group

Unit costs

Costs in
1,000 EURO

3.9%

139 €/t

151,141

8.3%
0.2%
2.0%

8,479 €/t

188,022

3.9%

4.90€/t

212

8.3%
0.2%

300 km
300 km

0.08776€/t km
0.04636€/t km

2,424
30

Note: (1,109,537 x (0.856+0.039+0.083+0.002) x 139 = 151,141)
14.58. Nationality of the carrier. A number of different sets of statistics are used as a basis to determine
the relevant partner country. The nationality composition of the road toll statistics is used for the
regionalization of lorry transport (adjusted to compensate for the disproportionately large share of
German trucks due to domestic transport). The ownership (not flag) share of the countries among the
world fleet is used for sea transport. Since only some sea transport is carried out using scheduled services,
the breakdown of the world fleet by country can be regarded as an informative approximation. The same
applies to air transport, i.e. the world market shares of the airlines are used. With respect to inland
shipping, information on the flags of the ships is the sole data available, hence only this information is
currently used. As cross-border rail transport has only been liberalized since 2007, no statistics are
available for this yet. A robust estimation for the rail system is carried out on the basis of the technical
systems of the national railway networks (e.g. track widths, line voltages, operating licences).
14.59. Transport insurance. The amount of transport insurance is also estimated using this procedure,
although no further details will be supplied regarding the methodological particularities of insurance
companies (service fee included in the insurance premium, etc.). Unlike transport costs, the insurance
costs are calculated as a share of the import value. This share is in line with the premium rates set by
transport insurance brokers for transportation from/to various regions. These rates were calculated parallel
to determining the transport costs by surveying several international transport insurance companies. The
reported country structure for transport insurance is used for the regional breakdown of these costs.
B.2.

Passenger transport and travel

B.2.a. Introduction
14.60. Passenger transport and Travel are service categories that are strongly related and where
compilation often involves the same or similar data sources. Passenger transport include fares and other
expenditure related to the carriage of passengers, including any taxes levied on passenger services, such
as sales or value added taxes (see MSITS2010 paragraph 3.95). Travel covers goods and services for own
use or to be given away, acquired from an economy, by non-residents during visits to that economy.
Hence, travel is transactor-based, which makes the compilation of data on travel challenging compared
with most other service categories that follow a product-based classification (see MSITS2010 paragraph
3.115).

175

B.2.b. Compiling statistics for passenger transport and travel
14.61. To compile statistics for passenger transport and travel, compilers can draw from a variety of data
sources. These can include ITRS (which can provide information on monetary flows between travellers
and tourism services providers and payments made through bank notes, traveller cheques or credit cards),
payments cards data and other administrative sources (such as immigration records), but surveys remain
an especially important data source. For passenger transport, these surveys may include dedicated
transport surveys (aimed at large transport-providing enterprises, such as airlines or railway operators).
14.62. For EBOPS travel item (exports), surveys can include survey of travellers; border surveys;
household surveys; and surveys of enterprises, including accommodation establishments and tourist
intermediaries (e.g., travel agencies and tour operators). Travel agencies are the traditional intermediaries
between those traveling and travel-related services providers, such as hotels, car rentals, cruise lines and
package tours. Direct reporting allows for a better measure of travel expenditures booked through these
intermediaries than what can result from the ITRS. As far as individual tours are concerned, the travel
agencies have detailed information that allows them to distinguish travel components only when
individuals pay for tours to visit foreign countries. For package tours, it is often the case that tour
operators are in a better position to provide the product classification of different services assembled in a
package. This information can be obtained through the implementation of a specific direct report for tour
operators.
14.63. Visitors’ surveys may provide more detailed insights, such as the purpose of travel without which
the trip would have not taken place, the type of person traveling, the structure of the expenses made
during the trip and how they pay for those expenses. Supplemental sources include tourism activity
statistics (accommodation providers’ statistics) along with accommodation prices in the consumer price
index, and, on the debits side, mirror statistics from partner countries.
B.2.b.1 Passenger transport
14.64. Transport services should be recorded on a gross basis. This means that the value of passenger
transport services includes fares and other expenditures related to the carriages of passenger. For example,
commissions of ticket sales should be included in services value.
14.65. Given prevalent practices of interlining and code sharing by airlines, compilers should obtain data
on revenue earned by an airline from residents of other countries rather than data on sales by an airline to
residents. It may be possible to collect such data as airlines keep records on revenue generated by country
of sale. However, as not all airlines earning revenue from residents of a particular country will have
offices in that country, it may be difficult for compilers to obtain a complete coverage of passenger fare
revenue earned by non-resident operators from residents of the home country. In such cases, a data model
based upon the number of non-resident passengers carried by resident operators, passengers’ countries of
origin and destination, and average fare rates could be used to produce an estimated value.
14.66. An alternative means of measuring passenger fare revenue earned by non-resident operators is to
collect information on the total value of tickets sold in the compiling country and deduct from this total
value the earnings of resident carriers. An estimate of ticket sales may be derived from a household
budget survey or other surveys of individuals, for example.
14.67. In some cases, passengers fares may be a component of package tour payments, and the compiler
may, in consultation with travel industry representatives, have to separate passenger transport and other
travel components.

176

Country experience: Ireland
14.68. In Ireland export sales data for passenger transport are obtained from resident airline and ferry
operators in terms of their receipts from non-residents for travel to and from Ireland.
14.69. Direct data on receipts for other types of resident transport companies are not available – such
expenditure by non-resident visitors to Ireland being captured indistinguishably in the travel and tourism
receipts (exports). Respondents may provide their best estimates’ in respect of the geographical
breakdown required because of the difficulty of knowing in all cases the precise country of residence of
their customers. Payments by Irish residents to non-resident transport enterprises in general cannot be
directly distinguished at present. Such payments (imports) are included in the travel and tourism
expenditure data. Receipts by resident airline and shipping companies for freight services provided
(exports) to non-residents are obtained from these enterprises, the geographical breakdown being
provided on a best estimates’ basis where necessary.
Country experience: Australia
14.70. In Australia passenger transportation services are obtained from the ABS's quarterly Survey of
International Trade in Services with estimates for freight debits derived from the ABS's International
Merchandise Trade Statistics. Survey responses for passenger transport are adjusted to ensure only
international services are captured as described below.
14.71. The SITS requests non-resident providers of passenger transport services to report revenue earned
from the sale of international tickets within Australia. Survey responses could therefore include tickets
sold to both resident and non-resident travellers, and regardless of whether it involves transport between
Australia and the rest of the world or within Australia as part of an international journey. Data from the
International Visitors Survey conducted by Tourism Research Australia and ABS Overseas Arrivals and
Departures data are used to estimate the value of international tickets sold within Australia to nonresidents on non-resident airlines and this value is deducted from the survey total for passenger transport
debits. Similarly estimates are made for the value of international tickets sold within Australia to nonresidents on resident airlines and this value is added to the survey total for passenger transport credits. It
should be noted that the survey requests the same information from resident providers of passenger
transport services for the sale of international tickets abroad, however no adjustment is currently made to
account for the purchases by Australian residents abroad.
14.72. The value of services provided to non-residents by Australian carriers in Australia, when sold
abroad as part of an international ticket (on-carriage), is collected from the carriers and allocated to travel.
Services provided on purely domestic travel in Australia by non-residents, whether pre-purchased abroad
or while in Australia, are also included in travel. No classification adjustments are made for non-resident
earnings from residents for internal flights abroad, and all earnings from sales in Australia for on-carriage
in a foreign country, or for pre-purchased domestic travel in a foreign country, are included
indistinguishably in transportation debits. Cruise fares are excluded from passenger services and included
in travel, although sea passenger services sold in Australia and provided to residents travelling from one
country to another and any resident sea passenger earnings are included in transportation.
B.2.b.2 Travel
14.73. The concept of travel, as it used in MSITS 2010, is closely related with the concept of tourism.
Thus, it is advised that trade in services compilers understand the conceptual framework of tourism
statistics, including Tourism Satellite Account (TSA). Detailed descriptions are available in two UN

177

publications (IRTS 2008 and TSA: RMF 2008) and in the Compilation Guide for Tourism Statistics
available at the World Tourism Organization website.193
14.74. In addition to the aggregate measure of travel exports (credits) and imports (debits), BPM6 and
EBOPS 2010 recommend the compilation of further breakdowns of travel. These breakdowns can be used
not only to better assess the scope of travel activities and to gauge their possible impact in terms of
economic activity, but also to ensure consistency between data on travel compiled in accordance with
MSITS 2010 and other related statistics, such as the tourism statistics, tourism satellite account or supply
and use table.
14.75. Some of these requirements are not new compared with the previous edition of MSITS and BPM.
That is the case of the mandatory split between business and personal travel and the breakdown of
personal travel into health-related, education-related and other motives. Other breakdowns, such as the
alternative presentation of travel according to the types of goods and services, are new challenges for the
balance of payments compilation. C To meet these requirements compilers should draw on the advantages
of the increased proximity with agents directly involved in the external operations and on the availability
of more detailed information. Table 14.7 shows complementary data sources that can be used to compile
different breakdowns.
Table 14.7 Travel breakdown: data sources (example)
Breakdown
Travel Purpose:
Business / Personal
Personal Purposes: Health-related, Educationrelated, Other
Group of Traveler: Border, seasonal and other
short-term workers
Types of Goods and Services

Data Sources
Payment Cards Database
Direct Reporting - Travel Agencies and Tour Operato rs,
visitor surveys and household surveys
Survey, Administrative Data
Mirror Statistics
Survey
Payment Cards Database
Direct Reporting - Travel Agencies and Tour Operato rs
Survey

14.76. When defining the specific level of detail for each breakdown, other specifications should be
considered in order to have an integrated and consistent framework, which would provide all details
required by other statistical domains, such as the tourism satellite account, national accounts and the
harmonized index of consumer prices. A needs assessment is essential to understand how the available
data sources can deliver the level of detail required. For this purpose, the cooperation between different
statistical authorities seems to be crucial in order to reduce compilation costs, as well as to integrate
diverse statistical systems and conceptual frameworks.
14.77. In particular, compilers who are responsible for estimating the receipts and expenses for travel
could work closely with the Ministry of Tourism or similar governmental agencies. When international
passenger surveys and other surveys are conducted or outsourced, thorough discussions should be made
between the compilers of trade in services statistics and the Ministry of Tourism. Such discussions would

193

See: http://statistics.unwto.org/en/content/methodology-0 and http://statistics.unwto.org/content/internationalrecommendations-tourism-statistics-2008-irts-2008-compilation-guide.

178

contribute to the consistency of primary data used for the compilation of these statistics and tourism
statistics, including tourism satellite account.
14.78. The interconnection of the different sources can be seen as rooms in different floors of a house
(see Figure 14.1, which is based on the experiences of Austria), where the SNA serves as the roof top
giving inputs to the TSA and receiving input from Travel item of balance of payments (T-BOP). The
basic tourism statistics, accommodation statistics and the sample survey are major inputs for compiling
the compound statistics TSA and T-BOP. The basic statistics should be mutually harmonized in major
conceptual fields and the outcomes should be reconciled where they describe the same pieces of reality.
Figure 14.1 Integrated Aspects of Tourism and Travel Statistics

Country experience: Austria
14.79. In Austria, OeNB and Statistics Austria are running a joint survey for tourism expenditures and
the travel item. For collecting or compiling tourism in its forms of domestic, outbound and inbound the
accommodation statistics (covering inbound and domestic) and the sample survey are used. These two
basic surveys cover all forms of tourism with respect to the physical flows. For bringing in expenditure
variables for inbound expenditure, T-Mona, which is basic tourism statistics, is used. These three basic
statistics are followed by the compound statistics TSA and the T-BOP. T-BOP statistics provide key
figures with the TSA.
14.80. As the travel components such as seasonal and border workers, students, medical patients and
those who are crossing the border but still are within their usual environment can be isolated, T-BOP data
can be used as key figures of the TSA. The TSA uses several other data sources, including the
accommodation statistics, the sample survey and T-Mona. The T-BOP that goes beyond the concept of
tourism statistics also uses several tourism statistics as inputs. The T-BOP is input for compiling the rest
of the world account, delivering bridge tables for the private consumption and identifying business travel,
which is seen as production input in the NA.

179

14.81. Reconciliation of basic tourism statistics. The selection of the T-Mona sample is based on data of
the accommodation statistics as these data are available on a detailed geographical level. The data are
further used for weighting and grossing up procedures. Also the questionnaires are harmonized in major
fields as, for example, the same categories of accommodation statistics are used in both statistics. As the
accommodation statistics do not consider stays at friends and relatives but as this information is also
included in the T-Mona survey, inputs are used from the accommodation statistics to establish proper
sample basis.
14.82. The questionnaires of the sample survey and T-Mona are harmonized according to
accommodation establishments and other definitions according to the tourism concept. The data sources
of both sample survey and accommodation statistics cover domestic tourism. Therefore the outcomes at
least for the physical flows of the number of overnights and arrivals of Austrian residents in Austria are
mutually checked.
14.83. Reconciliation between the BOP travel item and tourism statistics is performed on both credit and
debit sides of T-BOP.
i.

Credit side of the T-BOP. The first major data source for the compilation of the credit side of
the T-BOP is T Mona, as it delivers the average expenditure of non-resident tourists in
Austria in paid and non-paid accommodation establishments. The expenditure questions in
the T-Mona questionnaire meet TBOP needs and are harmonized with concept used in BPM6
and MSITS2010. The second major data source for the compilation of the credit side of the
T-BOP is the accommodation statistics which delivers the population of non-resident tourists
in paid accommodation establishments by country of origin. However due to survey
restrictions the number of countries, which are directly surveyed, is limited to 60. In contrast
to this limited geographical breakdown the T-BOP considers data from credit and debit card
companies which are delivered by the Austrian National Bank and therefore delivers detailed
information about the geographical breakdown. So data can be used to further splitting up the
geographical breakdown of the accommodation statistics based on expenditure data.

ii. Debit side of the T-BOP. Major data source for the compilation of the debit side of the TBOP is a sample survey. As this sample survey is based on the tourism concept harmonized
questions are implemented to meet the needs for the travel concept of the T-BOP. Besides the
differentiation between business and personal travel also same day visitors are included as TBOP does not distinguish between lengths of stay as only the expenditure variable is
important. Credit and debit card data which are implemented in the T-BOP are used for doing
plausibility checks with the outcome of the sample survey. The data are also used to create a
detailed geographical breakdown of the sample survey as for this survey the geographical
breakdown is limited to approximately 10 because of the sampling error. This reconciliation
exercise helps to overcome the limits of the sample survey.
Country experience: Portugal
14.84. The framework designed in Portugal to compile travel statistics is to a large extent based on the
instruments used to pay for travel expenditures, that is to say (a) payment cards, (b) traveler’s checks and
(c) cash. Regarding transactions settled in cash, the introduction of euro resulted in an additional
difficulty for the compilers, particularly in the euro area countries. It is necessary to estimate these
expenditures and this estimation can be quite challenging, since the use of cash must certainly differ
according to the type of good or service acquired by travelers.

180

14.85. The selection of the data sources above-mentioned to use in such a framework depends upon the
moment of payment of the expenditures: (a) pre-payment or (b) local payment, and upon the channel used
to book and/ or pay for travel arrangements: (a) direct reservation or (b) reservation made through a travel
agency or tour operator (resident or non-resident). The approach and sources selected to compile travel
credits and debits may be somewhat different, taking into account that the scope of data sources available
and the degree of coverage that is possible to obtain from each data source individually can differ for the
credit and the debit side.
14.86. Tables 14.8 and 14.9 present an overview of the main sources selected for each combination of
channel, moment of payment and payments instrument used. The combinations showed in the tables are
the ones that are more likely to occur. For example, the pre-payment of a trip booked through a direct
reservation channel most probably will not be performed in cash. Therefore, this alternative was excluded
from both tables.
Table 14.8 Main source for travel credit (supply side)
Channel used to carry
out travel arrangements

Direct reservation

Moment of
payment

Payments
instruments

Pre-Payment

Payment Cards
Payment Cards

Local payment

Cash
Traveler’s check

Pre-Payment
Reservation made
through a travel agency or
tour operator (resident or
non-resident)

Payment Cards
Cash
Payment Cards

Local payment

Cash
Traveler’s check

181

Data Sources
Payment Cards Database
Payment Cards Database
Direct Reporting - Banks and
MTO
Payment Cards Database
Survey
Direct Reporting – Banks
Tourist Activity Statistics;
Accommodation Prices;
Direct Reporting - Travel Agencies
and Tour Operators; Survey
Payment Cards Database
Direct Reporting - Banks and MTO
Payment Cards Database
Survey
Direct Reporting – Banks

Table 14.9 Main source for travel debit (demand side)
Channel used to carry
out travel
arrangements

Direct reservation

Moment of
payment

Payments
instruments

Data Sources

Pre-Payment

Payment Cards

Payment Cards Database

Payment Cards

Payment Cards Database
Direct Reporting - Banks and MTO

Local payment

Pre-Payment

Cash

Payment Cards Database
Survey

Traveler’s check

Direct Reporting – Banks

Payment Cards

Direct Reporting - Travel Agencies
and Tour Operators

Cash
Reservation made
through a resident travel
agency or tour operator

Payment Cards

Direct Reporting - Banks and MTO
Local payment

Pre-Payment
Reservation made
through a non-resident
travel agency or tour
operator

Payment Cards Database

Cash

Payment Cards Database
Survey

Traveler’s check

Direct Reporting - Banks and MTO

Payment Cards

Payment Cards Database; S urvey

Payment Cards

Payment Cards Database
Direct Reporting - Banks and MTO

Local payment

Cash

Payment Cards Database
Survey

Traveler’s check

Direct Reporting – Banks

Country experience: Hungary – health-related travel
14.87. The Hungarian Central Statistical Office (HCSO) compiles services transactions between
residents and non-residents. Data for business and transportation services are based on annual and
quarterly conducted surveys by HCSO. The statistical units are the resident enterprises, government and
non-profit organizations, which supply services to non-residents and use services of non-residents. The
surveys contain a list of 62 service types, of which health services. According to the methodology,
respondents should report all services of NACE 86 category between residents and non-residents,
provided by physicians, nurses, other qualified personnel working in health care. Even the services
provided from distance (ex. Laboratory, through internet) should be included.
14.88. The main data provider for health services is the National Health Insurance Fund (OEP). Among
the other companies who reported export data for health services were university hospitals, air emergency
rescue companies and pharmaceutical trade companies. According to an agreement between OEP and
HCSO (and based on the international methodology) the insurance fund export data cover emergency and
other demanded health services provided by Hungarian health institutes for foreigners in Hungary. The
invoices paid by foreign health insurance companies, or paid by foreigners at the site.

182

14.89. On the other hand, imports cover emergency and other demanded health services provided by
foreign institutes for Hungarians, the invoices paid by OEP directly, and the services paid by Hungarians
at the site, reimbursed by OEP later. The health services for Hungarians who travel abroad for the certain
purpose of medical treatment are considered as part of the tourism service and are not reported here. The
differences between the methodology of external trade statistics and SHA have as result different values
for the national health accounts of Hungary and international trade of health services. In the trade
statistics the difference is part of the tourism service import.
14.90. Data on tourism expenditure is based on two surveys: Questionnaire on foreigners visiting
Hungary, and Questionnaire on travel abroad. The sample, excluding lorry drivers, covers those outgoing
foreign and incoming Hungarian citizens, who cross the border, including participants in package tours.
The sample is a stratified non-probability sample. On a yearly basis, average sample sizes are ~120,000
questionnaires for foreigners and ~50,000 questionnaires for Hungarians. The selection of the days is
systematic and casual. Data collection is conducted through personal interviews. The surveyed places are
24 road border stations and Budapest-Liszt Ferenc International Airport. The survey is voluntary.
14.91. From the whole set of questions two contain information that can be related to health care,
namely the questions on purpose of visit and on expenditures. The purpose of visit, among others, can be
“medical treatment” and “spa, wellness”. In the latter should be accounted both the spending at spas
based on physician’s prescriptions, and spending for purpose of wellbeing or recreation.
14.92. However, a mechanical use of data from tourism statistics may lead to double-counting on
aggregate level, since the household survey data (the Hungarian out-of-pocket expenditure data being
based mainly on this survey) contain all the household spending no matter if it took place in the country
or abroad, even if in the household survey guidance there is not any reference to payments abroad. The
exports of health services to non-residents are more important than the imports, given the lower prices of
the Hungarian health care providers.
Country experience: United States – education-related travel
14.93. Exports of education services measure foreign students’ education expenditures in the United
States. Foreign students are defined as individuals enrolled in institutions of higher education in the
United States who are not U.S. citizens, immigrants, or refugees. Data on the number of students are
obtained from an annual survey of about 2,700 accredited U.S. institutions, conducted by the Institute for
International Education (IIE). Characteristics of the population used in the estimates include the
geographic area of origin (residence), type of institution (public or private), enrolment status (part‐time or
full‐time), and academic level of institution (2 year, 4 year, or university).
14.94. Estimates of average expenditures for tuition and for room and board are developed from annual
surveys of most U.S. accredited institutions conducted by the National Center for Education Statistics,
U.S. Department of Education, and matched by BEA to the characteristics of the student population. Data
on living expenses are based on Bureau of Labor Statistics, U.S. Department of Labor, estimates of low‐
income level family budgets in metropolitan and nonmetropolitan areas, reduced to a single person, and
adjusted for inflation each year. Estimates of foreign students’ expenditures are made by multiplying the
number of students by average expenditures.
14.95. Imports of education services measure U.S. students’ expenditures abroad. Students consist of (1)
U.S. residents who receive academic credit for study abroad from an accredited institution of higher
education in the United States and (2) students who enrol directly with foreign institutions, including

183

medical students, and receive no academic credit from U.S. institutions. The total of U.S. students’
expenditures abroad is the sum of the estimates for the two groups of students.
14.96. For students who receive academic credit from U.S. institutions, data on the number of students
are obtained from an annual survey of about 1,300 U.S. institutions conducted by the Institute of
International Education. Characteristics of the population used in the estimates include country of study,
type of institution (public or private), and academic level of institution in the United States (2 year, 4
year, or university). Data do not include students who study abroad without receiving academic credit
from a U.S. institution, or students enrolled for a degree overseas.
14.97. Student payments to U.S. colleges and universities for tuition and room and board are assumed to
be forwarded to foreign institutions. Estimates of average expenditures for tuition and room and board are
developed from an annual survey of most accredited U.S. institutions; the survey is conducted by the
National Center for Education Statistics, U.S. Department of Education. Average living expenses are
estimated by applying a ratio of U.S.‐to‐foreign living costs to the low‐income‐level family budget series
developed for foreign students who study in the United States. Estimates of U.S. students’ expenditures
abroad are made by multiplying the number of students by average expenditures for tuition and room and
board and for average living expenses.
14.98. For students who enroll directly in foreign institutions and receive no academic credit from U.S.
institutions, supplemental estimates of education payments for the United Kingdom, Australia, Canada,
and Ireland, based on national data from those countries, are used to capture U.S. students’ expenditures.
B.3.

Construction

B.3.1 Introduction
14.99. Construction covers the creation, management, renovation, repair or extension of fixed assets in
the form of buildings, land improvements of an engineering nature and other constructions such as roads,
bridges and dams. It includes related installation and assembly work, site preparation and general
construction, as well as specialized services such as painting, plumbing and demolition (see 3.132-3.135
of MSITS 2010).
14.100. Construction is valued on a gross basis, i.e. inclusive of all goods and services used as inputs to
the work (whether subcontract or not), other costs of production and the operating surplus that accrues to
the owners of the construction enterprise194 . Construction is also valued on a gross basis in the sense that
it can be disaggregated into construction abroad and construction in the compiling economy. The
construction exports (credits) would result from summing up the credit entries (i) from construction
abroad and (ii) from construction in the compiling economy. Similarly, the construction imports (debits)
would consist of debit entries (i) from construction abroad and (ii) from construction in the compiling
economy. 195 If the external operations of a construction company are substantial enough, a separate
branch, resident in the host economy, may be constituted which will usually give rise to a direct
investment relationship between the parent company and the branch.

194
195

BPM6 CG, paragraph 12.95.
BMP6 CG, Chapter 12.99, 12.101.

184

B.3.2 Compiling statistics for construction
14.101. With respect to the value of the goods and services used as inputs to the work, compilation
difficulties may arise when it is not possible to identify separately the goods purchased in the home
economy and the host economy. For practical reasons, the compiler may need to estimate a breakdown, or
otherwise attribute all goods purchased to either the host or the home economy of the construction
enterprise (see also Box 14.2). Also, it may not always be possible to identify the purchase of goods and
services separately from labor costs. In this case, the compiler will need to estimate a breakdown or,
alternatively, allocate all costs either as goods and services or as compensation of employees (for
employees that are residents of the host economy).
14.102. In order to correctly compile the item it is particularly important to precisely identify the
residence of the enterprise realizing the construction work. Indeed, a construction enterprise established in
one economy may undertake construction projects in another economy directly (no local entity created) or
through a branch, i.e. via a direct investment relationship. In the former case the construction activities are
regarded as export of services, whereas in the other case they are considered direct investment operations.
14.103. In the case of long-term projects where no local entity needs to be created, MSITS 2010 and
BPM6 suggest a number of factors to identify if a unit of one economy which is active in another has
operations substantial enough to consider that this unit has a branch resident in this other economy (see
MSITS 2010 paragraph 3.142). The operations therefore need to be analysed to identify if they are
substantial. In particular, compilers should avoid creating too many notional units.
14.104. First of all, the compiler should identify the duration of the construction project. If the project
extends over a period of at least one year, then it is a strong indication that the compiler should investigate
further on the project. If certain other factors are met, the construction work undertaken is to be treated as
if a separate institutional unit – a branch (subsidiary) – is created, which is resident in the economy where
the activity is being carried out. This branch would be considered as a direct investment enterprise. If
some or all of these factors are not met, the activity is to be treated as an export by the construction
enterprise. The other suggested criteria are:
i.

The maintenance of a complete and separate set of accounts for the activity (i.e., income
statement, balance sheet, transactions with the parent company, etc.);

ii.

The activity being subject to tax in the host country;

iii.

The existence of a substantial physical presence;

iv.

The receipt of funds for its work for its own account, etc.

14.105. The decision is based on the weight of the evidence for a set of criteria and not on any single
criterion; for example it would be very difficult to identify a branch if for the construction activity a
separate set of accounts cannot be prepared or maintained. Construction activities involving major
projects (bridges, dams, power stations, etc.) that are carried out through unincorporated site offices, in
many cases meet the criteria of a direct investment enterprise and thus are treated as part of the production
of the host economy, not as an export of services to that economy.196
196

Ibid., paragraphs 12.92-12.94.

185

14.106. In the case of construction projects conducted in the context of aid programmes, the treatment
should be to record a transaction under services if relevant (i.e. if project is small scale etc. using the
criteria defined in the previous paragraphs) between the economy where the aid agency financing the
project is located (credits) and the economy where the construction is taking place (debits). The
counterpart entry should be in the capital account. If the contractor is not a resident of the economy of the
donor, then there would also be some (construction) transactions to record between the economy of the
donor and the economy where the contractor is conducting its operations from (using a gross
recording).197
14.107. To compile statistics on construction, both ITRS and surveys can be used as sources of
information regarding construction. However, surveys may provide more detailed and relevant data than
the ITRS, in particular when it comes to gathering information on construction abroad debit and
construction in the compiling economy credit. The complexity of cross-border construction activities is
also another reason to prefer a survey, in particular if the compiling economy has many international
construction projects (either exports or imports). If countries should decide to use ITRS data, particular
care should be taken to measure transactions involving bank accounts of construction companies in the
host economy, because some of these companies may meet the criteria for treatment as residents.198

197

Such a treatment is not limited to construction.
A survey collecting data on construction could also jointly collect information on foreign direct investment and
other resident/non-resident transactions BPM6 CG, table 12.4, p.205.
198

186

Box 14.2
Numerical example of measurement of construction
A construction enterprise resident in economy A starts a construction project in economy B on the 1st of February.
The end date of the project is the 10th of April of the same year. The gross construction value is 100,000. The
project is considered as a construction service, as it lasts less than one year (69 days). The enterprise is requested to
report the construction project in the questionnaires related to both the first and the second quarter, specifying the
project start and end dates, the counterpart country and the gross value of the construction.
In order to undertake the construction project the enterprise purchases inputs (materials, services and labour) during
the first quarter. The purchases are reported as follows:



Goods purchased in Italy: 20,000;
Goods, services and labour purchased/acquired abroad: 50,000.

The gross construction value pertaining to Q1, to be allocated in A’s BOP as constructions abroad - export with
counterpart country B, is computed as follows:
(100,000 / 69 total days * 59 days in the quarter) = 85,507
The reported goods, services and labour purchased/acquired abroad are allocated as constructions abroad - import
with counterpart country B. The reported goods purchased in Italy are deducted from the goods exports, again with
partner country B, of the BOP. The complete recording for Q1 is shown in the table that follows:
Q1 BOP
Export
85,507

Constructions abroad

Import
50,000

-20,000

Goods (adjustment)

In Q2 the enterprise does not purchase any input, neither in Italy nor abroad, thus, it has only to report the
construction start and end dates, the counterpart country and the gross construction value. The construction value
pertaining to Q2, to be allocated the BOP as construction abroad - export with counterpart country B, is computed as
follows:
(100,000 / 69 total days * 10 days in the quarter) = 14,493.
Hence, the BOP recording forQ2 is the following:
Q2 BOP

Constructions abroad

Export
14,493

187

Import

Country experience: Italy
14.108. In Italy’s system, the construction activity is either regarded as an FDI-related operation, if the
construction work extends over a period of at least one year, or regarded as a service transaction if the
construction work takes less time. In Italy the gross construction value of projects lasting one year or
more largely exceeds that of short-term works, both for constructions abroad (98% vis-à-vis 2% as an
average in the period 2008-2012) and for constructions in Italy (80% vis-à-vis 20%).
14.109. Constructions lasting less than one year. The quarterly non-financial transactions questionnaire
(TTN questionnaire) collects specific information needed to compile the construction item. In particular,
only for constructions abroad, firms are required to report the following transactions, in relation to the
reference quarter (see Chapter 6 Section C.8., paragraphs 6.78-6.87 for data collection of construction):
i.

Goods, services and labour purchased/acquired abroad, used to compile the construction abroad
debits. According to MSITS 2010, the goods and services acquired by the construction enterprise
from third economies should be recorded under the appropriate general merchandise or services
item. However, in order to reduce the reporting burden, firms are not required to split the inputs
purchased in the host economy from those acquired in third countries; as a consequence the
inputs purchased abroad are all allocated to construction abroad debits and to the host economy as
the partner country. Moreover, another approximation is that also labour costs are included in
construction abroad debits, as it proved not feasible for firms to separately identify this cost
component.199

ii. Goods purchased in Italy, used to adjust the BOP goods item, by deducting the corresponding
amounts from merchandise exports. This is necessary in order to avoid duplications, as the goods
purchased in Italy by the resident construction enterprise are also recorded in merchandise
exports, based on foreign trade statistics.200 In principle, also the services acquired by the resident
construction enterprise from residents of the home economy should be excluded; however, in
order to simplify the reporting, it is assumed that all services are acquired from the host economy.
14.110. Table 14.10 shows the different sources used for each component of construction services. The
construction abroad as exports, and the construction in Italy, as imports are computed assuming that the
gross construction value is uniformly distributed throughout the duration of the work.
14.111. For constructions in Italy, detailed information is usually not available to the reporting enterprises
receiving the service. In order to fill this information gap, an assumption is made that the cost structures
of constructions abroad and in Italy are similar. Therefore, the missing information (i.e. the goods,
services and labour purchased/acquired in Italy and the goods purchased in the country of residence of the
construction firm) is estimated on the basis of the ratio of the symmetric items reported for constructions
abroad to the gross construction value.

199
200

Ibid., paragraph 3.141.
Ibid., paragraph 3.140, footnote 14, and BPM6, paragraph 10.22 (d).

188

Table 14.10 Methodology for the compilation of the construction services item in Italy
Export
Import
Total contract value
Goods, services and labour
Constructions abroad
pertaining to the reference purchased/acquired abroad
quarter
Goods, services and labour
Total contract value
Constructions in Italy
purchased/acquired in Italy pertaining to the reference
(estimated)
quarter
14.112. Constructions lasting one year or more. Also in the case of constructions lasting one year or more
the gross construction value is uniformly distributed throughout the entire duration of the work. The quota
pertaining to the reference quarter is considered an increase of FDI equity stock (FDI abroad - or assets in the case of construction abroad and FDI in the reporting economy - or liabilities - in the case of
construction in the compiling country).
14.113. For constructions abroad, in the quarter in which the construction project ends, the reporting
agent has to report the additional information on the net margin, i.e. the difference between the gross
construction value and all costs incurred by the construction enterprise in relation to the project. For
constructions in Italy, the net margin, which is not directly available to the reporting agent, is estimated
by the compiler applying to the gross construction value the average rate of return observed for
constructions abroad.
14.114. Consequently, in BoP for the quarter in which the construction ends: (i) the gross construction
value is recorded as FDI disinvestment (on the assets side in the case of constructions abroad, on the
liabilities side in the case of constructions in Italy), since the construction is delivered to the client; (ii) the
net margin is recorded as FDI income (on the credit side in the case of constructions abroad, on the debit
side in the case of constructions in Italy).
B.4.
B.4.a.

Insurance, pension and financial services
Introduction

14.115. This sub-section describes the compilation of insurance and pension services as well as financial
services, maintaining the consistency with the recommendations of the BPM6 CG 201 . Some of those
services are provided without explicit fees, which are included in property income and other flows. Thus,
calculation of services needs certain degree of estimations or imputations, based on balance sheet and/or
profit/loss, which are often provided through supervisory reports. Other services, in particular auxiliary
services, are provided with explicit fees.
14.116. According to the MSITS 2010, insurance and pension services are disaggregated into four
separate sub-components: direct insurance; reinsurance; auxiliary insurance 202 ; and pension and

201

The BPM6 CG provides recommendations on the treatment of insurance, pension and financial services, in
particular, in its Appendix 2 “Insurance Transactions and Positions, Pension and Standardized Guarantees Schemes”
and Appendix 3 “Financial Intermediation Services Indirectly measured”.
202
According to MSITS 2010, auxiliary insurance services are included in insurance and pension services. This fact
should not be confused with the sectoral classification of SNA 2008, whereby insurance brokers and alike are
classified under “financial auxiliaries.”

189

standardized guarantee services203 . Direct insurance is further broken down into life insurance, freight
insurance and other direct insurance. Pension and standardized guarantee services is further broken down
into pension services and standardized guarantee services.
B.4.b Insurance and pension services
B.4.b.1. introduction
14.117. The insurance and pension services reflect the output of specific industries whose calculation is
best described in the context of balance of payments recordings. BPM6 Appendix 6c provides some
background to the operation of the insurance and pension schemes and the BPM6 CG, Appendix 2
provides more details on the overall recording of transactions in the balance of payments, including the
relevant services. This section summarizes the main aspects related to the calculation and data sources for
the insurance and pension services.
B.4.b.2 Compiling insurance and pension services
14.118. The estimation methods and necessary data for various insurance and pension services, if they are
provided without explicit fees, are basically the variation204 of those for direct insurance as shown below
and in the example in box 14.3. Their insurance services are calculated based on premiums earned and
losses incurred pertaining to the accounting period:

Plus
Minus
Equals

Gross premiums earned (from direct business)
Net income from investments attributable to policyholders (premium supplements)
Estimated claims incurred (adjusted for claim volatility, if necessary)
Insurance service charges

Box 14.3
Numerical example of direct insurance services
For resident insurers with separate data on policyholders abroad:
Premiums earned from abroad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Claims payable abroad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Income attributable to policyholders . . . . . . . . . . . . . . . . . . . . 20 (premium supplements)
Insurance service charge is . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 [=100+20-95]
14.119. For life insurance and pension services, the net increase in life insurance actuarial reserves or
pension reserves (pension entitlement) should be further deducted as such increase is regarded as asset
accumulation to their policyholders.

203

The MSITS 2010 does not recognise fees paid to administrators of pension funds as a separate category, while
the 2008 SNA recommends that pension fund managers, (administrators of pension funds, which only manage the
activities of pension funds without taking ownership of the assets or liabilities), be classified as financial auxiliaries.
204
The value of the output of standardized guarantees providers can be calculated in the same manner as that of nonlife direct insurance. For reinsurance, commission payable should be further deducted from the premiums earned
and the claims adjusted for profit sharing in excess of loss reinsurance.

190

Plus
Minus
Minus
Equals

Gross premiums earned
Premium supplements
Benefits due
Net Increase in life insurance actuarial reserves
life insurance service charges

14.120. To compile the exports of insurance services, compilers can generally obtain comprehensive data
from surveying resident insurance enterprises, particularly in countries where surveys are the major
source data for TIS data collection (see Chapter 6 F (iv)).
14.121. Alternatively, if comprehensive surveys are not feasible, compilers may be able to obtain
information directly from domestic insurance corporations that would allow them calculate an
approximate insurance service in a given reporting period. Specifically, if they can collect data on the
breakdowns of premiums received from resident clients and those from non-resident clients, insurance
services provided to non-resident clients can be calculated assuming that the service-to-premium ratio is
the same between resident and non-resident clients. Such a ratio could also be used for estimating imports
of insurance services if payments of insurance premiums are captured through general enterprise surveys,
as shown in Box 14.4. Nevertheless, compilers should pay attention to possible differences between the
service-to-premium ratio of domestic insurance corporations and that of foreign insurance corporations.
Box 14.4
Numerical example of exports of insurance services
For resident insurers with separate data on policy-holders abroad for premiums only:
Total insurance services (to residents and non-resident clients combined) . . . . . . . . . . . . . .50
Total premiums . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
Of which: Premiums from residents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .120
Premiums from non-residents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .80
Estimated insurance services provided to non-residents are. . . . . . . . . 20 [= (80/200)*50]
14.122. Compilers, depending on the ITRS as a main source data, may not be able to compile a
comprehensive set of accounts in order to approximate the insurance services exports. However, from the
ITRS, compilers may obtain settlement data for insurance premiums received from abroad and insurance
claims paid to abroad. Also, compilers may also obtain settlement data for insurance premiums paid
abroad and insurance claims received from abroad to estimate insurance services imports. Insurance
services can be estimated by multiplying these data by a service-to-premium ratio (Box 14.5). Premiums
are a better indicator than claims for determining the service charge. The reason is that claims are
contingent on events incurred to trigger payments, and there may be periods without claims or with
regularly large claims. The service-to-premium ratio needs to be fixed and revised periodically by means
of checking financial reports of domestic insurance enterprises or direct inquiries to them.
Box 14.5
Numerical example of imports of insurance services
Premiums from residents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . .. . . . . . 40
Ratio of service charge to premiums (average from data on insurers abroad). . . . 25 %
Estimated insurance services from non-residents . . . . . . . . . . . . . . . . . . 10 [40*0.25]

191

14.123. The concept of accrual recording is an important aspect in calculating the insurance service as
insurance contracts may spread over different periods of time than the ones used in the compilation of
statistics. In estimating the insurance services on an accrual basis, compilers would need to differentiate
between the earned premiums and the written premiums. The estimation of the insurance service should
be based on the earned premiums, which are consistent with the accrual recording. At the time a policy is
first written, the total of the premium may be unearned, as premiums are often fully paid at the inception
of the policy. Direct written premiums are the amount charged to and actually paid over the life of a
contract by the policyholders for insurance coverage. Each day thereafter, the premium amount accrues to
the insurance unit until the end of the policy. At the end of the reporting period, the insurance corporation
assesses the premiums reserves representing the unexpired terms of the policy. The earned premiums plus
the unearned premium for a policy equals the written premium.
Figure 14.2 The relationship between earned premiums and written premiums

14.124. For insurance service, major catastrophes that may require large payments of claims are expected
to occur once in several years. When they do occur, the payments of claims in the year of the catastrophe
could exceed the value of premiums. If only the claims incurred in the accounting period are used in the
formula, the value of insurance services could be highly irregular and, in some cases, could even be
negative. Therefore, an adjustment in claims due is required to reflect a more long-term view the
functioning of the insurance sector, in line with the decision-making process in the insurance industry.
The adjustment for claims volatility shows the difference between the actual claims in the period and the
normally expected value of claims, where the expected value of claims removes the effects of claims
volatility. In periods when large values of claims are incurred, the adjustment, if it is to be made, would
be negative, while in other periods the adjustment would be positive.
14.125. Regarding the adjustment for smoothing the amounts of claims by policyholders on insurance
corporations, BPM6 (Appendix 6 c paragraph 22) proposes three methods: the expectation approach, the
accounting approach and the sum of cost plus “normal profit” approach are proposed in the.
14.126. In the expectation approach, output is calculated as: premiums plus expected premium
supplements minus expected claims. This approach replicates the ex- ante model used by insurer
corporations to set their premiums on the basis of their expectations. In accepting risk and setting
premiums, insurers consider both their expectation of loss (claims) and of income (premiums and
premium supplements). Ideally, the micro data of the insurance corporations accounts could be used in
the expectation approach for estimating the output of the insurance corporation, however this information
is seldom available to the compiling agencies. In the absence of such data, a statistical technique to
simulate this approach can be applied by using macro statistics, to smooth past data to forecast the
expected claims.
14.127. In the accounting approach, output is calculated as: actual premiums earned plus premium
supplements less adjusted claims incurred; where adjusted claims are determined by using claims due plus

192

the changes in equalization provisions and, if necessary, changes to own funds. In the sum of cost plus
“normal profit” approach, the output is calculated as the sum of costs (including intermediate costs, labor
and capital costs) plus an allowance for “normal profit”.
Country experience: United States – insurance services
14.128. The U.S. method for measuring trade in insurance services most closely conforms to that outlined
in example 4 in MSITS 2010.205 In addition to premiums minus a proxy measure of expected claims
(actual claims payable with an adjustment for claims volatility), called normal losses, and premium
supplements as outlined in table 14.11, the U.S. measure of insurance services also includes a measure of
auxiliary insurance services. The U.S. measure includes separate estimates of trade in direct insurance and
reinsurance. Although both this manual and BPM6 recommend estimating life and non-life insurance
separately, the United States treats all direct insurance as non-life insurance because U.S. cross-border
transactions in life insurance are thought to be insignificant.
14.129. The U.S. measure of trade in insurance services is compiled using data from a variety of sources.
The main source is a survey of U.S. insurance enterprises. This survey, conducted by the Bureau of
Economic Analysis (BEA), collects quarterly data on reinsurance premiums sold to and purchased from
abroad and annual data on reinsurance claims paid and received, primary insurance premiums sold and
claims paid, and auxiliary insurance services. Every 5 years, BEA conducts a benchmark survey of
insurance enterprises to collect information on c enterprises that fall below the reporting threshold on the
quarterly survey. A separate survey of U.S. businesses, also conducted by BEA, collects data on primary
insurance premiums purchased and claims received by non-insurance enterprises and additional data on
auxiliary insurance. Data on the income generated by insurance enterprises’ reserves, used to calculate
premium supplements, are from Best’s Aggregates and Averages: Property-Casualty by A.M. Best
Company.
14.130. Insurance services include three components: premiums less normal losses, called ‘risk-pooling
services’; premium supplements; and auxiliary insurance. The calculation of risk-pooling services
requires information on current premiums and a measure of what insurers expect their claims to be.
Premiums are collected on BEA surveys as outlined above. The proxy measure for insurers’ expected
claims, called normal losses, is estimated by applying a ratio based on historical premiums and claims to
current period premiums. This method assumes that insurance enterprises base current premiums on their
expectation of current period losses and that their expectation of losses is based on their loss history.
14.131. The U.S. method assumes that insurance enterprises plan for two basic types of losses, regularlyoccurring losses that occur every period and catastrophic losses that occur at infrequent intervals.
Separate estimates are made for these two types of losses. To calculate separate estimates, catastrophic
losses must be separated from regularly-occurring losses in the loss data reported by insurance enterprises.
When a catastrophe such as a major hurricane occurs, the magnitude of the related loss is estimated using
data from the survey of insurance enterprises and publically available reports from insurance enterprises
impacted by the event. Losses other than catastrophic losses are considered regularly-occurring.
14.132. Expected regularly-occurring losses are estimated by applying to current period premiums a 6year arithmetic moving average of each prior period’s ratio of regularly-occurring losses to premiums.

205

MSITS 2010, box III.7, page 61.

