ST/ESA/STAT/SER.

F/74

Department for Economic and Social Affairs

Statistics Division

Studies in Methods Series F, No. 74

Handbook of National Accounting

HANDBOOK OF
INPUT-OUTPUT TABLE
COMPILATION AND ANALYSIS

United Nations
New York, 1999
 NOTE
Symbols of United Nations documents are composed of capital letters combined with figures.

The designations employed and the presentation of material in this publication do not imply the
expression of any opinion whatsoever on the part of the Secretariat of the United Nations concerning the legal
status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers
or boundaries.

Where the designation “country or area” appears, it covers countries, territories or areas.

Litho in United Nations, New York

24174—October 1999—2,860

ST/ESA/STAT/SER.F/74
United Nations publication
Sales No. [Link].4
ISBN 92-1-161416-3
Copyright © United Nations 1999
All rights reserved. No part of this publication may be reproduced, stored in a public retrieval system, or
transmitted, in any form or by any means – electronic, mechanical, photocopying, recording or otherwise –
without the prior permission of the United Nations.

ii
 Preface

The Statistical Commission of the United Nations has been concerned with conceptual or practical
difficulties in the implementation of the System of National Accounts (SNA) since it was introduced in 1968.
At its 1983 session, the Commission agreed that the SNA should be revised with the specific objectives of
clarifying it and harmonizing it with related statistical systems rather than modifying it. Since 1986, many
expert group meetings have been convened by the United Nations jointly with other international organizations
including the International Monetary Fund (IMF), the Organization for Economic Cooperation and
Development (OECD), the World Bank and the Statistical Office of the European Communities
(EUROSTAT), to make specific recommendations to revise the 1968 version of the System of National
Accounts, Studies in Methods (ST/STAT/SER.F/2/[Link] No. [Link].3). Their recommendations, a
major advance in national accounting, have been incorporated into the System of National Accounts 1993
(ST/ESA/STAT/SER.F/2/[Link] [Link].4). Adoption of the 1993 SNA was unanimously
recommended to the United Nations Economic and Social Council by its Statistical Commission at its twenty-
seventh session, held in New York from 22 February to 3 March 1993. The Council recommends that Member
States consider using the 1993 SNA as the international standard for the compilation of national accounts
statistics to promote the integration of economic and related statistics, as an analytical tool, and in the
international reporting of comparable national accounting data.

This Handbook is issued as part of a series being developed by the member organizations of the Inter-
Secretariat Working Group on National Accounts (ISWGNA)1. Their general objective is to support the
implementation of the 1993 SNA through further elaboration of concepts with regard to specialized areas of
national accounting, discuss existing and new compilation methodologies to measure the concepts of the
specific field dealt with in each Handbook, and finally present illustrative examples of analytical and policy
uses of the concepts. The relative extent to which each of the three elements is dealt with differs between the
Handbooks: Some focus more on conceptual issues, others detail compilation methodologies, while a third
group may focus on particular analytical and policy uses.

This Handbook is more than a revision of the United Nations publication entitled Input-output Tables
and Analysis (ST/STAT/SER./14/Rev.1, Sales No. [Link].11), which was published by the United Nations
in 1973. It was almost entirely rewritten by Vu Quang Viet, a staff member of the United Nations Statistics
Division (UNSD), in the Department of Economic and Social Affairs (DESA), under the general guidance of
Jan W. van Tongeren, taking into account more recent developments. A draft of this Handbook had been sent
to international organizations and many I/O experts for comments. The draft was then revised taking into
account these comments as far as possible. Valuable comments were received from Erling Flottum of Statistics
Norway; Jean-Etienne Chapron of Institut national de la statistique et des études economiques (INSEE),
France; S.J. Keuning of Statistics Netherlands; Norbert Rainer, Austrian Central Statistical Office; Thijs ten
Raa of Tilburg University, the Netherlands; Felizardo Suzara of UNSD and Willem van den Andel. The
Handbook deals with an area of concepts, compilation and analysis which has been developed in great detail
by many economists, who generally were not national accountants. However, it is not the objective of the

1
To date the following have been published: Balance of Payments Manual (IMF, 1993); Balance of Payments
Compilation Guide (IMF, 1994); Handbook on Integrated Environmental and Economic Accounting
(ST/STAT/SER.F/61. Sales [Link].12); Handbook on Inflation Accounting (OECD, 1996); Handbook on System
of National Accounts for Transition Economies and (ST/STAT/SER.F/66. Sales [Link].11); A System of
Economic Accounts for Food and Agriculture (FAO, 1996).

iii
Handbook to give an overview of all these developments. Instead, its focus is on the much more recent
conceptual and analytical integration of input-output tables within the national accounting system, which was
largely accomplished in the 1968 SNA, but was further developed and refined in the 1993 SNA.