193

Loss data for the current period are not included in the average in order to achieve an ex ante concept of
regularly occurring losses.
14.133. Because catastrophic losses occur much less frequently than regularly-occurring losses, they are
assumed to affect loss expectations over a much longer period. To account for this, catastrophic losses are
removed from current period losses, and spread over the 20 years following their occurrence in equal
increments. Similar to regularly-occurring losses, expected catastrophic losses are estimated by applying
to current period premiums a 6-year arithmetic moving average of the ratio of each prior period’s share of
catastrophic losses to premiums. Thus, only a small fraction of catastrophic losses is factored into each
year’s calculation of expected claims.
14.134. Normal losses are the sum of expected regularly-occurring and catastrophic losses. Separate
estimates of normal losses are calculated for primary insurance and for reinsurance and for credits and
debits. For the United States, the ratio of losses to premiums is lower for primary insurance than for
reinsurance because administrative and financial intermediation services differ for these two types of
insurance. Primary insurance is more retail in nature—selling and writing a large number of individual
policies to customers—and, thus, may have higher administrative and other costs than reinsurance, which
involves fewer, larger transactions between insurance enterprises.
14.135. BEA does not directly collect information on the technical reserves of insurance enterprises on its
surveys because it is deemed too burdensome for enterprises to report these data. Due to the lack of data
on technical reserves, it is not possible for BEA to use a relationship between investment returns and
technical reserves to estimate premium supplements. As a result, BEA developed a ratio of expected
investment gains to premiums and multiplies that by current premiums to estimate premium supplements.
The ratio of investment income to premiums is from Best’s Aggregates and Averages: Property-Casualty
by A.M. Best Company. A.M. Best provides data on investment gains that are attributable to insurance
transactions, as opposed to investment gains attributable to the insurers’ own funds. The ratio is a
weighted moving average of the previous ratios of actual investment gains to premiums. In the crossborder trade data, the expected investment gains-to-premiums ratio is estimated separately for primary
insurance and reinsurance, in recognition of the fact that reinsurers may have different ratios of net gains
to premiums than primary insurers. The different ratios may arise because reinsurers hold larger reserves
than primary insurers or because reinsurers hold reserves for a longer period of time.
14.136. Once these ratios have been calculated, they are applied to the estimates of premium receipts for
direct insurance and reinsurance obtained from BEA surveys to derive premium supplement receipts.
Because similar data on investment income of foreign insurance enterprises are not available for
payments, the ratio used for receipts is applied to the estimates of premium payments in order to estimate
premium supplement payments.
14.137. Auxiliary insurance services cover items such as agents’ commissions, actuarial services,
insurance brokering and agency services, and salvage administration services. Data are collected on BEA
surveys. Auxiliary insurance is a component of primary insurance; there are no auxiliary services
associated with reinsurance.

194

Table 14.11 Numerical Example of Measuring Insurance Services in United States
1 Primary insurance receipts (line 2
plus line 3 plus
2
Risk-pooling services (line 5
minus line 6)
3
Premium supplements (line 5
times line 9)
4
Services auxiliary to insurance (from survey)
Primary insurance supplementary information
5
Premiums received (from survey)
6
Normal losses (line 5
times line 8)
7
Actual losses paid (from survey)
8

Normal loss ratio

line 4)

1
2

9
10

Premium supplement ratio
Catastrophic losses paid (from survey and public
Non-services entries in balance of payments account

Current transfers (line 6 minus line 7 plus line 10

Capital transfers (line 10 with sign reversed)
Income payments (line 3 with sign reversed)

reports)
s
)

2010
1,090
825
165
100

2011
1,575
1,200
260
11 5

2012
1,350
1,015
210
125

1,500
675
500

2,000
800
3,000

1,750
735
800

0.45

0.40

0.42

0.11
0
175

-165

Based on a 6-year moving average of premiums and l

Based on a moving average of investment gains to p

remiums, see paragraph 14.63

Negative sign denotes outflow from U.S. insurance

company to foreign policy holder.

0.13
1,750
-450
-1,750
-260

0.12
0
-65
0
-210

osses, see paragraphs 14.57-14.61.

Country example: Japan – recording insurance services on an accrual basis
14.138. Since Japan employs an ITRS as a major data source, recording insurance services on an accrual
basis has been challenging. To deal with this problem, Bank of Japan (BOJ) adjusts ITRS data in
implementing a method that most closely conforms to that outlined in example 3 in the box III. 7 of
MSITS 2010; the method measures insurance services by multiplying actual premiums earned by the
insurance service ratio.
14.139. The ITRS captures premiums actually paid rather than actual premiums earned that accrue to the
accounting period. Thus, data are adjusted under the following assumption. The BOJ assumes that
insurance premiums payable cover 12 months risks, and as insurance premiums are generally paid at an
inception of a policy, actual premiums earned are calculated by equally allocating premiums in
subsequent 12 months (figure 14.3). This adjustment is made for other direct insurance and reinsurance.206
14.140. The insurance service ratio for other direct insurance and reinsurance, which is the ratio of service
charges to gross premiums earned, are estimated from resident nonlife insurance enterprises’ financial
statement and applied to both imports and exports. Operating and administrative expenses are regarded as
costs for providing insurance services, and service ratio is estimated by dividing aggregated operating and
administrative expenses by corresponding premiums. This ratio is fixed for a year, and is updated when
new financial statements become available. The insurance service ratio for freight insurance is captured
separately and obtained from resident insurance enterprises.

206

Since freight insurance policies cover a single contract, which is usually a short-term, the same adjustment is not
applied to freight insurance premiums. Life insurance and pension services are excluded from estimating services, as
cross border life insurance and annuity contracts are negligible in Japan.

195

Figure 14.3 Estimation of actual premiums earned
Premium
actually payable
Y Jan.

120

Feb.

120

Mar.

240

Apr.

120

May

120

June

360

July

120

Aug.

120

Sept.

120

Oct.

120

Nov.

120

Dec.

480

Y
Jan.
10

Feb. Mar.

Apr. May June

July

Aug. Sept.

Oct. Nov. Dec.

Y+1
Jan.

Feb. Mar.

Apr.

May June

July

Aug. Sept.

Oct. Nov. Dec.

Actual Premiums Earned
forDecember, Y: 180

Premiums Actually Payable
in December, Y: 480

Equally allocated to subsequent 12 months <40 premi

ums per months from Dec., Y to Nov., Y+1>

14.141. Multiplying the service ratio and monthly actual premiums earned, 207 insurance services are
calculated. Premium supplements are regarded as zero, as retained incomes for nonlife insurance reserves
are negligible. Claims are recorded when paid, under secondary income for normal claims and under
capital transfer for high claims when catastrophic events occurred.
Box 14.6
Numerical example of actual premiums received and c laims paid to non-residents

Actual premiums received for December, Y

180

Insurance service ratio for year Y

40%

Claim paid to non-resident in December, Y

Insurance service (export)

72 (=180*40%)

14.142. This method is likely to stabilize outputs. Compared with a simple cash-based method, which is
premiums minus claims, volatility of the outputs is substantially reduced. More importantly, negative
figures, which are mainly due to lumped claims, are avoided.

207

Freight insurance premiums receivable on both exported and imported goods are obtained through direct report
from resident nonlife insurance companies. The corresponding values for imported goods are also used to derive a
premium ratio. Freight insurance premiums paid from resident to non-resident on imported goods are estimated by
excluding those earned by resident insurance companies from the total value. The total value is derived by
multiplying premium ratio on imported goods to total value of imports.

196

Country Example: Germany – smoothing of insurance premiums paid/received
14.143. Calculation of Insurance imports. In Germany, there is a lack of any kind of administrative data
for compiling imports of insurance services. Therefore the data available from the external sector
statistics are those to be used and are immediately available on a monthly basis. They are, however, not
detailed enough and need to be supplemented by several types of estimates:
i.
ii.
iii.
iv.

Adjustments for premiums paid to premiums payable
Estimates for premium supplements
Estimates for changes of insurance technical reserves
Estimates on the stock of insurance technical reserves.

14.144. The first step is to smooth the premiums paid, which are relatively volatile, to premiums payable
by using a 12 months moving average. The decision is related to the fact that premiums usually are paid
in advance for one year. The second step is to apply the ratio of premiums earned to service derived from
the export side to the smoothed premiums of the import side, which is in line with the BPM6 suggestions.
Figure 14.4 Premiums paid from Germany to insurance companies abroad

14.145. Premiums received but not yet earned. The technical reserves represent the amounts identified by
insurance companies to account for prepayments of premiums and claims occurred, but not yet paid.
Germany intends to use the same approach as for smoothing premiums paid to premiums payable for the
two components which contribute to the development of technical reserves, the premiums paid to
insurance companies abroad and the claims received. If the premiums paid are bigger than the premiums
payable the technical reserves increase, if it is the other way round, they decrease.
Figure 14.5 Claims received in Germany from insurance companies abroad

197

14.146. If the claims received are bigger than the claims receivable the technical reserves decrease, if it is
the other way round, they increase. The sum of the changes of technical reserves will have to be recorded
in the financial account.
Figure 14.6 Development of technical reserves in Germany

14.147. There is still a need to estimate premium supplements as the interest on technical reserves. There
are two ways to make this calculation, namely (1) calculation interest on the estimated stock of technical
reserves or (2) applying a ration on the premiums payable derived from the export side. Germany’s
intention is to make use of the second possibility since it is immediately available.
B.4.c. Financial services
B.4.c.1 Introduction
14.148. Financial services may be charged for explicitly or implicitly, and some transactions in financial
assets may involve both explicit and implicit charges. The most common implicit charges relate to (1)
margins on buying and selling transactions; (2) asset management costs that are deducted from property
income receivable, in the case of asset-holding entities; or (3) margins between the interest rate and the
reference rate on loans and deposits (FISIM). EBOPS 2010 identifies FISIM separately from all other
financial services.
14.149. Among financial services provided without explicit fees, margins on buying and selling are
regarded as service changed by securities dealers. In theory, margins are identifiable for each transaction
within securities dealers. Thus one way is to process such micro data if they can be collected. Another
way, which appears more realistic, is to estimate margins on an aggregate basis, for example by
multiplying their transaction volumes by spread ratio. Transaction volumes can be obtained from financial
accounts of balance of payments statistics and spread ratio can be obtained from financial market data
providers. BPM6 CG Chapter 12 paragraphs 114 – 116 provide additional detail on the data collection for
such services.
B.4.c.2 Compiling FISIM
14.150. The 2008 SNA and BPM6 recognize FISIM produced only by certain financial corporations and
only on the loan and deposit instruments on their balance sheets. Thus, FISIM exports are calculated on
the loan assets and deposit liabilities of resident financial corporations for which the counterparty is a
nonresident unit. is calculated as follows:

198

 For loans, it is the difference between interest receivable by a financial corporation and the
interest cost of funds calculated at a reference rate on the loan balance;
 For deposits, it is the difference between the interest payable at the reference rate on the
deposit balance and actual interest payable to depositor.
14.151. FISIM imports are calculated on the resident institutional sectors (mostly nonfinancial) loan
liabilities and deposit assets with nonresident financial corporations.
14.152. The 2008 SNA refers to the cost of funds at the reference rate as “SNA interest”. Determining the
reference rate is a key element in compiling FISIM. The reference rate should contain no service element
and should reflect the risk and maturity structure of deposits and loans. The rate prevailing for interbank
borrowing and lending may be a suitable choice as a reference rate. However, different reference rates
may be needed for loans and deposits in other currencies, in particular in the case of transactions with
non-resident financial institutions. The reference rate will change over time with market conditions.
14.153. For FISIM exports, the reference rate is, in principle, the cost of funds from the liability side of
resident financial corporations’ balance sheets. However, exports of FISIM on loans to nonresidents can
be estimated using the reference rate for domestically produced FISIM as interest receivable less the
product of the loan position and the (domestic) reference rate, if it can be assumed that transactions are
mostly in national currency. Exports of FISIM for deposits of nonresidents (excluding financial
corporations) can be estimated as the product of the deposit position and the domestic reference rate, less
interest payable (BPM6 CG Appendix 3, paragraph A3.11).
14.154. For FISIM imports, the reference rate is, in principle, the cost of funds from the liability sides of
nonresident financial corporations’ balance sheets by economy of residency.
Imports of FISIM for loans received from nonresidents can thus be estimated as interest payable to
nonresident financial corporations less the product of the loan position and the reference rate for the
applicable funds lent. Imports of FISIM for deposits with nonresident financial corporations can be
estimated as the product of the deposit position multiplied by the reference rate for the funds deposited,
less the deposit interest receivable from the nonresident FISIM provider (BPM6 CG Appendix 3,
paragraph A3.13).
14.155. The most comprehensive source data for exports and imports of FISIM come from (a) resident
financial corporations (surveys or administrative data collections or financial supervision authorities),
which can identify the deposits and loans with non-residents, as well as (b) surveys of selected
nonfinancial corporations, households, and NPISH, which could provide data on residents’ accounts with
financial corporations abroad to support compilation of FISIM import (see Chapter 6). FISIM imports by
general government is often derived from administrative data (see Chapter 9) or government financial
accounts. Information on the interest flows and deposit and loan positions can alternatively be sourced
from the compilation of balance of payments and the international investment position. Statistics on
interest rates and/or data on interest flows are used to calculate the reference rate as well as
deposit/lending rates.
14.156. The data sources available for economies’ own financial corporations sectors and for those of
their FISIM trading partners will tend to control the specific approach to determining the relevant
reference rates for international trade in FISIM.

199

14.157. Additional source of data difficulties may be encountered in compiling FISIM imports. Thus,
partner country data, including data sourced from international organizations (most relevant here, BIS),
may provide useful information. In this context, the domestic reference rates of the economies supplying
financial intermediation services may be used. To ensure data consistency, it would be helpful if
economies disseminate their domestic reference rates for possible use by non-resident compilers.
14.158. For economies where cross-border FISIM is small, it can be measured with relatively simplified
methods based on aggregated data. For example, FISIM for deposits with financial corporations is
calculated by deducting deposits interest flows from the value obtained by multiplying deposit stocks by
the reference rates. FISIM for loans from financial corporations is calculated by deducting the value
obtained by multiplying loan stocks by the reference rates from loan interest flows. More elaboration on
the calculation of the imports and exports of FISIM is provided in BPM6 CG Appendix 3.
14.159. In general, FISIM of financial corporations (calculated on the loan assets and deposit liabilities) is
expected to be positive. However, the calculation of FISIM of financial corporations can be negative if
the reference rate becomes higher than lending rate and/or lower then deposit rate temporarily. In this
case, for practical reasons, the compiler may wish to assume the value of FISIM is zero. A negative
FISIM may reflect erroneous calculations and inappropriate reference rates. If the negative values
continue for a long time, compilers should review the calculation of the reference rates.
Reference rate in FISIM
14.160. Given the complexities involved in estimating imports and exports of FISIM the Advisory Expert
Group (AEG) on National Accounts (at its May 2013 meeting) provided a series of recommendations
which are applicable to the international transactions. These recommendation are summarized below and
trade in services compilers are encouraged to follow them.
14.161. For estimating imports and exports of FISIM, FISIM should be calculated by at least two groups
of currencies (national and foreign currencies). The reference rate for a specific currency need not be the
same for FISIM providers, which are residents in different economies, however they are normally
expected to be relatively close. Use of partner country information or other relevant information is thus
encouraged, where national estimates are not available.
14.162. The calculation (definition) of the reference rate should be determined according to national
circumstances, using preferably any of the following approaches:
i a reference rate based on a single observable exogenous rate for a specific instrument, such as
interbank lending rates;
ii a reference rate based on a weighted average of observable exogenous rates of maturities with
different terms (weighted by the stock of loans and deposits in each maturity); or
iii a weighted average of the endogenous interest rates on loans and deposits.
14.163. Estimation of FISIM over periods of volatile movements in reference rates should be carefully
monitored, in particular if the results turn out to be negative, particularly for depositors, but also for
borrowers. In such cases, countries are encouraged to review the applicability of the underlying reference
rate for that period to calculate FISIM.

200

14.164. Research continues in the area of credit default risk related to FISIM and the merits of its
exclusion from FISIM, including for practical reasons.
Country experience: Estonia – calculating FISIM
14.165. For Estonia, external FISIM has a relatively small importance in relation to internal FISIM. This
is due to the fact that most of the banking sector belongs to foreign banks and loan resources for local
economy are made available through interbank transactions. The latter form the overwhelming majority in
external transactions of loans and deposits. Having considered the size of the external FISIM, costeffective approach and simplifications have been taken into compilation practice instead of additional
data collection with increase of burden for data providers.
14.166. The compilation process is based on the stock and interest income figures that are available
through the balance of payments surveys and banking statistics. In order to compile FISIM, the average
interbank rate is needed to calculate and apply for stock figures. FISIM is the difference between the
interest actually charged and the adjusted interest or vice versa, depending on the lender or borrower point
of view.
14.167. For stocks data, the balance of loans and deposits is needed. In case of the assets related to loans
and liabilities related to deposits, stock data are needed only for institutional sectors S.122 and S.125.
14.168. The main difficulty arises due to the fact that assets and liabilities between financial
intermediaries should be excluded from the stock data. Therefore, the counterpart data have to be taken
into account while determining the stock from which FISIM is derived. This does not concern banking
sector statistics that already have data by counterpart sectors. For other financial intermediaries (S.125),
no loan data are available by counterpart sectors or the data are too aggregate level (leasing companies).
Compilation of stocks for S.125 includes the assumption that they have loan liabilities against financial
intermediaries. On the asset side of loans at the moment, all stocks of S.125 have been taken into account,
reflecting also those with other financial intermediaries.
14.169. Another issue concerns stock data of households where the main data source is the ITRS. The use
of ITRS does not give a good indication about the positions. One of the solutions to tackle the issue is to
use mirror data, i.e. other countries banking sector statistics by counterpart sector. However, households
sector still remains on the estimation level.
14.170. The reference rate is derived by using credit institutions report on the balance of loans and
resources from which the stocks against non-residents financial intermediaries are taken by currencies and
by maturities. Each stock is multiplied by the corresponding contractual interest rates from the same
reports. The amounts to be paid or received are then divided by the stocks. The result from this
compilation represents an average weighted interbank rate which is used as a reference rate.
14.171. Before multiplying stocks with the reference rate, both the stocks and unadjusted interest income
are distinguished by counterpart sectors. The process is taken on in order to exclude the stocks and
interest income between financial intermediaries. Data by counterpart sectors for both items could be
derived by using credit institutions report on the balance of loans and resources. However, the stocks and
income could not be divided for other financial intermediaries due to the lack of proper data sources.
14.172. Experiences up to the present indicate that preconditions still exist to eliminate stocks and income
between financial intermediaries properly. Therefore, data sources have to include data by currencies and
by counterpart sectors also for other financial intermediaries. For households sector further estimations

201

for stocks is needed to be developed. Estimations were needed to be added in case of negative FISIM.
Usually it happens due to huge transactions in deposits by general government sector where the stocks in
both the beginning and the end of the period were zero while at the same time interests were earned. In
order to keep a cost-effective approach to data providers, detailed sectorized data requirement in different
currencies have not been applied. Instead the weighted average reference rate itself reflects the weights of
the currencies and maturity while implementing it for stocks by maturities.
Country experience: Japan – margins on buying and selling transactions
14.173. Dealers implicitly charge services by incorporating a spread between their buying and selling
prices. Japan will estimate margins on debt security transactions by multiplying their "transaction
volumes" by corresponding "spread ratios", "Transaction volumes" and "spread ratios" are estimated as
follows.
14.174. In most cases, a resident dealer exports services through inward investment transactions.
Correspondingly, a resident investor imports services through outward investment transactions.
14.175. Therefore, Japan assumes that export of services occurs only in inward investment, and that
import of services occurs only in outward investment. Some inward (outward) investment transactions are
conducted with non-resident (resident) dealers and such transactions are excluded. Volumes (sum of
buying and selling transactions) of inward investment and outward investment by type of securities are
obtained from direct reports from financial institutions and ITRS.
14.176. A spread ratio is the difference (as a ratio) of an ask-price and a mid-price, where a mid-price is
the average of an ask-price and a bid-price. The spread ratio varies by every transaction, but there is no
perfect data source. Therefore, Japan chooses the most common products and uses their spread ratios for
approximation. For inward investment (export of service), spread ratios of Japanese Government Bonds
(JGB) are used. For outward investment (import of service), spread ratios of six major countries'
government securities that account for majority of outward portfolio investment are used and are applied
to investment in corresponding countries. As they are significantly different, spread ratios of short-term
securities and spread ratios of long-term securities are separately measured. Data are obtained from
Bloomberg.
Table 14.12 Numerical example of exports involving margins on buying and selling
Bid price for 10Y JGB
Ask price for 10Y JGB

101.675
101.695

Mid-price [(bid + ask) / 2]:
Spread [(ask – mid) / mid]:

101.685
0.01%

202

Table 14.12 continued
(Billion yen, %)
Transaction Volume (Inward Investment)
Spread on JGB
(d)

Margin
(c) x (d)

82,000

0.01%

8.2

25,000

26,800

0.01%

2.68

9,000

5,000

14,000

0.01%

1.4

16,000

2,000

18,000

0.01%

1.8

Total

─

14.08

Country
of
investor

Credit
(a)

Debit
(b)

Total
(c) = (a) + (b)

65,000

17,000

1,800

Table 14.13 Numerical example of imports involving margins on buying and selling
(Billion yen, %)
Transaction Volume (Outward Investment)

Spreads on
government
bonds
(d)

Margin
(c) x (d)

Country
of
issuer

Credit
(a)

Debit
(b)

Total
(c) = (a) + (b)

1,000

1,500

2,500

0.04%

1.0

1,500

2,700

4,200

0.05%

2.1

7,000

1,800

8,800

0.03%

2.64

9,000

2,000

11,000

0.01%

1.1

Total

─

6.84

Country experience: Luxembourg ̶ E-commerce (merchandising) services
14.177. Luxembourg hosts an increasing number of subsidiaries of international groups208 that provide
worldwide e-commerce services and are active in well diversified domains (distribution of music, audio
books, and audio-visual products, online trading platform services, operating of websites for the purpose
of selling, auctioning, renting or otherwise distributing products and services, etc.). Thus, “Service
Merchandising” has become one of the important components of telecommunications, computer and
information services, audiovisual and related services, royalties or other business services.
14.178. Given the importance of international transactions of these subsidiaries, Luxembourg’s aim was
to include all of them in its monthly BOP reporting for services. While there are a few international
groups that operate through different local operational entities, in most cases, the international groups act

208

By combining the advantages of both European directives (2002 / 38 / CE and 77 / 388 / EEC) from July 1st,
2003, the subsidiaries from international groups providing e-commerce services were allowed to apply t he
Luxembourg rate of 15 % VAT on services provided, and not the higher VAT rates applicable in the country of
residence of European clients.

203

through a single operational legal entity with a low physical presence through which services are provided
to a multitude of clients, especially inside the European Union, whereas expenses (i.e., imports) are often
intercompany retrocessions in different form of services (including royalties), both within and outside the
European Union. Therefore, while the gross flows may be sizeable, in many cases, on a net basis, the
international trade in services transactions nearly cancel out because the overall accounting result of the
resident legal entities is often around zero, due to the intercompany retrocessions. Thus, because the value
of “Merchandising services” exported from and imported to Luxembourg is recorded on a gross basis, the
resulting values may seem inflated, given the fact that there is a low physical presence of the international
groups in Luxembourg. This treatment is applicable because most of the firms hosted in Luxembourg
trade services that are classified to the appropriate specific service classification (such as
telecommunications, computer and information services, audio-visual and related services, charges for the
use of intellectual property n.i.e. or other business services), and if the firms act as an agent on a
commission basis, then only the commission is recorded as the service provided.
14.179. Luxembourg covers the external transactions through its survey and completes monthly figures
with an estimate for FDI income.
B5.

Services related to intellectual property products

B.5.a. Introduction
14.180. This sub-section discusses the compilation of services related to intellectual property products
(IPP), and covers EBOPS category Use of intellectual property n.i.e.; as well as parts of other main
EBOPS categories including telecommunication, computer, and information services; other business
services, and personal, cultural, and recreational services. This Guide combines these components in this
sub-section because of the partial overlap and interrelationships between the categories.
14.181. Services derived from IPP cover following items in the 2010 Extended Balance of Payments
Services Classification (EBOPS 2010):






Charges for the use of intellectual property n.i.e. (main EBOPS category)
Computer and information services (included in telecommunication, computer, and
information services)
Research and development services (included in other business services)
Architectural, engineering, scientific and other technical services (included in other
business services)
Audio-visual and related services (included in personal, cultural, and recreational
services)

14.182. Recent decades have seen an explosion in transactions related to these categories. Many of the
transactions relate to the use of an underlying produced asset (typically research and development,
software, databases and audiovisual originals). However these categories present significant measurement
challenges. Firstly, because the distinction between the categories from a reporter’s perspective may not
always be clear (for example software versus research and development, or software versus audiovisual).
And, secondly, because intra-firm transactions may be affected by transfer pricing phenomena or indeed
more general tax planning issues, meaning that the distinction between flows recorded as trade in services
and flows recorded in the primary income account of the Balance of Payments as property income may
not always be clear cut.

204

14.183. This section provides a detailed description of each of the service categories derived from IPP. It
focuses on the particular challenges concerning flows relating to IPP recorded as assets in the system of
national accounts. Special mention is also made for the treatment of transactions related to franchise and
trademark licence fees. These fees usually also reflect payments for the use of marketing assets, which are
considered as non-produced assets. As a consequence, in theory transactions should be recorded in the
primary income account in the balance of payments. However given that the payments also includes a
service element and if it is not possible to make the distinction with the income element, then by
convention the full value is recorded under franchise and trademark licence fees.
14.184. Despite the guidance offered below, it is important to note that it is unlikely that compilers will
ever be able to fully account for transactions in intellectual property in a way that is consistent with the
underlying principles of economic ownership. However, it is at least possible to make progress on
achieving more coherence: both internationally, by encouraging compilers to engage in asymmetry
reconciliation exercises, and nationally, by improving the coherence of capital stock and productivity
estimates. Ultimately, pragmatic approaches to measurement are necessarily advocated here.
Improvements in a few key areas could significantly improve quality.
B.5.b Use of intellectual property n.i.e.
14.185. Charges for the use of intellectual property n.i.e. are defined in paragraph 3.214 of MSITS 2010,
and include franchises and trademarks licensing fees, licences for the use of outcomes of research and
development, licences to reproduce and/or distribute computer software, licences to reproduce and/or
distribute audio-visual products and licences to reproduce and/or distribute other products.
14.186. Franchises and trademark licensing fees are in theory related to charges for the use of nonproduced assets (Marketing Assets). As such and, if one were to follow the underlying principles of the
National Accounts, payments would not be recorded in production accounts of the National Accounts nor
in the goods and services account of the Balance of Payments. Rather the payments correspond to
property income. However, often, the payments are bundled with additional service items that make it
difficult to disentangle the pure payments for the use of the underlying brand and, as such, both the
2008SNA and BPM6 recommend that such payments are recorded as payments for services. Nevertheless
outright purchases of the entire brand (the Marketing Asset) or indeed rights to use the brand in certain
regions, such that the licence has the characteristics of a licence to reproduce, should not be recorded in
the goods and services account and should instead be recorded in the Capital Account.
14.187. Licences for the use of outcomes of R&D include payments for licences to use and licences to
reproduce. The latter reflects a transaction in a pre-existing produced asset in the SNA corresponding to
negative Gross Fixed Capital Formation (GFCF) for the unit selling the licence and positive GFCF for the
acquiring unit.
14.188. Care should be taken to differentiate between payments included in this category from those
included in EBOPS category 10.1, which includes payments for customized (made to order) R&D as well
as acquisitions of entire R&D originals. Differentiating between sales/purchases of licenses to reproduce
and sales/purchases of entire originals may not always be trivial but the latter will usually, at least for
transactions between unaffiliated parties, be accompanied by sales/purchases of legal instruments, such as
a patent or copyright.
14.189. Since the 1993 SNA, when software was recognized as a produced asset for the first time,
significant improvements have been made in measuring software and software related transactions, and
the proposed changes for more details in the latest EBOPS classification system should consolidate those

205

advancements. The inclusion of a separate category Licences to reproduce and/or distribute computer
software, which specifically relates to licences to reproduce and not licences to use (which are included in
EBOPS category 9.2) was requested by the National Accounts community to assist in efforts to estimate
GFCF in software. Licences to reproduce (with a contract period of more than one year) are recorded as
GFCF by the purchaser and negative GFCF by the seller.
14.190. Similar improvements have been made with respect to other intellectual property products such as
databases, audio-visual, literary and other artistic originals. Licences to reproduce and/or distribute
audio-visual products and Licences to reproduce and/or distribute other products refer only to licences to
reproduce and not licences to use.
B.5.c. Computer services
14.191. MSITS 2010, in paragraph 3.224, defines Computer services as covering “hardware- and
software-related services and data processing services”. Compilers shout take particular care with respect
to the coverage of computer software, in particular its distinction with licences to reproduce and/or
distribute computer software. (see also paragraph 3.225 of MSITS 2010).
14.192. The key difference vis-à-vis licences to reproduce concerns the recording of flows in the national
accounts. Whereas licences to reproduce (of more than one year) are not considered as reflecting new
output, transactions in computer software, with the potential exception of software originals (which may
have been produced in prior periods) are. This item does not include the value of software embodied in
other products, (such as a computer) and sold as a bundle where the software component and value is not
separately invoiced. Moreover, with the exception of sales to households and expenditures on licences
with one year or less, all domestic expenditures in this category are typically recorded as GFCF in the
national accounts.
14.193. Also of note is the inclusion of software provided on a physical storage device with a periodic
licence fee as transactions in services, whereas software acquired on physical media with a one-off
payment are recorded as transactions in goods.
14.194. It is clear that the blurring of the lines between a good and a service, magnified by the difficulties
that may exist in differentiating between the two categories in practice, will complicate measurement. A
complementary grouping (C.3) in EBOPS is included for Computer software transactions, which includes
all related transactions whether embodied on physical media or not (see below). An additional grouping
C.3.1 is included to separately identify those transactions that concern licences (explicit or otherwise)
whose duration is for more than one year, which are recorded as GFCF when acquired by producers.
14.195. Other computer services (EBOPS 9.2.2) include all other related transactions that are generally
not recognised as investment in the national accounts. The detailed coverage of other computer services is
given in paragraph 3.230 of MSITS 2010.
B.5.d Information services
14.196. MSITS 2010 divides information services into news agency services and other information
services. These are defined in paragraph 3.232 of MSITS 2010. One particular issue worthy of mention
here concerns database related transactions. The use and outright purchase or sale of originals and copies
of databases are included in this category. However some care is needed to ensure that software
development costs in developing databases are not included in this category but in Computer software.

206

B.5.e Research and development services
14.197. Research and development services are defined in paragraph 3.234 of MSITS 2010. EBOPS 2010
recommends a breakdown of research and development services into two subgroupings: work undertaken
on a systematic basis to increase the stock of knowledge (reflecting the coverage of research and
development within a 2008 SNA context) and other research and development services.
14.198. It is important to note that the categories above within provision of customised and noncustomised R&D services 10.1.1.1 primarily relate to ‘new’ research and development, such as newly
produced customized software, or transactions in ‘originals’, where ownership and the concomitant rights
are transferred to the purchaser. Care is needed in this context to differentiate between licences to
reproduce and originals. Most, if not all, of these expenditures will be recorded as investment in the
accounts of the importing country. Note too that not all acquisitions necessarily need reflect patented,
copyrighted or other protected forms of R&D. In many cases, companies may deliberately opt not to
patent the outcomes of (some of their) R&D . This may be motivated by a desire to preserve secrecy, the
costs of patenting may in some instances be perceived as outweighing the benefits, or the R&D may not
be a patentable subject matter.
14.199. Other research and development services primarily reflect other R&D related expenditures that
are not expected to add to the stock of knowledge as defined in the OECD Frascati Manual and
recognised in the SNA, and so, will not be treated as investment in the national accounts.
14.200. There are other borderline issues to consider however. These chiefly concern design originals and
the potential overlap with architectural, engineering, scientific and other technical services, covered
below.
B.5.f. Audio-visual and related services
14.201. Audio-visual and related services relates to the production of motion pictures (on film, videotape,
or disk or transmitted electronically), radio and television programmes (live or on tape) and musical
recordings (see 3.256 of MSITS 2010). As noted above care should be taken to ensure that licences to
reproduce and/or distribute audio-visual products are not included in this category. There may
however be borderline issues concerning the rights acquired by transmitters such as TV companies and
radio stations, when the contractual arrangements allow for multiple transmissions. Where the fees are
paid on a ‘pay as you go’ basis, for example a royalty payment is made every time a song is played on a
radio station, these should be recorded under 11.1.1 (audiovisual). If however a one-off fee is paid that
provides for unlimited air-time it should be recorded as being equivalent to a licence to reproduce
(intellectual property).
14.202. Performing arts and other live entertainment event presentation and promotion services (namely,
live performances such as concerts and plays) are excluded from audiovisual services and are instead
included in artistic related services. The transaction corresponding to the performance of resident actors,
musicians or other artists for the shooting of a movie (or other types of visual programs) or musical
recording by a non-resident entity (and vice-versa) is also included under artistic services, under the
condition that artists are not in an employer-employee relationship with the recording entity. However,
the subsequent transactions for the result of the recording will be included in audiovisual services. If for
the recording the services of an independent recording studio or similar services are outsourced, then
transactions will be included under audiovisual services if between residents and non-residents (and there
is no employer-employee relationship if these correspond to the services of an independent service
provider). Also included in audiovisual services are amounts receivable or payable for rentals of

207

audiovisual and related products and charges for access to encrypted television channels (such as those
offering cable and satellite services).
14.203. Like software, a complementary grouping is provided for audiovisual transactions in recognition
of the grey borderline between transactions in goods and services. In the same way a complementary
grouping is included to account for those transactions that reflect investment expenditure in the national
accounts (i.e., licences for more than one year), when undertaken by producers.
14.204. For artistic related services, care should be taken to ensure that transactions are only included if
the service providers are not employees of the entity making payments (otherwise, they should be
recorded as compensation of employees).
14.205. Sales of ownership rights of literary or other artistic originals (excluding those covered under
audio-visual), should be recorded under 11.1.2 (artistic-related services). There may be borderline issues
for rights that restrict reproductions to certain markets or languages, where transactions should be
recorded under EBOPS 8.4.2 (intellectual property).
14.206. Other borderline issues may concern transactions relating to the selling of exclusive rights before
any originals have been created (for example, the exclusivity of a publisher to publish future literary
works of an author). These types of rights should be recorded as contracts for future production, and so
payments should be included in the capital account. However at the point in time when the ‘original’ is
produced, an imputation should be recorded for either (a) the transfer of the entire ownership from the
‘author/producer’ to the ‘funder’ or (b) a licence to reproduce.
B.5.g. Compiling IPP related services
14.207. Despite some of its current shortcomings in identifying international transactions in trade in R&D
services, with the addition of supplementary questions, the Frascati209 -based survey approach provides
perhaps the best mechanism to improve the measurement of trade in R&D services. The OECD
Handbook on Deriving Capital Measures of Intellectual Property Products describes a prototype
questionnaire with a comprehensive list of questions that could inform the design of new or modified
surveys (see also below).
14.208. The key challenge for measuring R&D flows, however, concerns transactions between affiliated
parties. R&D-related transactions between affiliated enterprises are not always observable, and, often,
payments that are implicitly related to R&D are instead recorded in the primary income account of
BPM6. At the same time, flows may also be channelled through convoluted chains of affiliates, including
Special Purpose Entities, that have been set up to maximize post-tax profits for the controlling MultiNational Enterprise (MNE).
14.209. The Task Force on Global Production210 and its predecessor, the Task Force on Globalisation in
the National Accounts, have the goal of developing guidance on the issue of ensuring that flows for the
funding, performance and use of R&D align with the concepts of ‘economic ownership’ in the National
Accounts. Since the conclusions reported in the Guide to Measuring Global Production are still under

209

OECD Frascati Manual: Proposed Standard Practice for Surveys on Research and Experimental Development,
6th edition. Note that at the time of writing the Frascati Manual is being revised, in part to provide better alignment
with the National Accounts.
210
See Chapter 4 of Global Production, UNECE, forthcoming.

208

discussion in other inter-governmental bodies at the time of writing this Compilers Guide, more concrete
guidance on these issues will be provided on the on-line version of this Guide. In the Interim Guide to
Measuring Global Production,211 the Task Force proposes that the most expedient approach to ensuring an
underlying consistency between payments for the use of an asset and economic ownership would likely
be through adjustments to balance sheets and, where possible, through the recording of transfers of assets
from one affiliated party to another.
14.210. Central to this proposition is a desire to remain close to recorded and observed flows. This partly
reflects practicalities, in particular the difficulties in making corrective imputations in the absence of
complementary data, but it also reflects a desire to retain a close consistency between recorded profits and
taxes paid, even if these payments suggest departures from the principles of economic ownership.
Nevertheless, in two specific cases, the Task Force on Global Production suggests imputations to correct
for flows recorded ‘incorrectly’ as property income, where evidence is available.
14.211. The first case relates to affiliates using an underlying asset, owned by the parent (or other
affiliate), but where no explicit payment for use is recorded, and, instead, related flows are recorded as
property income. In these circumstances, the Interim Guide to Measuring Global Production212 proposes
that, in principle, an imputed payment for the use of the underlying asset should be recorded as charges
for the use of intellectual property n.i.e.213 The imputation could be based on observable flows in property
income that can be related to the underlying asset, or failing that, the imputation should be based on the
affiliate enterprise’s share of total relevant MNE output multiplied by the total income generated by the
underlying asset. However, the data requirements of such an approach are arduous as it requires not only
information on the entire MNE, but also on the income generated by the asset, or proxies, such as the
value of capital services provided by the asset. As such, more often than not, it is unlikely that such
imputations will be made in practice. Nevertheless, if compilers do make such imputations, they should
ensure that counterpart transactions (in particular international transactions) should be coherent, requiring
coordination with other statistics agencies.
14.212. The second case concerns dedicated affiliates engaged in the production of intellectual property
products for use by other affiliated parties in production but where the affiliate producing the intellectual
property does not itself use the asset in production nor does it receive any revenue through sales of licence
to use or reproduce the asset or related income. Often these units have no recorded output beyond the
production of the asset itself, which is recorded as own-account production of GFCF. In these
circumstances the Interim Guide to Measuring Global Production214 proposes that an imputation is made
such that the produced asset is transferred (exported in cases of international transactions) to the parent
company. Sources of such information are scarce but some potential sources or approaches exist.
14.213. The first source relates to firm-level data and, in particular, data collections on Activities of
MNEs and Foreign Affiliates. Units classified to the Research and Development activity (ISIC Rev 4: 72)

211

http://www.unece.org/fileadmin/DAM/stats/documents/ece/ces/bur/2014/Guide_to_Measuring_Global_Productio
n_-_CES.pdf.
212
http://www.unece.org/fileadmin/DAM/stats/documents/ece/ces/bur/2014/Guide_to_Measuring_Global_Productio
n_-_CES.pdf.
213
Since the conclusions reported in the Guide to Measuring Global Production are still under discussion in other
inter-governmental bodies at the time of writing this Compilers Guide, more concrete guidance on these issues will
be provided on the on-line version of this Guide. [Insert web address of on-line version of CG MSITS 2010].
214
http://www.unece.org/fileadmin/DAM/stats/documents/ece/ces/bur/2014/Guide_to_Measuring_Global_Productio
n_-_CES.pdf.

209

but with no identifiable output, except for own-account production of R&D originals, or with no
expectations of revenue through sales of licences (which can be indicated if historically this is also the
case) can be considered as satisfying the criteria established in the Interim Guide on Measuring Global
Production 215 for imputing unrecorded or misreported exchanges of R&D assets to their parents (see
footnote 42). In these cases, the value attributed to the asset should be equivalent to the valuation used in
estimating the own-account production (typically based on the sum of costs with an estimated mark-up
for gross operating surplus).
14.214. A second potential source is the surveys used for the collection of R&D data in the Frascati
framework. Data on external funding of business R&D are collected in a large number of countries as part
of this framework. The information includes funding from abroad by affiliated and non-affiliated
enterprises. Although business enterprises mainly perform market transactions, these flows of funding
might include donations and subsidies and not necessarily represent acquisitions of R&D, so some care is
needed in interpreting this information. Nevertheless, the source can provide a useful proxy or diagnostic
to investigate flows recorded as acquisitions of R&D originals or payments for customized R&D services
within the EBOPS category for R&D services.
Country experience: Germany ̶ Charges for the use of intellectual property
14.215. Germany collects all BOP relevant service transactions via a cut-off survey directly from
enterprises, public authorities and natural persons on a monthly basis. Thereby, a generic questionnaire
has to be used by the respondents to submit all their service transactions to the Deutsche Bundesbank, as
long as the value of a single transaction is above the reporting threshold of 12,500 Euros. Thus, the
collection of data on “Charges for the use of intellectual property” builds only one block in the frame of
the general approach. Instead of asking explicitly what kind of intellectual property has been used and
charged for in the reporting period, respondents have to report only a transaction code indicating for
which licences payments have been made/received (use or reproduction).
14.216. The list of transactions codes for filling out the forms is part of the basic law about the collection
of BOP data. However, as the list might not always be self-explaining enough for respondents, the
Deutsche Bundesbank has published in addition an explanatory note on the coding list. Like for other
BOP items, the note describes, in the case of intellectual property with over three pages and in an
understandable way, which kind of transactions should be reported under a specific code and explains
differences between the codes.
14.217. From the following excerpt of the note, it can be seen that, at the beginning of the section on
intellectual property, a small table summarizes all relevant codes falling under this category216 . Thereby,
the structure of the table follows mainly the structure given by Table III.1, Treatment of Intellectual
Property, in Chapter III, subparagraph 8 of the MSITS. Subsequently, for each category (licences,
distribution rights and purchase/sale) at first a general “definition” of the category is given, followed by
detailed remarks and examples about transactions to be reported under the respective code. All of them
are in line with the recommendations of the MSITS and the BPM6.

215

Ibid.
It becomes obvious from the table that, under this section, not only codes belonging to the item “charges for the
use of intellectual property” are listed, but also codes for transactions with intellectual property which have to be
assigned to other EBOPS items. This is done for practical reasons because respondents usually get confused if
transactions in intellectual property are spread over various service items.
216

210

Figure 14.7 Treatment of Intellectual Property

14.218. Currently the reports of the reporting agents are the only source used to compile the item
“Charges for the use of intellectual property”. Besides the transaction code, the respondent has to provide
for each transaction (credit and/or debit) the partner country and the amount which has been paid.
14.219. The incoming reports are checked by staff members for plausibility and completeness, mainly by
cross-checking them with the reports of former periods. In case of doubts about the correctness,
declarants are contacted to clarify the doubted transaction. In the event that an open request could not be
217
answered in due time by the respondent, imputations are made at least for important reporters. The
monthly values for the total and each partner country are then calculated by aggregating the transactions
codes belonging to the item “Charges for the use of intellectual property” of all individual reports. Codes
for transactions with intellectual property belonging to other EBOPS items, for example the use of
software (code 613 in the table above), are separated automatically and assigned to the appropriate item
like software services.

217

Usually they are replaced by the confirmed or corrected transaction in the following month.

211

B.6.

Education and health services

14.220. For the compilation of data on resident/non-resident transactions in education or health services,
additional sources of information from administrative sources (e.g., the Ministry of Education and the
Ministry of Health), specialized entities, or other third party sources may be necessary or other types of
sources. In particular, on the receipts side, relevant ministries usually gather such information (i.e., for
mode 2, but also other modes) or would have an interest in doing so. Another option is to collect data
from all higher education institutions and universities operating in the country with the aim of collecting
data on foreign students studying in the compiling economy. Other questions relevant to mode 2 could be
also added to the survey questionnaires. Data from health insurance companies could be used and
combined with administrative sources and information from travel surveys for further analysis on health
services.
14.221. Embassies and consulates may also hold information for both exports and imports of such
services. For health services, it is advised that compilers use administrative data from health and social
insurance programs. Other breakdowns could also be encouraged for sectors of particular interest to an
economy; e.g., cultural services or leisure.218
14.222. Given the importance, for example, of increasing demand for information on health-related or
cultural services, and the complexity of some existing questionnaires, other methods could be envisaged
to compile more details on such services, such as combining household surveys with administrative data,
or health insurance and credit card data expenditures to health care providers, as this information could be
extracted by the appropriate merchant code categories.219
14.223. Some information on health services could be derived from the travel purpose variables used for
education-related travel, by including a breakdown for “Study and courses” and “Medical treatment”.
Additional information could be derived from linking the aforementioned sources with administrative
data. such as VAT data (from the National Tax and Customs Administration).
B.7.

Government goods and services n.i.e.

B.7.a. Introduction
14.224. While services supplied by and to governments should be classified to the relevant service
category (business services, health, etc.), if possible, services related to government functions that cannot
be classified to another specific service category are classified as government services. Transactions
covered by government goods and services n.i.e. include trade of goods and services between the
government unit and the territory in which it is physically located; i.e., as government and international
organization enclaves are not residents of the territory in which they are physically located, their
transactions with residents of the territory of location are international transactions. Other official entities

218

For example, given Australia's interest in the exports of education services, the Australian Bureau of Statistics
compiles a breakdown of education travel data by type of expenditure (i.e., education fees and other expenditures)
and type of institution.
219
For example, Austria is deriving information on health-related services which residents consume in Hungary as a
neighboring country from a household survey, together with administrative information from the project “health
region – Regional Network for the Improvement of Healthcare Services,” which is run by the European Union.
Both relevant information sources are combined to establish sound estimations.

212

are also considered government transactors, such as aid missions, government tourism information and
trade promotion offices, and international organizations. Other examples of such transactions include
charges for visas; payments for police-type services; technical assistance under certain circumstances (see
paragraph 14.115); and government supply of a licence or permit that is classified as provision of a
service (see paragraph 14.116).
14.225. MSITS 2010 recommends that government goods and services n.i.e. be further classified
according to the following breakdown based on the transactor: embassies and consulates; military units
and agencies and other government goods and services n.i.e.. Administrative costs incurred in the donor
economy as a result of providing technical assistance or aid should be included under the specific services
provided. Technical assistance provided by a government or an international organization is classified
under government goods and services only when not specified to a specific service and if the technical
assistance personnel are employed by the donor government or an international organization.
14.226. If the issue of government licenses involves little or no work on the part of the government, with
the licenses being granted automatically on payment, it is likely that they are simply a device through
which to raise taxes and should therefore not be considered a service. For example, by convention,
amounts payable by households for licences to own or use vehicles, boats or aircraft, along with licences
for recreational hunting, shooting or fishing, are treated as taxes. On the other hand, cases in which the
licensing is used to check the competence or qualifications of the person concerned, to check the efficient
and safe functioning of equipment, or to carry out some other form of control, the payments made should
be treated as purchases of services from the government, unless the payments are clearly out of all
proportion to the costs of providing the services.220
14.227. All expenditure on goods and services by diplomats, consular staff, and military personnel, and
dependent members of the same household, in the economies in which they are located is also included in
government goods and services n.i.e. However, the expenditure of locally engaged staff of embassies,
military bases and so forth, and international organization staff is not included in government goods and
services n.i.e. Moreover, all staff of international institutions staying in the host economies for 12 months
or more should be regarded as residents of those host economies and their expenditures are therefore not
included.221
B.7.b Compiling government goods and services n.i.e.
14.228. As discussed in Chapter 11, the majority of government services transactions are most commonly
compiled using administrative records. Data on government expenditure abroad should also be available
from an ITRS. 222 Moreover, estimates of expenditures of diplomats and other government personnel
posted abroad could be based on the wages paid to these persons—details of which should be available
from administrative records—and an assumption about the percentage of wages spent on such
expenditures.223 It may be more difficult to capture expenditures by foreign governments and international
institutions located in the compiling economy using an ITRS, in which case an enterprise survey of

220

MSITS 2010, paragraphs 3.277 – 3.279.
On the other hand, staff of international institutions staying in host economies for less than 12 months should be
regarded as residents of the economies in which they maintain permanent households—typically their economies of
origin. BPM6 CG, chapter 12, fn. 37.
222
BPM6 CG, paragraph 12.150.
223
BPM6 CG, chapter 12, paragraph 12.150.
221

213

nonresident bank accounts or a survey of foreign embassies and international institutions could be used.
When comprehensive data collection is difficult, existing replies to past surveys could be used as
representative values.
14.229. For provision and receipt of aid, compilers of the donor country could obtain information on the
costs and type of provided services from official sources. In the recipient country, compilers could obtain
information from the embassy of the donor country or the relevant domestic ministry or agency. An ITRS
can also provide information on several related transactions (e.g., current transfers to government
received through the banking system) or from customs (data on imports of materials and equipment). An
alternative source is OECD’s Official Development Assistance (ODA) records.224
14.230. If data on government expenditure abroad are not timely, or source data only provide broad
aggregates or partial data, it may be necessary to extrapolate certain series or create data models based on
past survey data or historical trends. In such cases, government expenditure policies, budget planning and
decisions, and observed statistical relationships among historical indicators should be considered.
Country experiences: Denmark
14.231. In Denmark, information for the expenditure side of government goods and services n.i.e. is
provided by the Ministry of Foreign Affairs and the Ministry of Defence. Data on officials’ (ambassadors,
military personnel, etc.) personal expenditure in the host country in which they are located is based on
their wages. Importantly, Statistics Denmark assumes that these government officials spend 50% of their
salary on personal expenditures in the host country in which they are located. The data for military units,
however, is based purely on expenditure data rather than data on wages.
14.232. The revenue side of government goods and services n.i.e. is compiled by combining information
from different sources. The Ministry of Foreign Affairs provides lists of embassies and international
organizations located in Denmark, which Statistics Denmark contacts to obtain their number of
employees of Danish and foreign origin. Non-responses are supplemented by information from the
Ministry of Foreign Affairs. It is assumed that government officials of foreign origin located in Denmark
spend a similar amount on personal expenditures in Denmark as Danish government officials spend when
abroad. Thus, the number of staff of foreign origin per embassy or international organization located in
Denmark is multiplied by the average expenditure on goods and services of Danish government officials
located abroad, which Statistics Denmark has previously estimated as described above.
Country experiences: Japan
14.233. Japan relies on the ITRS as the data source for most of its balance of payments (BOP) statistics,
including government goods and services n.i.e when possible. However, there are certain transactions of
government goods and services that are carried out with no monetary settlement, which therefore cannot
be captured by the ITRS and must be estimated.
14.234. Following BPM6 convention,225 the value of transactions that occur without monetary settlement
in the absence of appropriate market prices is regarded as the sum of the relevant costs of providing the
good or service (in this case, government expenditures for providing the goods and services in question).
Actual government expenditure data in Japan have certain limitations that preclude their effective use in
224
225

BPM6 CG, chapter 12, paragraph 12.159.
BPM6, paragraph 12.49.

214

the statistical compilation process; namely, such data are not timely226 and are not reported according to
official BOP definitions. Therefore, the Bank of Japan (BOJ) uses government budget data, which are
more timely than government expenditure data227 and follow BOP classifications, in place of actual
government spending data in order to compile government goods and services n.i.e. when the ITRS does
not suffice. The discrepancy between the budget data and the actual expenditure is usually insignificant,
according to BOJ research. The budget data are then allocated to the correct time period in accordance
with the duration covered by the budget. If a supplementary budget is passed, it is also incorporated into
the budget data used to compile government goods and services transactions.
C.

Allocation of resident/non-resident trade in services to modes of supply

C.1. Introduction
14.235. MSITS 2010 recommends that trade in services statistics are also broken down by modes of
supply, based on the resident/non-resident trade in services and FATS statistical frameworks. FATS
output data for services should be used to measure mode 3, whereas the resident/non-resident trade in
services data needs to be broken down by modes. Different options exist to compile such information,
either via a simplified allocation of EBOPS2010 categories, or on the basis of collected data. This section
describes these options in more detail (it should be noted that deriving modes of supply using a data
model is also a possibility, this is described in chapter 17).
14.236. This section starts with explaining how the MSITS 2010 recommendation, that in the absence of a
special data collection, compilers use a simplified (mechanical) allocation of FATS and balance of
payments data to modes of supply228 , can be applied in practice (C.2). Subsequently, given that the goal is
to ultimately implement a data collection and compilation of a breakdown of trade in services by mode of
supply based on collected data, section C.3 describes the steps compilers can take to set up such a data
collection in the most efficient and policy relevant manner. Section C.4 concludes with suggestions on
how more detailed information regarding especially mode 2 and 4 can be compiled.
C.2 Resident/non-resident trade in services data by mode of supply: a simplified allocation
14.237. Given that the compilation of resident/non-resident trade in services data broken down by mode
of supply is a relatively new area of data compilation, and given that amending existing data collection
mechanisms or creating new ones may be difficult, it is suggested to adopt a step-by-step approach for
compiling such a break down. In the absence of relevant data, or if only some sporadic information is
available, it is first proposed that compilers conduct a simplified allocation of existing statistics as given
in table V.2 in MSITS2010.229 This method consists of attributing EBOPS service categories to either one
dominant mode or to several modes using a distribution, based on an assumption on how specific EBOPS
2010 service items are most probably supplied by exporters (or to importers) of the economy.
14.238. This method provides a first set of estimates on modes of supply comparable at the international
level (which also could be disseminated in a common manner; see chapter 20 for more information). Such
an allocation has the advantage of being a relatively low-cost solution, as the compiler can start working

226

Data on government spending are not released in Japan until the end of the fiscal year, which is too late for
inclusion in the compilation of the relevant statistics.
227
Government budget data in Japan are released before the beginning of the fiscal year.
228
See also MSITS 2010 Chapter V and Table V.2 and chapter 1, paragraph 1.5.
229
Ibid.

215

with the available balance of payments services data and gradually build his/her knowledge of how
services are supplied internationally. However, compilers should treat such an allocation only as a first
rough approximation of resident/non-resident services transactions by mode of supply as this technique
has important limitations. This Guide strongly encourages compilers undertake efforts to develop more
precise estimation procedures at a later stage.
14.239. Using table V.2, compilers are suggested to make the allocation in three steps: 1) allocate 2)
evaluate and 3) refine, as described below:
14.240. Firstly, compilers can allocate each service item to one of the columns identified in table based on
an assumption of how a specific service item is most probably supplied by exporters (or to importers) of
the economy. In order to provide a first approximation in a comparable way, all compilers are strongly
encouraged to conduct this generic allocation.
14.241. Secondly, compilers should evaluate if the "generic" allocation as conducted at the first step is
relevant for their economy, and review results accordingly. For example, it may be worthwhile for the
compiler to discuss with the institution in charge of trade in services negotiations if the results reflect their
knowledge of how services are supplied abroad and to their national economy, as far as it relates to
transactions recorded in the balance of payments.
14.242. Thirdly, and based on the results obtained in the second stage, compilers can refine their
allocation by gathering additional information to improve the knowledge of some specific service sectors.
Such additional information can be gathered in cooperation with the institution in charge of trade in
services negotiations and might validate the assumptions made earlier by statisticians or negotiators.
14.243. Various ways of gathering more information may be envisaged, such as contacting major services
providers or trade or consumer associations; conducting qualitative interviews with one or two relevant
services providers in a specific sector (e.g., legal services, computer services, consultancy, construction
etc.), or conducting interviews with employment agencies which have international services transactions
with clients abroad. Compilers can also approach relevant ministries, in particular sectors where
internationalization is known to be important (e.g., the ministry of industry, education, and health), or
approach compilers in other statistical domains to obtain further information on particular sectors and to
adjust the allocation if needed (e.g., through micro-data linking).
14.244. When making refinements to the initial allocation, compilers should also consider other factors,
such as the business structure of the compiling economy (e.g., the dominance of large enterprises or
SMEs/micro-enterprises), the type of service traded (some specific/technical services require the physical
presence of the service provider), the geographical location of the compiling country and/or the distance
to the trading partner (i.e., the less distant the partner, the more likely trade in mode 2 or mode 4 can
occur), language barriers, the evolution of business strategies and tradability over time (e.g., technological
advances).
C.2.a. Limitations to the simplified allocation
14.245. The compiler should always keep in mind that table V.2 and the methods described above, are
simply a theoretical guide for classifying resident/non-resident services transactions according to the most
likely predominant modes of supply. In each specific economy, other modes than those indicated may be
involved for some specific services categories, considering their nature. For example, in table V.2
personal, cultural and recreational services are shown as deemed to be provided (or consumed) through
modes 1 (cross-border) or 4 (temporary presence of service provider, either himself/herself, if self-

216

employed, or his/her employee). However, for example, in the case of countries that are important
destinations for the shooting of films, mode 2 (presence of consumer abroad to consume services) may
also need to be considered. In other words, these assumptions may need to be reviewed to identify how
well they respond to national needs. It would also be useful to conduct some analysis of contracts for
specific groups of persons to better understand how they operate in the context of trade in services.
14.246. There are also other shortcomings to be considered when following the procedures outlined
above. Compilers should ask themselves questions such as the following: how is the most significant
mode allocated? Regarding enterprises which report their main economic activity, how should secondary
economic activities be treated? Regarding manufacturing enterprises also providing production services
or services packages around high-value goods, how should these services be treated? Even if this
framework provides an approach for a first rough measurement of trade in services by modes of supply
with a minimum use of resources, relevant qualitative background information and research is needed.
Compilers need to consider these aspects when performing the conceptual allocation.
14.247. Compilers are encouraged to make use of their own information about modes of supply and
possibly different country-specific distributions among services than the general allocations. Allocations
to the modes may be based on the knowledge compilers have about the provision of services from their
close contact with respondents or based on their knowledge of the business structure gained in the
enterprise survey design process. More generally such an exercise is encouraged for improving the
knowledge of compilers with respect to international trade in services.
C.3. Resident/non-resident services transactions by mode of supply: towards full data collection and
compilation
14.248. As indicated above, the ultimate goal is to implement data collection and compilation of a
breakdown of resident/non-resident services transactions by mode of supply. This compilation can be
done in much detail for (a) specific service sector(s) of interest to the compiling economy (i.e., using the
simplified allocation for the other sectors of less interest), or through the compilation for all services
sectors. The compilation of such data may entail a number of challenges, mainly because of the fact that it
may be difficult to collect the raw data itself. However, it should be noted, that when the relevant details
are collected, there are no additional compilation issues, besides the general considerations (i.e., grossingup, dealing with non-response, quality of information obtained, data confidentiality etc.).
14.249. A variety of options have been developed (or can be envisaged) by countries to set up a new data
collection (or add questions to existing data collections) regarding trade in services by mode of supply.230
Both a broader data collection exercise, as well as collaboration with other agencies and institutions, can
be envisaged as described below. Alternatively, data on modes could be derived through the combination
of information; e.g., through the use of a data model, as described in chapter 17.
C.3.a. Broad data collection and generic surveys
14.250. For collecting data on the value of supply of services by mode of supply, the most relevant source
would be through surveys, mainly enterprise surveys (see chapter 6). By, for example, amending existing
surveys or integrating appropriate questions in survey forms, data from existing enterprise surveys can be
230

Another possibility would be for the data compiler to establish a data model for certain services categories or
modes (see chapter 17).

217

used as a way forward to estimate the value of the international supply of services by mode. Hence,
potentially, such estimates can be generated without establishing a new and additional data collection,
while still being of acceptable quality. Even efforts targeted towards a limited number of relevant services
categories in the compiling economy can ensure the production of policy relevant results. Especially
because the expected respondents’ burden could be high, it would be a good practice to conduct at the
initial stages sample surveys of enterprises, which should address questions needed to estimate or allocate
services by modes. At a later stage, modes of supply could be integrated in the existing questionnaires as
a mandatory data element.
14.251. Such an approach (i.e., amending existing surveys in a step–by-step manner) would necessitate
some knowledge building from the compilers side. As a starting point this could be done by following the
considerations outlined in the section on the conceptual allocation. It could also be achieved by
conducting a "screening" survey to identify the sectors and enterprises that should be targeted, or by
exploring academic research (literature review) to identify the way in which services are generally
rendered. Relevant enterprises could be extracted from the existing business registers (see also chapter 5,
section E and chapter 6) focusing as a first step only on some key (business) services e.g. legal services,
engineering services, computer services.231
14.252. Structural business statistics provide compilers with additional information regarding the
structure of the businesses, size-class, employment, etc. which could be useful for national industry
policies.232 Important service sectors could be identified as a starting point from the shares of the services
to the GDP or share in total services. Some countries compile jointly international trade in services and
structural business statistics, therefore the whole business population is available and the relevant subpopulation of interest for a specific mode could be derived from this source. In some cases the
information is also available in internally maintained registers, so-called BOP-registers.
C.3.b. Specialized surveys
14.253. A focused or specific data collection could be developed on the basis of economic, politicaleconomic and socio-economic issues; for example, mode 4 could be integrated in a broader concept of
cross-border movements and related international transactions in globalization theories. In addition to
enterprise surveys, surveys of persons and households (see chapter 7) could also be (partially) useful to
for certain mode 4 niches; e.g., for services provided by self-employed persons or for households as
consumers of services provided by contractual service providers, although the latter may not be
significant for the majority of compiling countries (small proportion of population, small transactions as
compared to mode 4 transactions of enterprises). Surveys of persons and households could also be used to
identify mode 2 (i.e., travel).
14.254. Finally, a pluralistic-integrated approach, i.e., a sector-specific survey that integrates various
statistical domains including trade in services (but also e.g. FATS, SBS, innovation or others), can
provide relevant statistical information for different statistical domains. Focusing on different user

231

See also Eurostat manual on the Production of Business Services, Annex VIII of the European Parliament and
Council Regulation No 295/2008. For more information see:
http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2008:097:0013:0059:EN:PDF.
232
Information on small and medium-sized enterprises and specific characteristics of enterprises related to
particular breakdown of activities might be important for national policy to promote cross-border services trade for
enterprises which are not exporting services or to support enterprises which are already active in the export of
services.

218

interests, specific sector studies would serve several purposes; e.g., policy makers could also be interested
in specific studies, such as those incidental to agriculture or environment services. A more detailed
approach (i.e., modes for more detailed services categories) in such surveys may be implemented, but
often the policy needs have to be identified by the national authorities for this to be developed.
C.3.c. Specialized surveys in collaboration with other agencies
14.255. As described in chapter 6, the collection and compilation of resident/non-resident trade in
services statistics by mode of supply may also be established in collaboration with agencies that have
specific interests in obtaining such data. For example, ministries of economy or trade may be interested in
a special collection for this type of information. In these circumstances, it is important that the agency in
charge of the collection and compilation of official trade in services statistics (generally the NSO or
central bank) coordinates with the specialised agency so that the international recommendations for the
collection and compilation of trade in services statistics are followed, and to ensure that the information
that is collected can be used in a broader context.
14.256. An example of such a data collection is presented in chapter 10 (paragraphs 10.63-10.67). This
example shows that the Australian International Legal Services Advisory Council (ILSAC) has for a
number of years been conducting its own compilation of statistics on the international supply of legal
services to non-residents. This approach benefited from the assistance of the Australian Bureau of
Statistics. Such collaboration has highlighted the desirability, from ILSAC's point of view, of increasing
the focus on compiling data that is more closely aligned to the specific modes of service delivery
recognized by economies in trade negotiations, and although it does not mention it, to ensure that the data
collected will be following as closely as possible international standards. This experience can also be seen
as an example of positive collaboration of the body in charge of compiling trade in services statistics and
an institution which has a strong interest in collecting sector-specific data. Although from the statistical
compilers perspective it would seem difficult to replicate this for all services sectors, such experience can
prove useful for improving the quality of statistics as well as providing some first estimates according to
the four modes.
14.257. Similarly, large employment or recruitment agencies233 that (could possibly) have important
cross-border activities could also be targeted. This would be for example agencies e.g. working with
persons from new EU-members for instance or working with persons in so-called regional (border)
clusters). For example, it may be possible to use the data from these agencies on the types of contracts to
identify those services relevant for measuring mode 4.
C.4. Compiling more detailed mode of supply statistics for resident/non-resident trade in services
C.4.a. Specific considerations regarding mode 4
14.258. It is suggested that the compiler investigates how to collect/compile more details for those items
where mode 4 (presence of natural persons) is deemed to be important for the compiling economy. For

233

For example, since 1994, recruitment agencies in Germany can recruit in all occupational fields before there were
some exceptions in management and artistic related activities.

219

example, professional and management consulting services (mainly business services) is generally
considered to be predominantly provided (or consumed) through modes 1 (cross-border supply) or 4 as
these services often entail a proximity to the consumer for the service to be rendered. Alternatively,
providing processing services or maintenance and repair services may most likely involve mode 2
(consumption abroad) or mode 4. So including a question asking for the share of mode 4 for specific
services categories may be sufficient to greatly improve the statistics on trade in services by mode.
C.4.b.

Specific considerations regarding mode 2

14.259. As presented in MSITS 2010, some EBOPS components are strongly linked to the supply of
services through mode 2. These include for example manufacturing services, maintenance and repair
services, travel and waste treatment. Travel has the most obvious link with mode 2 supply of services. In
this context BPM6234 and MSITS 2010235 propose an alternative breakdown of travel into: goods, local
transport services, accommodation services, food-serving services and other services. This level of detail
would already serve many information needs, including those related to mode 2. It is in particular
important to identify goods separately as these products are of minor interest from a GATS/trade in
services perspective.
14.260. Various possibilities exist for compiling these breakdowns for the Travel item, such as the use of
credit card data, border surveys or household surveys (which are often used in the context of data
models). Tax-free purchases could be used as complementary sources to separately estimate the goods
purchased by persons going abroad for travel reasons, because those traveling can request a refund for the
VAT paid on the goods when exporting the goods.
14.261. When using credit card data, the use of information included in merchant codes is suggested. This
could enable a more detailed analysis of travel/tourism data and would also allow to identify e.g. transport
services and to extract data split by goods and services. For example payments of or to a company which
provides maintenance and repair services would be reported since this company is providing services and
the payments have to be reported. Also merchant codes categories like Hotels/Motels/Inns/Resorts, car
rentals or tourist attractions and exhibits are subject of reporting. Using merchant code information from
credit card data has several advantages, including the relatively low costs for compilers, and the fact that
in addition to the EBOPS travel item, also other services categories (other business services,
communication services, government services) could be compiled or verified.
14.262. In addition, customs data could help in the identification of thresholds in order to adjust travel
and goods (EBOPS concept) accordingly (i.e., durable goods and valuables).236 For that, totals on travelrelated inflow from travel survey data or credit card data has to be calculated. The value of the total
amount of valuables and durable goods in excess of custom thresholds from customs data also has to be
calculated and efforts need to be made to avoid double counting.

234

BPM6, paragraphs 10.85-10.100.
MSITS 2010, paragraphs 3.115-3.131.
236
BPM6/MSITS 2010 and ITRS 2008 have a different treatment concerning some goods purchased by those
traveling: the BPM6/MSITS 2010 travel item excludes purchases of valuables and consumer durables above a
customs threshold, whereas tourism statistics includes all such purchases irrespective of the threshold. See MSITS
2010 Box III.5 for the relationship between data on travel and tourism statistics.
235

220

Country experience: New Zealand on collecting data on modes of supply
14.263. The user is invited to read the experience of New Zealand collecting data on modes of supply as
presented in the country example of chapter 6. In New Zealand respondents were asked in the 2011
Census of International Trade in Services and Royalties to estimate the percentage of the export value that
was delivered through mode 1, mode 2, and mode 4. Testing showed that most respondents understood
the modes of supply concepts, although feedback suggested the guide which was provided together with
the trade in services questionnaire was helpful as well.
C.7.

Country experience: Turkey

14.264. Turkey’s compilation system entails some advantages for modes of supply analysis. To be more
precise, the first advantage stems from already having the following breakdown of travel by type of
product for Mode 2 purposes both for travel credit and debit as well as for personal and business travel
(except for the expenditure on package tours: the share pertaining to Turkey for credit and the share
pertaining to non-residents for debit) almost on an EBOPS 2010 classification: breakdown a) Goods, b)
Local transport services, c) Accommodation Services, d) Food-serving Services, e) Health services, f)
Sports, education and culture.237 In addition, some of the Mode 4 type of transactions may be identified
through the ITRS forms.
14.265. At this point, it may be useful to give more information on the TURKSTAT travel surveys
conducted in cooperation with the CBRT and the Ministry of Culture and Tourism. In connection with
Mode 4 purposes. These are face to face surveys conducted at the border gates, on the nationality basis,
four times a year, so as to cover quarterly periods. For instance, the “Departing Non-resident Visitors”
survey for travel credit is carried out at 25 border Gates, which covers 90 percent of all departing visitors
according to the departure way (Air, Road, Rail and Sea), and as regards years and terms, new gates have
been added or excluded. The purpose of this survey is to determine the profile (age, sex, education level,
occupational status), travel characteristics (purpose of visit, the place stayed, accommodation type, night
spent, types of expenditures) and to estimate Turkey’s travel income of foreigners and citizens residing
abroad. The number of foreign visitors is based on the administrative border statistics of the Directorate
General for Security for the related periods which cover all arrivals and departures in all border gates in
the details of citizens, foreigners, nationality, month and border gates (overnight visitors and
excursionists).
14.266. The survey is carried out for overnight visitors and excursionists by 0.5% sample rate to estimate
on the basis of departure way and nationality and estimations are given quarterly in the detail of 26
selected nations and 10 country groups.238 The sample survey results estimate the average expenditures
with the breakdown of overnight stays and the excursions, which are expanded with the related
Directorate General for Security border statistics. In order to estimate average expenditure figures for
Turkish travelers abroad, sample surveys are also conducted on a quarterly basis for resident visitors
arriving in Turkey.
14.267. In the light of the above considerations, the “Departing Non-resident Visitors” survey has
questions on general occupational status and purpose of the visit with one of the options being “Business
(conferences, meetings, assignments etc.)” albeit with no further breakdown. This option may further be
broken down into i) conferences, meetings, trade fairs and exhibitions etc. and ii) as a contractual service
237
238

In the EBOPS 2010, the distinction regarding other services is between education services and health services.
Turkish Statistical Institute (2011), Tourism Statistics.

221

supplier and iii) other. The next step would be to modify the questions on general occupational status to
identify the type of the employer-employee relationship for those whose purpose of visit is business and
professional activities as such if (i) their employer is in Turkey, (ii) outside Turkey. The resulting matrix
of the mentioned questions may then prove to be useful, provided that a statistically meaningful
expansion can be achieved, which should be assessed with statistical scrutiny.
14.268. The compilation of the additional item on tourism services related expenditure in travel and
passenger transport should also be encouraged to establish a clearer link between the BOP and tourism
statistics. Finally it is necessary that BOP and tourism statisticians cooperate, and in particular investigate
if more detailed services categories are of interest. Also some data for the tourism satellite accounts could
probably be integrated.
C.8.

Country experience: Portugal

14.269. A separate alternative breakdown of travel into types of goods and services is recommended
according to BPM6/MSITS2010. Box 14.8 shows the level of product detail required as supplementary
breakdowns for travel in according to BPM6 and EBOPS 2010 integrated with additional requirements
included to provide the necessary level of detail for other statistical domains. This also responds to mode
2 information needs.
Box 14.8 Level of product detail required for supplementary
breakdown of travel

14.270. The forthcoming payment cards database can provide important information to meet these new
requirements, using the activity sector code of the goods and service provider. For payments made in
Portugal by cards issued abroad in Portugal the NACE of the POS owner is available, while for payments
made abroad by cards issued by resident institutions the MCC of the POS owner is provided. These
variables are used as proxies. A correspondence table between NACE or MCC and the different travel
expenditures on goods and different types of services has to be developed. Furthermore, to identify local
transport services other than rental services separately from international transport services the border
survey is going to be used.

222

14.271. In addition, when travel expenditures are prepaid to resident travel agencies, both in terms of
travel credits and travel debits, the direct reporting by these companies will provide the breakdown by
good and type of service for travel credits as well as for travel debits. This information complements
payment cards data. For the same breakdown of payments made to non-resident travel agencies, on the
debits side, the border survey is being considered as a possible data source. In terms of travel expenditure
on goods, BPM6 recommends the acquisition of valuables, consumer durable goods, and other consumer
purchases for own use or to give away acquired by travelers in excess of customs thresholds to be
registered under general merchandise and not under travel. To identify the acquisition of such goods,
different variables from the payments card database have to be combined, namely, the activity
classification of the goods provider is restricted to jewelry, art, cars and electronic goods, and considering
some minimum threshold for the value of the operation.
D.

Service transactions between related (affiliated) enterprises

14.272. MSITS 2010 acknowledges that information on the value of all transactions between affiliated
enterprises is helpful in understanding the degree to which globalization of services is taking place. Such
intra-firm trade in services can take up a substantial share of total services trade; for example, transactions
between related (affiliated) enterprises accounted for 28 per cent of both U.S. exports and U.S. imports of
private services in 2013.239 . In addition, statistics on intra-frim services trade can highlight the role of
services, in particular those related to IPPs, provided by one unit but used throughout the MNE. For
example, again the information for the United States shows that intra-firm trade accounted for 59 per cent
of receipts of royalties and license fees and 71 per cent of payments of charges for the use of intellectual
property in 2013.
14.273. MSITS 2010, therefore, recommends that data on resident/non-resident transactions in services
separately identify transactions with related and unrelated enterprises. Although such a breakdown would
be most informative at the level of the detailed EBOPS 2010 classification, it is recognized that this could
place an additional burden on both suppliers and compilers of data and could raise issues of
confidentiality. In this connection, MSITS 2010 recommends that such a breakdown be carried out at the
aggregate level for total services transactions (under the complementary grouping of EBOPS entitled total
services transactions between related enterprises), although countries willing to provide additional detail
are encouraged to do so for some (relevant) aggregated EBOPS 2010 categories.240
14.274. The MSITS 2010 associates related enterprises (intra-firm trade) to direct investment
relationships241 . Should the criteria for related enterprises depart from this recommendation, it is good
practice to clearly describe the criteria chosen in a country’s trade in services metadata.
14.275. It is good practice to carefully examine the valuation of services transactions between related
enterprises , as the recorded transactions could be under- or overestimating and misrepresenting the real
flows of trade (as measured by market prices) if such transactions between related enterprises are based
on transfer prices (see also chapter 1 for transfer/market prices).

239

U.S. cross-border trade in private services excludes transactions by the U.S. government (including the military).
Trade in private services is the most appropriate basis of comparison because intra-firm trade covers only trade by
businesses.
240
Ibid., paragraph 3.56. Note that for some transactions/services items compiling such a breakdown would not be
relevant, e.g. transactions where individuals are c onsuming services, education services.
241
See MSITS 2010, paragraphs 4.58 – 4.60 for further details.

223

Country experience: the United States
14.276. For conceptual reasons or due to the source data limitations, the U.S. Bureau of Economic
Analysis (BEA) only presents statistics on intra-firm trade for some selected types of services.
Transactions between individuals and businesses are considered to be between unaffiliated parties. Thus,
personal and business travel services, which are transactions by individuals who travel to foreign
countries; education, which consists of expenditures for tuition and living expenses by students studying
in foreign countries; medical services, which cover expenditures by patients in foreign countries; and
expenditures by non-resident workers, are all considered to be unaffiliated transactions. Passenger
transportation, which covers transactions between individuals and foreign airline or vessel operators, are
also considered to be unaffiliated transactions. Transactions in insurance services are deemed to be
unaffiliated even when they are between affiliated companies because the services are deemed to be
provided to the policyholders who pay for the insurance premiums and who are unaffiliated with the
multinational company. Transport services other than passenger transport are treated as unaffiliated
because the source data do not allow transactions between affiliated parties to be separately identified
from transactions between unaffiliated parties.
14.277. The remaining categories—including charges for the use of intellectual property;
telecommunications, computer, and information services; financial services; and other business
services—can have affiliated transactions. The source data for these statistics are BEA surveys.242 On the
surveys, U.S. companies are asked to report their transactions in a specific service type by country and,
for each country, by whether the transaction was with their foreign affiliates, with their foreign parent or
with foreign affiliates of their foreign parent, or with unaffiliated parties.
14.278. The definitions of affiliated parties are the same as those used to identify a direct investment
relationship. A foreign affiliate is a foreign business enterprise in which a U.S. person directly or
indirectly owns or controls 10 per cent or more of the voting stock in an incorporated business enterprise
or an equivalent interest in an unincorporated business, including a branch. A foreign parent is the first
person outside of the United States that owns or controls 10 per cent or more of the voting stock in an
incorporated U.S. business enterprise or an equivalent interest in an unincorporated U.S. business,
including a branch. Foreign affiliates of the foreign parent are defined according to a more restrictive
criteria than above as they are any foreign persons proceeding up the foreign parent’s ownership chain
that owns more than 50 per cent of the person below it up to and including that person that is not owned
more than 50 per cent by another foreign person and any foreign persons, proceeding down the ownership
chains of each of these members, that is owned more than 50 per cent by the person above it.
14.279. By collecting data on transactions with foreign affiliates separately from those with foreign
parents and foreign affiliates of the foreign parent, transactions within U.S. multinational enterprises, i.e.,
between U.S. parent enterprises and their foreign affiliates, can be distinguished from those within foreign
multinational enterprises, i.e., between U.S. affiliates and the foreign multinational enterprises that invest
in them.

242

Financial services are collected on the BE-185 Quarterly Survey of Financial Service Transactions between U.S.
Financial Service Providers and Foreign Persons and the BE-180 Benchmark Survey of Financial Service
Transactions between U.S. Financial Service Providers and Foreign Persons. Telecommunications and most
business, professional, and technical services are collected on the BE-125 Quarterly Survey of Transactions in
Selected Services and Intangible Assets with Foreign Persons and the BE-120 Benchmark Survey of Transactions in
Selected Services and Intellectual Property with Foreign Persons. These surveys can be found on BEA’s website,
www.bea.gov, by looking under “International” and clicking on “Survey Forms and Related Materials.”

224

14.280. Multinational enterprises sometimes allocate expenses across various divisions or parts of the
enterprise rather than billing them separately. These allocations, often called allocated expenses,
headquarters services, or miscellaneous charges, are sometimes for a designated service, such as for
research and development (R&D), but sometimes no specific service is designated. It is important to note
here that management of patents and license fees, which may be of a similar nature to allocated expenses,
should be recorded in other business services, n.i.e. If the type of headquarters service is known, BEA
asks reporters to include those allocated expenses in its data for that type of service. If the type of service
is not known, BEA asks reporters to include them in the category of “management, consulting, and public
relations (including allocated expenses).”

225

Chapter 15 Compilation of Foreign Affiliates Statistics and th e international supply

of services
15.1. Scope. This Chapter deals with the compilation of FATS and consists of the following sections:
Compilation of foreign affiliates statistics: summary of good practices (Section A), general purpose and
description of FATS compilation (Section B), FATS variables and their compilation (Section C), selected
additional data compilation issues (Section D). Specific country experiences have been included in the
main text, and full texts of country contributions will be made available on the website dedicated to the
compilers' guide.
A.

Summary of good practices

15.2.
FATS describe the activities of foreign-owned or foreign-controlled enterprises in host
economies. This statistics relates to commercial presence in GATS (mode 3), but also it is more broadly
of interest in the context of globalization. 243 As described in MSITS2010 the interest in FATS is on all
variables, with a particular focus on variables related to the supply of services. This Guide advises on the
compilation issues in relation to FATS and provides more details on items where services measurement
deserves more attention. Partial information on the presence of natural persons may also be obtained in
the process of FATS compilation if data on employment by foreign affiliates is collected and if the
foreign employees (such as corporate transferees) can be separately identified. For these purposes, FATS
are of interest in their own right, but it will often only be possible to ascertain their full significance when
they are viewed in conjunction with other information, such as comparable information on total investorcountry or host-country economic activity and on services supplied through modes other than commercial
presence.
15.3. Consequently collection of data based on balance sheet information and activity measures such as
sales, employment, imports, exports or value added figures, etc. provide a more complete picture of the
international supply of services. While the production and use of services provided by foreign affiliates
are part of the national accounts of the host country (including exports of such services to other
countries), these measures will not appear in the national accounts of the investing country.
15.4. Good practices. Given the complexity of the FATS compilation framework, a step-by-step
approach is advised as follows:
xii. While setting up the FATS compilation programme a thorough analysis of data needs for FATS
has to be conducted in the context of various possible options for the organization of data
collection and data compilation process. Given that inward FATS are generally easier to obtain
than outward FATS and that this information is directly connected to international negotiations, it
is a good practice that countries first concentrate on the compilation of inward FATS;244
xiii. A review of all available sources should be conducted (e.g. FDI statistics, enterprise surveys,
registers, administrative sources). Procedures for the identification of the relevant population,
either directly or through linking exercises should be developed and tested. The identification of
the Ultimate Controlling Institutional unit is particularly important for compiling FATS;

See chapter 1. 244 It is important to note that a country’s outward FATS conceptually should match its partner's
inward FATS.
244
It is important to note that a country’s outward FATS conceptually should match its partner's inward FATS.

226

xiv. A list of the most important variables for compilation has to be established early on based on the
list of core (and eventually additional variables) identified in MSITS2010. The guide suggests as
a starting point to concentrate on sales/turnover or output (the latter being the preferred measure
out of the two), employment, and number of enterprises;
xv. The choice of the statistical unit needs to be made, although in many cases this will be determined
by the existing definitions used in other domains of national economic statistics (e.g. for business
statistics, national accounts);245
xvi. Ideally data should be compiled annually and, if possible, on a calendar year basis;
xvii.
Compilation of data broken down by economic activities is advised. Compilers are
encouraged to do so for all activities, with sufficient details to respond to the needs related to
services negotiations (see ICFA Rev. 1 in MSITS 2010). For output or sales/turnover, feasibility
of a disaggregation into total sales/output of goods and total sales/output of services for each
activity should be investigated to respond better to information needs related to the measurement
of the international supply of services;
xviii.
Given the complexity of the FATS framework and the often high level of detail in of the
collected information, compilers should pay particular attention to ensuring data confidentiality
when cross-classifying information by activity/product and partner country; proper procedures for
evaluating what data can be publicly disseminated should be developed and systematically
applied.
B.

General Description

15.5. FATS data should be compiled for as many variables as possible. Some are more closely related
to the needs of GATS (output or turnover, employment, number of enterprises), but for broader analytical
purposes other variables would be of interest such as assets, R&D expenditure, etc. The main elements of
the FATS conceptual framework and its scope for services are described in Chapter IV of MSITS2010.
For additional guidance, compilers are advised to refer to the OECD Benchmark Definition of Foreign
Direct Investment 4th edition chapter 8 as well as the OECD Handbook on Economic Globalisation
Indicators chapter XX.246 All sources identified above also include some practical information on FATS
compilation.
15.6. FATS are the main data source for assessing mode 3, or commercial presence.247 In the majority
of cases this mode of supply corresponds to the activities of entities that are controlled or owned from
abroad in the territory where they are established. The supply of services through mode 3 by non-resident
suppliers to residents of the compiling economy is made by the resident affiliate of this supplier (i.e.,
recorded as transactions between residents of the same economy).

245

Compilers should note that this may differ between inward and outwards FATS, see section D.1
Another valuable source of information for compilers is the Eurostat Foreign Affiliates Statistics (FATS)
Recommendations Manual, 2012 edition, produced by Eurostat Available at:
http://epp.eurostat.ec.europa.eu/portal/page/portal/product_details/publication?p_product_code=KS-RA-12-016
247
Mode 3 also refers to cases where there is an establishment of a non-resident entity (e.g. for less than one year)
delivering services to customers of the economy of establishment. In these cases international transactions in
services will be recorded in the balance of payments statistics. However, MSITS2010 recommends considering
these only if it is believed that such cases are important for the compiling economy.
246

227

15.7. The main interest from a mode 3 perspective is in compiling data broken down by
destination/origin of control and by service type. However in the absence of data by service type or if
such a breakdown is difficult to compile, it is possible to estimate the most likely type of service being
provided according to the primary activity of the firm. For example, a company in the telecommunication
services industry is likely to have been set up abroad to provide mainly telecommunication services.
15.8. FATS will include variables such as output or turnover/sales, employment, etc. Although the
interest from a mode 3 perspective is not limited to this output, output is the measure of mode 3 supply of
services, which, in a GATS context, is analogous to exports and imports of services recorded in the
balance of payments (corresponding to modes 1, 2 and 4). In the absence of output data, a solution could
be to use FATS sales/turnover of services data, keeping in mind the mode 3 measurement issues
associated with wholesale and retail trade, financial services, and insurance services (see section C.1
below).
Chart 15.1 Output of foreign affiliates and mode 3 supply of services: a statistical view

15.9. It is also important to consider what has in fact been supplied to consuming entities of the
economy where the affiliate is established. The measure of mode 3 supply of services is a subset of a
foreign affiliate's output figures, i.e., output that is not exported to third economies or back to the
economy of the controlling enterprise. As shown in box 15.1, company B’s production and delivery of
services to consumers of B is supply of services through mode 3 for economy B (originating from A).
Company B’s exports of services to the rest of the world are part of output figures as a FATS variable, but
are not included as economy B’s mode 3 consumption of services. It could be supply of services through
modes 1, 2 or 4. Similarly, Company C’s production and delivery of services within economy C is supply

228

of services through mode 3 to consumers of C (from economy A). The services output produced by
company B (or company C) but provided to non-resident consumers outside their economy of
establishment should be accounted for as trade in services (either mode 1, 2 or 4) between the economy of
residence of this company and the economy of residence of the consuming entities. In other words this
information is already covered in balance of payments services data of respective economies. When
analysing the international supply of services by mode, only those data derived from the FATS
framework that correspond to mode 3 should be considered, to compare to data derived from the BOP
services account.248
15.10. Within the list of variables recommended for FATS compilation, employment is also one of great
interest particularly if compilers are able to separately identify the portion of employees from abroad (i.e.
non-residents). This would be useful in relation to mode 4 commitments, as further described in section
C.4 of this chapter as well as chapter 16. If possible, data on the number of intra-corporate transferees
would be useful particularly given the high level of commitments made in trade negotiations for this
category of persons.
15.11. The variables should be broken down according to:

i. The country of location of the foreign affiliate for outward FATS or the origin of ultimate
control/FDI for inward FATS;

ii. The activity of affiliates;
iii. The type of products produced by the foreign affiliates for the variables for which such a
breakdown is possible, at a minimum distinguishing services from goods. Although this is
generally seen as a longer term goal, some compiling economies have already compiled such
data, or such a breakdown is feasible; and

iv. For output or sales, Distinguishing between those within the host economy, with the home
economy and third economies.
15.12. Supplementary data may also need to be compiled as a separate dataset in the specific
circumstances of some economies such as the existence of significant joint ventures, or where control
cannot be determined solely using the guideline of more than 50% ownership of voting power, or in
economies where control is not possible by law.
15.13. FATS often require integrating multiple data sources, which can present measurement challenges.
Different data sources are often necessary to identify the population covered by the framework, as well as
the breakdowns by partner country.
C.

Selecting data sources

15.14. Data source for FATS are described in chapters 5, 6 and 10. Comparison of data sources is shown
in chapter 11. When beginning FATS compilation, to select data sources one should take into
consideration the fact that mode 3 supply of services as measured through FATS is most likely to occur in
the context of multinational groups, in particular those with important and numerous FDI relationships.
In other words, in particular on the outward side, large companies are more likely to be the most

248

Keeping in mind other considerations with respect to the comparability of these data, see chapter 20 on data and
metadata dissemination.

229

important players for supplying services abroad. As a starting point, it may therefore be useful to consider
focusing on larger (services) companies or those with important FDI transactions and positions, which
could provide an intermediate low-cost solution for compilation. However, one should not underestimate
the role of smaller companies because (i) in the context of inward FATS, foreign affiliates of large
multinational groups may not necessarily be large themselves and (ii) in the context of outward FATS, the
international role of small and medium enterprises is becoming increasingly important. This means that
although such an intermediate solution may be useful in the short-run, in the longer-run compilers should
consider compiling a FATS dataset representative of the entire population. In the context of employment,
it is also important to note that larger (services) companies/affiliates are more likely to recruit many nonresident employees (i.e. from the perspective of the compiling economy) or have important intracorporate movements of their personnel.
15.15. Collecting source data and compiling FATS implies the need for various stakeholders to
cooperate. For compiling FATS, as done in many economies, the data sources should be used in a
complementary way, in particular for inward FATS. For example, although the main data source to
collect data may be a survey under the responsibility of one agency (e.g. statistical office, central bank),
often the information relating to the identification of (part of) the population to be covered and/or the UCI
is available from another data source (e.g. other surveys, enterprise registers) that can be under the
responsibility of another agency (e.g. central bank or another administrative body). In addition, in many
economies the collection of data from financial institutions is conducted by the central bank whereas the
collection of data of other types of units is conducted by the statistical office.
15.16. It is therefore a good practice to assess whether cooperation of various stakeholders is necessary
and if so, to identify possibilities and responsibilities for linking information (e.g. common identifier in
registers/administrative sources) before beginning FATS compilation. This process has been implemented
by many economies, and often enables the compilation of FATS with little or no additional burden for
reporters or work for compiling agencies. Information on legal frameworks and institutional arrangements
is provided in chapters 2 and 3 respectively. If there is a difference in statistical units underlying (i)
information on ownership or financial information at the enterprise level and (ii) operational data at the
establishment level, then compilers must account for this difference in their compilation procedures.
C.1

Which data source to choose?