By integrating input-output tables and analysis with national accounting, an additional use of input-
output in national accounts compilation is added to the many analytical uses that have been developed since
Leontief created the analytical I/O model. This is the statistical use of so-called supply and use tables, a special
non-symmetric format of I/O tables first developed in the 1968 SNA, which is used to check the internal
consistency of data on production and products.

Part One of the Handbook is devoted to the analytical foundation of I/O analysis, the SNA accounting
and statistical foundation of I/O tables, and the methods of converting the supply and use tables into a single
symmetric I/O table. In this context, a bridge is made between the terminology used by input-output specialists
and the terminology developed in the 1993 SNA.

Part Two focuses on the compilation aspects of the supply and use tables with further elaborations
on the SNA concepts. Included here are compilation of production accounts (output, intermediate
consumption, value added), of final demand and imports, table balancing and updating. This part tries to be
as comprehensive as possible in presenting SNA concepts within the context of compiling the supply and use
tables so as to lessen the need by compilers to go back and forth checking with the 1993 SNA. However, this
part does not enter into the data resources in much detail; for that aspect the reader is referred to a handbook
of national accounting issued by the United Nations in 1986.2

Part Three presents a few applications of input-output tables and models dealing with price deflation
and annual national accounts estimates, impact analysis and pollution study. The scope of this part of the
Handbook is obviously not complete in respect of the extensive literature in this field. It is generally restricted
to the use of input-output in dealing with basic economic issues and does not enter into much detail on I/O
topics such as productivity and price analysis, analysis of energy consumption, use of I/O in economic
projections and forecasting, regional, interregional and international I/O tables or their use in production and
trade modelling, dynamic input-output analysis, etc. These types of studies normally require highly advanced
mathematical techniques that cannot easily be covered in depth in this Handbook. For information on these
advanced uses of I/O techniques the reader is referred to the numerous textbooks, books of readings on input-
output economics and numerous articles published in economic journals.3

The Handbook includes numerous examples which clarify many difficult concepts and methods of
economic compilation. Readers are advised to go through the examples in order to obtain a better
understanding of these concepts and methods. Furthermore, since economic analysis using input-output tables
involves the manipulation of large matrices, readers are encouraged to work through these examples using a

2
Handbook of National Accounting, Accounting for Production: Sources and Methods, (ST/STAT/SER.F/39. Sales
[Link].11), is outdated in concepts but it contains valuable discussions on data sources and estimation techniques.

3
To look for references of articles published in economic journals and books, readers should consult the Journal of
Economic Literature, classification 222, published by the American Economic Association. Current economic research
using I/O, besides other economic journals, may be found in Economic Systems Research, published by the International
Input-Output Association.

iv
computer with the help of APL, LOTUS and EXCEL. The two latter softwares can easily manipulate matrices
with up to 60 sectors.

Main Abbreviations
C.i.f./f.o.b. Cost, insurance and freight / free on board
CCIS Cross-classification between Industries and Sectors
COFOG Classification of the Function of Government
COICOP Classification of Individual Consumption by Purpose
COPNI Classification of the Purposes of Non-profit Institutions serving Households
COPP Classification of Outlays of Producers According to Purpose (COPP)
CPC Central Product Classification
FISIM Financial Intermediation Services Indirectly Measured
GDP Gross Domestic Product
I/O Input-Output
ISIC International Standard Classification of All Economic Activities
NPISHs Non-profit Institutions serving Households
PIM Perpetual Inventory Method
R&D Research and Development
SAM Social Accounting Matrix
SIOT Symmetric Input-Output Table
SNA System of National Accounts
SUT Supply and Use Tables
VAT Value Added Taxes

v
CONTENTS

PART ONE:
ANALYTICAL AND STATISTICAL FOUNDATION OF INPUT-OUTPUT MODEL

CHAPTER I
BASIC INPUT-OUTPUT SYMMETRIC MODEL (SIOT)