15.17. A comparison of data sources for FATS is suggested in Chapter 11. Chapter 6 provides more
detail on the information that should be sought from these data sources and how they could be used.
15.18. Inward FATS variables can in general be obtained from four main sources:

i. Survey used to collect information on a compiling economy's business structure, identifying the
foreign-controlled population of firms, as well as the country of the UCI. This information could
be sourced from the survey itself by including the appropriate items, or by cross referencing with
information available from other sources (e.g. business register, FDI register). Coverage of
(services) activities is an important consideration when choosing such a source, as in many cases
it "only" covers business activities (i.e. excluding education, health, recreational service activities,
agricultural, mining activities).

ii. Survey used to collect foreign direct investment information, where questions would be included
to identify the variables of interest. Again it would be necessary to identify within the population

230

of foreign direct investment enterprises those that are foreign-controlled as well as the economy
of the UCI249 .

iii. A dedicated FATS survey.
iv. Administrative sources (business registers, tax return or regulatory reports such as for banks or
insurance companies, or more generally on foreign investment, privatization monitoring, etc.)
15.19. Economies that have business statistics this source may be preferred for inward FATS. This
option has been implemented by many economies. Some others have preferred using (existing) FDI
surveys to collect FATS. In fact, this may be an interesting solution as the FATS population may be
relatively easy to isolate (directly from the survey, or from the register used for establishing the FDI
survey). But one needs to keep in mind the need not to overload an FDI survey, in particular if it needs to
be run on a quick turnaround.
15.20. The outward FATS characteristics are usually collected based on (1) FDI surveys, and (2)
dedicated FATS surveys. Outward FATS information is believed to be more difficult to obtain as it
relates to the activities of firms established outside the compiling economy. As a starting point it may be
useful to consider using an FDI survey to obtain data on outward FATS, keeping in mind the pros and
cons of using such a source. A number of economies have chosen this option, particularly when starting
to collect the most important outward FATS characteristics. Alternatively, a FATS survey could be
developed. 250 In general, it would be advisable to conduct a periodic census for outward FATS (i.e.
including all units identified as belonging to the target population of reporting units). Between censuses,
sample surveys could be conducted. Administrative data could also be considered for compiling outward
FATS if the data include relevant variables.
15.21. For some of the variables supplementary sources may be needed, such as merchandise trade
statistics or BOP data on goods or services for the trade variable, research and development surveys for
relevant variables, or administrative or other sources for data for some specific (services) activities
(finance, insurance, oil, mining, other major industries). If needed compilers may also refer to other
sources to gather information on structures or operations of multinationals, such as industry associations,
free zone authorities, stock exchange and securities commissions, special registers of foreign companies
such as those maintained by some international organization (e.g. UNCTAD, OECD, Eurostat) or private
companies (e.g. Dun and Bradstreet), the Internet, and partner country statistics. Again, compilers need to
be cautious of how all this information, including the data collected, aligns with the FATS
recommendations. These sources could also be useful to verify or validate some of the FATS information
compiled. For counterpart statistics it is important to note that the information is likely to be confidential
and should not be able to provide information on individual businesses. Aggregate information may be
available but care should be taken over the interpretation of concepts and the implication of different
compilation methods.
15.22. Although not advised, some countries apply cut-off thresholds. For practical reasons (reduction of
the costs and the burden) such thresholds can, nevertheless, be considered acceptable as long as they are
kept at minimum and estimations are provided for the population under the threshold.

249

The identification of the UCI is discussed in sections E.5.
For example if the FDI survey needs to be run on a quick turnaround, or if compilers do not want to unduly
overburden FDI firms that are not part of the population to be covered by FATS.
250

231

15.23. To summarize, compilers may have various options for sourcing their data for compiling FATS,
and the choice will depend on the situation in each country. However:
i.

For inward, it may be preferable to consider sourcing data from business statistics keeping in
mind that these statistics may need to be complemented for activities that may not be covered by
this dataset, e.g. agricultural activities or various types of social services). Another solution, at
least as a starting point, could be to consider an existing inward FDI survey.

ii. For outward, an FDI or dedicated FATS survey could be used, although the former may be easier
to implement as a first solution.
iii. If an FDI survey is used, response burden should be considered, particularly if the survey is
conducted on a quick turnaround and because a portion of FDI firms are not part of the FATS
universe (i.e. firms with influence relationships, but not control).
iv. The link with FDI, in particular to identify the statistical population, is an important point that
should be considered when choosing a data source and when identifying the population to be
covered and the UCI.
C.2 Country experience: Austria
15.24. The two major sources of the inward FATS statistics in Austria are the comprehensive structural
business statistics that Statistics Austria collects and the direct investment survey that the OeNB conducts.
Foreign-controlled enterprises are a subset of direct investment enterprises, which also include minority
holdings above a 10% threshold and below the 50% majority holding level. The fact that the OeNB has,
for many years, compiled direct investment information on a per-enterprise basis and has included a
question on the “ultimate controlling investor” (UCI) in its survey, allows the OeNB to analyze multiple
minority ownership structures and to establish the country in which the ultimate parent is located, as is
required for FATS statistics.
15.25. As the OeNB’s direct investment survey includes only enterprises above a certain threshold,
instances of foreign-controlled ownership below the threshold need to be identified for the inward FATS
statistics in a second step. This is done by an automated analysis of the administrative Company Register
data. However, the information available does not allow for determining whether the corporate
headquarters of these generally small enterprises are located in a third country; therefore the country in
which the institutional unit exercising immediate control is located is, by default, considered the country
of ultimate control.
15.26. In a third step, the enterprises under direct foreign control determined on the basis of OeNB
surveys and IT-based analyses of Company Register data are checked for any first and second-tier
affiliates they may have in Austria. To this end, an algorithm is applied to Company Register data to
establish, in a step-by-step procedure, all majority holdings in Austria of the enterprises identified thus
far. The process described above enables the OeNB to submit an exhaustive list of all resident foreigncontrolled enterprises to Statistics Austria.
15.27. Statistics Austria’s first task is to link the OeNB’s list of enterprises with the entries in its own
business register. This is feasible using three common identifiers, namely the OeNB’s internal key, the
NSIs unit identifier, and the official company code which is stored in the public Company Register. These
identifiers are available in both data-bases, thanks to a long standing practice of the mutual exchange of
register information (including name address, and economic activity of the unit). Usually 100% of the
units can be matched automatically, without any manual intervention. Next, Statistics Austria once more

232

checks the data, thereby eliminating “inactive” business units from the OeNB’s list. Such inactive units
may appear if they have not yet started economic activity or if enterprises are being liquidated. In some
cases, control relationship may be unclear – e.g. where a change of ownership occurs during the reference
year – and may be corrected at this point in time.
15.28. The most important step is then to retrieve the required characteristics for inward FATS from the
existing database for structural business statistics. The values in this database may be either a direct result
from the SBS survey, if the enterprise in question was part of the sample, or they may be imputed data,
based on regressions with employment and turnover as exogenous variables, if the relevant enterprise was
not part of the sample. In any case, the data used for reporting inward FATS statistics are exactly the
same as for SBS statistics, which is an important aspect of quality. Every second year, the data is
supplemented by information about the R&D activities of foreign affiliates in Austria. These R&D
characteristics of foreign affiliates are to be derived from the available R& D statistics by linking data sets
at the individual enterprise level.
C.3 Country experience: Spain
15.29. The Spanish INE has long recognized the importance of indicators of Spanish subsidiaries of
foreign companies and has carried out several pilot studies or surveys on subsidiaries abroad in the years
before entry into force of the European regulations. Reducing the statistical burden is a strategic objective
of the INE and that, within this line of action, one of the theoretical principles to consider when starting
and developing new statistical projects is promoting the use administrative data and prioritizing them in
relation to the collection in the field, whenever a reliable and timely administrative source from which
they can effectively meet research objectives is available.
15.30. Consistent with this view, and in parallel with some pilot studies, the INE explored the possibility
of obtaining the information from the Foreign Investment Registry of the Ministry of Economy and
Competitiveness (MINECO). This Register is formed from statements by the Directorate General for
Trade and Investment, and more specifically the Directorate General of International Trade in Services
and Investment, collected about foreign investment in Spanish companies and Spanish investment in
foreign companies, both as regards flows and stock.
15.31. Specifically, in the case of Spanish investment stock in foreign companies, Spanish resident
investors that make investments in foreign companies whose net worth exceeds a threshold and in which
the investor's share capital or total voting rights are equal to or greater than 10%, must submit an annual
report on the development of foreign investment in the first nine months of each calendar year. Also
included in this report is the investment in foreign companies whose activity is the holding of shares in
the capital of other companies, regardless of the amount of the investment. Holders of foreign branches
also report the size of the investment in the annual report. Information is obtained from the companies in
which the investor invests and in turn, investee companies. Finally the third level of the chain of
participation in subsidiaries is reached, although less detailed information from the latter is collected.
15.32. The Foreign Investment Registry covers Foreign Direct Investment in Spain and Spanish Direct
Investment Abroad, published annually and included in the National Statistical Plan (PEN), and presents
Foreign Direct Investment in Spain and Spanish Direct Investment Abroad quarterly and bi-annually as
well as the PEN. Foreign Investment Registry statistics are also obtained on Spanish Investment in
Foreign Securities and Foreign Investment in Transferable Securities Spanish, both flows and position. In
meetings between the two institutions involved (MINECO and INE), officials discussed the
methodological alternatives, implement procedures and legal issues arising from the eventual transfer of
the relevant files. Further analysis of the information provided by the Register and comparative studies

233

conducted with pilot surveys highlighted the quality and wealth of information, as well as the role it could
play as a source of primary data to meet the objectives.
15.33. Finally, after the evaluation of the various advantages and limitations associated with each of the
procedures, the administrative source was chosen as a more effective means of achieving the objectives of
this research. However, it should be noted that while the data derived from said Investment Register are
considered as the primary basis for generating key information, the data has been supplemented with
information from the European Register of Groups (*) in order to obtain the final results of the
investigation. It should be noted that the coverage of branches in the Investment Register is not complete
as far as the population of subsidiaries are concerned, because according to the current rules, companies
are not required to declare investments in companies with assets of less than a threshold. The results of
the subsidiaries from the third level of the chain of participation are not shown either. It has therefore
been necessary to perform additional processing and estimation taking into account European Register of
Groups (information on enterprise groups from private databases and records units in the statistical offices
of the individual Member States and the countries of the EFTA) to measure the activity of these
subsidiaries and through appropriate adjustments, to derive the final results of the statistic.
15.34. In conclusion, the statistical design for the primary data generated from the Foreign Investment
Registry of the Ministry of Economy and Competitiveness, can reduce the statistical burden on
enterprises while allowing for national dissemination of the main results on Spanish subsidiaries abroad.
D.

General compilation issues
D.1

Which statistical unit to consider?

15.35. For inward FATS, the statistical units are all the enterprises (or establishments) and branches that
are under foreign control and for outward FATS, the statistical units are all the enterprises (or
establishments) and branches abroad that are controlled by an institutional unit resident in the compiling
economy.
15.36. MSITS 2010 makes no recommendation as to the choice between enterprises or establishments
for statistical units and explains that both approaches have strengths and weaknesses (see chapter 6 and
MSITS 2010 paragraph 4.13-4.14). In addition, if the collection takes place at the enterprise level, then it
is possible that it is also done at the local enterprise group level.251 There is, therefore, an even greater risk
of interpretation of the activities of the local enterprise groups as it will regroup units engaged in multiple
activities, making data classification more difficult when it comes to compilation. The statistical unit for
FATS will in fact heavily depend on the statistical units of the existing national statistical systems if
defined, or on the limitations that exist in the data collection system. Because the statistical units can
have an important bearing on how the statistics are interpreted, both in isolation and in comparison with
other data sets, the compiler should consider this when compiling FATS. In addition, it is a good practice
that metadata on the statistical units used in collecting FATS be disclosed in explanatory notes given that
the level of consolidation may affect users’ interpretation of the statistics.
15.37. Compilers should also consider that units may differ between inward and outward FATS. For
example, to the extent that for inward FATS, structural business surveys are the basis, the unit for this
compilation is likely to be an establishment (recognising the need for a degree of industry homogeneity

251

It is advisable that these enterprises in the host
economy that are part of a common enterprise group
are
identified and see how this consolidation question
can be addressed.

234

within the unit); for outward FATS, the use of FDI surveys as a potential basis will tend towards an
enterprise or local enterprise group as the unit for compilation. The use of different units for inward and
outward FATS will make within country comparisons and balances difficult, particularly for business
counts. This is why compilers in the longer term should identify ways, if possible, of aligning the
statistical units used in both inward and outward FATS.
D.2.
Choice of reference period, periodicity of compilation and valuation of monetary
variables
15.38. MSITS2010 does not make any recommendation as to the choice of the reference period for
compiling FATS.252 In practice, economies that are compiling FATS use years as a reference period. At
the time of this writing, such a choice seems reasonable from the perspective of users, as well as with
respect to the reporting and compilation burden. However, in the future there may be a need for data
referring to shorter periods of time, e.g., quarters, at least for some main FATS aggregates.
15.39. MSITS2010 also does not make any recommendation as to the periodicity of update of FATS
data. In practice, agencies producing FATS do so every year, at least for the variables that are deemed to
be the most important (i.e., priority variables). In some cases it may be reasonable to consider compiling
FATS less frequently (e.g. every odd year) for the variables/breakdowns of minor importance or for those
that are most complicated to produce.
15.40. In principle, data for a given reference year should correspond to the calendar year, but firms may
report on a fiscal or accounting year basis if that is what is available. In addition, it should be noted that
some characteristics are compiled from already existing data, in particular for inward FATS. For these
characteristics, the reference period may have to be taken as a given and might not always coincide with
the calendar year. Compilers should clearly inform users of the reference period used, and if there are any
deviations for some variables.
15.41. Consistent with other economic statistics, activity should be recorded as it occurs rather than
when the related payment is made (accrual basis). Flow variables should be recorded for the whole
reference period, while stock variables should, if possible, be recorded as of the end of the reference
period. An exception is the employment variable, for which use of a period average is preferable (see
also section D.4) if the employment is subject to strong seasonal variation. Data for a given reference
period are mainly valued at the average prices and, if relevant, average exchange rates of that period. For
stock variables (for example, property plant and equipment), values of source data might be recorded at
historical cost and will largely reflect prices at the time the asset was acquired rather than prices of the
reference period. In such a case, revaluations to reflect current-period prices should be made by the
compiler.
D.3.

Treating the Activities of Special Purpose Entities in a FATS context

15.42. Special Purpose Entities (SPEs) are engaged in various activities and take various forms. Thus,
compilers should pay attention to the treatment of SPEs in compiling FATS. Although there is no precise
definition of SPEs in the 2008 SNA, BPM6 and BMD4, these documents include some elements that help
to better identify SPEs and their activities.
15.43. A unit can be considered an SPE, if it meets the following criteria:

252

HEGI recommends that the calendar year is used as a reference period.

235

It is a legal entity:

i. Formally registered with a national authority253 and
ii. Subject to fiscal and other legal obligations of the economy in which it is resident;
SNA 2008, paragraph 4.6.for definition of a legal entity.

ii.

The entity is ultimately controlled by a non-resident parent, directly or indirectly;

iii.

The entity has no or few employees, little or no production in the host economy and little or
no physical presence in the economy in which it is created by its parent, which is typically
located in another country;

iv.

Almost all the assets and liabilities of the entity represent investments in or from other
countries;

The core business of the entity consists of group financing or holding activities; i.e.,
channelling of funds from non-residents to other non-residents. However, in its daily
activities, managing and directing plays only a minor role.

15.44. According to the 2008 SNA, judgment has to be made regarding the independence of an SPE as
an institutional unit. However, non-resident SPEs owned by residents (and resident SPEs owned by nonresidents), which are the main focus of FATS, are treated as independent institutional units by convention.
When those SPEs transact by their own decision, their transactions, in principle, should be recorded in the
same manner as those of non-SPE foreign affiliates.
15.45. At the same time, however, it should be considered that SPEs are often established to facilitate
cross-border ownership of enterprises and thus generally do not have any substantive (services) activities.
Thus, special treatment of SPEs in FATS (and FDI statistics) might provide economically more
meaningful data and thus enhance their usefulness.254 For example, it would be relevant to separately
identify SPEs and to compile supplementary data not considering certain types of SPEs.
15.46. Nevertheless, some SPEs are engaged in services activities in addition to passing through funds,
and such types of entities should not be ignored in FATS. The appropriateness of considering SPEs in
FATS depends on their activities. Identifying their activities is not straightforward but one way of doing
so is to classify SPEs according to the 2008 SNA sector classification assuming that such classification is
conducted by statistics compilers of host countries.
15.47. Among various types of SPEs, holding corporations, shell companies and conduits are often
established only to channel the funds from non-residents to non-residents. Thus, supplementary data
looking through such pass-through entities could be useful for users in relation to FATS. In contrast,
royalties and licensing companies, merchanting companies, and securitisation companies are engaged in
non-captive financial and non-financial activities. The other types of SPEs provide financial services
although most of them are captive financial corporations. Thus, these non-pass-through entities should not
be looked through.

253

This excludes non-resident units that have been registered for VAT purposes only, which may be the case in
some EU-countries acting as a hub for the imports of goods to the European Union.
254
See BD4 for more information.

236

15.48. Compilers should also consider that it may be difficult to collect information on SPEs given their
little or no physical presence or how they are accounted for in the national statistical system. In fact the
treatment of these in national statistics will strongly influence the possibilities for the FATS compiler. See
chapter 6 for more information.
15.49. More guidance on SPEs and their treatment in statistics may be found in Eurostat's FATS
Recommendations Manual, as well as the UNECE/OECD/Eurostat publication on "The impact of
globalization on national accounts" in chapter 4.
D.4.

European Union practice

15.50. In the European statistics, special purpose entities are relevant for both inward and outward
FATS. The Eurostat FATS Recommendations Manual further clarifies that "They should be excluded
from the target populations of statistical or reporting units only if they had no turnover and no
employment during the reporting period."
15.51. There is no single approach to how SPEs should be treated, even though the main principle to be
followed is that the SPEs should not be automatically excluded from the target population as they carry
out important economic transactions with their respective parents or associated enterprises. SPEs should
be excluded only if they had no turnover and no employment during the reference period.
D.5.

Other considerations

15.52. It is good practice that compilers evaluate the validity of reported data and establish checking
procedures in the context of FATS (e.g. registers, etc.). Survey processing staff will need to evaluate the
reported data to ensure accuracy, consistency, and reasonableness, and this is particularly relevant for
FATS. Many of these techniques will be related to the usual work expected from compilers in other
statistical domains. For compilers that have chosen such a solution to gather FATS data, the strong link
between the structural business statistics and inward FATS should be taken into consideration when
validating the data. Consistency of information, in particular for the identification of the population or
UCI could be ensured by checking with other registers that some compilers may have at hand, for
example registers of enterprise groups or foreign-owned enterprises. More information on evaluating the
validity of reported data are available in chapter 19.
15.53. In most countries there are legislative requirements for confidentiality that necessitate avoiding
disclosing information, and these would also naturally apply to FATS. Multinational companies that by
definition are involved with FDI and FATS may be particularly sensitive about such matters for
competitive reasons. Even countries in which FDI and FATS data collection are mandatory maintain
strict confidentiality. This is particularly relevant given the granularity of the data obtained through the
compilation system, i.e., breakdowns by detailed country, by detailed activity and by product.
15.54. For example, when compiling data by activity (e.g. according to ICFA Rev.1 categories)
categories may sometimes have to be suppressed (that is, not separately shown) so as to preserve the
confidentiality of those data for individual companies. This need for suppression occurs most often at the
most detailed level of the classification, when smaller countries are involved or in cases where the data
are cross-classified by country or area. Other possibilities exist to ensure confidentiality of information,
such as perturbation techniques which may be sufficient to mask individual responses and increase the
utility of the statistics for the analysts. The compiler should therefore identify appropriate techniques to
ensure the confidentially of data reported by individual companies. (See Chapter 20 for further details on
dissemination and confidentiality issues).

237

Compiling FATS Variables

15.55. Economic variables — operational and financial — with regard to FATS should be collected
primarily for their usefulness for trade policy needs and for analysing the globalization phenomena.
Practical issues associated with obtaining data must also be considered. With such considerations in mind,
and in the interests of harmonization with other international guidelines (see paragraph 15.6), MSITS
2010 recommends that the FATS variables to be collected include at least the following basic measures of
foreign affiliate activity:
i.

Sales (turnover) and/or output;

ii.

Employment;

iii.

Value added;

iv.

Exports and imports of goods and services;

Number of enterprises.

15.56. In the context of this Compilers Guide, output and/or sales (turnover) are the variables of most
interest, followed by employment and number of enterprises. Although the five variables identified above
constitute a basic set that can provide answers to a variety of questions, additional variables may prove
useful in addressing specific issues (among these variables are compensation of employees, gross fixed
capital formation, total assets, and research and development expenditures). MSITS 2010 suggests several
additional measures that might be considered for collection by countries that are able to compile such
information. Most of the “basic” and the “additional” variables, as well as their definitions, have been
drawn from the 2008 SNA. 255 It may not always be specified, but there is a particular interest in obtaining
detailed data for services sectors (either activities or products if relevant) for all the variables listed below.
E.1.

Sales (turnover) and output

15.57. Sales and turnover are terms which are used interchangeably in this Guide. They refer to the
totals invoiced for the sales of goods or services rendered, including all duties and taxes, excluding
consumption and sales taxes (i.e. VAT type taxes). Sales/turnover also includes all other charges passed
on to the customer (transport, packaging, etc.), even if these charges are listed separately in the invoice.
Reductions in prices, rebates and discounts as well as the value of returned packing must be deducted.
Charges for the use of intellectual property are included. Income classified as other operating income,
financial income and extra-ordinary income in company accounts is excluded from turnover. It is
important to note that revenue of long-term contracts, such as building contracts should be recognised by
reference to the stage of completion of the contract. For insurance, turnover/sales correspond to gross
premiums written and for pension funds they refer to total pension contributions. For financial services
excluding insurance, pension funding, and auxiliary services, turnover refers to the explicit fees charged
for the services provided (fees and commissions receivable) as well as interest and leasing income,
income from shares, net profit on financial operations, income on foreign exchange transactions and other
operational income.

255

Other variables not mentioned in the Manual on Statistics of International Trade in Services as basic and
additional variables may also be collected in order to cover important needs, in particular “hours worked” (SNA
2008, paragraph 19.78) which are relevant in the context of the measurement of productivity. More details on
variables and their definitions may be found in OECD's BD4 and HEGI.

238

15.58. The variable sales or turnover is usually easy to compile because it is easy to collect and besides
the usual adjustments for nonresponse, under-reporting or grossing-up, no specific calculations are needed
to produce the final data. In addition, sales or turnover generally offer more options for disaggregation, in
particular for having a breakdown by service product.
15.59. As indicated above, output 256 is a preferred variable considering mode 3 supply of services.
Output is a superior and more refined measure than sales for most purposes and is recommended as the
preferred variable for compilation. Following the 2008 SNA, output differs from sales mainly because it
includes changes in stocks of finished goods and work-in-progress. In general, these considerations do not
apply to services. There are also differences in measurement applicable to wholesale and retail trade,
insurance, and financial services industries. For affiliates in wholesale and retail trade, insurance, and
finance, sales may include non-service elements or may exclude the value of services provided without an
explicit charge.
15.60. For the wholesalers and retailers, their output of services should be measured as trade margins—
wholesale or retail sales of goods or services less the goods and services purchased for resale. While not
in the core list of variables to compile, and even though it is only mentioned as a suggestion for
compilation in MSITS 2010 (paragraph 4.65), compilers will need to make sure that data on purchases of
goods and services for resale in the same condition as received are collected or can be estimated.
15.61. For insurance, international economic accounting guidelines recommend measuring output as
gross premiums earned plus premium supplements minus estimated claims incurred (either estimated
claims or benefits due) and, in the case of life insurance, minus the net increase in life insurance actuarial
reserves. For financial intermediaries, in addition to the explicit fees charged, compilers should include a
measure of implicit service fees as well as financial intermediation services indirectly measured.
15.62. In general, output statistics can be compiled directly from data collected (i.e. services sales or
turnover) or extracted from existing sources, making necessary adjustments and taking the above points
into account. The latter may entail that additional data collections, data from outside sources, and
estimation methodologies must be used. The suggested practice is as follows:
i.

Output is the preferred measure, but sales/turnover is easier to compile as a first step and may
offer more options for disaggregation. For wholesale and retail trade, insurance, and finance,
some additional calculations are needed to compile output.

ii. Activity/product breakdown257 :
a. Output or sales/turnover should first be compiled by activity.
b. A breakdown of output or sales/turnover into total goods and total services products for
each activity should be provided.
c. As a long term goal, compilers should strive to produce product detail, if possible
compatible with EBOPS2010.

256

Output can be valued at producer or basic prices. The latter is generally preferred.
More information on breakdowns by partner or service activity and products is provided in relevant sections
under E.
257

239

iii. Output or sales/turnover should not only be broken down by country of UCI for inward or by
establishment of affiliates for outward, but a distinction should be made as to the output actually
delivered to clients in the host economy. This is further elaborated in the section on trade
variables.
E.2.

Example of United States

15.63. The U.S. BEA compiles data on sales/turnover by industry, with a distinction between sales of
goods and sales of services for each industry. To estimate the output of wholesale and retail trade,
insurance, and finance, the BEA initiated new data collections, developed new compilation
methodologies, and identified data from other sources.
15.64. To construct estimates of distributive services supplied through affiliates, the U.S. BEA collects
data on the cost of goods sold and the beginning- and end-of-year inventories of the goods for resale on
its censuses of FATS, which are conducted every five years.258 In between its censuses, BEA uses data
on the domestic wholesale and retail trade industries to produce estimates of distributive services supplied
through affiliates (based on movements in trade margins).
15.65. To estimate the value of insurance output supplied through affiliates, the U.S. BEA deducts
“normal” losses—a proxy for insurers’ expectations—from premiums earned. To do so it uses data
collected on its FATS surveys on the premiums earned and losses paid by majority-owned affiliates with
operations in insurance. BEA uses a six-year moving average of the relationship between premiums
earned and losses paid to estimate normal losses. When it first introduced this compilation method, to
avoid having to wait until 6 years of data had been collected on premiums and losses, BEA used data on
the relationship between premiums and losses from the domestic insurance industry and gradually
replaced them with data collected from affiliates. In addition, information collected on investment income
of majority-owned affiliates with operations in insurance collected on its FATS surveys are combined
with data on the domestic insurance industry to estimate the value of investment income earned on
technical reserves (i.e. premium supplements). This is done using data on the domestic insurance industry
to calculate the share of total investment income that is income on technical reserves and applying this
share to affiliates’ investment income to derive an estimate of premium supplements.259

258

Originally the data were collected through the annual surveys but moved to collecting only on benchmarks to
reduce burden on reporters and BEA resources.
259
BEA uses data from a global credit rating agency which has a focus on the insurance industry. It breaks out total
investment income earned by the domestic insurance industry into categories, including one called “investment
gains on funds attributable to insurance transactions”—that is, on their technical reserves. BEA takes the percentage
that this category accounts for in total investment income and multiplies this by the investment income attributable
to insurance for the affiliates to derive their premium supplements. The break out by destination is used to define
the destination of their sales of services.

240

Figure 15.2 Impact of compilation of services output by U.S. Affiliates of Foreign multinational
enterprises, 2002

15.66. The U.S. BEA collects and compiles FATS data for banks. It constructs its estimates for financial
intermediaries based on data collected on both banks’ explicit fees and services charges, and on their total
interest paid and received to estimate the value of output by affiliates in banking.
15.67. The impact of the compilation of output /supply of services as opposed to sales of services by
foreign affiliates in the United States (inward FATS) is illustrated in the chart below. Although the impact
may differ from one country to the other, it is interesting to note that using the more refined variable
output to measure the supply of services through mode 3 may substantially change the picture as
compared to the use of sales/turnover of services. The example below shows that the preferred measure of
output increases by almost one third the estimate for mode 3 supply of services as opposed to data on
sales of services.
E.3.

Value added

15.68. Value added (in services production) is also a variable of interest for analysing the impact of
mode 3 of services supply. The 2008 SNA defines the gross value added of an establishment, enterprise,
industry or sector as the amount by which the value of the outputs produced exceeds the value of the
intermediate inputs consumed. A related concept—net value added—is defined as gross value added less
the consumption of fixed capital.260 Gross value added can provide information about the contribution of
foreign affiliates to the gross domestic product of a host country, both in the aggregate and in specific
industries. Gross value added receives higher priority than net value added and is often easier to compute
because it does not require estimation of capital consumption.
15.69. Value added cannot be directly observed from accounting records. Two ways are available to
derive value added. First, as gross output (revenue) less its intermediate inputs or consumption

260

SNA 2008, paragraph 6.74.

241

(purchased goods and services used in production). The valuation of value added is linked to the gross
output. Second, value added can be derived as the costs incurred (except for intermediate inputs) and
profits earned in production. Costs generally fall into four categories: compensation of employees, net
interest paid, taxes on production and imports, and the costs of capital consumed in production.
Alternatively it can be calculated from gross operating surplus by adding personnel costs. Choosing one
or the other method of computation will depend on the availability of data, as well as the method used to
compute value added in national business statistics. For inward FATS, value added will often be easily
derived from information available from regular industrial or enterprise surveys, but for outward FATS,
value added will often have to be derived from various variables collected in FATS (or FDI) surveys.
15.70. The valuation of value added will depend on the valuation of output (basic prices or producer
prices) while intermediate inputs are always valued at purchasers' prices. Alternatively, value added can
be valued at factor cost. For more information see 2008 SNA, 2.65.
E.4.

Trade variables

15.71. International goods and services transactions of foreign affiliates constitute another basic
indicator of activity. As previously described, on the exports side this variable is related to the
sales/turnover or output variables. Imports are a part of the purchase of goods and services. MSITS 2010
recommends that the trade variables reflect international trade in goods and services between residents
and non-residents, i.e. are compiled on a balance of payments basis. It should be noted that, for the above
definitions to be applicable to both inward and outward FATS, the concept of ‘residents’ has to be
interpreted in different ways. While for inward FATS, ‘residents’ refers to the compiling country where
the foreign affiliate is resident, for outward FATS it is defined as the host country where the foreign
affiliate is located.
15.72. Ideally, data would need to be compiled for exports and imports for each activity and economy of
operation of the affiliate (for outward) or UCI (for inward), by type of product for goods and services
(with a particular focus on services given the interest of this information for services trade
negotiations/analysis), as well as the origin or destination of the international trade conducted by the
affiliate with a distinction between trade with related enterprises and trade with unrelated parties.261 This
is a complex dataset to build.262 But beyond the specific focus on services provided by MSITS 2010, there
is a strong interest for such information for the analysis of globalisation and more particularly on global
value chains and trade in value added.
15.73. As outlined in the section on sales/output, for measuring mode 3, having an estimation of total
services (separated from total goods) that are sold by affiliates to entities in the economy of establishment
separated from total services (and total goods) that are exported outside the country of establishment
would provide useful and often sufficient information from the perspective of the measurement of the
international supply of services. The compilation of such variables could be done if this distinction is
possible in the data sources (most likely for the sales/turnover variable), both for inward and outward
FATS.
15.74. In the broader context of globalisation analysis, more information could be considered for the
compilation of these variables. This would mean having total goods and total services broken down as
follows:

261
262

The distinction between related and unrelated parties is made according to the definitions used in FDI statistics.
It is also important to note that exports and imports cannot normally be isolated in standard company accounts.

242

Exports and imports of affiliates with parent enterprise.

ii.

Exports and imports of affiliates with other related enterprises (in third countries,
without distinction)

iii.

Exports and imports of affiliates with unrelated parties in the country of the parent
enterprise and third countries (without distinction between third countries).

The compilation of such variables could be done for exports in the same line as described above, that is if
the distinction is possible in the data sources for the sales/turnover variable, both for inward and outward
FATS. However, for imports one would need to develop appropriate questions so that the information
necessary for the compilation of these variables is available.
15.75. The possibilities for the compilation of the more detailed product-origin/destination of trade
breakdown will largely depend on the sources used to obtain the data. Data could be derived from
information provided by reporting units in sources used for FATS (see section 15.C), whether business
activity surveys (only for inward), FATS or FDI surveys. However, these approaches may require a
significant increase in the level of detail requested from reporting units, which is something compilers
must consider before pursuing one of these approaches. If relevant, some separate questionnaires could be
used to collect the details, in particular for the larger companies for which affiliates are more likely to
conduct international trade.
15.76. However, linking information with balance of payments trade in goods (or merchandise) and
services data could ensure that these additional detailed statistics can be compiled. This can be
implemented for inward FATS.263 When the data are obtained through linkages with primary data sources
for balance of payments transactions, breakdowns by product and by origin or destination will often be
possible. If this is the case, exports and imports of services (or goods) may be disaggregated, not only by
the primary activity of the affiliate according to ICFA Rev.1 but also by product, on a basis compatible
with EBOPS 2010.264 However, to be able to do this the compiler needs to be able to distinguish in the
balance of payments data the transactions between related and unrelated parties. For services this is in line
with the text provided in MSITS 2010 paragraph 3.56.
15.77. For the inward FATS trade variables, relevant information could be obtained through the linking
of trade and business statistics, such as the in the context of the Services Trade by Enterprise
Characteristics (STEC) that some countries have been developing in recent years. STEC describe trade
flows between countries broken down by the characteristics of traders such as type of trader, activity
sector, size class, or type of ownership265 . The latter variable would be of particular interest to establish
this link with the FATS trade variables and therefore produce relevant breakdowns as described above266 .

263

If merchandise data are used as a source instead of goods, then compilers need to make necessary adjustments to
comply with the recommendations regarding resident- nonresident transactions in goods. If this is not possible, then
the differences in concept should be clearly documented. See MSITS 2010 Box III.1 or EU FATS
Recommendations Manual for further information. The OECD HEGI Chapter 5 also provides useful information
with respect to the measurement of intra-firm trade.
264
Naturally a similar consideration can also be made for goods/merchandise.
265
External trade by Enterprise Characteristics, European Communities, 2007
266
Precondition for establishing TEC statistics are the availability of business and trade registers in a country.
Through linking both, normally at the level of statistical unit (legal unit) and a common identifier, trade flows can be
associated to trader characteristics. For reliable statistics, a high matching rate between records in the linked
datasets is required.

243

15.78. Compilers should be mindful of the high burden for reporting and compiling this information
when considering the production of data for the trade variables. Consequently, it is advised to approach
this in stages:
i.

First produce data to measure mode 3, by distinguishing for each activity (or services product if
this is available) between sales or output of services in the country of the affiliate from services
exported.

ii. Compile in a second stage trade variables, breaking down the exports and imports data into (i)
total goods and total services (ii) trade with the parent enterprises, other affiliated enterprises and
unaffiliated parties (iii) trade with the country of the parent enterprise and with third countries.
iii. In a third stage compile the more detailed information by product (group) and by country.
Synergies should be found for the production of detailed data for the trade variables, in particular for
inward FATS, e.g. STEC.
15.79. In the European Union for outward FATS, two sets of trade variables are suggested for
compilation (subject to pilot studies): "exports and imports of goods and services" and "intra-group
exports and imports of goods and services", intra-group referring to trade between enterprises controlled
by the same unit, and the controlling unit itself. The recommendations/definitions provided are in line
with those of MSITS2010 and related guidelines (BD4, HEGI). The EU FATS Recommendations Manual
also provides additional clarifications as to the trade conducted by foreign affiliates: "Exports/imports of
goods and services can be performed directly by the foreign affiliate or indirectly by means of other firms,
such as wholesalers, that are resident in the same country and act on behalf of the foreign affiliate. In this
latter case no property right transaction must occur between resident firms.
E.5.

Employment

15.80. The MSITS 2010 recommends measuring the number of persons on the payroll of foreign
affiliates, whether full-time or part-time. The enumeration should specify the period covered (such as an
annual average), or, in the absence of strong seasonal or other fluctuations, a point in time, such as the
end of the year. The compilation of this variable should generally follow the national practices. Although
data on a full-time equivalent basis would be preferable, given that this measure is not widely available,
MSITS 2010 recommends that data on the number of persons employed is compiled (i.e. inclusive of
employees, working proprietors).267 To serve the information needs for mode 4, a useful extension would
be the separate identification of employment of non-residents of the compiling economy working for the
affiliate, of which those that are intra-corporate transferees (i.e. employees transferred between affiliated
enterprises).
15.81. Data on employment can be derived from the information collected through FDI or FATS
surveys, or from business statistics (for inward). Again, it could be possible to derive the data by linking
to other data sources, such as administrative sources, by using common company identifiers. Data should
be compiled as a period average, but in the absence of strong seasonal and other fluctuations it may be
compiled at a certain point in time (e.g., end of the year). The data should be derived from information
that is generally readily available from company accounts.

267

Although not recommended in MSITS 2010 other measures may also be compiled in order to cover importan t
needs, in particular “hours worked” (SNA 2008, paragraph 19.78) which are relevant in the context of the
measurement of productivity.

244

E.6.

Number of enterprises

15.82. The preferred measure is number of enterprises, but data should reflect the number of statistical
units, that is enterprises or establishments, which correspond to the data described for other variables. It is
a basic indicator of the prevalence of control by foreigners in the host economy. It should be made clear
that the enterprises to be taken into account in these statistics are those with a genuine economic activity.
Firms which have no activity, but simply set up an office in a country for the purpose of transferring
funds or taking advantage of certain tax breaks, should be excluded from this category of data.
15.83. Although the data compilation of this variable is considered as a by-product of the FATS data
collection and compilation, a number of considerations need to be taken into account, such as determining
if units have been active during the period under consideration. In addition, the number is likely to be
affected, often significantly, by the level of company consolidation and by thresholds for reporting on
surveys.268 Given that for outward FATS the data will often be reported by one unit on behalf of a group
of affiliates (e.g. country head office, regional office), it is important to ensure that data are reported and
compiled on a non-consolidated basis. Consequently, to avoid any problem of interpretation that there
may be if data do not reflect the above principles, it is important that any deviation is clearly documented
in available metadata. . Whatever measure is retained, to assist users in interpreting counts of units
(whether enterprises or establishments), compilers are also encouraged to document how the numbers
were derived.
15.84. The EU FATS Recommendations Manual provides further indications as to how this item could
be compiled. It states that "only active units which had turnover or employment at any time during the
reference period should be included. Dormant (temporarily inactive) and inactive units are excluded. This
statistic should include all units active during at least a part of the reference period." It also indicates that
not only separate legal entities which are dependent on foreign enterprises but also "local units (branches)
which do not constitute a separate legal entity and which are dependent on foreign enterprises …" should
also be included. In other words if they are considered as branches (i.e. they meet the branch criterion, see
for example chapter 14 on construction) then their operations in the country of establishment should be
reflected in FATS, and they should be counted. But if they do not then these operations should not be
accounted for in FATS.269 It also states that in the context of outward FATS the number of enterprises is a
characteristic that in principle is obtained by counting the units in the target population, although in
practice there can be collection problems when a resident parent enterprise supplies information on the
activity of many affiliates abroad. The respondent may, for example, tend to group affiliates located in the
same foreign country as a single enterprise.
E.7.

Other FATS variables

15.85. Other FATS variables may, for certain countries, hold equal or even greater importance than
some of the priority variables as described in MSITS 2010. The list of additional variables provided in
MSITS2010 is the following: assets, 270 compensation of employees, net worth, net operating surplus,

268

Note that HEGI also recommends the compilation of the number of consolidated enterprises. See HEGI
paragraph 267 for more information.
269
And its sales will be recorded as imports/exports under the relevant BoP services item.
270
Often referred to as the "balance sheet total." Th e OECD was at the time of writing working on the issue of
linking (real) economic variables from FATS with financial variables from balance sheets and income accounts. See
Harmonisation and integration of financial and economic measures of multinational enterprises - Final report,
OECD working Group on International Investment Statistics, March 2013, DAF/INV/STAT(2011)10/FINAL.

245

gross fixed capital formation, taxes on income, research and development expenditures271 and purchases
of goods and services. For some of these variables, data are already being compiled by some countries in
their business statistics, national accounts, etc. The definitions of these variables are provided in MSITS
2010 (paragraph 4.65). These definitions are drawn from the 2008 SNA, which may be consulted for
additional information.
15.86. Data are usually derived directly from information collected, with necessary adjustments to the
statistical process. This means that the production of such data will necessarily entail an increase in the
reporting burden of respondents (except for inward FATS if data are derived from existing data collected
on business statistics). Only one of the additional variables listed in MSITS 2010, purchases of goods and
services (intermediate consumption), can be directly derived from the basic measures recommended for
compilation. It can be derived from gross output and gross value added if intermediate consumption
cannot be collected directly.272
15.87. The list of additional variables and their definitions will also depend on those used in the
compilation of national business statistics and national accounts. In general, the practice is to compile
most of these additional items for inward FATS rather than outward, at least in the first stages of
development of the framework at the national level. The choice for compiling additional variables will
depend on the priorities and feasibilities in a given country. If the data for inward FATS are derived from
the source used for structural business statistics, then the list of variables will most likely be driven by
what is included in it. Countries could also provide variables that are not listed in MSITS2010 or in other
international guidelines, but that are of interest in that country.
E.8.

Country experience: Vietnam

15.88. In Vietnam data on foreign direct investment enterprises are collected and compiled through an
Annual Enterprises Survey (AES) conducted by the General Statistics Office (GSO) since 2003. AES
cover all FDI enterprises in Vietnam. According to the results of Establishment Census 2012, number of
FDI enterprises is 9500. The separation out of FATS enterprises from the list of FDI enterprises could be
presented using the criterion of over 50% foreign ownership.
15.89. The statistical unit of the inward FATS statistics in Vietnam is the enterprise. Twelve inward
FATS variables can be obtained from the Annual Enterprise Survey data. The FATS variables which are
presently collected by the Annual Enterprise Survey include: basic variables such as sales, turnover,
output, employment, and additional variables such as: assets; net worth, operating surplus, gross fixed
capital formation, taxes on income; R&D expenditure, compensation of employees. Beside these
variables, some more detailed indicators of main activity of FATS enterprises can be also compiled based
on the data filled in questionnaire of result on main activity as manufacturing, construction, distribution,
transportation, financial services, insurance and some others. All above variables can be compiled, are
published annually comprising different detailed tables breakdown by country/kind of activity; kind of
activity/country.
15.90. Data on exported and imported goods are collected from the Customs resource through matching
customs database and enterprises survey database based on the tax code of each enterprise which is
provided by the tax office/register office whenever the enterprises finished all register schedules. The data

271

This variable is important as FDI is deemed to be a source of technology transfer. Note that the OECD Handbook
on Economic Globalisation Indicators also suggests "Technology payments and receipts." More related in formation
on R&D definitions, etc., may be found in the Frascati Manual.
272
As noted previously, it is also linked to the imports variable.

246

of service exports and imports of some enterprises are collected through ITS sample survey
quarterly/annually.
15.91. The classification of enterprises by type of activity was determined in accordance with the
Vietnamese Standard Industry Classification of Economic Activities 2007 (VSIC 2007). The data cover
all sectors except Agriculture and forestry.
F. Attribution of FATS variables
15.92. MSITS 2010 recommends that FATS variables are broken down: by activity of affiliates, by type
of service rendered and by partner country. Compilation guidelines for these three types of breakdowns
are provided below.
F.1.

Activity breakdowns

15.93. Although the focus in this Guide is on services, in general the compilation should cover all
economic activities (i.e., goods and services activities). The breakdown should be made according to the
main activity of enterprises using the ISIC, Rev. 4, or using an activity classification compatible with
ISIC Rev.4. This basis of presentation allows activities of services enterprises to be viewed within the
context of the activities of all enterprises. In general, it will be easier to produce a more detailed ISIC
Rev. 4 breakdown for inwards FATS than it is for outward FATS. Compilers should, at minimum,
produce data for the activities that are of importance for their economy or international trade/investment
negotiations. In the context of information needs of trade in services negotiations, compilers should aim at
distinguishing and providing more details for services activities, and having in mind the categories as
suggested in ICFA Rev.1, which is recommended for presenting data alongside the resident/non-resident
trade in services data (see Chapter 20).
15.94. The ICFA Rev. 1 categories cover all activities but provide more detail for services than for
goods. Annex II on page 149 of MSITS 2010 offers general guidelines. However, if countries are in a
position to provide more detail than is presented in annex II, this supplementary breakdown should be
compatible with ISIC, Rev.4. A total for services activities should be compiled or estimated to better
respond to information needs on the international supply of services. It is important to note that a
presentation by activity will show data for variables according to the primary activity of enterprises.
Because a given firm will often have secondary activities in industries other than the activity of their
primary classification, the value recorded for any given activity must be interpreted as an indication of the
total activity of enterprises for which the given activity is the most important, rather than as a precise
measure of the value of that activity itself. This needs to be made clear to users (i.e. metadata).
15.95. Compiling data for some activities may sometimes be challenging, in particular if not covered in
the data sources, or given the difficulty to gather information for particular industries (e.g., financial
activities). This is particularly true for inward FATS in that if structural business data are used as a
source, then some activities may not be covered, such as agriculture or some personal or social services.
In this case compilers should derive this information from other sources or estimate the variables for these
particular sectors. Alternatively, at least as a starting point, compilers may wish to concentrate solely on
(services) activities of importance to their economy.
15.96. The activity breakdown should follow the activity of the "affiliate enterprise." For inward FATS,
the activity of the resident affiliate should be used. Frequently for outward FATS, the activity of the
resident investor is incorrectly used by compilers as an approximation for the activity of the affiliate.
Assigning an activity code - contrary to a widespread misconception - is not an easy task. For outward
FATS the only way to compile the information is by using information obtained directly from the

247

respondent. In order to compile information on activity breakdowns, the compilers are encouraged to find
out if affiliates are engaged in production, trade or any other service activity; be able to assign at least 2digit ISIC codes; and obtain a verbal description of the activity. This should be obtained through the data
collection process (see chapter 6). By carefully assessing this information along with consultation of the
internet and public registers, compilers can derive a meaningful allocation of activity. While this
procedure is burdensome at the time the affiliate is initially accounted for, the information typically will
not need to be compiled again for a number of years.
F.2.

Particular treatment for certain activities

15.97. The specific treatment of wholesale and retail trade; and finance, and insurance enterprises was
described in Section E.1, as it mainly related to the variables sales/turnover and output. Particular
attention should also be paid to other activities relating to enterprises operating ships and other mobile
equipment, as well as some leasing arrangements.273 Compilers need to pay particular attention to the
treatment of construction activities with no subsidiary or branch clearly identified in the country where
the construction project is taking place particularly when some large-scale projects have operations that
are not substantial enough for the entity in the host country to qualify as a branch. A consistent treatment
with the balance of payments/trade in services or national accounts statistics should be applied. The
criteria for recognizing if activities are substantial are described in Chapter 14. If it is identified that these
criteria are met, then the operations of the activities of this unit should be recorded in the FATS
framework. In that context, compilers may find it difficult to gather the necessary data, and may need to
turn to alternative data sources or establish new data sources (for example in co-operation with residentnonresident trade in services statisticians, see Chapter 6 for more information).
15.98. This Guide advises to compile information for all economic activities, with details classified
using ISIC Rev.4. The activity breakdown should follow the activity of the "affiliate enterprise" for both
inward and outward FATS. Given the needs for trade in services negotiations, compilers should aim at
distinguishing and providing more details for services activities, keeping in mind the ICFA Rev.1
presentation and at compiling a total for services activities.
F.3.
Secondary activities and compilation of a breakdown by product for sales/turnover,
output and trade variables
15.99. As outlined above, MSITS 2010 recommends that FATS variables be classified by the activity of
the affiliate according to ISIC Rev. 4. Details should be produced having in mind ICFA, Rev.1 and its
more detailed breakdown for services (see above section). Using ICFA Rev.1 for presentation purposes
allows activities of services enterprises to be viewed within the context of the activities of all enterprises
and provides a link for the presentation of services FATS data and resident/non-resident trade in services
data. Within the context of this Guide there is an additional interest in compiling information on services
produced as secondary activities of enterprises, i.e., services produced by enterprises primarily engaged in
goods producing activities (e.g. agriculture, manufacturing). 274 For example, services supplied by
enterprises primarily engaged in manufacturing activities may be of particular interest in the context of
the supply of manufacturing services on physical inputs owned by others. Besides providing a bridge with
the EBOPS 2010 classification primarily used to classify resident/non-resident transactions, ICFA Rev. 1

273
274

See also section D.3 on the treatment of special purpose entities in FATS.
There is also interest more generally in knowing which services activities produce goods, in what proportion etc.

248

provides a framework for displaying services produced as a secondary activity by enterprises classified as
goods producers.275
15.100. In this context, and as a longer-term goal, MSITS 2010 encourages compilers to work towards
disaggregating some of the variables by product—including sales (turnover), output, exports and imports.
This is the type of information that would be of most interest for trade negotiators and analysts. If
implemented for services, this should be done using a product classification system compatible with
EBOPS 2010. Product-based statistics are more likely to be free of problems of interpretation related to
secondary activities and are consistent with the basis on which GATS commitments are made and with
the basis of classification used for trade between residents and non-residents. In addition such a
compilation practice is important to achieve because goods-producing industries can be important
suppliers of services.
15.101. MSITS 2010 and this Guide recognize that this may be a difficult task, in particular with respect
to the data collection as it will necessarily increase substantially the reporting burden for respondents as
well as the compilation work. MSITS 2010 recognizes this and encourages compilers in a first step to at
least achieve for output (or sales) a breakdown between total goods and total services for each activity.
Such a solution, while providing more relevant data to users, would imply fewer burdens for
respondents/compilers than establishing a more detailed breakdown by product. This would be much
easier to implement, however, to achieve this compilers need to ensure that such a breakdown is feasible
or can be derived from the data sources. Some countries have been able to produce such data because the
information (or estimates) on sales of goods and sales of services was gathered through the data sources
used to collect FATS. It is also important to note that the investment income element should be separated
out from sales of services (and goods).
15.102. In structural business surveys, information is often requested on turnover with a breakdown in a
number of goods and services product groups. Consequently, if inward FATS are compiled using this
source, then a further breakdown of sales (or output) of services can be established or could be easily
estimated. Similarly, this could be achieved for the trade variables if they are included in the survey.
15.103. As described in section E.3, for the trade variables a breakdown by product may be derived from
information provided by reporting units in sources used for FATS, whether business activity surveys
(only for inward), FATS or FDI surveys. However, linking information with balance of payments trade in
goods (or merchandise) and services data could ensure that additional more detailed statistics can be
compiled. Section E.3 further describes the possibilities offered by this linking exercise. Data obtained
through this work could also help in first steps to estimate a more detailed breakdown of output or
turnover, in particular if enterprises are (services) export oriented, or if it is assumed that the pattern of
(services) exports by product is similar to that of (services) sales or output in the host economy.
15.104. It is advised that a breakdown by (service) product of output and/or sales or turnover be derived,
as well as trade variables from data sources (e.g. breakdown in FDI or FATS surveys, or for inward FATS
structural business surveys). Given the difficulty to produce a breakdown by product, for each activity
countries are encouraged to at least compile a breakdown of output or turnover into sales of goods and
sales of services (excluding investment income). For trade variables, it is considered a good practice to
investigate possibilities offered for linking FATS and resident/non-resident trade in services statistics in
order to develop a breakdown.

275

Establishing a correspondence between the two bases of classification may in particular be useful, for activities
that tend to be carried out only by enterprises that are specialized in the activity and generally do not engage in
significant secondary activities.

249

F.4.

Country practices: United States

15.105. The U.S. BEA compiles data on output by industry, with a distinction between goods, services,
and investment income for each industry. It can do so through the collection of such a breakdown of sales
in its FATS surveys. As described above, it then makes a number of adjustments for some industries to
arrive at the concept of supply (i.e., output) of services. A product type of breakdown, at least
distinguishing total services from total goods sales for each industry is an important achievement for trade
analysis and beyond. In particular enterprises primarily engaged in the production of goods may be
important services suppliers. As illustrated in Figure 15.3, being able to produce such a dataset enables
the United States to have a better sense of the actual services supplied to foreign markets, as compared to
the total sales (i.e., of goods and services) of foreign affiliates of U.S. multinational enterprises engaged
in services activities. At the total services level, i.e. activities versus products, the approaches show very
different results. The main factors behind these differences are the fact that sales of goods are removed for
services producing enterprises (the biggest activity contributing to this difference being wholesale and
retail trade) and services produced by goods producing enterprises are included. For 2012, data show total
sales/supplies of firms engaged in services activities 2 times higher than total sales of services/supply
products by all firms.
Figure 15.3

United States Outward FATS sales/supplies to foreign persons

3500
3000

298

2500
2000
1500

FATStotal sales of firms
primarily engaged in services
activities
129
0

1000
500

1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011

FATStotal sales of
services products (all firms)

For total sales/supplies of firms primarily engaged in services activities: beginning in 1999, industry
classification based on NAICS rather than SIC. Beginning in 2009 sales by bank affiliates and by the nonbank
affiliates of U.S. banks are included in the statistics. For 2009-2012, values for goods and services supplied by
majority-owned foreign affiliates in services industries to all foreign countries are used. For 1986-2008, values for
total sales by majority-owned foreign affiliates in services industries to all foreign countries are used. Total sales are
comprised of 3 components: goods supplied, services supplied, and investment income. In the 5 years leading up to
2009, investment income accounted on average for 3% of total sales.
For sales/supply of services products, beginning in 1999, sales by foreign affiliates were classified as goods or
services based on industry codes derived from NAICS rather than the SIC system, which resulted in a redefinition of
sales of services and a net shift of sales from goods to services. See SURVEY 81 (November 2001). Beg inning in

250

2004, services provided by bank affiliates and by the nonbank affiliates of U.S. banks are included in the statistics.
Also beginning in 2004, the statistics are presented as "services supplied"(equivalent to services output) rather than
"sales of services.” Compared with sales of services, services supplied adds 1) wholesalers’ and retailers’
distributive services, 2) insurers’ premium supplements, and 3) banks' implicitly-charged services; it subtracts a
proxy measure of insurers' expected losses. For more information, see SURVEY 89 (October 2009).

F.5.

UCI and partner country

15.106. Determining the country of the ultimate controlling institutional (UCI) is the most important task
in the FATS compilation work. The UCI is the recommended basis for classification by country for
inward FATS, but it is also relevant in some respects for outward FATS. (See MSITS 2010 Box IV.2 and
Chapter 1, Section C of this Guide).
15.107. Although many countries were able to implement this concept, it may be difficult to define
control relationships using the criteria as defined in international guidelines. Some countries have used
the percentage of equity shares as a first estimation of the percentage of control of the voting power. In
this context compilers should note that large enterprises usually have complicated structures. Small and
medium-sized enterprises (SMEs) tend to have simple structures, and the immediate controller is often the
ultimate controller. However when there is indirect control the UCI may be more difficult to identify.
15.108. Compiling inward FATS. For inward FATS, variables for a given affiliate are attributed in their
entirety to a single country of control. As descriptors of the operations of affiliates, they should not be
factored down by share of ownership of voting power (or equity). Nor are the values of the variables to be
apportioned between the controlling enterprise and any foreign minority owners of voting power.
15.109. Data on the UCI should be directly available as a by-product of FDI statistics.276 Alternatively
information on the UCI can also be drawn from the FATS or other data sources, or derived from a
combination of sources and/or registers. If the UCI cannot be extracted directly from existing data, the
decision about the UCI should be based on a step-by-step analysis of control relationships up the
ownership chain in the enterprise group. MSITS 2010, BD4 and HEGI suggest some factors that could be
used to decide how to define the country of the UCI for specific examples, in particular in the case of a
group of affiliated investors, (i.e. determining the status of the different investors, e.g. direct relationships
vs. indirect, government entities, SPEs etc.).
15.110. MSITS 2010 concludes that in the absence of any such factor that could be used as a basis of
attribution, the value of FATS variables may be allocated evenly among the foreign countries of control.
However, data so allocated may pose problems of interpretation, and efforts should first be made to
determine a basis for allocation to a single country. See MSITS 2010 section H.1 for more information.
The Eurostat FATS Recommendations Manual provides useful additional considerations on the
determination of the UCI. If necessary, it is advisable to check with important partners about how the
decision on the UCI has been achieved for bigger players. This can in particular be relevant at the

276

BD4 "strongly encourages" reallocation of all FDI positions (voting power of 10% or more) to the country of the
UCP of the direct investor. To the extent that the UCPs of all direct investors in resident enterprises are identified,
the UCPs of those direct investors with controlling interests in residents will be available. In order to apply the
directional principle, FDI statistics will identify whether the resident direct investment enterprise has a foreign or
domestic ultimate controlling parent (noting that this parent will be controlled by itself and so will have a resident
ultimate parent if there are no controlling links above it).

251

regional level, where a multinational enterprise may have established an affiliate/regional headquarters in
one country which controls affiliates in other countries of the region.
15.111. Frequently, the first foreign parent and the ultimate investor are the same. Although for inward
FATS the recommendation is to allocate data to the country of the UCI as a first priority, in practice this
may be something difficult to achieve, in particular in the initial stages of development of a FATS
national compilation framework. In practice, many economies initially developed their inward FATS
classifying information on the basis of the immediate controlling country, before moving to a
classification according to the country of the UCI. This guide proposes such an approach for the
compilers that are not able to determine the UCI when they start compiling FATS. In fact, that
information on immediate controllers are available by referring to FDI data. If FATS entities are
identified through linkages with FDI data or collected directly on FDI surveys, data on the country of
immediate foreign parent should be readily available. To facilitate comparisons with those data, MSITS
2010 encourages compilers to make available some data classified by the country of the first foreign
parent on a supplementary basis.277 However, in light of the relevance of the ultimate-controller basis and
the demonstration by a number of countries that compilation on this basis is feasible, this Guide advise to
compile data on an ultimate controller basis as soon as achievable (i.e., as data sources, linking
procedures or compilation procedures permit).278
15.112. For certain countries, it may be of interest to compile some operations data beyond the control
relationship, e.g., for associates (share of control between 10 and 50 per cent). This may be driven by the
fact that in these particular countries the law does not allow foreign investors to own a majority equity
stake (for all activities, or strategic ones for the compiling economy). In such cases, and if of interest to
the compiling economy, information could be compiled on operations, but with separate breakdowns for
the activities of majority-controlled affiliates (over 50 per cent owned, i.e. FATS) and minority owned
affiliates (e.g. 10 to 50 per cent).279 However, a clear distinction should be made between both datasets in
order that users are informed as to differences with international guidelines.
15.113. Compiling outward FATS. Outward FATS are statistics on foreign affiliates controlled by
residents of the compiling economy and should include all controlled foreign affiliates, regardless of
whether the control in the affiliate is held directly or indirectly through a chain of ownership and
regardless of whether the direct investor in the compiling economy is the UCI unit or is, instead, an
intermediate investor in an ownership chain. The data should be allocated to the country where the
operations of the affiliate are taking place.280 It is consistent with the treatment of foreign-controlled
enterprises in the 2008 SNA, in that the value added in production by the enterprise is attributed in all
cases to the economy of location of the enterprise. In other words, it is included in the GDP of the
economy where the enterprise is located. In addition variables should not be apportioned according to the

277

As a matter of comparison, BD4 encourages that supplemental inward FDI position statistics be compiled on an
ultimate investing country (UIC), and the corresponding industry sector..
278
Identification of the UCI can also be used to identify cases of round-tripping in inward FATS statistics. Roundtripping refers to the channeling abroad by direct investors of local funds that are then subsequently returned to the
local economy in the form of direct investment.
279
It is important to note that for the latter this will lead to double counting when comparing with data compiled by
other countries.
280
This is in contrast to the guidelines for FDI. According to BD4, if the ownership is through a directly held
affiliate located in another country, the variables should be attributed to the country of that affiliate. BD4 included
in its research agenda the presentation of FDI positions by the country where the affiliate is located (ultimate host
country). Because of the difficulties in tracking financial flows from the immediate counterparty to the ultimate
destination, most countries only present FDI data attributed to the immediate counterparty.

252

level of control (e.g. 100% of sales should be reported, even if the mother companies controls less than
100% of the voting power).
15.114. However, because the activities of an affiliate held through an ownership chain could be recorded
in the FATS of both the ultimate and the intermediate investors, and in order to facilitate international
aggregation without double counting, compilers are strongly encouraged to identify the aggregate share of
FATS variables accounted for by enterprises for which the compiling country is the ultimate controller.
Of particular interest will be data on this set of foreign affiliates belonging to ultimate investors resident
in the compiling economy. See section G for more information on UCI and regional aggregations.
15.115. It is advised that FATS data should be attributed geographically as a first priority to the economy
of the UCI unit for inward FATS and to the economy where affiliates' operations are taking place for
outward FATS. If control relationships cannot be clearly established, compilers should adopt a step by
step approach to determine them. For big players it is a good practice to investigate ways to coordinate
with compilers in major counterparty countries. This Guide further advises that ownership of 50% or
more of equity shares could be used in the compilation process as an approximation to determine control.
However compilers should aim at working on the basis of control relationships. Although the first priority
for inward FATS is to allocate variables to the UCI, in first instances compilers can consider producing
data according to the country of the immediate controller. In fact this is suggested as a supplementary
dataset in MSITS 2010. Outward FATS should be compiled for all affiliates controlled by enterprises
resident in the compiling economy. For reporting to regional/international organisations, outward FATS
data should also be compiled for the subset of affiliates of UCIs of the compiler's economy (i.e.
controlling units in the compiling economy that are not themselves controlled from abroad).
15.116. European Union. Identifying the UCI plays a crucial role as all data for foreign affiliates under
direct or indirect control should be attributed to the country of the UCI. If the information on the UCI is
not available from the existing data, the European statistics on FATS follows a general rule, which says
that the identification of the UCI should be based on a step-by-step analysis of control relationships up the
owner chain in the enterprise group. Once the top enterprise of the group is identified, its decision-making
ability must be examined.
15.117. An important source for identifying affiliates and allocating them to the UCI should be the future
EuroGroups Register (EGR), which, at the time of this writing, is under development. The EGR should
cover step-by-step all multinational enterprise groups operating in the European Union. This will help to
identify the correct country code of the UCI to which the enterprises belong. With a full and well
developed EGR, no double counting or gaps would exist and the same data sets from different compilers
would be linked.
G.

Regional aggregation

15.118. This section is mainly for compilers of regional/international organisations, but also provides
more insights as to the reason why outward FATS data are also needed on a UCI concept (i.e. for outward
FATS only data where the parent entity is not itself controlled from abroad), as opposed to a national
concept (where all outward FATS are considered whether the mother company is a UCI or not). Indeed in
this context data will be compared from one country to another, and in certain cases it will be necessary to
compute regional or world aggregates. If all countries apply international guidelines in the same way, in
principle, if UCIs are detected in a consistent way, inward FATS data compiled by countries will be
duplication free. This will not be the case for outward FATS, as the main recommendation is to compile
data on foreign affiliates of all controlling entities established in the compiling economy, whether UCIs or
units who are themselves controlled from abroad. However, following international recommendations,
compilers should also compile information focusing on affiliates of UCIs of the compiling economy. This

253

is the information that should be used by regional or international organisations when constructing
aggregates for outward FATS.
15.119. In producing such aggregates it is important that common rules for collecting and compiling
FATS are adopted by the countries participating in the group (i.e., beyond the international agreed
definitions). An important aspect to consider when producing these aggregates is the possibility of
distinguishing intra-regional FATS from FATS with the rest of the world. To do this, it is necessary that
countries compile information according to agreed intra- and extra-regional definitions and/or have the
necessary partner country detail. For outward FATS, it is important to use the data compiled by countries
referring only to the affiliates of ultimate controlling units of the compiling economy (i.e., foreign
affiliates of resident units not themselves controlled by non-residents). A set of guidelines on the basis of
international recommendations and the level of detail provided should be established. Finally, it is
necessary to ensure that confidentiality of information is still ensured when computing regional
aggregates (see the description of the EU experience above).
G.1.

European Union Experience

15.120. The legal framework for the provision of foreign affiliate statistics in the European Economic
Area (EEA) is the Regulation No. 716/2007 of 20 June 2007. In addition, Eurostat publishes the FATS
Recommendations Manual, which provides a common methodological framework on the definitions and
concepts to national compilers in order to produce a harmonized set of FATS data across the Member
States. Having a comparable set of the FATS data is a pre-condition to compile meaningful, reliable and
high-quality European aggregates.
15.121. IFATS and OFATS data are collected from the European Member States, candidate and EFTA
countries on an annual basis; data. Data collection has been mandatory starting with the reference year
2007. One of the most important concepts when looking at the European statistics is that FATS should be
compiled according to the UCI concept. The UCI is the institutional unit (enterprise, branch) proceeding
up a foreign affiliate's chain of control that is not controlled by another institutional unit (enterprise,
branch).

254

Chapter 16 Compilation of additional indicators on the international supply of
services
16.1. Scope. The purpose of this chapter is to focus on the compilation of additional indicators on the
international supply of services, in particular data on the number of persons/trips relevant to modes 2 and
4, the use of data on services production, prices, employment or sectoral performance indicators. Also,
linking of trade and business registers offers the potential for the compilation of new types of data.
Section A introduces the general purpose of this chapter and Section B provides a summary of good
practices with respect to additional indicators on the international supply of services. Section C contains
an overview of the compilation of variables on the movements of natural persons under modes 4 and 2 of
supply of services. Section D describes the linking of business and trade registers. Finally, section E lists
possible existing additional indicators and the level of detail necessary for a full analysis of services
sectors in the context of the international supply of services.
A.

Introduction

16.2. While most of this Guide provides good practices for obtaining the value of the international
supply of services (i.e., the value of trade in services or sales/output of services of affiliates), trade
negotiators and analysts need additional information on the market performance of individual service
sectors, such as foreign direct investment flows and stocks, research and development expenditure in
affiliates, income flows relating to foreign affiliates active in services sectors, etc. In the same way as
there is interest in the monetary and non-monetary measures of merchandise trade, non-monetary
information drawn from various quantitative indicators is also important to assess trade commitments and
to conduct further in-depth analysis on the international supply of services (e.g., number of modes 4 and 2
persons, number of foreign affiliates established abroad in the context of mode 3, etc.).
16.3. This chapter aims at presenting these additional data items many of which are already defined
within other statistical frameworks and may be extracted without additional cost if the level of detail
needed is readily available (i.e., the frameworks for foreign affiliates statistics, foreign direct investment
statistics, and sectoral statistics; e.g., tourism and education). These additional data items will be further
detailed in section E. Linking trade and business registers is also a new line of work that compilers should
be aware of, given the potential that such an exercise offers for compiling new sets of data of interest to
policy makers with little additional effort or costs (section D).
16.4. However, meeting some user needs requires the development of new data compilation systems, or
the extension of existing ones. For example, in the case of mode 2, availability of information on the
number of persons/trips (section C.2) who are traveling abroad and consuming services, whether for
personal or business reasons, is in high demand. Such extension is even more valid for a variable of
greater interest to negotiators and analysts -- the number of persons moving to provide services under
arrangements covered by mode 4 (section C.1).
16.5. Knowing how many people are providing services abroad (flows, stocks, as well as the number of
associated trips) is very important for analysis of both GATS and international mobility. First, as stated
above, like for merchandise trade, it is non-monetary (quantitative) statistics that may usefully
complement value data about mode 4 supply of services. Second, given the interest that mode 4 generates
amongst officials and research institutions involved in the international supply of services and/or
migration issues, quantitative data provide a useful gauge about the relative importance of mode 4
movements of persons within wider economic international mobility. This is why this chapter focuses
particularly on mode 4 number of persons and trips. The chapter identifies sources which could be used

255

and elaborates how relevant statistics could be produced. In the context of mode 4, a subset of the
population of business travellers will be of particular interest, but other categories of internationally
mobile individuals will also be of relevance.
B.

Summary of good practices

16.6. The identification of mode 4 movements and stocks in existing sources is a new area of work.
This Guide suggests dealing with this issue in stages. For incoming and outgoing mode 4 movements,
compilers have to take into account the needs of users (in particular for categories of primary interest) and
various possibilities offered by each data source (and possible drawbacks).
16.7. As a starting point it is suggested to compile rough estimates of the size of mode 4 movements in
terms of the number of natural persons or, if easier, to collect data on number of trips. This could be done
if appropriate sources are identified, grossed-up or estimated using an appropriate data model. However,
the objective is to have more pertinent and detailed data on number of persons. Breakdowns by purpose,
type of service supplied, country of origin or destination should be compiled, for inbound and outbound
flows and stocks. To enhance the usefulness of the compiled data they should be produced, in the long run,
with a breakdown by length of stay and occupation/skills of individuals (e.g. using ISCO-2008 or ISCED
2011).
16.8. Compilers should strive to obtain, as a first priority, information for the mode 4 categories of
persons which are of most interest to their economies (generally contractual service suppliers, whether
employees or self-employed, or intra-corporate transferees), whether incoming or outgoing. This could be
done by analysing the needs for information. A longer term goal would be to have data for all mode 4
categories and all service sectors, both for incoming and outgoing persons/trips, therefore enabling overall
international comparisons.
16.9. The compilation of mode 2 data on inbound and outbound flows of persons should be given more
priority than to the compilation of data on stocks, except if deemed necessary for specific services sectors
or categories of persons(for example, education/students or health/medical patients). Data should be
broken down by purpose of movement, by type of service consumed (using EBOPS 2010), as well as by
destination/origin of persons or trips. Although of less priority in the context of MSITS 2010, a
breakdown according to length of stays could be compiled.
16.10. For mode 4 and mode 2 quantitative indicators, it is a good practice to concentrate on obtaining
annual statistics first, as this should be sufficient to serve most analytical and monitoring needs. If in
future compilers identify the need for more frequent statistics, their compilation could be envisaged but,
maybe, at a less detailed level.
16.11. Given the policy relevance of such information, compilers are encouraged to analyse the national
possibilities for compiling trade in services data by enterprise characteristics, by linking trade in services
data with information drawn from the statistical business register. Although foreseen as a longer term
objective, compilers should keep in mind the needs to be identified with users for additional indicators in
relation to services trade analysis and negotiations. Needs should be identified with users, but a list is
suggested to serve as a starting point in section D. Given that this information is linked with needs of
other statistical frameworks this Guide strongly advices that trade in services compilers cooperate with
compilers in charge of the data collection and compilation in respective statistics domains.

256

Mode 2 and mode 4 movements and stays

16.12. The supply of services involving movements of physical persons is defined in the GATS as
follows: (i) movements and temporary physical stay abroad of consumers of services (mode 2); and (ii)
movements and temporary physical stay abroad of persons to supply services (mode 4)281 . Mode 2 is
generally well understood as it refers to any potential consumer traveling or temporarily staying abroad
for any purpose (e.g., leisure, studies, health, business or employment) and consuming services while
abroad. Mode 4 movements and stays are concepts more difficult to grasp. As described in Box 1.1, mode
4 can, in general, be further described as covering foreign natural persons entering the host economy to
directly fulfil service contracts (contractual service suppliers, either self-employed or employees of a
foreign service supplier);282 work in a foreign affiliate that delivers services (intra-corporate transfer or
directly recruited by the affiliate);283 or negotiate services contracts or the establishment of services
related commercial presence without actually carrying out service transactions).
16.13. The availability of data on the number of natural persons moving (flows) between countries to
consume (mode 2) or to supply (mode 4) services and on foreign natural persons present at a certain point
in time (stocks) is highly important for trade policy making, monitoring and for the analysis of global
supply of services, as well as broader tourism and migration policy.284 Data on the movement of natural
persons should include detailed information on the characteristics and activities of such persons. This
could also be useful for estimating the value of services consumed or supplied by them in the cases when
value data cannot be obtained from other sources.
16.14. Taking into account the information needs, an indicative list of the variables covering movements
(and stays) of natural persons under modes 2 and 4 is provided in the following sections (sub-sections B.1
and B.2). The prioritization of the variables is also discussed. It is advised that compilers consider the list
and incorporate it in their data collection and compilation programmes taking into account their specific
needs and circumstances. In this context this Guide strongly encourages compilers to coordinate with the
entities in charge of the other (and broader) statistical frameworks used to collect and compile such data
(e.g., tourism, migration or employment statistics).
16.15. Compilation of number of persons/trips does not necessarily always demand sophisticated
compilation methods, 285 but it requires availability of a rather focused and efficient data collection
mechanism. Due to this entanglement of the collection and compilation of the number of mode 2 and
mode 4 movements and stays, this chapter mainly presents the types of breakdowns required and
complements chapter 11 by presenting the potential sources of information. The collection of number of
persons moving abroad (or trips) and/or on the characteristics of trips is dealt with in the context of
surveys of businesses (chapter 6), of persons and households (chapter 7) and administrative sources
(chapter 9), respectively.

281

Further information on definitions and recommendations on mode 2 and mode 4 may be found in chapter 1 of
this guide as well as chapters 2 and 5 of MSITS 2010.
282
In general the transactions would be measured as trade in services (mode 4) in the balance of payments services
account, see chapter 14, section C.
283
The supply of the service to the client is through mode 3, and the transactions would be recorded by FATS as
services output of foreign affiliates, see Chapter 15
284
See also MSITS 2010, chapter 5 Section A.
285
Either data would need to be grossed up to be representative of the whole population, or alternativel y data could
be used in a data model to be grossed up.

257

16.16. It should be noted that the information on number of persons traveling during a given period of
time is difficult to collect as many sources rather provide information on the number of the trips (or on
fractions thereof).286 This means that a same person may be counted more than once in a reference period.
In other words, one will not be observing the number of persons, but rather the number of
trips/movements or number of permits. However, while in certain cases, it may be possible to identify
individual persons and link them to trips/movements or permits, such identification will most likely not
be feasible in the majority of data sources. For mode 2, this issue may not necessarily be important, as
statistics based on trips/movements are consistent with the information needed; but for mode 4, this issue
may be important, as the primary interest may be more on number of persons, with the number of
movements and/or permits being of secondary relevance. However, if data on trips are of a good quality,
then such data can be used to estimate the number of persons travelling under these two modes.
16.17. Quality issues need to be considered in the context of the compilation of data on the number of
persons and or movements/trips in relation to modes 2 and 4; namely: do the primary observations cover
the whole population of interest or the most important sub-groups; can we observe the characteristics of
the relevant statistical units; do we observe (obtain good approximations of) the characteristics of interest
using particular data sets? Are the observations reliable and the resulting statistics sufficiently accurate for
both levels and changes/differences? Do we get the statistics with the required frequency and timeliness?
16.18. Building stronger institutional arrangements is, therefore, a precondition for success in compiling
quantitative indicators on mode 2 and mode 4 movements and stays. No single source can provide
statistics on all these categories, and the direction of movements (e.g. incoming/outgoing) adds a
difficulty for data producers. Thus, it will often be necessary to combine statistics based on different
sources and data collection methods. This is an additional argument for encouraging different institutions
to work together. The development of clear guidance for the administrative forms and statistical
questionnaires will be very important given that this subject is often difficult to understand. It should be
noted however, that the challenge is to make respondents understand what information is requested. This
does not necessarily mean that the statistical concepts have to be fully explained as the relevant statistics
may be derived by the application of algorithms to the responses to questions that are easy to understand
and concern information that respondents can easily recall or obtain from records readily available to
them.
16.19. Flows and stock data need to be compiled for modes 2 and 4 quantitative indicators. Compilation
of flow data requires attention to such issues as differences in coverage of data sources methodology of
making necessary adjustments, grossing up, etc., of gathered information so that the resulting statistics are
a fair representation of the target population and are of sufficient quality. Stock data are connected with
flow data, as they result from the accumulation of flows of previous periods and they are changed by the
inflows and outflows in the period under consideration. However, compiling a base or benchmark figure
for the stock at a given time may be difficult, in particular given the high mobility of persons concerned.
For stocks related to mode 2 or mode 4, presence in the compiling economy, arriving of inbound persons
or trips would add to the inward stock, whereas those departing would subtract from the inward stock.
For stocks related to mode 2 or mode 4, presence abroad, returning outbound persons or trips would
subtract from the outward stock, whereas those departing would add to the outward stock. Compilers will

286

For example the border surveys capture information about a particular individual on a given trip. Since a person
can make several trips during the same reference period the number of trips and the number of persons travelled are
usually not the same.

258

need to ensure that these calculations are made at a specific (and regular) point in time within the period
under consideration (e.g. beginning, middle or end of the period).
C.1 Movements and stays of persons related to mode 4
16.20. As described in chapter 11, the potential data sources for mode 4 movements are enterprise
surveys (see chapter 6 for more information), administrative sources (chapter 9), and household and
labour force survey, population censuses and border/passenger surveys (chapter 7). There are no
dedicated and comprehensive data sources for mode 4 movements and stays, which means that compilers
could use one or the other source depending on the type/category of inward or outward stocks of mode 4
for which information has to be compiled. This could be established based on identified needs and
potential sources. For example:
i.

Trade in services surveys would probably be the best source to collect data on contractual
services suppliers alongside the data on the value of the contracts:

ii.

FATS data sources could be useful on intra-firm movements 287 or direct recruitments by
foreign affiliates;

iii. Other enterprise surveys, covering specific services sectors, or temporary employment
agencies could also be specially tailored to capture information of interest on mode 4.
iv. Border/passenger surveys should be considered for obtaining characteristics of those
traveling for business, work or employment purposes, combined with counts of border
crossings;
v. Household/labour force surveys could help to obtain information for outgoing mode 4
persons, in particular the self-employed, and if relevant combining this information also with
counts of outgoing persons/trips;
vi. Business registers (see Chapter 5) might contain information needed to identify potential
mode 4 self-employed service suppliers;
vii. Administrative sources (migration records, registers) might be used, in particular in
receiving countries, and could offer useful records, for example, on counts of persons or trips.
Work permits or documents issued in the context of social security coordination and
employment services could also be considered. In other words there may already be relevant
administrative information on short term movements of workers. However, compilers need to
ensure that in administrative sources, the category(ies) that could be of interest for mode 4 are
defined in a way close enough to the coverage (and if possible breakdowns) recommended in
MSITS 2010 and this Guide. If well-defined and used appropriately, entry/exit cards could be
used to make a first selection of border survey respondents that could potentially be of
interest in the context of mode 4. These respondents could then be surveyed to obtain more
refined information;

287

Whether the employer-employee relationship remains with the sending firm, i.e. corresponding to the rendering
of services by the sending firm to the receiving one; or the employer-employee relationship is with the receiving
firm.

259

viii. . Similar considerations could be put forward for other sources. Finally, partner country data
may provide useful information given that it may be difficult/not possible to collect details on
characteristics of incoming persons/trips through surveys of enterprises or households, or
using administrative records for outgoing persons. In the same way population sources tend
to reflect outbound mobility better than inbound, which contrasts with migration data where
emigration is much harder to capture than immigration. As a consequence partner country
information could turn out to be essential and be used/compared through the use of a
clearinghouse for such data statistics. This however implies that the data collection and
quality of the compiling country depends on partners’ willingness and capacity to collect and
provide quality data, as this normally involves coordination, cost and in the case of trade
negotiations also certain interests, a central brokering institution may be important. Compilers
therefore need to be cautious in using partner country data (definitions, national
laws/regulations in place may differ, etc.). Refer to chapters 5, 6, 7, 9 and 10 for a more indepth description of the respective data sources.
All sources identified above offer some potential to collect data. However, it is important to
identify/coordinate with those in charge of these respective data collection mechanisms to see if it is
possible and not too burdensome to identify, collect and compile data on the mode 4 categories of interest
to the economy (both inward and outward) and their breakdowns. A comparison of data sources by
purpose of mode 4 movement/stay and direction of movement is provided in chapter 11. Alternatively,
the compilers are advised to explore a possibility to combine obtained information and develop an
appropriate model to derive the data and breakdowns of interest (see chapter 17).
16.21. For compiling data on mode 4 movements and stays, two groups of persons should be
distinguished: those who will not be in an employer-employee relationship in the country in which they
are temporarily present and those who will be in an employer-employee relationship in the country in
which they are temporarily present (this refers solely to part of intra-corporate transferees and those
directly recruited by foreign affiliates). This Guide advises the following statistical treatment of these two
groups
Persons who will not be in an employer-employee relationship in the country in which they are
temporarily present the mode 4.
The variables are flows and stocks and should be broken down as follows:
Outbound:
i.

Flows: the number of natural persons of the compiling country who departed to other
countries to render services (or negotiate a contract or establishment of commercial
presence), as well as on the number of their trips.

ii.

Stocks: the number of natural persons of the compiling country who were present in
other countries at a certain point in time in the reference period (e.g. beginning,
middle, or end of period) to render services (or negotiate a contract or establishment
of commercial presence).

260

Inbound:
i.

Flows: the number of natural persons of other countries who entered the
compiling country to render services (or negotiate a contract or establishment of
commercial presence), as well as on the number of their trips.

ii. Stocks: the number of natural persons of other countries who were present at a
certain point in time in the reference period (e.g. beginning, middle, or end of
period) in the compiling country to provide services (or negotiate a contract or
establishment of commercial presence).
The information on number of inbound/outbound flows (persons and trips) and stocks (persons)
will have to be broken down by:

Purpose of (mode 4) movement/stay abroad(outbound) or in the compiling
economy(inbound) with at least a separate identification of the four main
categories of mode 4 movements: service contract either by self-employed or
employee -intra-firm trade or not-, or sales/commercial presence negotiation;

ii.

Service category;

iii.

Country of destination (outbound) or origin (inbound);

iv.

Duration (length of stay);

Skills/occupation of persons.

Persons who will be in the employer-employee relationship in the country in which they are
temporarily present. ,
The variables to be collected are the same as for the first group, except that the breakdown by
purpose of stay should identify those relating to the movements between affiliated companies
(intra-corporate transfer) from those who refer to a direct recruitment by a foreign affiliate in the
country of temporary stay. The status of employment for both categories is being an employee of
the foreign affiliate. In addition for the service category, since the employment of these persons is
not necessarily directly related to the supply of the service by the affiliate to clients, it is
recommend compiling data broken down by type of service activity of the affiliate/employer
(using ISIC Rev.4 /ICFA rev.1).
This Guide advises that the compilation of data on stocks should be given the same priority as the
compilation of flow data in view of their great importance for the GATS mode 4 analysis.
16.22. Compilers should strive to obtain information for the mode 4 category of most interest to their
economy (generally contractual service suppliers, whether employees or self-employed, or intra-corporate
transferees), whether incoming or outgoing. Identifying categories of interest could be done by
conducting an analysis of the potential size of the population based on the types of agreements signed by
a country (e.g. free movement of persons, etc.), the estimated value of mode 4 trade, if certain visas that
are issued are strongly related to trade in services, which services sectors are assumed to be a mode 4

261

comparative advantage of the economy (and reversely how big could mode 4 imports be) etc. The
different circumstances in each country determine locally specific forms of trade in services through
presence of natural persons. A methodological approach to the production of statistics on mode 4 persons
should take into account, not only market specifics, but also for incoming persons/trips, the national
regulations of immigration and labour policies, as well as those of foreign investment and trade in
services supply through presence of natural persons as well as effectiveness of regulations and policies.
For instance, if foreign investment in a service industry is strictly controlled then intra-corporate
transferees could not be considered for estimation or if low skilled workers cannot get work permits then
estimation of mode 4 person number could be concentrated on other higher skilled occupations. Therefore,
a step by step approach should be adopted. In early stages each economy could establish a necessary
statistical capacity for compilation on mode 4 movements and stays depending on their economy's
circumstances, identifying simplifying features to determine key categories of data and major authorities
to cooperate with, and compilation should be based on facts and in responding to recommendations of
MSITS2010. A longer term goal would be to have data for all mode 4 categories and all service sectors,
both for incoming and outgoing persons/trips, therefore enabling overall international comparisons.
16.23. With respect to the periodicity, compilation of (detailed) data on an annual basis should serve
most information needs. However, given the strong link of this information with trade/labour mobility
policy, it could be of interest to also envisage the compilation of some information for shorter periods, for
instance on a quarterly basis, at least for main aggregates.
16.24. As indicated previously, depending on the source used, the compilation of data on trips is
significantly less resource intensive and provides a satisfactory information basis for policy purposes and
estimation of missing value data. Therefore, as a general rule, this Guide advises that compilers first
concentrate on the compilation of data on inbound and outbound trips and use these data for the
compilation of data on persons as necessary.
16.25. For the country of destination or origin, the primary focus should be on the country of the
supplier and that of the consumer of the service. In the case of receiving countries (where mode 4 data
will be of primary interest given direct relation with commitments made), priority should be given to
identifying the country of origin of the supplier, in particular for contractual service suppliers-employees
where this can be more difficult to establish as it can differ from the country of the individual(s). When
the source data does not enable this, the country of origin of the person could be used as a reasonably
good indication of the country of origin of mode 4 trade in services.
16.26. The identification of purposes of business, work or employment related movements using
internationally agreed definitions ensures the compilation of comparable data needed for trade and other
purposes. Adopting such an approach can also facilitate the estimation of some value data for
resident/non-resident trade in services transactions. Compilers are advised, as a starting point, to make use
of the list of purposes of trip/migration provided in table V.3 of MSITS 2010, which is based on the
classification of purposes of tourism trips provided in IRTS 2008 and from the list of migrant and nonmigrant categories provided in RSIM Rev.1.288 This list contains two main groups: the first one refers to
the tourism visits/trips, and the second refers to all the other RSIM Rev.1 categories. Within tourism
visits/trips, two main purposes are identified: (a) personal and (b) business and professional. The first
group is broken down into 8 items (none is of interest in the context of mode 4), while the second is not
subdivided. In view of the information needs related to mode 4, it is advised that compilers subdivide this
latter group as identified in MSITS2010 table V.3:

288

See IRTS 2008, paragraphs 3.15 – 3.21 and paragraph 5.26 (5th bullet point); and RSIM Rev.1.

262

contractual service suppliers, self-employed,289 ;

ii. contractual service suppliers, employees of a juridical person (of which intra-firm trade in
services which is equivalent to intra-corporate transfers where the employer-employee
relationship remains with sending entity);
iii. services sellers and persons responsible for setting up commercial presence.
Within the RSIM categories of migrant work/employment-based settlement (i.e. where the employeremployee relationship is in the country of destination of the person), a separate identification of those
intra-corporate transferees or those directly recruited by services producing foreign affiliates would
complement the mode 4 information needs.
16.27. The compilers should take into account the MSITS 2010 recommendation that notwithstanding
the difficulties involved, they should break down relevant statistics between permanent and nonpermanent stays according to their national definition of residence, regardless of whether stays may be
substantially longer than the one year guideline generally suggested by statistical systems. In this
connection, it should be recalled that in tourism statistics the duration of trips is measured in terms of the
number of overnight stays, and only will include those not exceeding one year. In the case of mode 4
where the employer-employee relationship stands in the country of temporary stay, one should turn to
migration statistics. Migration statistics only refer to short-term (3 months to 12 months) and long-term
(more than 12 months) stay. In both frameworks there are no recommendations on groupings as identified
in MSITS 2010. MSITS 2010 recommends,290 as a long term goal, that the breakdown for mode 4 stays
should be provided by:
i.

stays of less than three months;

ii. stays of between three months and less than one year;
iii. stays of between one and three years;
iv. stays of between three and five years, and;
v. stays of more than five years.
When it comes to the length of stay, the information needs triggered by the GATS for the variable number
of persons or for number of trips go beyond those used as guidelines in international statistical standards.
Consequently compilers should adapt this classification to their national needs and statistical systems (and
laws) including the needs of national tourism statistics and migration statistics, as well as other types of
statistics.
16.28. The detailed multipurpose international classification of services is provided in CPC, Ver.2, while
more aggregated service categories adopted for use in statistics on service transactions between residents
and non-residents are contained in EBOPS. Compilers should select the services classification for use in
the context of movements of natural persons/trips under mode 4 depending on their needs and
circumstances, but are advised to do so on a basis compatible with EBOPS 2010 to facilitate the analysis
of this information as well as link, if possible and relevant, with the compilation of some balance of

289

This category is to be compiled for mode 4 purposes if there is not an employer-employee relationship in the host
economy. See Box V.2 of the MSITS 2010.
290
MSITS 2010, paragraph 5.26, 5th bullet.