A. Historical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

B. A simple input-output framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

C. The inverse matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1. Solution of an input-output model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2. Economic interpretation of the inverse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

D. Measurement unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

E. Prices and costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1. Price as costs of production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2. Price analysis in input-output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3. Some rules in price analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Appendix: Mathematical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

CHAPTER II
THE SYSTEM OF NATIONAL ACCOUNTS (SNA) FRAMEWORK OF SUPPLY AND USE TABLES
(SUT): OVERVIEW

A. Basic concepts used in (SUT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

B. Structure of SUT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1. The supply table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
The c.i.f./f.o.b. adjustment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2. The use table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
(a) The adjustment for trade and transport margins . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
(b) The adjustment for purchases of residents abroad . . . . . . . . . . . . . . . . . . . . . . . . . . 33
(c) The adjustment for purchases of non-residents at home . . . . . . . . . . . . . . . . . . . . . 33
3. The advantage of using basic prices in SUT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4. Balances between value added and final demand in the use table . . . . . . . . . . . . . . . . . . . . . 34
(a) GDP as sum of value added (the production approach) . . . . . . . . . . . . . . . . . . . . . . 34
(b) GDP as sum of primary incomes (the income approach) . . . . . . . . . . . . . . . . . . . . . 35
(c) GDP as net final demand (the expenditure approach) . . . . . . . . . . . . . . . . . . . . . . . 36
(d) The commodity flow approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

vii
 5. Final consumption expenditure versus actual final consumption . . . . . . . . . . . . . . . . . . . . . . 37

C. Link between SUT and production accounts of institutional sectors in the SNA . . . . . . . . . . . . . . . . 38

D. The statistical units and classification schemes for data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

1. Statistical units and classification scheme for an activity or industry . . . . . . . . . . . . . . . . . . . 41
2. Statistical units and classification scheme for products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3. Correspondences between ISIC and CPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4. Statistical units for institutional sectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Appendix A: Industry classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Appendix B: Central product classification, Version 1.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Appendix C: Classification of the functions of Government (COFOG) . . . . . . . . . . . . . . . . . . . . . . . 51

CHAPTER III
VALUATION IN I/O TABLE

A. Concepts of valuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

B. Valuation and the tax system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

1. Non-VAT taxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2. Value added taxes (VAT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

C. The derivation of the supply and use tables (SUT) at basic prices, VAT and non-VAT . . . . . . . . . . 58

D. The derivation of SUT at producers’ prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

E. The supply and use tables with VAT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

1. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2. Producers’ prices, gross and net treatment of VAT and basic prices . . . . . . . . . . . . . . . . . . . 62

CHAPTER IV
CONVERTING SUPPLY AND USE TABLES INTO A SYMMETRIC I/O TABLE: TREATMENT OF
SECONDARY PRODUCTS

A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

B. Types of secondary products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

1. Secondary products resulting from statistical practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
2. Secondary products resulting from production technology . . . . . . . . . . . . . . . . . . . . . . . . . . 77

C. Partitioning enterprises into units of homogenous production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

1. General SNA rules for creating a separate establishment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
2. Separation of own-account construction and vertically integrated activities . . . . . . . . . . . . . 79

viii
 3. Cost allocation of ancillary activities to establishments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4. Ancillary activities with sales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5. The redefinition method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

D. Treatment of secondary products resulting from production technology . . . . . . . . . . . . . . . . . . . . . . 82

1. The negative transfer method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
2. The aggregation or positive transfer method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3. Transfer of outputs and inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

E. Transferring secondary outputs and associated inputs by mathematical methods . . . . . . . . . . . . . . . . 86

1. General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
2. The industry technology assumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3. The commodity technology assumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4. Evaluation of the two assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

Appendix A: Desirable properties of symmetric input-output coefficient matrices . . . . . . . . . . . . . . . . 100