263

payments services items and FATS. If a product classification cannot be used, then the field of activity of
the service supplier could be used (i.e. using ISIC Rev. 4 or ICFA Rev. 1).
16.29. Although not completely consistent with GATS definitions, such an integrated approach with
tourism and/or migration statistical definitions will strengthen the institutional arrangements in data
collection and data compilation and will ensure a more efficient use of limited statistical resources, while
providing information that can be used as reliable indicators on mode 4. Another potential benefit of
cooperation or coordination could be to ensure consistency between enterprise and household/labour force
surveys that have work conditions/policy as common subject matters, and could, therefore, be combined
to obtain data on mode 4. 291
16.30. The breakdown by skills/occupations is entirely optional. However, it is advised that countries
follow the International Standard Classification of Occupations (ISCO-08) and/or the International
Standard Classification of Education (ISCED) 2011 if such data are compiled.
Country experience: China - compilation of mode 4 number of persons
16.31. China was developing the methodology to estimate the number of mode 4 persons for many
years. This work was focused on two areas: overseas contracted projects and overseas labour service
cooperation. The former refers to overseas construction projects carried out by Chinese firms and their
foreign affiliates. It includes persons working in projects where the establishment of a branch or affiliate
is not necessary (either for short-term projects, or those which do meet the criteria for the establishment of
a branch), i.e. referring to contractual service suppliers in the construction sector (most of the income
refers to these projects). It also covers employees sent by the mother (construction) company to work in
affiliates or branches abroad (intra-corporate transferees). The overseas labour service co-operation refers
to contractual service suppliers. They are employed by the China Labor Service Co-operation firms which
offer services to overseas firms and organization (i.e. clients abroad) based on service contracts. Data on
outflow and stock of mode 4 persons by the type of contract is almost completely covered by China's
direct data collection. Self-employed contractual service suppliers are not covered, as well as intracorporate transferees in other sectors.
16.32. According to the Regulation of Overseas Contracted Projects and Regulation of Overseas Labor
Service Cooperation of China, qualified enterprises to be able to sign contracts should apply and get
approval for the right to trade services through presence of natural persons. The relevant authority is not
only responsible for applications of enterprises or service suppliers, but also participates in the collection
and aggregation of data on outflow and stock of mode 4 persons in the category of contractual services.
Two statistical programs have been jointly established by the Ministry of Commerce and the National
Bureau of Statistics. According to these programs, relevant enterprises are obligated to report required

291

A similar approach to th
e work of Eurofound could be established for mode 4 information purposes In its work
Eurofound aims at providing information on the quality of living and working conditions in Europe. In order to
realise this goal they combine three surveys, each targeting a different population to get specific information. The
European Working Conditions Survey questions workers to gain insight on the quality of work and employment.
The European Company Survey targets managers and employee representatives in companies to gather information
on workplace practices. The European Quality of Life Survey interviews European citizens to get a perspective on
living conditions and perceptions of quality of life. Despite these differences in terms of target population,
Eurofound aims to harmonise survey methodology whenever possible, to ensure that lessons learned in one survey
are implemented in the others.http://www.eurofound.europa.eu/surveys/methodology/index.htm.

264

information, such as the type of service project, outflow and stock of persons, occupation or overseas
work, length of stay, revenue, country of destination, etc. Administrative sectors at all levels collect data
and check it. The national authority is responsible for aggregation and annual publishing of the data in the
China Statistical Yearbook and the Chinese Statistics of Trade in Services, as shown in table 16.1 below.
Data are also broken down into 8 sections of the Industrial Classification of the National Economy of
China (agriculture, forestry, animal husbandry and fishing; manufacturing; construction; transport;
computer services and software; accommodation and restaurants; scientific research, education, culture,
sanitation and sports; others.
Table 16.1Outflow and stock of mode 4 Chinese persons in 2012 (contractual services)
Unit: persons
Variable
Country (region)

Total
Outflow

Contractual Projects
Stock

Outflow

Stock

Labour Contracts
Outflow

Stock

Total

511745

850181

233365

344618

278380

505563

Angola

23525

43604

17445

31905

6080

11699

Russia

15354

18822

4502

4088

10852

14734

...
Note: Stock indicates the total number of persons sent overseas by the end of the year.

C.2 Movements and stays of persons related to mode 2
16.33. The possible sources for quantitative data on mode 2 movements and stays are in most cases
closely related to the tourism/travel data collection and compilation mechanism. Household surveys
(including labour force surveys) can capture information on individuals who have travelled abroad
(outbound flows). Border or passenger surveys could be used to gather information on persons who have
travelled abroad (outbound flows) as well as on those who have travelled to the compiling country
(inbound flows). These will mainly provide information on the characteristics of the travel/trips, while
border counts (or counts in ports of entry-exit) will provide figures on the number of entries or exits. As
described in the previous section data from this source could also be used to satisfy (some) mode 4
information needs. Chapter 7 further describes data collection through surveys of households and persons.
Administrative sources are also a source that could be used in this context, although it may be difficult to
clearly identify the categories of interest (see chapter 9). Enterprise surveys could also be used to obtain
the mode 2 demand side statistics (if employers are asked to report on the travel abroad of their
employees), but most likely they will be used to generate the supply side data (surveys targeting
accommodation services, tourism attractions, or specific services sectors such as health or education
establishments). Other sources may also be of interest depending on the situation of countries. A data
model could be used to combine information from multiple sources in order to obtain mode 2 statistics
with relevant breakdowns (see chapter 17). A comparison of data sources by purpose of mode 2
movement/stay is provided in chapter 11.
16.34. Quantitative data should be compiled on mode 2 for inbound and outbound flows and stocks as
follows:

265

Outbound:
i.
ii.

Flows: the number of natural persons of the compiling country who departed to
other countries to consume services, as well as on the number of their trips.
Stocks: the number of natural persons of the compiling country who were present
in other countries at a certain point in time in the reference period (e.g. beginning,
middle, or end of period).

Inbound:
i.
ii.

Flows: the number of natural persons of other countries who arrived in the
compiling country to consume services, as well as on the number of their trips.
Stocks: the number of natural persons of other countries who were present in the
compiling country at a certain point in time in the reference period (e.g. beginning,
middle, or end of period).

The information on number of inbound/outbound flows (persons and trips) and stocks (persons)
will have to be broken down by:
i. Country of destination (outbound) or origin (inbound);
ii. Purpose of stay abroad (outbound) or in the compiling country (inbound);
iii. Types of (services) products consumed;
iv. Duration (length of stay).
In the majority of cases, mode 2 stock information will be of minor interest to users/policy makers. It is,
therefore, a good practice that compilers identify with users the relevant categories of persons/purposes
where stays will most probably be of a more long-lasting nature and consequently stock data of more
interest, for example in the case of students.
16.35. Compilers should select the services classification for use in the context of movements of natural
persons under mode 2 depending on their needs and circumstances, but are advised to do so on a basis
compatible with EBOPS 2010 (in particular the travel breakdown by product consumed) to facilitate the
analysis of this information as well as link if possible and relevant with the compilation of some balance
of payments services items and FATS. In view of the information needs related to mode 2 and tourism it
is also advised that countries build their breakdowns in a way which is also useful for tourism statistics.
Such an integrated approach will strengthen the institutional arrangements in data collection and data
compilation and will ensure a more efficient use of limited statistical resources.
16.36. For the purpose of stay abroad or in the compiling country, compilers are advised as a starting
point to make use of the classification of purposes of tourism trip provided in IRTS 2008. This
classification contains two main groups of purposes: (a) personal and (b) business and professional. IRTS
2008 suggests a breakdown of the first group into holidays, leisure and recreation; visiting friends and
relatives; education and training; health and medical care; religion/pilgrimages; shopping; transit; and
other. The second group is not subdivided.292 Note that from the perspective of this Guide it may also be
useful to gather information on purposes other than tourism related (e.g. education purposes for more than
one year, or those who are temporarily abroad for employment purposes).

292

See ITRS2008, section B.1 for more information.

266

16.37. Compilation of data on an annual basis should serve most information needs. However, it may
also be of interest to obtain some information for shorter periods, for instance on a quarterly basis at least
for some main aggregates. This would, in principle, also be linked to the needs of other statistical domains
(e.g., tourism).
16.38. The table below shows an example for presenting data on mode 2 number of persons, mainly
relating to tourism purposes. It is drawn from the UNWTO Compendium of Tourism Statistics. An
extension that would serve the needs as identified in MSITS 2010 would be gather information of
purposes other than tourism related (e.g. education purposes for more than one year, or those who are
temporarily abroad for employment purposes).
D.

Linking services trade data with the business register

16.39. Nowadays many compilers have started projects to link merchandise trade statistics to the
business register at the micro-level, in order to produce so called Trade by Enterprise Characteristics
(TEC) indicators (as published e.g. by Eurostat and the OECD293 ). These data indicate which firms – by
for example industry, size class, foreign ownership, or geographical (sub-national) region – are actually
engaged in international trade. This information is considered very policy relevant as it shows the
importance for example of SMEs, or foreign owned firms, in total trade, and gives insight into which
industries use what kind of imported products, which is vital for the analysis of global value chains.294
16.40. Similar data-linking activities can be conducted in the area of trade in services and are called
Services Trade by Enterprise Characteristics (STEC). These data are compiled by combining the
information from enterprise or establishment trade in services surveys (generic or specialized), with the
business register that contains information on industry classification, size class and other relevant
stratifying variables. A well-maintained statistical business register (see chapter 5) is essential for such a
data linking exercise. Microdata linking is greatly facilitated when the trade in services surveys are
sampled directly from this register and use the same statistical unit. Similarly, it is recommended to
identify the full population of firms engaged in trade in services trade, for example via a multi-annual
census or via VAT registers, in order to facilitate the grossing up of the data.
16.41. It should be noted that given the diverse nature of trade in services data sources, the total value of
trade in services cannot be fully attributed to individual enterprises. Compilers should not see this as a
weakness, but rather as a consequence of the fact that some service categories, notably travel, as well as
for some others, are collected through certain methods which renders the linking to the business register
conceptually impossible.
16.42. It is a good practice that compilers carefully consider confidentiality issues when disseminating
STEC data, as a more detailed breakdown of services trade by industry and size-class may make it easy to
identify individual firms. In such instances, it is preferred to publish a set of tables that break down
services trade according to only one enterprise characteristic (e.g., either industry OR size class), rather
than a more detailed table that contains many confidential cells.
16.43. Further, as described in chapter 5, statistical business registers could prove useful in the
compilation of additional data to support the analysis of service industries and enhancing comparability

293

See for Eurostat:
http://epp.eurostat.ec.europa.eu/statistics_explained/index.php/International_trade_by_enterprise_characteristics and
for OECD: http://www.oecd.org/std/its/trade-by-enterprise-characteristics.htm
294
See e.g. the OECD-WTO TiVA project, www.oecd.org/t rade/tiva.

267

and consistency across collections, e.g. the particular example of potential mode 4 self-employed service
suppliers was indicated above.
E.

Other indicators of interest for the global analysis of service industries

16.44. Other indicators of interest for the global analysis of service industries are listed in MSITS 2010
paragraphs 5.83-5.86 and 5.111-5.112. This chapter has already dealt in detail with the compilation of
variables related to the persons crossing borders and staying temporarily abroad in the context of mode 2
or mode 4. Beyond the FATS variables of direct interest on the international supply of services (i.e.,
turnover/sales or output, employment, in particular intra-corporate), Chapter 15 has also extensively
described the compilation of other variables and their breakdowns. Out of these variables, the number of
enterprises and assets of foreign affiliates are of direct interest with commitments that could be made in
trade or investment negotiations.
16.45. For the other indicators, MSITS 2010 does not suggest new recommendations for compilation,
but rather indicates which statistics and with which (additional and reasonable) breakdowns they could be
compiled in order to be useful for trade in services policy makers and analysts. It builds upon existing
data already compiled by many countries and collected by many regional and international agencies. The
set of other indicators detailed below is in no way exhaustive, rather, it contains the indicators that will
most likely be of interest to the majority of users:
i.

Foreign direct investment statistics
a. Financial transactions, income and positions
b. By destination and origin of investment (immediate and ultimate)
c. By type of service activity (if possible with a total for services)

ii. National accounts statistics and Structural business statistics by industry
a. Value added, output, employment, capital formation
b. By type of service activity (if possible with a total for services)
iii. Employment statistics by type of service activity (if possible with a total for services)
iv. Sectoral statistics:
a. Tourism (beyond number of inbound and outbound visitors which is discussed above)
1. Number of establishments, hotels and similar establishments, food and beverage
serving activities, travel agencies and other reservation services activities
2. Number of rooms, number of bed-places (including occupancy)
3. Number of employees by tourism industries
b. Research and development: Gross domestic expenditure on research and development
(total, financed by business enterprise, by government, by higher education)
c. Audiovisual services
1. Number of internet radio stations, radio channels, television channels

268

2. Number of indoor cinemas (of which multiplexes)
3. Number of distribution companies (of which foreign controlled), number of film
exhibitors (of which foreign controlled)
4. Number of national feature films produced (of which 100% nationally produced,
and internationally coproduced)
5. Gross box office receipts of feature films exhibited (of which foreign feature
films)
6. Number of admissions of feature films exhibited
7. Number of feature films exhibited (of which national, and foreign)
d. Postal and courier services
1. Number of permanent post offices (of which staffed by administration officials)
2. Income for letter post, parcels and logistics services, postal financial services
3. Number of letter-post items, of ordinary parcels (of which domestic service, and
international service – dispatch)
e. Construction
1. Number of contractors (of which foreign controlled)
2. Construction of new building (of which dwellings, non-residential, residential)
3. Cement production
f.

Finance and insurance
1. Domestic credit provided by the banking sector
2. Insurance density
3. Insurance penetration
4. Interest rate spread
5. Market capitalization

g. Telecommunications: Number of fixed (wired-) broadband internet subscriptions,
Internet users, secure internet servers, telephone subscriptions, mobile-cellular telephone
subscriptions
h. Transport: air, sea, road, rail etc.
1. Fleet (by country of ownership, by country of registration etc.)
2. Network
3. Freight transport (national and international)

269

4. Passenger transport (national and international)
5. Container port traffic
i.

Education services
1. Number of students (primary, secondary, tertiary education)
2. Outbound and inbound mobile students (students from a given country studying
abroad)
3. Public, private current expenditure on education (primary, secondary, tertiary
education)

Environmental services
1. CO2 emissions, NOx emissions
2. Population connected to wastewater collecting system
3. Population served by municipal waste collection
4. Hazardous waste generation
5. Organic water pollutant (BOD) emissions
6. Other greenhouse gas emissions, HFC, PFC and SF6
7. Water pollution

k. Health related and social services
1. Number of hospital beds
2. Number of physicians, dentistry personnel, nurses and midwives
3. Life expectancy at birth
4. Health expenditure, public and private
5. Out-of-pocket health expenditure
16.46. The sources and recommendations on the above statistics are available from the relevant
international manuals and guidelines. Some of these publications are included in the list of references in
box 16.1. To respond to these information needs, compilers are strongly encouraged to discuss with users
what type of information would be necessary, as well as discuss with those in charge of the collection and
compilation of these statistics to identify what would be used or could be envisaged for compilation (these
could cover diverse units of the statistical office as well as various government agencies).
16.47. In addition to indicators that describe the performance of services industries, indicators to judge
services flows on an inflation-adjusted base would be useful. However, export and import price indices
for services are not easy to compile. While the Voorburg Group focuses on the measurement of producer

270

price indices, the IMF's Export Import Price Index Manual295 offers guidance for measuring prices of
internationally traded services.
16.48. Services for which prices can be compiled using existing data collection systems fall into the
domain of transport. Among price-related time series, the US Bureau of Labour Statistics (BLS), for
example, is compiling air freight, air passenger or ocean liner freight rates. On travel and tourism the US
BLS derives price information from a subset of the CPI for lodging, food, beverages, etc. from foreign
visitors by country of their residence. However, beyond transport and travel, price information on
internationally traded services is scarce and further research is needed on developing the respective
information.296
Box 16.1 References for other indicators of interest for the global analysis of service industries















International Recommendations for Tourism Statistics 2008, UNWTO
Manual on Statistics of International Trade in Services 2010, chapter 5.
SNA 2008
BPM6
BD4
System of Health Accounts, OECD
Framework for cultural statistics, UNESCO UIS
Measuring cultural participation UNESCO UIS
ISCED 2011, UNESCO UIS
ISCO 2008, ILO
Frascati Manual : Proposed Standard Practice for Surveys on Research and Experimental Development, OEC D
WTO's Integrated-Trade Intelligence Portal ( http://i-tip.wto.org/services/), statistics module
IMF, Export and Import Price Index Manual, Theory and Practice
http://epp.eurostat.ec.europa.eu/statistics_explained/index.php/International_trade_by_enterprise_characteristics
and for OECD: http://www.oecd.org/std/its/trade-by-enterprise-characteristics.htm
 OECD-WTO TiVA project, www.oecd.org/trade/tiva

295
296

Available at http://www.imf.org/external/np/sta/xipim/pdf/xipim.pdf
IMF (2009), Export and Import Price Index Manual, Theory and Practice, p. 282.

271

Chapter 17 Estimation and modelling of missing data , forecasting or back-casting
17.1. Scope. This chapter describes the use of models and estimates to complement the observed data
required for the compilation of statistics on the international supply of services, which are frequently
needed because: (a) part of the required information cannot be collected or can be gathered only at an
unsustainable cost, and/or (b) the available sources cannot provide the required coverage, detail,
frequency and/or timeliness foreseen by the international standards, and/or (c) the various sources to be
combined in the collection system are partially overlapping. The chapter consists of the following
sections: summary of good practices (section A); imputation for filling data gaps and for data editing
purposes (section B); forecasting, back-casting and revising time series (section C); allocating to
individual EBOPS items and trading partners (section D); and model-based estimates (section E).
A.

Summary of good practices

17.2. In order to properly carry out imputations needed for filling data gaps and for data editing
purposes, it is a good practice that compilers first check and load data from primary and secondary
sources followed by integration, processing and analysis of data. If the compiler believes that editing of
the source data might be needed, the source data provider should be contacted to edit or adjust data if
deemed necessary.
17.3. For dealing with unreported or aggregated transactions due to thresholds applied by ITRS,
different approaches could be used, such as collecting information on small value transactions from
periodic sample surveys or from an analysis of small transactions before the thresholds were raised. If a
threshold is established, it is a good practice that transactions below the threshold level are reported in an
aggregated amount and classified using the appropriate code. Alternatively, statistical estimations can be
applied which can be updated periodically with actual data.
17.4. When data are not available on a sufficiently timely basis, the compiler may extrapolate certain
series from earlier periods or interpolate new data points within a range of known data points. The choice
of extrapolation or interpolation method should be based on the characteristics of the past series and the
range of information available at the time of compilation. If other, more frequent, indicators provide
evidence of seasonality in the series, then the data models and interpolation techniques should take these
into account.
17.5. For back-casting, in absence of relevant indirect measures, compilers may consider using a
constant percentage change between the start and end points if no better approximation is feasible. It is a
good practice that these relationships be analysed to verify that they hold over time in order for the
compiler to decide the appropriate back-cast time period. Again, if other, more frequent, indicators
provide evidence of seasonality in the series, then the back-casting techniques should take this into
account.
17.6. Model-based estimates can be used for various statistics on the international supply of services.
For instance, for estimating travel services, a model could be constructed using primarily the number of
visitors and other short-term individuals traveling (partially available from tourism statistics or
transportation operators) and estimates of per capita expenditure obtained from occasional surveys of
persons who travel. Additionally, for estimating the value of mode 4 and the number of persons moving
under a mode 4, model-based estimates could be developed using existing statistics on the international
supply of services, travel information (including model-based estimates of a number of travelers), as well
as existing data from, for example, tourism, migration, or employment statistics. In this context, it is

272

advised that compilers analyse available metadata, familiarize themselves with the methodologies behind
data from other statistical frameworks, cooperate closely and exchange the relevant microdata with the
other statistical domains and institutions involved.
17.7. Compilers should strive to systematically allocate all services transactions to the relevant
individual EBOPS services categories, at the most detailed level possible, as well as to appropriate trading
partners. If diverse transactions are bundled into a single payment or receipt, the compiler should estimate
the relevant share of services transactions and allocate estimates to items/partners as relevant.
B.

Imputation for filling data gaps and for data editing purposes

17.8. Data gaps may arise in preliminary stages of compilation for various reasons, including nonresponses from establishment surveys, lack of timely reporting or missing entries in secondary data
sources. Compilers should consider taking the following steps to identify data gaps, missing replies, and
suspect outliers and impute data.
i.

The first step is to check and load data from primary and secondary sources. Namely, the data
should be standardized to fixed formats in advance (produced in separate processes). When
loading data, checks on codes, the completeness of necessary fields, and value ranges should be
run automatically. For primary source data, or directly reported data, only incorrect records
should not be loaded. For secondary data sources, or data that are not submitted directly by the
reporter (including administrative sources), the complete file should not be loaded until reviewed
and fully corrected;

ii. The second step in imputing data gaps and missing replies is the integration and processing of all
data. After the data files have been checked, corrected, and loaded as described above, the data
should be tabulated and processed. Data that are missing or otherwise not reported by survey
respondents can be imputed using statistical procedures such as the mean value reported by
respondents with similar characteristics or the average change for such variables from the prior
period.297 Imputation procedures are especially important for universe (benchmark) surveys that
are designed to provide results for the entire population and that form the basis of future annual or
quarterly surveys. In some cases missing values, if not provided by respondents during regular
follow-up procedures, can be obtained from financial statements and commercial databases.
iii. The third step in this process is checking the data, which involves data analysis. As part of this
analysis, significant increases and decreases in time series, e.g., of imports or exports, also as a
share of net exports, of a particular service or with a particular country (or group of countries) are
examined. The analysis should be done step-by-step, using a top-down approach, meaning
starting from total services with partner world to more detailed levels of imports, exports, and net
figures. The aim of macro-editing is to focus on suspicious values influencing publication totals.
It also leads to gains in efficiency. Analysis is also done to trace a dubious enterprise or source
that may need to be edited and possibly adjusted. A similar approach can be used for other
statistics on the international supply of services;
iv. The fourth step is the editing of primary and secondary sources. After having traced an
enterprise or other source with dubious data, the unit/source should be assessed by the editing
team and be contacted by phone or email to edit or adjust the data if deemed necessary. This will

273

be done depending on the issue identified and the result of the investigation. Various methods can
be used to adjust or edit, for example using an average of previous periods, growth rates for
available data, etc.
17.9. Another issue related to filling data gaps is when a threshold is applied for reporting/collecting
data. In many ITRS, thresholds are established for reporting transactions to prevent undue reporting
burdens and processing costs, which may imply that a considerable amount of transactions may be
missing, especially small value service transactions.
17.10. Different approaches to dealing with transactions falling below the threshold could be applied,
such as carrying a small (ad-hoc) sample survey or other sources (e.g. credit card data). Also, estimation
can be carried out through an analysis of small transactions before the threshold was raised, or through
empirical research.
17.11. For example, as described in the country experiences of Chapter 8, research conducted in Japan
suggests that the frequency of transactions increases exponentially as a value of transaction decreases and
that statistically, a Pareto distribution fits the data well.298 Compared to using information on transactions
from before a threshold increase, this approach to estimating below-threshold transactions has the
advantage in that the estimations can be updated periodically with recent data. Further, statisticians are
able to choose another statistical approach if the implemented method does not fit the data well.
B.1.

Country experience: the Netherlands

17.12. Imputation for filling data gaps at Statistics Netherlands is conducted according to the four-step
process described above. After checking and loading primary and secondary response data, imputations
are made for non-responses. For non-responses from large enterprises (LE) and small and medium
enterprises (SME) and special purpose entities (SPEs), imputations are done on the basis of the average of
the last four quarters of responses. For LE, if there is no response for Q-1, Q-4 receives a higher weighted
average. For SME, if there is no response for the last four quarters, no imputation is made. For travel,
imputation is based on Q-4 data. Other sources are imputed entirely on the basis of Q-1. To gross up (e.g.,
for SME data), each stratum is weighed, based on the number of enterprises in the population and the
number of responding enterprises or imputed responses. Source data are at different level of detail in
terms of services and/or countries involved in the international transaction. LE source contains the most
detailed data; Travel is also detailed; and all other sources are less detailed. All these sources’ data are
‘converted’ to the most detailed levels on basis of the corresponding LE data (mean of last 3 years). After
ensuring all data are fully detailed, tabulation takes place by aggregating or summing up these
components to more condensed groups of services or countries.
17.13. The Netherlands checks its data according to a top-down approach. At Statistics Netherlands, this
analysis is carried out with the use of Macroview, an interactive tool for top down analysis of statistical
data. Major built-in analysis scans within Macroview include: (a) Quick scan (all services and aggregates,
in graphical form); (b) External deliveries scan (selected services and underlying enterprises); (c)
Complete scan (all services and aggregates and all country (groups), all sources, all underlying enterprises
and weighing factors).
17.14. The editing of primary and secondary sources is submitted to the editing team with specific
assignment information. If the assessment concerns a LE or SME enterprise, the enterprise is contacted by
298

Japan will employ statistical methods starting with the implementation of BPM6. See chapter 8 (paragraphs 8.42
– 8.48) for the full description of Japan’s country experience with its ITRS.

274

phone or email. If needed, the data of the enterprise will be edited or adjusted and loaded again. If the
assessment concerns another source, the source data provider is contacted and if needed, a new source file
will be made and loaded again.

Forecasting, back-casting, and revising time series

17.15. Not all sources may supply data on a timely basis. In those instances, the compiler may have to
extrapolate certain series. Compilers are advised to consult the BPM6 Compilers guide, which explains a
number of interpolation, extrapolation and forecasting techniques that range from very simple, such as
using the value from the previous period or same change as occurred in the previous two periods, to more
complex techniques that draw information from models, taking account of seasonality in sub-annual
series.299 These techniques can be applied to various types of statistics.
17.16. Similarly, historical statistics on the international supply of services are also important (e.g. for
analysis purposes). Compilers are often asked to provide long time series, especially when new
guidelines, new data sources or a compilation methodology are introduced. Since it is often difficult to
collect source data on the new basis for a long historical period, suitable overlap periods and the stability
of relationships over time need to be analysed to decide how far back in time the data can be revised. To
generate series for earlier periods similar techniques as for forecasting can be used (for example, by. a
constant (percentage) change, possibly accounting for seasonality.
17.17. Back-casting is also very important for providing long time series for new classifications based
on EBOPS 2010. A first step could be to use the EBOPS 2010 - EBOPS 2002 correspondence table,300
taking into account that some new EBOPS 2010 classifications are not directly related to those based on
EBOPS 2002 (such as manufacturing services). It may also be difficult to relate the two classifications if
only main services aggregates are compiled, although some new recommended breakdowns can be
derived from data for earlier periods based on relationships between old and new classifications in a
recent overlap period.

C.1.

Country experience: Australia

17.18. The Australian Bureau of Statistics (ABS) maintains long time series for national and
international accounts. A large proportion of these series are maintained in original, seasonally adjusted
and trend variations. The introduction of changes to compilation methods and treatments can result in
shifts to the levels of component and total series. If the shift in level is sufficient to distort the seasonally
adjusted time series, the ABS revises the historical series to make the time series as continuous as
possible.
17.19. In some cases there is not sufficient detail available to adjust the historical series directly. In these
cases, the ABS estimates the shift in level of the series by comparing estimates at one point in time for
both the current and new basis (although comparison for additional periods is desirable).
17.20. Ideally any change in the level of a series would be measured over a sufficient time period to
enable seasonal patterns to be observed. This is possible for some modelled estimates, but for estimates
based on surveys, the cost of producing two estimates for one or more time periods is expensive both in
processing costs and provider burden. In these cases, alternative methods may be needed.

299
300

See e.g.BPM6 Compilation Guide, paragraph 8.29 and 8.30.
Available on the UNSD website.

275

17.21. To ensure consistent treatment of time series, the ABS has established a standard approach to
measuring shifts in the level of series. The size of the level shift induced by a methodological or
measurement change is assessed using regression analysis techniques on ratios between the current
published estimates and actual or simulated estimates produced by the revised methodology. In cases
where the level shift was found to be significant in the seasonally adjusted series, the historical series is
back casted to make the time series as continuous as possible while maintaining, as far as possible, the
integrity of the period to period seasonally adjusted movements, taking into account real world changes.
For a small number of lower level series it may not be possible to create a valid time series and these
series are marked 'not available' for periods prior to when data collection commenced.
17.22. Where it is not possible or necessary to maintain a long time series, an approach of ‘bridging’ the
current published estimates and the estimates produced by the revised methodology is used. ‘Bridging’
means that estimates on both the current and new basis are produced for one point in time and both sets of
estimates are released along with analysis to help users understand the differences between the series.
This technique is particularly relevant for series where modelling beyond a certain time may not be
appropriate.
D.

Allocating to individual EBOPS items and trading partners

17.23 Compilers should make every possible effort to allocate collected data on services transactions to
the appropriate individual EBOPS items. Throughout the data collection and compilation process,
compilers should as much as possible avoid using existing or catch-all services categories (e.g., "other
business services", "other business services n.i.e.", “services not allocated301 ,” “undefined”) or catch-all
partner categories (e.g., “not allocated,” “undefined”) to assign services transactions for which precise
identification of one EBOPS 2010/partner category is not readily clear. Indeed, assigning unidentified
services transactions to EBOPS categories such as “other business services n.i.e.” (or one of its parents if
this level of disaggregation is not available, e.g. "other business services")302 will artificially inflate, and
sometimes in very large proportions, the value for this specific EBOPS category and its parent categories
(e.g., “technical, trade-related and other business services” and “other business services”). This practice
decreases the analytical usefulness and reliability of these statistics, and will also increase bilateral
asymmetries because unidentified services will be classified in different ways by different reporting
countries.
17.24 Consequently, compilers should strive to systematically allocate all services transactions to the
relevant individual EBOPS services categories, as well as appropriate trading partners. To the extent
possible, this allocation should be done at the most detailed level. If diverse transactions are bundled into
a single payment or receipt, the compiler should, to the extent possible, estimate the relevant shares of
services transactions and allocate estimations to items/partners as relevant. It is advised to use the most
appropriate modelling and statistical techniques, based on the information available to compilers. Such
practices should be clearly documented in the metadata to improve users’ understanding with respect to
the interpretation and quality of the data.

301

The Eurostat BoP Vademecum (2012 edition), defines “services not allocated” as follows: “This item was created
due to the fact that some Member States are unable to allocate certain amounts to specific services. This results in a
discrepancy between the sum of individual services and the total services item. If a Member State encounters this
problem, they should record such residual item under item ‘SN’ which corresponds to the code for ‘Services not
allocated’. Only services whose origin cannot be determined should be included under this label.”
(http://forum.europa.eu.int/Public/irc/dsis/bop/library).
302
As defined in MSITS 2020, paragraph 3.252 and the EBOPS 2010-CPC2 correspondence table 9
http://unstats.un.org/unsd/tradeserv/tfsits/msits2010/ebops2cpc_detailed.htm).

276

17.25 It is reiterated that compilers should make every effort to allocate all services transactions to
individual EBOPS 2010 services categories; however, if there is no possible solution in every case,
compilers should use a category to be labelled "not allocated," which will be shown at the same level as
the main services items (i.e., not included within any of the services items) or main partners.303 The
corresponding values should only be included at the total services level if such a category corresponds to
the classification shown, or at the total World level if it relates to the partner breakdown shown. Metadata
should clearly document when total services or total partners include not allocated elements (see chapter
18).
17.26 Compilers should also establish rules to make allocations across the correct geographical partners
if transactions are recorded under “services not allocated.” One possible method of performing this
allocation is to consider the shares of geographical distribution evident in total World trade in services. It
should be noted that the allocation by nature of transaction and by partner country are independent; e.g., it
is sometimes the case that transactions cannot be allocated to individual services items, but can be
allocated to trading partners.

Model-based estimates

17.27. When source data are not timely or otherwise adequate, compilers may also choose to create a
data model to estimate (part of) statistics on the international supply of services.304 The data model
elements should be based on the compiler’s judgment of the specific information available. Some data
models can be relatively simple. For example, they may be based on proven relationships between
different data sources and variables. An additional straightforward approach for monthly or quarterly
estimates of items with strong seasonal fluctuations, is to look at relationships of relevant data series from
the current period (quarter or month) from each of the preceding (pre-determined number of) years. When
quarterly results become available, monthly estimates would then be reconciled so that the three months
equal the corresponding quarter. Models can also be more complex, based on mathematical and
econometric approaches, including exponential smoothing, linear and logistic regression, polynomial
interpolation (or cubic splines), and multilevel models to account for hierarchical relationships among
variables.
17.28. For estimating travel services, a model could be constructed using primarily the number of
visitors and other short-term individuals traveling (partially available from tourism statistics or
transportation operators) and estimates of per capita expenditure. However, it should be noted that these
data are not always accurate, as the transport operator may not always collect information on the
residence of the customer. Furthermore, when collected, the information may relate to the nationality of
the customer and not the residence. Estimates of per capita expenditure could be obtained from occasional
surveys of persons who travel. If necessary, separate data models may be designed to measure education
and/or medical-related travel based on information from relevant institutions or special surveys (e.g.,
survey of students).305 However, in order to enable the use of tourism statistics, the differences in terms of

303

It is important to note that the role of EBOPS 2010 items labelled as "other" or "not included elsewhere (n.i.e.)"
is not to include transactions for which the compiler cannot determine to which EBOPS item they belong to, but
rather corresponds to a residual list of services which are not identified as belonging to specific EBOPS items. In
other words these "other" or "n.i.e." items are act ually defined, and correspond to a specific list of CPC Version 2
products (see EBOPS 2010-CPC Version 2 corresponden ce table available at
http://unstats.un.org/unsd/tradeserv/TFSITS/msits2010/ebops2cpc.htm).
304
BPM6 Compilation Guide, 8.17 and 8.25.
305
BPM6 Compilation Guide, 12.75.

277

the coverage between balance of payments definitions, on the one hand, and tourism statistics as used in
the International Recommendations for Tourism Statistics 2008 (IRTS 2008) (as well as in Tourism
Satellite Account: Recommended Methodological Framework 2008 ), on the other, need to be
understood306 (also see chapter 14 of this Guide).
17.29. For estimating the value of mode 4 and the number of persons moving under a mode 4, modelbased estimates could be developed, using existing statistics on the international supply of services, travel
information, as well as existing data from tourism, migration, employment statistics, etc. All of this
information could be used for building an estimation of mode 4. More information on combining different
data sources is provided in chapter 13 Model-based estimates are a cost-efficient way to use existing data
to compile modes of supply data.
17.30. For such an approach to be efficient, compilers need to analyse existing metadata, familiarize
themselves with the methodologies from other statistical frameworks, cooperate closely and exchange the
relevant microdata with the other statistical domains and institutions involved. Some adjustments to the
data or data collection tools will probably be needed, in particular to obtain the relevant breakdowns and
to clearly distinguish the population of interest. For example identifying or estimating the population
qualifying as visitors or traveling for short term-employment, including appropriate breakdowns would be
necessary as inputs for data models for estimating travel or the number of mode 2 persons/trips. For mode
4 this would entail distinguishing within the group of traveling persons that are employed, those that will
have an employer-employee relationship in their country of origin (mode 4) and those that will have this
relationship in the host economy.307
17.31. For example, tourism data sources could be used for collecting further information on the
characteristics of mode 4, while some adjustments on different concepts and terminology in both
statistical domains would almost certainly be necessary. Subsequently, an estimation of the number of
mode 4 persons and of the value of mode 4 trade could be derived by grossing up the figures by using a
model based on research.
17.32. A similar approach could be adopted for other types of sources, such as labour force surveys,
household surveys or even business surveys. Adding options in the questions asked to respondents could
serve to identify mode 4 persons in such data sources.
E.1.

Country experience: Austria - a model to estimate monthly BOP data

17.32. In Austria, the Oesterreichische Nationalbank (OeNB) is responsible for compiling and
disseminating statistics on the international supply of services, including the balance of payments. In
cooperation with Statistics Austria, the OeNB collects data on Trade in Services on a survey basis, which
is conducted quarterly. In contrast to financial transactions, the OeNB found it unfeasible to employ
monthly surveys on real economic transactions due to cost-benefit considerations, as data are not
available on a monthly basis for most of respondents. Accordingly and due to the lack of user needs for
monthly data, national figures on trade in services are published as quarterly time series only. At the same
time, the OeNB has to fulfil monthly reporting requirements according to EU regulation.308 To meet

306

Ibid., 12.76.
Actually, defining services contracts versus employment relationships is something that would better serve the
compilation of tourism statistics as well as migration statistics.
308
Regulation (EU) No. 555/2012 amending Regulation (EC) No. 184/2005 on Community statistics concerning
balance of payments, international trade in services and foreign direct investment; ECB’s Guideline of 9 December
307

278

these ends, the OeNB has introduced a monthly estimation model to derive monthly Balance of Payments
data from quarterly reports.
17.33. The model produces total figures – credits and debits - for goods, services, compensation of
employees and current transfers. In addition to global figures, a regional breakdown is available for Intraand Extra-EU-trade, as well as trade with Intra- and Extra-Euro area. When assessing the model described
below, it has to be kept in mind that the results are not intended to be published as such; rather, they enter
into EU- and Euro area aggregates.
17.34. In principle, reports have to be sent to the ECB and EUROSTAT around the 10th day of the
second following month of the reporting period. Therefore, estimation is based on quarterly time series
which are updated according to revision standards as well as exogenous variables if available. After
quarterly figures have become available monthly data are reconciled with the quarterly results.
17.35. For calculating monthly figures, the original quarterly time series is first seasonally adjusted by
calculating the smoothing component, the seasonal component and the irregular components. Three
different methods of smoothing are employed: moving average, basic exponential smoothing and
exponential smoothing according to HOLT (Holt-Winters Forecasting Method). After the original time
series has thus been prepared, three different estimation models are employed: linear regression, quadratic
regression and cubical regression. A period of either12, 18 or 24 months is used for forecasting.
Therefore, 27 different modelling techniques are in principle available. For each item, monthly results are
estimated for 24 test periods according to the different models and the quality of the results is quantified
by comparing them with the actual figures by calculating relative absolute and relative quadratic
differences.
Figure 17.1Example of Austria’s monthly estimates distribution
160

inital time series
result after step 1 and 2
140

distribution of the difference between
monthly estimates and quarterly results

120

100

60
1

2011 on the statistical reporting requirements of the European Central Bank in the field of external statistics
(ECB/2011/23).

279

17.36. For services exports, an estimation model employing exponential smoothing according to HoltWinters Forecasting Method (and quadratic regression over a period of 12 months has proven to be most
suitable) 309 . The regression also incorporates overnight stays of foreign guests in Austria as an
independent variable. Still, in-coming travel makes up for approximately one third of Austria’s services
exports. Therefore the development of overnight stays is an influencing variable of the development of
overall services exports.
17.37. For services imports, a model with basic exponential smoothing and linear regression over a
period of 18 months has proven to be most effective.
17.38. After global values are estimated, they have to be divided into Intra-EU-, Extra-EU-, Intra-Euro
area and Extra-Euro area exports and imports. For estimating the current month of trade in services, the
regional division according to the respective quarter in each of the two preceding years is considered. In
contrast to linear regression, this method is employed for those items for which strong seasonal
fluctuations are observed, which is the case for trade in services.
17.39. When quarterly results become available, monthly estimates are reconciled so that the three
months equal the corresponding quarter. The adjustment is based on the Cubic Splines method by which
the monthly data are first seasonally adjusted and smoothed and the actual quarterly results are adjusted
for smoothing components. Then a multiple regression model with the Cubic Spline function is applied,
which serves as the basis for the monthly results to be estimated again. As a third step, the difference
between the actual quarterly result and the quarterly result based on the Cubic Spline function is
determined and distributed between the monthly estimates. This process is again accomplished by
applying the regression coefficients of the Cubic Spline function. As a last step, the monthly estimates
have to be adjusted for the seasonal and smoothing components which have been determined at the
beginning.
E.2.
Country experience: India: estimation of non-response in the survey on exports of
Information Technology and Information Technology Enabled Services
17.40. In the annual survey of exports of ‘Information Technology and Information Technology Enabled
Services’ conducted in India by the Reserve Bank, non-responses occur from a number of small as well as
medium size companies. The exports figures of the non-responding units are estimated using a welldefined methodology described below:
i.
Using the observed proportions of ‘nil’ and ‘closed’ units, first the number of companies
‘reporting no export of software products’, i.e., ‘nil exports’ and ‘closed’ out of the nonresponding companies are estimated. These are then removed from the universe of non-reporting
companies to obtain the number of operating non-reporting companies.
ii.
Since no information is available on the business activities of the non-reporting
companies, these are classified into four groups namely, IT services, Software product

309

For description of this method see: http://www.ons.gov.uk/ons/search/index.html?newquery=HoltWinters+Forecasting+Method+

280

development, BPO Services and Engineering services based on the observed proportions
corresponding to these four categories derived from the responding units.
iii.
As most of the small companies do not have onsite operations and an overwhelming
number of non-responding units are small companies, only offsite software exports of these
companies are estimated. For this purpose the offsite exports reported by the respondent
companies are used.
iv.
The distributions of offsite exports of the responding companies for the above mentioned
four groups are observed to be highly positively skewed and hence, instead of mean, the median
of the distributions are used for estimating software exports for each group. The estimated
software exports for ith group of non-respondent companies are computed using the formula:

�no. of reported companies in i th group �
Median (export) of i th group  �
�
� total no. of reported companies �
 no. of nonresponding operating companies.

v.
The total software exports of India is then derived adding the reported software exports of
responding companies and the estimated software export for non-respondent companies in each
of the four groups.
E.3.

Country experience: UK - experience with modelling outward FATS

17.30. Since reference year 2009, the UK has been obligated to provide Eurostat data on the majority
shareholdings outside the EU held by UK Ultimate Controlling Institutions (UCIs). This regulation
requires the UK to produce Outward Foreign Affiliate Statistics (OFATS) on an annual basis reporting on
the economic activity, turnover, number of persons employed and country of residence of subsidiaries
where a UK UCI has a majority shareholding. In order to compile these statistics, the UK adopted a
model-based approach using data from survey returns in conjunction with auxiliary information from the
Euro Groups Register (EGR) and the National Business Register.
17.31. The UK OFATS survey collects data from 200 enterprise groups where the UCI is resident in the
UK. Data items include employment, turnover and the number of foreign affiliates. These variables are
consolidated by country and industry for each parent company. By matching this returned information
with corresponding company data held on the EGR and national business register, a model is developed.
This is then applied to the remainder of the un-sampled population.
17.32. The process involves estimating affiliate employment (logistic regression, multilevel model) and
calibration:
i.

Logistic regression. An initial logistic regression is used due to the prevalence of zero
returns which affects the fit of the model. Sample returns are re-coded to a binary variable
to indicate if the employment is zero or non-zero. This model then calculates the probability
that the survey response is non-zero;

ii.

Multilevel model. Having developed a model to estimate if a record is non-zero, a multilevel
model is then used to estimate a value for the employment of an affiliate. The explanatory

281

variables that are used in our model have a hierarchical relationship and this model attempts
to account for that. For both models, the coefficients are estimated using the sample returns
and are then used to predict the estimates when applied to the un-sampled population;
iii.

Calibration. The final stage involves calibrating estimates to employment of global groups
outside of the UK. Information on global employment is available from the EGR, whilst
information on the groups UK employment is available from the national business register.
Model estimates are then calibrated to the calculated employment outside of the UK.

17.44. The estimates produced for UK OFATS are currently branded as experimental statistics by ONS,
as they are new official statistics undergoing evaluation.

282

Part IV

Cross-cutting topics

Part IV consists of four chapters which provide guidance on metadata (Chapter 18), quality management
(Chapter 19), data and metadata dissemination (Chapter 20), and on the uses of information and
communication technology (Chapter 21).

Chapter 18 Metadata
18.1. Scope. This chapter describes concept and structure of metadata (or data that define and describe
other data and processes310 ) for use within the statistical framework for describing the international supply
of services as well as good metadata compilation practices. It underscores that metadata are relevant for
the correct understanding of the content, coverage and limitations of the data, and should guide users on
their correct interpretation. The Chapter consists of the following sections: a summary of good practices
(section A); an overview of the basic concepts and definitions and the role of SDMX (section B);
indicative list of metadata items (section C); metadata standards of international and regional
organizations (section D) and country practices (section E).
A.

Summary of good practices

18.2. This Guide recognizes that metadata (i) play a crucial role in the statistical production process, as
they enable and facilitate sharing, querying, understanding and using data over the different stages of their
collection, compilation and dissemination, and at their various levels of aggregation and (ii) are
indispensable for assessment quality dimensions of data, as their availability and wide dissemination
constitute a precondition of the correct interpretation of publicly available statistics and their effective
use.
18.3. It is advised that compilers take into account that statistical metadata cover the following
items311 : statistical description, unit of reference, reference period, institutional mandate, confidentiality,
release policy, frequency of dissemination, dissemination format, accessibility of documentation, quality
management, relevance, data accuracy and reliability, timeliness, comparability, coherence, cost and
response burden, data revision and statistical processing.
18.4. The use of standard terminology for metadata across the various statistical domains and use of the
SDMX information model is also advised, as this will facilitate further integration of statistics, the
standardized sharing of data and the international comparability of data.
18.5. It is further advised that compilers design their metadata systems in the most efficient way so that
metadata items can be conveniently retrieved from the relevant databases, be used in the generation of the
intermediate and final datasets or in production of other metadata, and be updated and synchronized. In
this context, compilers are encouraged to design and use a data warehousing system of data and metadata
by which the dissemination of data and metadata becomes integrated with the collection and processing
components of the statistical production process.

310

See ISO/IEC FDIS 11179-1 "Information technology - Metadata registries - Part1: Framework", March 2004.
https://www.iso.org/obp/ui/#iso:std:iso-iec:11179:-1:ed-2:v1:en
311
Guidelines on IES, paragraph 5.91.

283

18.6. With regard to compiling metadata, compilers are encouraged to follow standardized metadata
concepts; make use of the metadata developed in related statistical domains and already being applied in
their national statistical system; define layers of metadata; establish metadata registries; incorporate
structural metadata items into the data processing as early as possible; establish clear links between data
and metadata; and compile reference metadata.
18.7. In the case of countries with less developed statistical systems it is a good practice to begin by
setting up an exhaustive, consistent and detailed repository (possibly in the form of a metadata registry)
with both structural and reference metadata, adopting as much as possible metadata concepts that are
standardized acrossall statistical domains, both nationally and internationally. This Guide strongly advises
that the next immediate priority should be granting equal, easy, extensive and timely access to metadata to
all user groups including general public.
18.8.
The structural metadata items promoted by this Guide are those defined within the framework of
the Balance of Payments Data Structure Definition (DSD), defined by MSITS 2010 and this Guide for
FATS and for additional indicators on the international supply of services. Metadata items should cover
both monetary and non-monetary (quantitative) data items which compilation is encouraged by this
Guide.
18.9. The metadata standards of international organisations should be carefully considered by
compilers, both to improve their metadata collection and compilation, as well as to ensure better
compliance with their international and regional data and metadata reporting obligations.
18.10. It is good practice that any deviations from international standards and use of estimations and
modelling to compile certain data series are clearly documented in metadata.
B. Metadata: basic concepts and definitions, the role of SDMX
18.11. Definition of metadata. Metadata are data that defines and describes other data and processes.312
Thus, data become metadata when they are used to describe other data.313
18.12. Scope of statistical metadata. Statistical metadata, according to the UN Statistical Commission,
describe various elements of the statistical processes including collection, processing and production of
statistical data and indicate the data sources and tools that are instrumental in statistical production like
statistical standards and classifications, business registers and frames, statistical methods, procedures and
software. Section C provides an indicative list of structural and reference metadata items relevant in the
context of statistics on the international supply of services.
18.13. Institutional arrangements for metadata compilation. To reduce the burden associated with
projects on metadata for statistics compiled within the framework for describing the international supply
of services, it is a good practice for compilers to closely cooperate with the specific units within the
national statistical system responsible for ensuring that metadata is produced, that it adheres to a standard
format, and that it is properly maintained and updated.

312

284

18.14. The UN Statistical Commission recommends the use of standard terminology for metadata across
the various statistical domains as this will facilitate the international comparison of data314 ; increasingly
encourages countries to treat metadata compilation and, subsequently, their dissemination as an integral
part of the statistical process in any statistical domain and promotes the standardization of the compilation
and dissemination of metadata315 .
18.15. The way forward: metadata warehousing. Statistical agencies have traditionally developed
separate databases for each statistical output. While this practice may simplify development processes, it
can be a hindrance to the successful integration of statistics, especially if there is no effort to standardize
variable definitions, labels and formats. Use of a centralized data warehousing system for data and
metadata can make creating, maintaining and accessing metadata more efficient and can contribute to the
integration of economic statistics. This process is facilitated as better information and communication
technology (ICT) tools become available316 .
18.16. With well-designed data warehouses, the dissemination of data and metadata becomes integrated
with the collection and processing components of the statistical production process. A data warehouse
should establish a simple and efficient process for accessing data to provide:

Comprehensive metadata to facilitate understanding and analysis;

ii.

Consistent and coherent long-term time series;

iii.

Reliable information about the availability of data;

iv.

Information about the availability of updated versions of published series;

Contact details for the people who can provide more information about a statistical

output.
18.17. The implementation of a more comprehensive metadata system is an important prerequisite for
developing an integrated questionnaire in the statistical system. The metadata will eventually provide the
necessary coherence between the various estimates and data collection tools leading to the production of
the statistical information. For sophisticated users, metadata are not only about concepts related to units,
variables and classifications; rather, metadata are also about quality.
B.1.

The role of SDMX

18.18. The Statistical Data and Metadata Exchange (SDMX) project is developed by an international
consortium317 for use in data and metadata management. The SDMX information model is applicable for
much of the information stored and processed within statistical organizations and its use by such
organizations is promoted by the UN Guidelines on integrated economic statistics.318

314

Guidelines on Integrated Economic Statistics (UN), page 40.
The important role in this respect was played by UNECE publication "Terminology on Statistical Metadata",
Conference of European Statisticians Statistical Standards and Studies, No. 53, Geneva, 2000.
316
See chapter 21 for more information on the use of ICT in the statistical process.
317
The BIS Guidelines, paragraph 5.123.
318
Guidelines, paragraph 5.123.
315

285

18.19. The use of the standardized information management model is very important for compilers of
statistics on the international supply of services as various agencies participate in the data collection and
compilation at different stages of the statistical production process and the establishment of a standardized
data sharing between them results in additional efficiency gains.
18.20. The development of global Data Structure Definitions (DSDs), defining the structure to data to be
exchanged (seesection C), by the SDMX consortium and international organizations enable the broader
adoption of the SDMX standard for data collection, exchange, and dissemination.
Box 18.1Statistical Data and Metadata Exchange (SDMX)

SDMX. Statistical Data and Metadata Exchange (SDMX) is an international co-operation initiative aimed
at developing standards and the employment of more efficient processes for the exchange and sharing of
statistical data and metadata among international organisations and their member countries.

The rationale of SDMX is standardisation for statistical data and metadata access and exchange. With the
ever increasing ease of use of the Internet, the electronic exchange and sharing of data is becoming easier,
more frequent and important. This heightens the need for the development of a set of common standards
for exchange and sharing of statistical data and metadata, and for making processes more efficient. As
statistical data exchange takes place continuously, the gains to be realized from adopting common
standards are considerable both for data providers and users.

The objective is to establish a set of commonly recognised standards, adhered to by all players, making it
possible not only to have easy access to statistical data, wherever these data may be, but also access to
metadata that makes the statistics more meaningful and usable. The standards are envisaged to help
national organisations to fulfill their responsibilities towards users and partners, including international
organisations, more efficiently. Among other things they are seen as facilitating use of Internet-accessible
databases in order to be able to retrieve data as soon as they are released. Several quality dimensions can
also be improved through the use of SDMX standards, such as timeliness, accessibility, interpretability,
coherence, as well as cost-efficiency.

Indicative lists of metadata items

18.21. This section contains the indicative lists of structural and reference metadata items which
countries are encouraged to compile as part of the production of their statistics within the framework for
describing the international supply of services. (The issues specifically related to the dissemination of
metadata are discussed in Chapter 20.) The list of structural metadata provided below stems from the
Balance of Payments DSD. The reference metadata are based on the recommendations contained in
MSITS 2010.
18.22. In the context of the SDMX, structural metadata refers to identifiers and descriptors of data, such
as concepts and attributes of variables, which are essential for discovering, organizing, retrieving and
processing statistical datasets. They can be thought of as the ‘labels’ that need to be associated to each
data item in order for it to have a meaning at all. Reference metadata are of a more general nature and
may refer to specific statistical data, to entire data collections or even to the institution that provides the
data.

286

C.1.

Structural metadata - The Balance of Payments Data Structure Definition (DSD)

18.23. The structural metadata contains the list of concepts and attributes of variables that are necessary
to codify the reporting requirements of four international agencies 319 for data collection exercises of
external sector statistics, including international trade in services320. The BOP DSD structure is based on
the methodology defined in the IMF Balance of Payments and International Investment Position Manual,
sixth edition (BPM6), the Manual on Statistics of International Trade in Services (MSITS 2010), and the
OECD Benchmark definition of foreign direct investment-4th edition. In order to code trade in services
data, the extended Balance of Payments Services classification, including the complementary groupings,
is included in the “international account item” dimension of the DSD.
18.24. The BOP-DSD, presented on the SDMX website321 , includes 16 concepts and 13 attributes (see
box 18.2). Concepts are used to uniquely identify a time series and, when joined together, they provide
the series code or “time series keys,” which are the unique identifiers for a time series. When defining a
time series key using SDMX, a valid code must be assigned to each concept of the DSD. Attributes are
used to further describe the data.
18.25. When coding detailed trade in services by partner country statistics, a number of the BOP-DSD
concepts are fixed; e.g., the reference and counterpart sectors defined as the total economy (S1) when the
data refer to total trade between both related and unrelated parties322 . However trade in services between
related parties can also be coded by using in the counterpart sector dimension the code S1A “affiliates”
whereas S1B should be used for unaffiliated parties.
18.26. Other concepts of the DSD are not fixed; e.g., the case of the “counterpart area” which is used to
identify the territory of the non-resident entity of individual time series. The country code list in the
counterpart area follows the ISO classification and is a "cross-domain" code list,. The codes used for
various regional groupings were harmonized across international agencies that use the BOP-DSD,
wherever possible.

319

The four agencies are the European Central Bank, Eurostat, IMF, and the OECD.
At the time of writing, there is not yet a DSD ava ilable neither for FATS neither for non monetary indicators of
modes of supply. A FATS DSD would include specific dimensions such as FATS characteristics: e.g., number of
enterprises, number of persons employed, etc.; Resident economic activity vs. non-resident economic activity; Items
specific to non-monetary indicators on modes of services supply would include Direction of trips: Inbound,
outbound; Country of origin or destination;
321
http://sdmx.org/?page_id=1747
322
In order to eliminate the possibility of having multiple ways of coding the EBOPS 2010 complementary grouping
“total services transactions between related entities”, this item is not coded in the international accounts dimension
of the DSD, but is coded in the Annual International Trade in Services dataflow as follows (example on the credit
side): A.N.#.%.S1.S1A.T.C.S._Z._Z._Z.$._T._X.N.
320

287

Box 18.2 List of concepts and attributes in the BOP-DSD
The second column provides information on the coding that should be used in the context of an annual trade in
service data submission:
1.Frequency
The code for the annual periodicity is ‘A”.
2. Adjustment indicator
The code for no adjustment is “N”
3. Reference country or area
For a reference country, a ISO 3166 code should be used
4. Counterpart Area
For the partner country, a ISO 3166 code should be used
5. Reference sector
The code for total economy which is used for ITS in all cases is “S1”
6. Counterpart sector
Code “S1” for total economy, “S1A” for related enterprises, “S1B” for
unrelated enterprises. ref para. 18.38.
7. Flows and stocks indicators
The code for Transactions which is used in all cases for TIS is “T”
8. Accounting entries
Credit “C”, Debit “D”or Net “N”
9. International accounts item
“S” for total services, “SC” for transportation, etc.
10. Functional category
Identifies functional categories applicable of financial accounts. It is not
applicable for trade in services. Code: “Z”
11.
Instrument
and
assets Identifies the type of financial instrument which is reported in the external
classification
sector time series. It is not applicable for trade in services. Code: “_Z”
12. Maturity
Identifies the types of maturity of the financial instrument of the external
sector statistics time series: It is not applicable for trade in services. Code:
“_Z”
13. Unit of measure
Refers to a currency unit.
14. Currency of denomination
Identifies the currency of denomination of the financial instrument. For
ITS, a constant “_T” is applied.
15. Valuation
Identifies the method of valuation for selected transactions and positions
data. For ITS, this is coded “unspecified “_X”.
16. Compilation methodology
Distinguishes between time series which are compiled according to the
methodology applied for national statistics by opposition to similar time
series which follow the specific methodology applied for economic or
currency union statistics. For ITS, this is coded as national “N”.
The BOP-DSD also uses the following 13 attributes:
I. Time format
II. Observation status
III. Confidentiality Status
IV. Pre-break value
V. Comments to the observation value
VI. Detailed description (title complement)
VII. Short title
VIII. Unit multiplier
IX. Decimals
X. Time period collection
XI. Reference period detail
XII. Compiling organisation
XIII. Underlying compilation

C.2

Example of a trade in services time series key coded according to the BOP DSD

18.27. The times series key [A.N.US.FR.S1.S1.T.C.SC._Z._Z._Z.USD._T._X.N] provides an example
of codification for a trade in services series. This times series key stands for a series in annual Frequency
“A”; with no adjustment indicator “N”; the Reference country is the United States “US”, the counterpart
area is France “FR”; the reference sector is the total economy “S1”; the counterpart sector is the total
economy “S1”; the Flows and stocks indicator dimension indicates a Transaction “T”; the accounting
entry dimension indicates that we are looking at a credit entry “C”; the international accounts item
specifies that the series relate to transport services “SC”; the functional category is not applicable “Z”; the

288

instrument and assets classification is not applicable “Z”; the maturity is not applicable “Z”; The Unit of
measure is US dollar “USD”; the currency of denomination is set to “All currencies of denomination” _T;
the Valuation is unspecified _X; and the compilation methodology is National “N”.
18.28. As noted above, attributes are used to qualify observations further; e.g., they provide information
on the “confidentiality” status or the “compiling organisation.” Attributes do not contribute to the
identification of a time series, as this is already done by using the dimensions.
C.3.

Reference metadata

18.29. The following items are typically part of the reference metadata associated with statistics on the
international supply of services:
i. Legal framework and institutional arrangements (e.g., references to relevant laws and
regulations, role of all institutions involved in compilation and description of coordination of
dissemination of statistics and data-sharing agreements among these institutions, either distinctly
or as part of broader statistics (e.g., balance of payments and other external sectors statistics));
ii. Underlying concepts and definitions (e.g., definition of residence; non-residence; residence
of units, as applicable; definition of statistical value; scope of statistics on the international supply
of services and their relationship to national accounts and international merchandise trade
statistics; distinction from other international transactions; classification under relevant services
item according to BPM6/EBOPS 2010 and any deviations from international standards, if any;
ultimate controlling institutional unit (UCI) concept; definition of a foreign affiliate; definition of
direct or indirect control; definitions of statistical and reporting units, etc.);
iii. Description of core data sources (e.g., ITRS, enterprise/establishment surveys, surveys of
households or persons, administrative records, statistical models, partner country data, or
combination of sources), including specific notes on services categories or activities for which
particular data collection arrangements or combination of sources are employed and comments on
limitations of source data in terms of coverage, frequency, level of detail, reliability, availability,
etc.;
iv. Description of data collection, data compilation methods and data-processing procedures,
including frequency of data collection; description of specific procedures used for data collection,
validation, editing, aggregation, etc.; adjustments made to source data, such as imputations,
misclassification, adjustments for non-response or under-coverage; adjustments to standard data
processing procedures, such as coding, tabulation errors, etc.; indications of departures from
international standards, if any;
v. Estimation methods, such as descriptions of methods for estimating non-reported
transactions or transactions falling below customs and/or ITRS thresholds (e.g., CIF-FOB
adjustments for transportation item);
vi. Dissemination policy, including release and revision schedules; indication of the
presentational format of data; level of disaggregation, eventual commentaries accompanying the
data; etc.;
vii. Additional explanations and footnotes concerning the data as required (e.g., explanatory
notes on revisions, breaks in series, definitions of confidentiality flags, etc.);

289

viii. Quality reporting, including publication of regular quality reports that use the quality
dimensions in the Template for a Generic National Quality Assurance Framework (NQAF)323 and
that include definitions of such quality dimensions (e.g., timeliness, accessibility, comparability,
etc.);
ix. Confidentiality, including descriptions of confidentiality rules and indications of how much
data is affected by such rules.324
18.30. Compilation of metadata. Metadata are compiled at all stages of the statistics production process.
This Guide encourages countries to use the following good practices, as applicable, in metadata
compilation:

i.
Use of standardized metadata concepts. In the same way as any data item, metadata items
also have to be clearly defined. Even though each statistical domain, including statistics of
international trade in services on the international supply of services, has its specific metadata
items, it is a good practice to use applicable standardized concepts that are relevant across
statistical domains (e.g., by adopting cross-domain concepts from the SDMX framework or
OECD Glossary of Statistical Terms). The aim should be to promote harmonization of statistical
information and their related high-level metadata across various institutions and statistical
domains, even if some specific metadata concepts are not applicable or are organized differently
in different domains or institutions;
ii.

While developing metadata for statistics compiled within the framework for describing
the international supply of services, make use of the metadata developed in the related statistical
domains and used in your country. Statistics of international trade in services is a relatively new
statistical domain in many countries. It is very likely that the metadata policy is already in place
in related statistical domains. Compilers are advised that such metadata is carefully reviewed and
utilized;

iii.
Define layers of metadata. It is a good practice to compile metadata in layers of
incremental detail and provide clear links between high-level and specific metadata concepts.
Such a layered structure of metadata will allow data users and analysts to access necessary
metadata items and to minimize the risk of misinterpretation of data content while, for example,
compiling data from various data sources, as well as to ensure clear presentation of metadata to
diverse groups of users.
iv.
Establish metadata registries. A metadata registry is a central repository (usually a
database itself) with information that allows for linking of the detailed definitions (semantics)
with the codes (representations) of the metadata items used to describe a particular statistical
dataset. It is a good practice that compilers put special emphasis on the development,
maintenance and dissemination of metadata registries in order to improve the harmonization,
standardization, use, re-use and exchange of their metadata.325

323

https://unstats.un.org/unsd/dnss/QualityNQAF/nqaf.aspx. See chapter 19 for more information on NQAF and
quality reporting and management in general.
324
See chapter 20 for more information on statistical confidentiality.
325
The Euro SDMX Registry includes harmonized structural metadata, the DSDs designed for the statistical
domains, metadata structure definitions; e.g., ESMS and other related information.

290

Confidentiality and access to metadata during the compilation process. As metadata for
statistics on the international supply of services might be compiled by various units of the same
agency or by units located in different organizations, there might be cases when metadata
describes data subject to confidentiality rules. It is a good practice, in this context, that
confidentiality rules are set up is a such way that they will allow compilers to obtain nonconfidential data aggregates with the same metadata content;

vi.
Incorporate structural metadata items into the data processing as early as possible (e.g.,
as parts of the records structure). This step will facilitate data processing including the
identification of viable options for data aggregation and subsequent presentation. It is advisable
that structural metadata are made an integral part of the database containing statistics compiled
within the framework for describing the international supply of services in such a way that it can
be extracted together with any data item and used in data processing to obtain meaningful
combined data sets.
vii.
Establish clear links between data and metadata. As metadata are generated and
processed during every step of the data compilation process, there is a strong requirement to
ensure that the appropriate metadata retain their links with data. In this connection it is a good
practice to implement metadata-driven management along the various stages of the statistical
production process.326
viii.

Compilation of reference metadata. Reference metadata can be presented as a detailed
explanatory note describing the scope, coverage, and quality of a dataset and can be made
available electronically alongside the database or in special publications.
18.31. Priorities in metadata management. Although ideally the management of metadata would take
into account all of the recommendations highlighted thus far, countries with less developed systems for
statistics on the international supply of services should begin by setting up an exhaustive, consistent and
detailed repository (possibly in the form of a metadata registry) with both structural and reference
metadata, adopting as much as possible metadata concepts that are standardized cross-domain, nationally
and internationally. The next immediate priority should be granting easy, extensive and timely access to
metadata to the general public. In subsequent phases the system could be improved by gradually
incorporating more advanced features; e.g., implementing a layered presentation of metadata, actively
linking data and metadata, etc.
D.

Metadata standards of international and regional organizations

18.32. International availability of appropriate metadata is of great interest to all the organizations
having global or regional responsibilities. These organizations have made efforts to standardize their
requirements on scope and structure of the data and metadata which they would like to obtain from
countries. Those requirements should be carefully studied by countries, both to improve their metadata
collection and compilation, as well as to ensure better compliance with their international and regional
data and metadata reporting obligations.

326

There are several information model specifications that can contribute to achieve this goal (most notably SDMX
and DDI ), which are designed to perform different functions but can be used together in the same system, or
complement each other in the compilation and exchange of data and metadata.

291

D.1.

SDDS/GDDS metadata of the IMF

18.33. The Special Data Dissemination Standard (SDDS) and General Data Dissemination System
(GDDS) are part of the IMF’s data standards initiative aimed at enhancing member countries’ data
transparency and promoting development of sound statistical systems. A dedicated electronic bulletin
board on the IMF website 327 posts information that SDDS countries provide to the IMF on their
dissemination practices and offers direct links to the economic and financial data that countries
disseminate under the SDDS and information that GDDS countries make available to the IMF on their
statistical practices
18.34. For both standards, metadata are organized by country and by topic. The SDDS metadata are
available in two presentations: the current SDDS format and the DQAF format, while the GDDS uses the
DQAF presentational format. Revisions to metadata are deemed to be made regularly and are available on
the website. Metadata aspects related to statistics compiled within the framework for describing the
international supply of services are embedded in the various quality dimensions of the balance of
payments framework.
D.2.

The SDMX content-oriented guidelines on metadata by Eurostat

18.35. Based on the SDMX information, model data structure definitions (DSDs) can be created for data
on the international supply of services (including FATS).328 The SDMX content-oriented guidelines have
been used to define reference metadata for the European Statistical System (ESS).329 Table 18.1 lists the
main components of the ESS reference metadata. This Guide advises compilers of statistics on the
international supply of services of other regions to take the EU recommendations into account, as
applicable, while setting up the conceptual structure of their own reference metadata for such statistics.
D.3.

Ensuring consistency of metadata and data reported to international organisations

18.36. Reporting of trade in services data and metadata to all international organisations should be
consistent. This can be achieved by ensuring that the total services and main items are transmitted with
identical values to all international organisations, even if a "not allocated" category is used in a compiling
country.
18.37. As mentioned in chapter 17 (section E), compilers should make every effort to allocate all
services transactions to the relevant individual EBOPS services categories and to appropriate trading
partners and should avoid, as much as possible, using catch-all services categories (e.g., "other business
services", "other business services n.i.e.", “services not allocated330,” “undefined”) or catch-all partner

327

http://dsbb.imf.org.
Available at: http://epp.eurostat.ec.europa.eu/portal/page/portal/statistics/metadata/metadata_structure; inward
FATS metadata are available at: http://epp.eurostat.ec.europa.eu/cache/ITY_SDDS/en/fats_esms.htm; while outward
FATS metadata are presented at: http://epp.eurostat.ec.europa.eu/cache/ITY_SDDS/EN/fats_out_esms.htm
329
See also the Commission Recommendation 2009/498/EC of 23 June 2009 on reference metadata for the
European Statistical System (OJ L 168,30.6.2009, page 50).
330
The Eurostat BoP Vademecum (2012 edition), defines “services not allocated” as follows: “This item was created
due to the fact that some Member States are unable to allocate certain amounts to specific services. This results in a
discrepancy between the sum of individual services and the total services item. If a Member State encounters this
problem, they should record such residual item under item ‘SN’ which corresponds to the code for ‘Services not
allocated’. Only services whose origin cannot be determined should be included under this label.”
(http://forum.europa.eu.int/Public/irc/dsis/bop/library).
328

292

Table 18.1 Main components of the European Statistical System reference metadata
Concept Name

Descriptions
Individual or organisational contact points for the data or
metadata, including information on how to reach the contact
points.
The date on which the metadata element was inserted or modified
in the database.
Data description, classifications used, concepts and definitions
etc.
The unit in which the data values are measured.
The period of time or point in time to which the measured
observation is intended to refer.
Set of rules or other formal set of instructions assigning
responsibility as well as the authority to an organisation for the
collection, processing, and dissemination of statistics.
A property of data indicating the extent to which their
unauthorised disclosure could be prejudicial or harmful to the
interest of the source or other relevant parties.
Rules for disseminating statistical data to interested parties.

Contact

Metadata update

Statistical presentation

Unit of measure

Reference period

Institutional mandate

Confidentiality

Release policy

Frequency of dissemination

The time interval at which the statistics are disseminated over a
given time period.

Dissemination format

Media by which statistical data and metadata are disseminated.

Accessibility of
documentation

Quality management

Relevance

Accuracy and reliability

Timeliness and punctuality

Comparability

Coherence

Cost and burden

Data revision

Statistical processing

Comment

Systems and frameworks in place within an organisation to
manage the quality of statistical products and processes; quality
assurance and quality assessment.
The degree to which statistical information meet current and
potential needs of the users.
Accuracy: closeness of computations or estimates to the exact or
true values that the statistics were intended to measure.
Reliability: closeness of the initial estimated value to the
subsequent estimated value.
Measurement of the impact of differences in applied statistical
concepts, measurement tools and procedures where statistics are
compared between geographical areas or over time.
Adequacy of statistics to be reliably combined in different ways
and for various uses.
Cost associated with the collection and production of a statistical
product and burden on respondents.
Any change in a value of a statistic released to the public.
Source data, frequency of data collection, data validation, data
compilation and adjustments.
Supplementary descriptive text which can be attached to data or
metadata.

293

categories. To conduct such allocations, compilers could use information from partner economies; if this
information is not available or reliable, then it is advised to use the most appropriate modelling and
statistical techniques, based on the information available to compilers. (See chapter 17 for more
information on estimation and modelling of missing data.) Such practices should be clearly documented
in the metadata.
18.38. If there is no acceptable way to allocate all service, compilers should use a category to be labelled
"not allocated," which will be shown at the same level as the main services items (i.e., not included within
any of the services items) or main partners.331 The corresponding values should only be included at the
total services level if such a category corresponds to the classification shown, or at the total World level if
it relates to the partner breakdown shown.
18.39. A "services not allocated" item is included in the BOP-DSD (because it is part of the reporting
requirements of ECB and Eurostat for the quarterly Balance of Payments (QBOP) and of the requirements
by Eurostat, OECD and UNSD for annual international trade in services332 data), whereas such a category
is not included in the QBOP data reporting requirement of the IMF. Consequently, for countries that use a
category "services not allocated," the individual services items will not add up to the total services in their
report to IMF. In this case, the compiler should indicate in the metadata provided to the agencies that that
total services do not correspond to the sum of main services items due to the presence of some
transactions that are impossible to allocate across services. Table 18.2 presents an example of data
reporting in such a case.

331

It is important to note that the role of EBOPS 2010 items labelled as "other" or "not included elsewhere (n.i.e.)"
is not to include transactions for which the compiler cannot determine to which EBOPS item they belong to, but
rather corresponds to a residual list of services which are not identified as belonging to specific EBOPS items. In
other words these "other" or "n.i.e." items are act ually defined, and correspond to a specific list of CPC Version 2
products (see EBOPS 2010-CPC Version 2 corresponden ce table available at
http://unstats.un.org/unsd/tradeserv/TFSITS/msits2010/ebops2cpc.htm).
332
By detailed partner countries for the former and by detailed EBOPS 2010 for the latter.

294

Table 18.2 Example of reporting trade in services data to IOs for the cases in which “services not
allocated” is (and is not) part of the data collection
Credit side

Submission of data to
international organizations
following EBOPS 2010
(“services not allocated” is part
of data collection)

Total Services
Manufacturing services on physical inputs
owned by others
Maintenance and repair services n.i.e.
Transport
Travel
Construction
Insurance and pension services
Financial services
Charges for the use of intellectual property
n.i.e
Telecommunications, computer, and
information services
Other business services
Personal, cultural, and rec.services
Government goods and services n.i.e.
Services not allocated

130
10

10
10
10
10
10
10
10

10
10
10
10

10
10
10
(Metadata need to explain the
difference between total
services and sum of subcomponents)

Submission of data to
international organizations
following BPM6 (“services not
allocated” is not part of data
collection)

Country experiences
E.1.

Italy

18.40. The Bank of Italy (BoI) adopts an integrated approach to the broad range of its statistical
programmes. BoI’s approach to management of metadata is based on two pillars: (1) an information
model, capable of fully describing the data, the processing steps and the elaboration algorithms, and (2) a
software platform, which supports the entire statistical production chain.
18.41. The BoI has designed a proprietary model called “Matrix,”333 which is based on mathematical and
statistical theory, to support all phases of statistics production process (data definition, collection,
compilation and dissemination) and all the data of interest (micro/aggregated, registers, questionnaires,
etc.). A fundamental infrastructural component of BoI’s system, representing a core part of the actual
implementation of the Matrix model, is the central statistical dictionary, a repository describing the whole
content of the statistical data warehouse, in terms of structural metadata (e.g. concepts, classifications,
data structures, processing rules) and reference metadata (e.g. methodological notes).
18.42. The Matrix model was also designed to take into account major international standards, so that,
for example, the Matrix data and metadata can be easily transformed into SDMX and other metadata
formats. Another essential feature of the Matrix model is that it enables a metadata-driven system by

333

Vincenzo Del Vecchio, Fabio Di Giovanni, Stefano P ambianco, The Matrix Model – Unified model for statistical
data representation and processing, Banca d’Italia, 2007.
(http://www.bancaditalia.it/statistiche/quadro_norma_metodo/modell_SIS/matrixmod.pdf).

295

utilizing a recently introduced software platform for statistical processing called INFOSTAT. 334
Consistent with the underlying holistic approach, INFOSTAT supports the statistical production chain
end-to-end, by providing the following services:
i.
ii.

Identity and access management (e.g., user registration, authentication, user profiling);
Metadata definition;

iii.

On-line data-entry and data upload;

iv.

Support for a secure data transmission, storing and versioning;

Validations and handling of reporters’ feedbacks;

vi.

Calculations;

vii. Data and metadata import, export and exchange;
viii. Event subscription and notification;
ix.

User environment for metadata prototyping and data production;

Reporting and publishing;

xi.

Search of information;

xii. Inquiry and download of data and metadata;
xiii. End-to-end monitoring of business processes.
18.43. An important advantage of this approach is that most of the changes in the statistical processes
(e.g., the establishment of a new survey, the production of new sets of statistics, the release of a new
publication, etc.) can be timely implemented by metadata administrators, avoiding the need for complex
software maintenance. In fact, due to an advanced user interface, metadata administration is rather userfriendly, allowing non-IT users to accomplish it directly, without the intervention of technical staff.
18.44. The system can handle both qualitative and quantitative indicators, micro and macro data,
questionnaires, registers and unstructured data (documents), thus allowing broad integration. INFOSTAT
also adopts the Matrix model language, called EXL (expression language), to define expressions used in
data validation and in data processing phases for calculations. EXL expressions are intuitive as they are
quite similar to spreadsheet formulas, and the language is conceived to be extensible, in order to support
the great variety and variability of statistical requirements.

334

Fabio Di Giovanni, Daniele Piazza, Processing and managing statistical data: a National Central Bank
experience, paper presented at the “International Statistical Conference” held in Prague on 14-15 September 2009
(http://www.czso.cz/conference2009/proceedings/data/process/piazza_paper.pdf).

296

Chapter 19 Quality management and quality reporting
19.1.

Scope. This chapter provides a summary of good practices in quality management and quality
reporting (section A); an overview of basic concepts and definitions used in quality management (section
B); a description of quality assurance frameworks (section C); good practices in quality measurement and
reporting (section D) and country practices (section E). The quality management issues pertinent to
managing statistical system and institutional environment were covered in Chapters 1-3. Dissemination of
information on quality management and statistical output is further dealt with in Chapter 20.
A.

Summary of good practices

19.2. Proper quality management and quality reporting on data compiled in such wide domain as
statistics on the international supply of services is especially important. Therefore, it is a good practice
that compilers responsible for statistics on services transactions between residents and non-residents,
FATS and additional indicators on the international supply of services maintain high, clearly defined and
measurable quality standards and regularly disseminate information on their implementation in order to
gain the trust and confidence of their users and data suppliers and to aid users in better understanding and
appropriately analyzing the disseminated statistics. This will ultimately raise the visibility and reputation
of the compiling agencies. The implementation of quality standards should include the application of a
formalised system that documents the structure, responsibilities and procedures put in place for satisfying
users, while continuing to improve the data production and dissemination process.
19.3.

Compilers are advised to develop their own quality assessment frameworks on the basis of the
Generic National Quality Assurance Framework (NQAF) endorsed by the UN Statistical Commission and
the quality assessment frameworks developed by international, supranational and regional organizations.
They should consider that the template for a NQAF is intended to be voluntary, and not meant to be
prescriptive or viewed as a replacement for other quality frameworks already adopted or in use by a
country’s national statistical office. In this context, examples of common quality dimensions or
components include relevance; accuracy; reliability; timeliness; punctuality; accessibility; clarity;
interpretability; coherence; comparability; credibility; integrity; methodological soundness; and
serviceability.

19.4. To ensure the quality of statistics on the international supply of services, compilers are advised to
use sound statistical methodologies based on the internationally agreed recommendations contained in
MSITS 2010, while proactively managing the respondent burden. It is a good practice that countries
develop a standard for quality reports that include quantitative and qualitative indicators covering the full
range of statistical processes (i.e., data collection, processing and dissemination) and their outputs based
on the quality dimensions listed in this chapter and that are issued regularly (e.g., at least every five years
or when significant methodological changes occur). For monitoring the quality of the processes and
effectiveness of quality-improvement actions, reviews should be conducted more frequently.
19.5.

Compilers are encouraged to facilitate continuous quality improvement. In particular, when
adopting new methods, compilers should assess the impact on the statistical series of adopting the new
methodology and subjecting the proposed methodology to peer review.

297

Quality management: an overview of basic concepts

19.6. Managing quality of official statistics is vitally important to compilers’ success in maintaining
the trust and confidence of their users and data reporters. Regular dissemination of the information on the
implementation of high and clearly defined quality standards will also aid users in better understanding
and appropriately analyzing the statistics and ultimately raise the visibility and reputation of the
compiling agency.
19.7. While there are several general definitions of quality, one of the most commonly used and
succinct is fitness for use or fitness for purpose.335
19.8.

The development of the Template for a Generic National Quality Assurance Framework (NQAF)
and the Guidelines to accompany the Template was undertaken by the Expert Group on NQAF in
response to a request by the United Nations Statistical Commission at its forty-first session in 2010. The
NQAF Template is intended to be used as a tool to provide the general structure within which countries
that choose to do so can formulate and operationalise national quality frameworks of their own or further
enhance existing ones. Other international organizations have also developed data quality systems.
Examples include the IMF’s Data Quality Assessment Framework (DQAF),336 which is used to assess the
quality of countries’ macroeconomic statistics, and as a standard presentation of metadata within the
GDDS and SDDS standards. Eurostat has developed its Quality assurance Framework (see box 19.1)
which contains a total of eight quality criteria for which reporting is defined in a Handbook on Quality
reports,337 while the OECD’s Quality Framework for Statistical Activities338 explicitly focuses the quality
of the data used, produced and disseminated by the OECD itself.

335

Statistical agencies have arrived at a consensus that the concept of quality of statistical information is multidimensional and that there is no one single measure of data quality. Several statistical organisations have developed
lists of quality dimensions, which, for international organisations, are being harmonised under the leadership of the
Committee for the Coordination of Statistical Activities (CCSA). Description of the CCSA activities in this respect
is available at: http://unstats.un.org/unsd/accsub-public/data-quality.htm.
336
http://dsbb.imf.org/pages/dqrs/DQAF.aspx
337
http://epp.eurostat.ec.europa.eu/portal/page/portal/ver-1/quality/documents/EHQR_FINAL.pdf .
338
http://www.oecd.org/std/qualityframeworkforoecdstatisticalactivities.htm.

298

Box 19.1 Quality assurance in the European Statistical System (ESS)
1.

In the European Statistical System (ESS) quality of statistics is managed in the framework of the European
Statistics Code of Practice which sets the standards for developing, producing and disseminating European
statistics.

In accordance with the 15 principles of the European Statistics (ES) Code of Practice and the provisions of
Regulation (EC) No 223/2009 on European statistics, quality is approached along 3 lines:
a. quality or characteristics of the institutional environment (6 principles):
1) Professional independence
2) Mandate for data collection
3) Adequacy of resources
4) Commitment to quality
5) Statistical confidentiality
6) Impartiality and objectivity
b. quality of the statistical processes (4 principles):
7) Sound methodology
8) Appropriate statistical procedures
9) Non-excessive burden on respondents
10) Cost effectiveness
c. quality of the statistical output (5 principles):
11) Relevance
12) Accuracy and reliability
13) Timeliness and punctuality
14) Coherence and comparability
15) Accessibility and clarity
Each of the 15 principles of the ES Code of Practice (1st level of quality assurance) contains specific
indicators which show how compliance with the principle can be demonstrated (2nd level of quality
assurance). Compliance with the ES Code of Practice is regularly monitored through the ESS-wide
exercise of peer reviews which start with a national self-assessment questionnaire – improvement actions
identified in the peer review exercise are then monitored and reported upon on an annual basis.

As a 3rd level of quality assurance, the ESS Quality Assurance Framework (QAF) has been developed in
2011-2012. Similarly to other existing quality assurance frameworks like UNSD's NQAF, the ESS QAF
provides methods and tools for implementation at institutional and process level for each of the indicators
of the ES Code of Practice as well as links to relevant reference documentation. Therefore, it provides
clear guidance to compliance assessors.

19.9. To ensure conformity of the use and interpretation of the meaning of quality dimensions by
compilers of all data sets within the statistical framework statistics recommended by MSITS 2010 for

299

measuring the international supply of services, all definitions in this chapter are taken from the NQAF
Glossary, which was endorsed by the UN Statistical Commission as part of the NQAF Guidelines.339

19.10. NQAF lists the following examples of common quality dimensions or components: relevance;
accuracy; reliability; timeliness; punctuality; accessibility; clarity, interpretability; coherence;
comparability; credibility; integrity; methodological soundness; and serviceability. The dimensions of
quality are overlapping and interrelated and, therefore, the adequate management of each of them is
essential if information is to be fit for use. SDMX defines eleven quality dimensions: relevance, accuracy,
timeliness, punctuality, accessibility, clarity / interpretability, comparability, coherence, integrity,
credibility, and methodological soundness.340
19.11. First of all, compiled statistics should be relevant, meaning that they should meet current and
potential users' needs. The compiling agency's challenge is to weight and balance the conflicting needs of
current and potential users to produce statistics that satisfy the most important needs within given
resource constraints. For a breakdown by mode of supply, such statistics should be produced for services
items that are important for the compiling economy and should preferably be developed in cooperation
with the users of such data (such as the ministries of trade, the economy or foreign affairs).
The relevant services could be identified through direct dialogue with the major users or, for the case of
data broken down by mode of supply, by examining the sector’s relative share of total exports/imports of
services.
19.12. Accuracy, or the closeness of computations or estimates to the true values that the statistics were
intended to measure, should also be ensured by compilers. It is usually characterized in terms of error in
statistical estimates and is often decomposed into bias (systematic error) and variance (random error)
components. Assessment of accuracy can either contain numerical measures of accuracy or qualitative
assessment indicators. It may also be described in terms of the major sources of error that potentially
cause inaccuracy (e.g., coverage, sampling, non-response, response error).

19.13. Statistics compiled within the framework for describing the international supply of services
should also adhere to standards of timeliness, measured as the length of time between data availability and
the event or phenomenon they describe. Timeliness is a crucial element of data quality, as it increases the
statistical information’s relevance and its ability to be used effectively by policy makers. For statistics on
trade in services, the data should preferably be produced at least on an annual basis. It is a good practice
that selected (e.g., aggregated) data series covering services transactions between residents and nonresidents, FATS or additional indicators on the international supply of services are produced and
disseminated more frequently depending on country’s needs and resources. The release date of data
should also be punctual, in that it follows the target date announced in the official release calendar.
Timeliness typically involves a trade-off against accuracy and cost.

19.14. Accessibility and clarity should also be ensured. Statistics on the international supply of services
should be presented in a clear and understandable form, and disseminated through a suitable and
convenient medium, with supporting metadata and guidance.

339

Definitions and their contexts are taken from NQAF Glossary (2012), Available at:
http://unstats.un.org/unsd/dnss/QualityNQAF/nqaf.aspx).
340
See SDMX, Metadata common vocabulary, page 109; available at: http://sdmx.org/wpcontent/uploads/2009/01/04_sdmx_cog_annex_4_mcv_2009.pdf.

300

19.15. Furthermore, statistics on the international supply of services should be comparable across
geographical areas (i.e., the degree of comparability between statistics measuring the same phenomenon
for different geographical areas); over time (i.e., the degree of comparability between two or more
instances of data on the same phenomenon measured at different points in time); and across domains (i.e.,
the comparability between different survey results which target similar characteristics in different
statistical domains). For data broken down by mode of supply, if a country is focusing on a particular type
of service, a description in terms of CPC of the service would be useful. Moreover, internal coherence (or
consistency) and coherence across domains should be ensured, as statistics are often obtained from
different sources or based on different approaches, classifications and methodological standards.
Metadata must convey such information that will help any interested party in evaluating the comparability
of the data, which is often the result of a multitude of factors.
19.16. Compiling agencies should also ensure integrity, or their values and related practices that
maintain confidence in the eyes of users in the agency producing statistics and ultimately in the statistical
product. One important aspect of integrity is the trust in the objectivity of statistics. It implies that
professionalism should guide policies and practices and is supported by ethical standards and by
transparency. Integrity is closely linked with credibility, or the confidence that users place in statistical
products based simply on their image of the statistical authority (i.e., the brand image).

19.17. Compiling agencies should also ensure that sound methodologies are used to compile statistics
that comply with the relevant international standards, including the professional standards enshrined in
the Fundamental Principles for Official Statistics.

19.18. At the same time, cost effectiveness should also be ensured. For the compilation of statistics on
the international supply of services broken down by mode, the mechanical allocation of EBOPS 2010 to
modes of supply presents a strong advantage, as this method is relatively inexpensive as it is based on
existing data and knowledge of the compiler. Finally, existing data transmission mechanisms and IT-tools
should be used to the extent possible.
19.19. Quality management and its components. Quality management is defined in SDMX as systems
and frameworks in place within an organisation to manage the quality of statistical products and
processes. Quality management refers to the application of a formalised system that documents the
structure, responsibilities and procedures put in place for satisfying users, while continuing to improve the
data production and dissemination process. It also includes how well the resources meet the requirement.
This concept can be broken down into: "assurance"; "assessment"; and "documentation".
i.
Quality assurance refers to all the planned and systematic activities implemented that can
be demonstrated to provide confidence that the processes will fulfill the requirements for the
statistical output. This includes the design of programmes for quality management, the
description of planning process, scheduling of work, frequency of plan updates, and other
organisational arrangements to support and maintain planning function.
ii.

Quality assessment contains the overall assessment of data quality, based on standard
quality criteria. This may include the result of a scoring or grading process for quality. Scoring
may be quantitative or qualitative.

iii.
Quality documentation contains documentation on methods and standards for assessing
data quality, based on standard quality criteria.
301

Focusing on quality assurance

19.20. Quality assurance is a critical part of producing the statistics which ensures that the methods have
been correctly applied and that the statistics are robust and fit for purpose. Given the complexity
surrounding the compilation of statistics within the framework for describing the international supply of
services, it would be particularly relevant that organisations produce, document, implement, monitor and
maintain a quality assurance strategy, policy and quality assurance procedures which apply to regular
publications, new outputs and changes to outputs, as well as statistical outputs derived from surveys,
administrative sources and other secondary sources.
19.21. Quality assurance should be built in at each step of the statistical process, including:
i.

The selection of the methods;

ii.

Ensuring issues related to the quality outcomes of the methods chosen are identified;

iii.

Careful checking of the outcomes of the applications of the methods;

iv.
Ensuring that a sufficient range of stakeholders are engaged in the quality assurance
process.

19.22. Compilers should adopt quality assurance procedures, including the consideration of each
statistical product against users’ requirements, and of their coherence with other statistical products. The
quality assurance policy should include aspects such as control, improvement processes, quality
measures, documentation and awareness-raising. It is a good practice that the quality assurance
procedures specify clear ownership and accountability for statistics and related products.
19.23. Appropriate validation to minimise the risk of errors should also be incorporated into the quality
assurance procedures, including:
i.

Validation built into the production processes wherever possible;

ii.
Internal validation checks – for example, checking against previously produced outputs
from the same source, or parallel running by two people where there is a large degree of manual
intervention;
iii.

External validation checks – for example, ‘sense-checking’ against other relevant sources

19.24. For all regular statistical outputs, a programme of periodic reviews should be planned and
undertaken, covering quality, methodologies and processes. More information may be found in
international guidelines on national quality assurance, such as the Guidelines for the Template for a
Generic National Quality Assurance Framework, which have been prepared and are available at the
UNSD website.341 At its forty-third session in 2012, the Statistical Commission fully endorsed the generic
national quality assurance framework template and encouraged countries to use it. The template is
provided in box 19.2.

341

Available at: http://unstats.un.org/unsd/dnss/QualityNQAF/nqaf.aspx.

302

Box 19.2 Template for a generic national quality assurance framework (NQAF)
1. Quality context
1a. Circumstances and key issues driving the need for quality management
1b. Benefits and challenges
1c. Relationship to other statistical agency policies, strategies and frameworks and evolution over
time
2. Quality concepts and frameworks
2a. Concepts and terminology
2b. Mapping to existing frameworks
3. Quality assurance guidelines
3a. Managing the statistical system
[NQAF 1] Coordinating the national statistical system
[NQAF 2] Managing relationships with data users and data providers
[NQAF 3] Managing statistical standards
3b. Managing the institutional environment
[NQAF 4] Assuring professional independence
[NQAF 5] Assuring impartiality and objectivity
[NQAF 6] Assuring transparency
[NQAF 7] Assuring statistical confidentiality and security
[NQAF 8] Assuring the quality commitment
[NQAF 9] Assuring adequacy of resources
3c. Managing statistical processes
[NQAF 10] Assuring methodological soundness
[NQAF 11] Assuring cost-effectiveness
[NQAF 12] Assuring soundness of implementation
[NQAF 13] Managing the respondent burden
3d. Managing statistical outputs
[NQAF 14] Assuring relevance
[NQAF 15] Assuring accuracy and reliability
[NQAF 16] Assuring timeliness and punctual
[NQAF 16] Assuring timeliness and punctuality
[NQAF 17] Assuring accessibility and clarity
[NQAF 18] Assuring coherence and comparability
[NQAF 19] Managing metadata

4. Quality assessment and reporting
5. Quality and other management frameworks

303

19.25. Within the NQAF, four components of quality assurance are particularly relevant for statistics on
the international supply of services:
i.
Assuring methodological soundness. This requires the use of the statistical methodologies
based on internationally agreed recommendation contained in MSITS 2010 and good practices
described in this Guide;
ii.
Assuring cost-effectiveness. Cost-effectiveness is assured by the implementation of
standardized solutions that increase effectiveness and efficiency, documentation of the costs of
data production at each stage of statistical process and carrying out the cost–benefit analyses to
determine the appropriate trade-offs in terms of data quality;
iii.
Assuring soundness of implementation. This is achieved by carrying out such activities as
staff selection and conducting training programmes that emphasize the importance of statistics
that are fit for purpose, building in the production process data quality checkpoints and (as
appropriate) sign-offs before proceeding to subsequent stages in the statistical life cycle,
documenting procedures for the design, development, implementation and evaluation of the
statistical compilations, and consulting with stakeholders, especially users and potential
respondents, at all appropriate stages of the statistical life cycle; and
iv.
Managing the respondent burden. The respondent burden is managed by raising
awareness that the requirement to collect information (user needs) should be balanced against
production costs and the burden placed on respondents (supplier costs). Compilers of statistics on
the international supply of services should be proactive in managing the respondent burden and
ensure that there are mechanisms in place to assess the necessity to undertake new statistical
surveys and that care is taken to reduce or distribute response burden. It is important that
compilers inform respondents about: (a) the purpose of the surveys (including the expected uses
and users of the statistics to be produced from the survey), (b) the authority under which the
surveys are taken, (c) the collection registration details, (d) the mandatory or voluntary nature of
the survey, (e) confidentiality protection, and (f) the record linkage plans and the identity of the
parties to any agreements to share the information provided by those respondents. Mechanisms to
maintain good relationships with individual providers of data and to proactively manage the
response burden are essential for improving quality.

19.26. Concerning the outputs produced within the statistical framework for describing the international
supply of services, NQAF lists six groups of activities, which should be carried out:
i. Assuring relevance in the context of the varying needs of users. Relevance can be assured
by consulting users about the content of the statistical work programme, prioritizing between
different users’ needs in the work programme, establishing an advisory council to assist in
setting overall statistical priorities, periodic review of the continuing relevance and costeffectiveness of individual statistical programmes/domains, ensuring good understanding of
the interdependencies between individual statistical programmes/domains and coordination,
harmonisation and full coverage of statistical information produced by the national statistical
system;
ii. Assuring accuracy and reliability of outputs by assessing and validating source data,
intermediate results and final outputs, using proper statistical techniques, comparing the
obtained data with other existing sources of information in order to ensure their validity,

304

clear identification of preliminary and revised data, and provision of the explanations about
the timing and reasons for and the nature of revisions;
iii. Assuring timeliness and punctuality implies establishing clear release policy (taking into
account the need for different dissemination formats and frequency of data); consideration of
the trade-off with other quality dimensions; clear identification of preliminary data so that
users are provided with appropriate information for assessing their quality;
iv. Assuring accessibility and clarity by releasing statistical results with readily accessible
and up-to-date metadata, which cover used concepts and deviations, scope, classifications,
basis of recording, data sources, compilation methods, statistical techniques, etc. to allow for
a better understanding of the data;
v. Assuring coherence and comparability by establishing well defined relationships between
statistics on residents/non-residents transactions in services, FATS and additional indicators
on the international supply of services is very important; also, the relationships between
statistics on the international supply of services and other economic statistics should be well
articulated. Compilers should ensure application of efficient and documented procedures for
combination of data from various sources; provide clear identification and explanation of
breaks in the series and the provision of methods for ensuring necessary data reconciliation;
and
vi. Managing metadata, which includes provision of information covering the underlying
concepts, variables and classifications used, methodology of data collection and processing,
and indications of the quality of the statistical information to enable the user to understand all
of the attributes of the statistics on the international supply of services, including their
limitations, for informed decision-making.
19.27. In the context of point (v) on assuring coherence and comparability, a number of compiling
agencies have created a unit in charge of ensuring coherence of data collection, data quality and
coordination of work across economic statistics concerning data on large multinationals enterprises. The
unit is often referred to as "Large and complex enterprises units" and may have different tasks assigned to
it. In many countries a relatively small number of multinational enterprises accounts for a major part of
total (services) production and trade. For this reason, they are generally included in most surveys carried
out in the area of economic statistics, and consequently they are covered in the work of the large and
complex enterprise unit. Several statistical offices also realized that a more proactive dialogue with
important respondents can improve the large enterprises’ understanding of the statistical data
requirements. This is why in countries where a small number of multinational enterprises account for a
large part of the national production, there are also likely to be important (services) traders. Compilers are
therefore encouraged to identify if there is a need to consider such a unit, and if it already exists to
approach it in order to identify how to collect and compile data in a coherent way.342
C.1.

Evaluating validity of reported data

19.28. It is important that due attention is paid to evaluating validity of reported data. Techniques for
this evaluation include comparing reported values for the current period to those for prior periods in order

342

More information can be found in the Guide to meas
uring Global Production (chapter 6).

305

to ensure consistency, calculating ratios of key items, such as sales per employee or value added to sales,
to ensure reasonable results, and establishing ranges and tolerances to identify outliers for review by
survey staff. Automated checks can be included in an electronic questionnaire or in the survey processing
system to detect unusual or large changes in the data, internal inconsistencies or invalid responses.343 The
questions about the validity of reported data can be subsequently resolved through consultation with
respondents and by other means (e.g., use of relevant financial statements, regulatory filings, and other
surveys). It is also a good practice to to have staff dedicated to evaluating the data reported by specific
respondents. Establishing a relationship between the compiler and the reporter can lead to improved
reporting.
19.29. Validated responses provided by respondents at the individual company (micro) level must be
aggregated to higher (macro) levels for further review and evaluation and ultimately public release. For
universe (benchmark) surveys, validated micro data (including estimated values for non-respondents) can
simply be summed over all units. For sample surveys or other non-benchmark surveys, aggregation of
micro data will depend primarily on how the sample was selected and the associated sample weighting
factors used to develop universe estimates. For a probability sample with weights, individual values are
multiplied by the weighting factor and the weighted values are summed. For cut-off surveys or other
surveys that are not based on probability samples, aggregation may be based on using growth factors to
extrapolate the aggregate value forward from the most recent benchmark year using values from a
matched sample for adjacent years. This method assumes that the growth rates for the firms in the
matched sample are representative of growth for firms that were excluded from the sample because they
fall below the sample cut-off threshold.
C.2.

Country experience: Ireland

19.30. The Central Statistics Office of Ireland (CSO) is the national office responsible for compiling and
publishing official Balance of Payments and related statistics and for implementing the related quality
assurance activities.
19.31. The surveys of manufacturing and non-financial service enterprises undertaken by the CSO are
designed to meet international conceptual and geographical requirements. Coverage is on a sample
selection basis, those surveyed being selected on the basis of statistical register information concerning
transactions with non-residents. Two types of grossing take place in the compilation of the results. For the
enterprises who report annually an estimate is made for non-response. This estimate can vary from the
last value carried forward to more complex trending/forecasting based on similarly sized and type of
enterprises and current market conditions. For non-coverage a separate grossing exercise takes place for
both services and profits. Using other data sources available (e.g. Annual Services Inquiry, Census of
Industrial Production) the enterprises and data are matched and compared and an imputation is made for
non-coverage based on a ratio derived by comparing the size of the overall data to the size of the
unmatched data. This exercise also serves as a useful quality indicator in comparing data collected via
different channels.
19.32. The survey information collected for all types of enterprises covers transactions with nonresidents concerning purchases and sales of services, income flows, transfers, as well as acquisitions and
disposals of foreign assets or liabilities. In order to facilitate compilation of the wider national accounts
statistics, the surveys also collect data on transactions of reporting enterprises with residents of Ireland.

343

See chapter 21 for more information on the use of information and communication technology.

306

This allows for a wide range of edit and plausibility checks; e.g., profits earned in relation to costs and
sales; dividends paid relative to earnings.
19.33. The type of quarterly survey form issued to a company depends upon the type of company. The
survey for manufacturing and non-financial service companies requests data on all resident and nonresident sales/purchases of services (including royalties, copyrights, licenses, etc.), assets and liabilities
(flows, stocks and reconciliation items) as well as related income transactions from companies
incorporated in Ireland and Irish branches of foreign companies. Sectoral and geographic details are also
required.
19.34. A qualified accountant is employed to provide advice on data queries from respondents and on
the various plausibility and edit checks in place. The accountant also examines and compares the data
reported on the BOP forms against the annual statutory accounts, where available. The accountant
accompanies the responsible staff when visiting respondents. The major enterprises are visited at least
once every two years while smaller enterprises are met less frequently. Enterprises are assessed on an ongoing basis for their suitability for quarterly/annual reporting.
19.35. Received data are loaded into the CSO’s relational database where they undergo various
automated consistency and plausibility checks. Further manual checking routines are carried out by
statistical staff. Depending on the materiality of the identified queries the CSO will make contact with the
enterprise concerned in order to obtain corrections/clarifications. Once verified and any further processing
has been completed, the data from all surveys are assembled and the combined data subjected to further
scrutiny and checking, at both a macro and micro level. If problems are detected at this stage the case can
be referred back to the data collection division who can contact the company if necessary to resolve the
issue. Apart from the general checking routines performed by BOP compilers additional checking is done
by a separate data consistency unit existing within the National Accounts Division. This unit carries out
checks on the consistency of monthly, quarterly and annual data returned by individual large
manufacturing and non-financial service companies to various divisions within the CSO (including the
BOP Division). Approximately 70 large companies or groups of companies are reviewed.
19.36. The work of the Consistency Unit has been extended to rationalise the data collection for the 50
largest companies/groups referred to above. The Large Cases Unit (LCU) was set up in 2010 to
coordinate the surveys issued to these enterprises. An important step was to examine the various CSO
data requests and to ensure that any particular item of data was only requested once from the respondent.
A combined survey form for these enterprise was designed which would collect not only the BOP data
but also production, turnover, services, stocks and profits. The LCU is the central point of contact for
these enterprise and works closely with both the enterprises and the various production divisions in the
CSO to ensure quality and consistency of data, while at the same time minimising the burden on the
respondent.
D.

Quality measurement and reporting

19.37. Quality management implies that countries undertake steps to measure quality and report the
results of such measurements to all participants of the statistical process and general public. In this
connection this Guide advises that countries develop a standard for regular quality reports that cover the
full range of statistical processes and their outputs including publication of statistics on residents/nonresidents transactions in services, FATS and additional indicators on the international supply of services,
as applicable. Such reports can be either producer-oriented, with the aim of identifying strengths and
weaknesses of the statistical process and leading to or containing the definition of quality improvement

307

actions, or user-oriented, with the aim of keeping users informed on the methodology of statistical process
and the quality of statistical output.
19.38. Quality reports should be prepared at least every five years, or more frequently if significant
methodological changes or changes in the data sources occur. For monitoring the quality of the processes
and effectiveness of quality-improvement actions, reviews should be conducted more frequently. It is a
good practice that countries base their quality reports on a set of quantitative and qualitative indicators, as
well as on a checklist covering data collection, processing and dissemination, in order to assess the
strengths and weaknesses in the statistical process and to identify possible quality-improvement actions.
19.39. Countries are advised to develop their own quality assessment frameworks on the basis of NQAF
and the quality assessment frameworks developed by international, supranational and regional
organizations.
19.40. Quality measures. Quality measures directly reflect a particular aspect of quality. For example,
the time lag from the end of the reference period to the release of particular data set is a direct quality
measure. However, in practice, quality measures can be difficult or costly to calculate. Instead, quality
indicators can be used in the quality assessment.
19.41. Quality indicators are summarized quantitative or qualitative evidences about the quality of the
data. They are generally defined with respect to some reference point and can assist in making different
types of comparisons. When countries define the quality indicators for their statistics on the international
supply of services, they are encouraged to ensure that the indicators satisfy the following criteria: (a) they
cover all dimensions of quality as defined in section B above; (b) are based on the consistent application
of a sound methodology; and (c) the indicators are easy to interpret both by internal and external users.
19.42. It is a good practice that countries maintain a balance between different dimensions of quality and
the number of indicators. The objective of quality measurement is to have a limited set (minimum
number) of indicators which can be used to measure and to follow over time the quality of statistics
compiled within the framework for describing the international supply of services and to ensure that users
are provided with a useful summary of overall quality, while not overburdening respondents with
demands for unrealistic amounts of quality metadata.
19.43. Suggested quality measures and indicators. Table 19.1 presents a possible set of indicators (and
measures) which countries might wish to consider for measuring the quality of statistics on the
international supply of services. The table is compiled on the basis of the quality measures and indicators
recommended by various international organizations including the IMF, OECD and Eurostat. The
utilization of such measures and indicators provides users with a clear and up-to-date overview of the
overall quality of these statistics.

308

Table 19.1 Suggested indicators for measuring the quality of data compiled within the statistical
framework for describing the international supply of services
Quality
dimension
Relevance

Quality measure and indicator
1. Gaps between key user interests and compiled trade in services statistics in terms
of concepts, coverage and detail;
2. Results of users’ satisfaction surveys and meetings with user groups.
The relevance aspect of quality could also be measured in terms of the availability of
the data to the final users. Data availability is measured through completeness and
confidentiality.
In general, data availability can be calculated as the number of reported data cells
divided by the total number of the data cells required in accordance with the adopted
data compilation methodology344.

Accuracy

Measure of confidentiality can be developed by taking into account the share of
values of the data cells containing confidential data in the total value of all data cells.
1. Differences between two successive releases of the same dataset.
2. Application of reporting thresholds.
3. Under-coverage (% of non-reporting due to thresholds, % of non-reporting due to
non-response);
4. Characteristics and frequency of revisions (e.g., as % of total value);
Mean Absolute Percentage Error (MAPE) may be used for characterising the size of
revisions. MAPE shows the average percentage difference between initial and final
estimates (between 0 and infinity inclusive). The higher is the value, the higher – the
average size of revisions. When the result equals z ero, there are no differences
between the first and final estimates.
MAPE is defined as follows:

X ft  X it
MAPE 
 100
X it
Where,
X it = the initial estimate for characteristic X in reference year t.

X tf = the last available estimate for the same characteristic in reference year t.
5. Use of data validation techniques and their impact.
In the case of sample surveys-based estimates, the accuracy can be measured using
the following indicators:

344

For example, such approach is adopted by the Eurostat. See Annex 1, Table 3 of the EU Commission Regulation
(EU) No 555/2012 of 22 June 2012 amending Regulation (EC) No 184/2005 of the European Parliament and of the
Council on Community statistics concerning balance of payments, international trade in services and foreign direct
investment, as regards the update of data requirements and definitions, http://eurlex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2012:166:0022:0066:EN:PDF. Also see BoP VADEMECUM,
Chapter 5.

309

6. Sampling errors.
The sampling error has to be calculated on the basis of the specific survey design
adopted.
7. Non-sampling errors:
- Unit response rate,
- Item response rate;
Total measurement error (sampling errors plus non-sampling errors) can be
expressed in terms of a coefficient of variation (CV), as follows.

CV 

estimate of the sampling variance
estimated value

A thorough description of the sample design is needed, containing all information
which is likely to affect accuracy and providing background information for the
assessment and interpretation of sampling errors.
Timeliness
Punctuality

Accessibility

Comparability

Coherence

8. Number and average size of revisions of particular datasets.
1. Time lag between the end of the reference period and the date of the first release
(or the release of final results) of data.
Punctuality can be calculated as the actual date of data delivery minus the scheduled
date of publication/transmission. It shows how many calendar days the release of
data was behind (positive value) or ahead (negative value) of the deadline (e.g. as
published in an advance release calendar).
1. Number and types of means used for dissemination of statistics;
2. Degree to which all detailed data sets are made available to users, as a percentage
of total number of data sets produced;
3. Degree to which detailed metadata are dissemina ted.
1 Differences that can be observed when statistics related to the same domain are
compared between geographical areas or over time. This component of quality can
be described by measuring asymmetries. For example The absolute and relative
values of asymmetries could be, calculated using partner country data, at different
levels of detail of service categories. Relative asymmetries could be calculated
giving equal weight to the datasets of the reporter and to the mirror data. Persistent
asymmetries are identified in case the top asymmetries for country A (for a given
service category and flow), are generated by the same partner country B for more
than one year.
1. Use of common concepts, classifications, data sources and methods;
2. Availability of appropriate bridging tables.
This component of quality could be described by measuring the internal consistency
(with respect to the application of integrity rules, coherence between the quarterly
and annual data, size of errors and omissions), and external consistency (e.g.,
coherence between statistics on trade in services between residents and nonresidents and similar statistics belonging to different statistical frameworks).

Integrity

Consistency with integrity rules. The integrity rules should include a set of
equations, which have to be respected in the datasets. For example: the sum of the
components should be equal to the aggregates.
1. Statistics are produced on an impartial basis.
2. Choices of sources and statistical techniques as well as decisions about
dissemination are based solely by statistical considerations.

310

3. The appropriate statistical entity is entitled to comment on erroneous
interpretation and misuse of statistics.
4. The terms and conditions under which statistics are collected, processed, and
disseminated are available to the public.
5. Internal governmental access to statistics prior to their release, if practiced in the
country, is publicly announced.
6. Products of statistical agencies/units are clearly identified as such.
7. Advanced notice is given of major changes in methodology, source data, and
statistical techniques.
8. Guidelines for staff behavior are in place and are well known to the staff.
Methodological
1. Number and degrees of divergences from the relevant international statistical
soundness
standards in concepts and measurement procedures which are used in the collection
and compilation of trade in services statistics (preferably in terms of the amount of
data affected).
* The quality dimensions “clarity/interpretability” and “Credibility” are not included in this table as there are
no identifiable quality measures or indicators for them.

19.44. One recognized good practice in relation to continuous quality improvement is process
measurement. A process is a series of actions or steps towards achieving a particular end; process quality
is an assessment of how far each step meets defined criteria; and process variables are factors that can
vary with each repetition of the process. It is a good practice that producers of official statistics define and
produce a selection of process quality measures to provide an indication of the overall quality of
processes and facilitate continuous quality improvement.
19.45. For official statistics produced from surveys, examples of process quality measures are the
percentage of ineligible sampling units found in the sample, the proportion of proxy interviews by survey,
travel time for interviewers, and scanning/keying error rates.
19.46. When dealing with statistics produced from administrative sources, process quality measures may
be the number of queries from the statistical producer to the administrative data supplier, and the
percentage of data items changed during quality assurance.
19.47. When adopting new or improved methods, quality assurance procedures should be used. These
should include assessing the impact on the statistical series of adopting the new methodology and
subjecting the proposed methodology to peer review. Various mechanisms exist to gain input from
experts into the suitability of new or improved methods. The setting up of specific peer review groups and
collaboration with users, academics and subject matter experts, for example industrialists or
demographers, can also be considered or simply peer review from a statistical colleague.345
19.48. The quality criteria as described in this chapter also apply to statistics on the international supply
of services by mode.

345

UK Statistics Authority, National Statistician’s Guidance, Quality Methods and Harmonisation
http://www.statisticsauthority.gov.uk/national-statistician/ns-reports--reviews-and-guidance/national-statistician-sguidance/].

311

Country experiences
E.1.

Spain - Automatic editing on statistics by modes of supply

19.49. In order to better control the quality of data on Modes of Supply in International Trade in
Services, INE has developed a software tool for recording and fieldwork treatment to facilitate the
automatic editing of modes data in the questionnaire. This software tool ensures that the editing rules
defined by the statistical expert are applied exhaustively and automatically in all used questionnaires. The
questionnaires are either automatically downloaded into the tool (in the case of web questionnaires) or
data are entered manually in the case of hard copy questionnaires.346
19.50. The editing rules in the tool are classified in two types: "strong" or "fatal" edits and "weak" edits.
Fatal edits imply that the questionnaire is not validated and INE staff has to call the respondent to solve it.
Weak edits do allow continuing with the questionnaire recording and subsequent validation processes but
an explanation must be included in the “Observations” field.
19.51. Four editing rules are applicable to modes of supply section in the survey:
i.

a fatal edit is implemented to when a mode has been associated to a non-service item (note: INE
informs respondents that only services items can have a mode of supply associated to them).

ii. a weak edit is implemented when a mode 3 association is (wrongly) made to a service (i.e. the
supply through mode 3 is not possible (by definition) in international transactions data covered in
a balance of payments trade in services survey; such data can rarely be associated to Mode 3,
except for construction, see MSITS 2010 Chapter 5 for more information.
iii. a weak edit is implemented when a mode 2 association is (wrongly) made to services that
theoretically cannot be supplied by this mode according to the guidance provided in MSITS 2010
(note that travel is not covered by the survey; see the MSITS for the non-travel services that could
be supplied by mode 2 (e.g. manufacturing, maintenance and repair)).
iv. a fatal edit is implemented as to admit only mode 1 for supplying the ex-service item of
merchanting.
E.2.

Luxembourg - quality management for compilation of financial services

19.52. Luxembourg is a small, very open economy anchored in foreign trade, thereby generating
significant international transactions and capital flows. Financial services (mainly investment fund
management and international private banking) are one of the mainstays of the national economy.
International trade in financial services, as reported by Luxembourg in its balance of payments, is very
significant. The Banque centrale du Luxembourg (BcL) and the Institut national de la statistique et des
etudes économiques (STATEC) are jointly responsible for establishing the balance of payments of
Luxembourg since January 2002 onwards.

346

According to INE policy, respondents can choose whether to answer questionnaires in hard copy or via the web
(although CAWI is strongly recommended).

312

19.53. Asymmetries with partner countries have always been a major concern for Luxembourg
compilers. Systematic mutual exchange of results and underlying data with partner countries in the past
has led to significant reductions in asymmetries in different areas. However, asymmetry tables produced
by Eurostat identified persistent bilateral asymmetries in the field of financial services between
Luxembourg and selected partners. An in depth analysis showed that the origin of the differences lies in
the treatment of asset management costs taken out of income.
19.54. The investment fund industry in Luxembourg is the largest in Europe and serves mainly nonresidents. Almost all the activity focuses on the external sector. In principle, domestic undertakings for
collective investment have no direct employment, no offices, do not directly charge fees and in most
cases, income earned (on assets) is capitalized (i.e., not distributed to investors). Resident undertakings
for collective investment incur expenses for management and administration, payable in a very large
extent to the resident management companies. A priori, these transactions between residents are outside
of the field of balance of payments statistics.
19.55. Based on a recommendation stemming from the European Monetary Institute and in line with
National Accounts concepts, Luxembourg introduced in its balance of payments time series a procedure
to consider the management fees that are implicitly passed on to the investors.347 The economic substance
is revealed by rerouting. All income (on foreign or domestic assets) earned by the undertakings for
collective investment is assigned simultaneously to the investors, regardless if distributed or not.
Therefore, property income earned by the fund determines the income of its unit holder (investor). The
offsetting of the income assignment to the investor is both the management fees assigned to the nonresident investors (financial services) and the reinvestment of the income.
19.56. The bilateral asymmetries observed have their origin in this specific treatment, partially because
partner countries do not always have the necessary information to perform the corresponding records.
Indeed, it is assumed that many shares of Luxembourg-based undertakings for collective investment are
held by nonresident households whose financial activities are probably not routinely covered by surveys.
19.57. The geographical breakdown comes from annual reporting to the supervisory authority that
requires an estimate of the percentage of shares or units of the undertakings for collective investment
invested in each country. However, uncertainty related to the geographical distribution is permitted; thus,
this uncertainty is a weakness of the procedure.
E.3. European Union – validation of data
19.58. Validation of data is performed by the national data providers as well as by Eurostat. Through
this process the plausibility of the data (e.g. the development of time series, possible outliers) and their
internal consistency (aggregates should match the sum of the sub-items) is verified. The strong link
between the Structural Business Statistics (SBS) and inward FATS also exists when validating the data.
19.59. In the EU countries inward FATS data are usually extracted from the data sources used for the
SBS and the results for both domains should be consistent. Some Member States have registers of
enterprise groups or foreign-owned enterprises that can be used as a basis for identification of the
population of foreign-controlled domestic enterprises and these registers are treated as the main source of
information used to identify the target populations. However, in practice, this consistency is not always
assured. One of the main reasons is that there are sometimes considerable differences between the frame
populations used for the inward FATS data collection and the foreign-controlled subset of Structural
347

BPM6, paragraph 10.124.

313

Business Statistics population. This should not be a problem if correct grossing-up methods are used. In
that case, the total number of all enterprises, including domestic-controlled ones should be equal in both
SBS and inward FATS.

314

Chapter 20 Data and Metadata Dissemination
20.1. Scope. This chapter consists of the following sections: a summary of good practices (section A);
an introduction to data and metadata dissemination concepts (section B); factors to consider in data and
metadata dissemination (section C); the presentation of statistics by mode of supply (section D); and
combined presentation of international merchandise and trade in services statistics (section E).
A.

Summary of good practices

20.2.
It is a good practice that countries continue their efforts in the implementation of the UN
Fundamental Principles of Official Statistics, according to which the dissemination of data and metadata
compiled within the statistical framework for describing the international supply of services should be
carried out with great care and attention to the needs of users while, at the same time, ensuring adequate
confidentiality of data providers. In particular, compilers should ensure the equal treatment of all users,
adequate user access, provision of advanced release schedules and pay due attention to metadata
dissemination. It is a good practice to discuss and agree with major user groups on the adequate balance
between timeliness versus reliability and accuracy. It is also a good practice to develop a revision policy
that is well-designed, transparent, well-publicized, and coordinated with other areas of statistics.
20.3.
More specifically, it is advised that compilers ensure consistency between disseminated datasets
covering services transactions between residents and non-residents, FATS and additional indicators on the
international supply of services, as well as they coordinate dissemination of trade in services statistics
with dissemination of balance of payments statistics. It is further advised to disseminate preliminary
estimates of available or priority aggregates on services transactions between residents and non-residents
monthly, with the main aggregates (by main EBOPS 2010 categories and by main trading partners) being
disseminated on a quarterly basis and the full detailed data sets on an annual basis. In particular, it is a
good practice that the value data on services transactions between residents and non-residents are broken
down by: (1) direction of trade flows (exports and imports); (2) EBOPS 2010, by 12 first level categories
as a minimum but preferably by all relevant sub-categories and (3) trading partners, (4) enterprise
characteristics (if available) and (5) modes of supply, if such data are compiled or if an allocation is
performed following the recommended procedures contained in Chapter 14 of this Guide.
20.4. For foreign affiliate statistics (FATS), the selection of variables which should be disseminated
depends on the quality assessment of the compiled variables and the information needs of the country.
However, efforts should be made to disseminate the variables identified in MSITS 2010 and elaborated in
Chapter 15 of this Guide. It is advised that dissemination the following annual data are treated as a matter
of priority: (i) sales (turnover) and/or output; (ii) employment; and (iii) number of enterprises. It is
advised that the data be broken down by: (1) direction of investment (outward and inward FATS); (2)
activity (and if possible, for output or sales(turnover) by product; or at a minimum, a disaggregation into
total goods and total services for each activity); and (3) partner (economy of affiliate operations for
outward and economy of UCI for inward). Moreover, if possible, data on sales(turnover)/output that are
rendered to residents of the economy of the affiliate should be separately identified.
20.5.
It is a good practice to disseminate statistics on service transactions between residents and
non-residents, FATS sales(turnover)/output data and selected additional indicators on the international
supply of services in a common report regularly (e.g., on an annual basis), presenting the international
supply of services broken down by mode, if possible. For non-monetary indicators, in particular on the
movements and stocks of persons covered modes 2 and 4, as a minimum requirement, countries are

315

advised to disseminate the full detailed data sets on an annual basis, using the breakdowns as
suggested in chapter 16. It would also be desirable to disseminate main aggregates on a quarterly basis.
20.6. In terms of maintaining statistical confidentiality, compilers should strive for a full coverage
of all flows and stock data which are in the scope of trade in services statistics, while applying
appropriate methods to keep, as relevant, certain information confidential.
20.7. This Guide stresses that provision of micro-data (or data about a data provider, including an
individual person, household, business or other entity) is consistent with legal and other necessary
arrangements that ensure that confidentiality is protected and that procedures for access to micro-data
are transparent and publicly available.
20.8. It is a good practice to adopt several formats and means of dissemination to ensure that data
and metadata are delivered efficiently to all user groups. In this connection it is advised to make most
use of electronic databases: (a) to allow free and equal access to all users to any data record considered
part of official trade in services statistics; (b) to store and deliver to users extensive metadata and
knowledge base; and (c) to allow users to make queries easily and with a user-friendly interface on the
entire database, and to download query results in the commonly used electronic data formats. It is also
recommended that statistical agencies consider the application of Statistical Data and Metadata
eXchange (SDMX) standards in the dissemination of data and its accompanying metadata in order to
ensure consistency and coherence.
B.

Data and metadata dissemination: an introduction

20.9. The availability of official statistics, including statistics compiled within the framework for
describing the international supply of services, is one of the cornerstones of public confidence in good
governance as such statistics can inform the debate and decision-making by governments and by the
wider community. This Guide highlights the importance of countries’ adherence to the UN
Fundamental Principles of Official Statistics which, inter alia, state that:348

i.
Official statistics “provide an indispensable element in the information system of a
democratic society, serving the Government, the economy and the public”;
ii.
These statistics should be made “available on an impartial basis by official statistical
agencies to honour citizens' entitlement to public information”;
iii.
The statistical agencies should “facilitate a correct interpretation of the data”, and,
therefore, have “to present information according to scientific standards on the sources,
methods and procedures of the statistics”;
iv.

The statistical agencies “are entitled to comment on erroneous interpretation and
misuse of statistics”.
20.10. In the light of these Principles, the dissemination of data and metadata should be carried out
with great care and attention to the needs of users while, at the same time, ensuring adequate

348

The Principles are available at http://unstats.un.org/unsd/dnss/gp/fundprinciples.aspx.

316

confidentiality of data providers. Dissemination enhances accessibility of statistical information and
constitutes an indispensable building block of the integrated statistical production process.349
20.11. Below is a brief description of several good practices which countries are advised to follow in
setting up their dissemination policy:
i.

Users should be treated equally, whether national or international;

ii.

Data should be made available to all users at the same predetermined time;

iii.

Adequate user access to data and metadata (including information on quality of data350 and
methodologies) should be ensured, in terms of making information publicly available in a
clear and easily understandable manner and in adequate forms of dissemination; making
statistics available on an impartial and timely basis; and providing prompt and knowledgeable
support services to users (e.g., “Frequently-Asked Questions”, contact information for
questions, and technical help). For a major statistical release, it is often helpful for the
statistical agency to organize a press briefing event. The highlights could be used to convey
significant findings, comparisons, and trends to assist the media, and other users, in
understanding and using the publications. Where feasible, special data services could be
provided including special or non-standard groupings of data items or outputs, and their
usefulness and their costs;

iv.

An advanced release schedule should be published. It is essential that the dates in the schedule
are met.

20.12. Importance of metadata dissemination.351 The statistical agencies responsible for statistics on
the international supply of services must ensure that users are able to access and correctly interpret the
information on statistical methods, concepts, variables and classifications used in producing statistical
results. Additional guidance on dissemination of metadata is provided below:
i.

Sufficient metadata should be made available to enable both the least and the most
sophisticated users to readily access data and understand their quality. It is a good practice to
structure metadata in layers of incremental detail.352

ii. Structural metadata should be presented as an integral part of the database and be published as
part of statistical tables (e.g., in the form of flags or footnotes identifying differences in
definitions, estimations and imputations, provisional values, confidentiality, break in the time
series, etc.). by default, unless explicitly removed by the user. Reference metadata can be
presented as a detailed explanatory note describing the scope, coverage, and quality of a
dataset and can be made available electronically alongside the database or in special
publications. In addition, compilers should make every effort to ensure that users have ready
access to metadata through multiple dissemination channels, both in printed and in electronic
format (whereby Internet dissemination plays a key role), and be free of charge (regardless of
whether the statistics they describe are disseminated for a fee according to the compiling

349

Guidelines on Integrated Economic Statistics, ST/ESA/STAT/SER.F/108.
http://unstats.un.org/unsd/nationalaccount/docs/IES-Guidelines-e.pdf.
350
See chapter 19 for more information on quality reporting.
351
Also see chapter 18.
352
See OECD, Data and metadata reporting and presentation handbook, 2007, page 22.

317

organization’s policies.353 Any deviations from international standards or discrepancies (e.g.,
trade in services, balance of payments, tourism statistics, FATS, etc.) should be adequately
explained to the user. Whenever feasible, it is a good practice that compilers disseminate
metadata using standardized concepts that are relevant across statistical domains (e.g., by
adopting cross-domain concepts from the SDMX framework, Annex 4).
C.

Factors to consider in dissemination of data and metadata compiled within
the framework for describing the international supply of services
C.1.

Variables to be disseminated

20.13. MSITS 2010 contains recommendations with respect to the kinds of data variables that should
be considered for dissemination. Taking into account the elaboration of the data and metadata
variables provided in this Guide, countries are encouraged to organize their dissemination policy as
follows:
i.

Statistics on value of trade in services between residents and non-residents. It is recommended
that the full detailed annual value data be broken down by: (1) direction of trade flows
(exports and imports); (2) EBOPS 2010, by 12 first level categories as a minimum but
preferably by all relevant sub-categories; (3) trading partners, (4) relation between the parties
and (5) modes of supply if such data are compiled or if an allocation is performed following
the recommended procedures contained in Chapter 14 of this Guide. It is good practice to
disseminate on a quarterly basis the main aggregates of such data (in particular for exports and
imports, by main EBOPS 2010 categories and by main trading partners). It is also a good
practice to disseminate preliminary estimates of available or priority aggregates monthly.
Compilers should ensure coherence between disseminated datasets and should closely
coordinate dissemination of the statistics with dissemination of balance of payments statistics.

ii. For foreign affiliate statistics (FATS), the selection of variables which should be disseminated
depends on the quality assessment of the compiled variables and the information needs of the
country. However, efforts should be made to disseminate the basic variables identified in
MSITS 2010 and elaborated in Chapter 15 of this Guide namely: sales (turnover) and/or
output; employment, value added, exports and imports of goods and services, and number of
enterprises. As a priority, given their relative ease in producing, and the primary interest of
users, data should be disseminated for (i) sales (turnover) and/or output; (ii) employment, and
(iii) number of enterprises. It is recommended that the data be disseminated broken down by:
(1) direction of investment (outward and inward FATS); (2) activity according to ICFA Rev. 1
or on a compatible basis such as ISIC Rev. 4; (3) if possible for output or sales by product,
using a classification system compatible with EBOPS 2010, but if not achievable as a
minimum compiling total goods and total services for each activity; (4) partner (economy of
affiliate operations for outward, UCI for inward). For sales/output, if possible, those services
rendered to residents of the economy of establishment of the affiliate should be
identified. .FATS statistics should ideally be disseminated at least on an annual basis for those
variables that are deemed to be the most important (i.e. priority variables). The corresponding
structural and reference metadata items should also be made available (see Chapter 18).

353

See OECD, Data and metadata reporting and presentation handbook, 2007, page 22.

318

iii. Countries are further advised to disseminate resident/non-resident trade in services and FATS
sales/output data in a common report.
iv. Non-monetary indicators, in particular on the supply of services by modes 2 and 4include data
on number of persons crossing borders (or trips) broken down by several classification criteria
appropriate for such modes. For countries that have not started compiling trade in services by
modes, a possible alternative would be to perform a “conceptual” allocation as presented in
chapter 14 C. It is a good practice to concentrate, at a minimum, on disseminating statistics on
an annual basis, using the breakdowns as suggested in chapter 16. However, given the strong
link of this information with trade/labour mobility policy, it could be of interest to also
envisage the compilation of some information for shorter periods, for instance on a quarterly
basis, at least for main aggregates. Countries are also advised to pay special attention to the
dissemination of the relevant structural and reference metadata in order to ensure the correct
interpretation of the data. For the other statistics described in chapter 16 section D, given that
these are compiled by statisticians in other statistical domains, no particular dissemination
practices are detailed in this guide other than the general principles described above (see also
section 20.D).
C.2.

Timeliness

20.14. Timeliness of the dissemination is one of the recommended quality dimensions and should be
fully taken into consideration when developing the release schedule. The delays in time between data
collection and publishing should be monitored, as well as how the release of these data fit with the
release of related data, such as BOP and National accounts354 .
20.15. When disseminating data, there is usually a trade-off between the timeliness, on the one hand,
and reliability, accuracy and level of detail of published data on the other hand. To help making
relevant decisions compilers should consider, user requirements (e.g.,. discussing trade-offs with major
user groups and making decisions public) and the timing of the collection of initial and revised data
from various sources.
20.16. Compilers are advised to publicly post (e.g., on the website of the national agency responsible
for the dissemination of the official trade statistics) before the beginning of each year an advance
release calendar of data with precise dates at which those will be released and possibly revised . It is
also a good practice to inform users about availability of such calendars using all appropriate means of
communication. In addition, to improve timeliness in the dissemination of statistics on the
international supply of services, it is a good practice to publish on a regular basis the provisional
estimates of totals, as well as of breakdowns by major service categories and main partners (see
section C.1 for more details), soon after the end of the reference period. Such estimates, by their nature,
would be based on relatively limited data content and are to be replaced by more accurate, but less
timely figures at a later date. Users must be aware of the trade-off between quality (size of revisions)
and timeliness (e.g., it is generally not a good practice to publish frequently large revisions); and
quality aspects need to be taken into account when deciding on the frequency of publication.
20.17. Compilers are encouraged to issue, as relevant, the first releases of resident/non-resident trade
in services data as follows: (a) monthly totals within 45 days after the end of the reference month, at
354

See BPM6 Compilation Guide, paragraph 17.4.

319

least by main categories of services (and by major trading partners, if possible); (b) quarterly data
within one quarter after the reference period; and (c) annual data generally within six to nine months
after the reference period. The release of certain data sets such as annual FATS can be extended.
20.18. In addition, compilers are advised that the data for the fourth quarter (or for the twelfth month)
need to be compiled and disseminated in their own right and should not be derived as the difference
between the annual totals and the sum for the first three quarters (or eleven months) in order to provide
undistorted data for all months and quarters.
20.19. Regarding coherence between monthly, quarterly and annual data, countries should ensure that
additivity is always present, possibly by systematic automated checks. In the case of non-additivity,
compilers should make available the reasons underlying it in the metadata or explanatory notes.
C.3.

Statistical confidentiality

20.20. Statistical confidentiality refers to the protection of information of individual statistical units and
has to be differentiated from other forms of confidentiality under which information is not disseminated
due to other considerations, for example due to national security concerns. It is a good practice to always
strive for a full coverage of all data which are in the scope of statistics on the international supply of
services, while applying appropriate methods to keep certain information confidential. This Guide
recognizes, however, the necessity of both statistical confidentiality and of balancing it against the need
for public information in cases where the application of statistical confidentiality would limit or make it
impossible to provide sufficient or meaningful information. It is also good practice to disseminate a
quantitative indicator of the amount of data subject to confidentiality.
20.21. The implementation of recommendations on statistical confidentiality depends to a large extent
on each country’s legislation and the general confidentiality policy adopted by its statistical system. The
assurance of confidentiality is essential to securing the cooperation of reporters and to maintaining the
integrity of the statistical system. Confidentiality concerns may also be a more serious issue in smaller or
less developed countries where there are fewer entities likely to be engaged in in the international supply
of services. An important challenge in the implementation of confidentiality rules is to ensure that
confidentiality is applied across all the different classifications in which data are disseminated.
20.22. Active confidentiality is advised, which implies that the statistical agency should take the
initiative to actively suppress or aggregate data the dissemination of which would enable the identification
of any individual person or entity.355
20.23. The Eurostat Handbook on Statistical Disclosure provides detailed examples and approaches to
applying sensitivity rules when applying active confidentiality and is recommended for further reading.356

355

It may be interesting to note that in the domain of international merchandise trade statistics, “passive
confidentiality” is recommended (i.e., data are tre ated as confidential only when the trader requests so on the
grounds that his or her interests would be harmed by the dissemination of their data and the statistical authority finds
the request justified based on the confidentiality rules) , unless the use of active confidentiality is already the
established, desired and accepted practice.IMTS 2010, paragraph 10.3. https://unstats.un.org/unsd/trade/egimts/IMTS%202010%20(English).pdf. IMTS 2010 Compilers Guide, paragraph 1.14.
http://unstats.un.org/unsd/trade/EG-IMTS/IMTS2010-CM%20-%20white%20cover%20version.pdf 356 Handbook on
Statistical Disclosure, Eurostat, 2007. http://unstats.un.org/unsd/EconStatKB/Attachment474.aspx.
356
Handbook on Statistical Disclosure, Eurostat, 2007. http://unstats.un.org/unsd/EconStatKB/Attachment474.aspx.

320

As shown in table 20.1, common sensitivity rules used are: the minimum frequency rule (data suppressed
for cells that have less than a pre-determined number of respondents reporting, typically three357 ); or the
dominance rule in which the preponderance of the value (the exact share of which to be pre-determined
by the compiler) is derived from the top one to three entities). It is important to keep the exact pre-defined
parameters confidential. In some cases, the contributor with the second largest contribution to a cell
which is non-sensitive according to the sensitivity rules cited above is able to derive a close upper
estimate for the contribution of the largest one by subtracting his or her own contribution from the
aggregate total.358 In such cases, the dominance rule would need to be adapted (see the p% rule in table
20.1).
20.24. If one data cell is made confidential, it is typically advisable to also suppress the next available
cell with the smallest value so that the value of the primary confidential cell cannot be calculated.
Table 20.1 Sensitivity rules359
Rule
Minimum frequency rule
(n,k-dominance rule)
p%-rule

Definition
A cell is consider sensitive when:
The cell frequency is less than a pre-specified minimum frequency of n
respondents
The sum of the n largest contributions exceeds k% of the cell total, e.g.,
x1+…=xn>k/100•X
The cell total minus the 2 largest contributions x1 and x2 is less than p% of
the largest contribution; e.g., X-x 2-x1
Source Exif Data:
File Type : PDF
File Type Extension : pdf
MIME Type : application/pdf
PDF Version : 1.5
Linearized : No
Page Count : 349
Tagged PDF : Yes
XMP Toolkit : Adobe XMP Core 4.0-c316 44.253921, Sun Oct 01 2006 17:14:39
Creator Tool : Outside In (C) Technology
Producer : PDF Export 8.5.3.70.1803290001

EXIF Metadata provided by EXIF.tools

Application Of Manual Statistics

Navigation menu

Versions of this User Manual:

Views

Navigation