PART TWO:
COMPILATION OF SNA SUPPLY AND USE TABLES

CHAPTER V
COMPILATION OF PRODUCTION ACCOUNTS OF INDUSTRIES

A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
B. General remarks on compilation of production accounts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
1. Outputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
2. Intermediate consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3. Value added at basic prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4. Statistical sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

C. SNA production concepts reviewed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

1. Production boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
2. Scope of output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
3. Output and input valuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
4. Work-in-progress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5. The treatment of second-hand goods and old scraps in input-output table . . . . . . . . . . . . . . 121
6. Operating leasing versus financial leasing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
7. Treatment of entertainment, literary and artistic originals and copyrights . . . . . . . . . . . . . . 124

D. Allocation of production activities of institutional units to establishments . . . . . . . . . . . . . . . . . . . . 125

1. Production activities of corporate and quasi-corporate enterprises . . . . . . . . . . . . . . . . . . . . 125
2. Production of government units and non-profit institutions serving households . . . . . . . . . 126
3. Production activities of unincorporated enterprises of households . . . . . . . . . . . . . . . . . . . . 129

E. Compilation of production accounts of selected industries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

ix
 1. Agriculture, forestry and fishery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
2. Wholesalers and retailers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
3. Transport services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
4. Owner-occupied housing services and government-subsidized housing . . . . . . . . . . . . . . . 134
5. The output of financial intermediaries and its treatment in the input-output framework . . . 134
6. Production accounts of insurance enterprises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
(a) Life insurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
(b) Non-life insurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
7. Production accounts of privately funded pension funds . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
8. Production accounts of research and development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
9. Production accounts of mineral exploration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
10. Production accounts of producers of government services and of non-profit organizations
serving households (other non-market output) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

Appendix A: Valuation of inventories in output calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

Appendix B: Measuring output of financial intermediaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

CHAPTER VI
TREATMENT OF IMPORTS

A. Compilation of imports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

B. Treatment of imports in the use table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155

1. Valuation of imports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
2. Types of imports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
(a) Non-competitive imports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
(b) Competitive imports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
3. Alternative presentation of imports in the use table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158

C. Analysis of imports in an input-output model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160

CHAPTER VII
COMPILATION OF FINAL DEMAND

A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

B. Final consumption expenditures and actual final consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167

1. Government final consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
2. Final consumption expenditures of NPISHs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
3. Household final consumption expenditures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Estimation and sources of final consumption expenditures . . . . . . . . . . . . . . . . . . . . . . . . 173
4. Final consumption expenditures versus actual final consumption . . . . . . . . . . . . . . . . . . . . 174

C. Gross capital formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

1. Gross fixed capital formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
2. Changes in inventories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

x
D. Exports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179

CHAPTER VIII
COMMODITY FLOW METHOD AND TABLE BALANCING

A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

B. Table construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

1. Industry input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
2. Commodity uses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
3. Sales ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
4. Flow and coefficient adjustment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186

C. Construction of margins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187

1. Construction of import margins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
2. Construction of the trade margin matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
3. Construction of the transport margin matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

D. Final balancing of the tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

CHAPTER IX
UPDATING INPUT-OUTPUT TABLES: RAS METHODS

A. Stability of I/O coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

B. The mathematics of the RAS method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

C. The iterative solution of the simple RAS method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

D. The modified RAS method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

E. Evaluation of the simple and modified RAS methods, based on a literature survey . . . . . . . . . . . . . 202

F. Conclusions for application of the RAS methods in country practices . . . . . . . . . . . . . . . . . . . . . . . 204

G. Other alternative methods for updating coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206

xi
 PART THREE:
SOME APPLICATIONS OF INPUT-OUTPUT TABLES AND MODEL

CHAPTER X
CREATING AN INDUSTRY NOT IDENTIFIED FROM ISIC CLASSIFICATION: ANALYSIS OF THE
INTERNATIONAL TOURIST INDUSTRY

A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211

B. Definition and classification of the international tourist industry . . . . . . . . . . . . . . . . . . . . . . . . . . . 211

1. Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
2. Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212

C. Modelling the international tourist industry in I/O table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215

1. Creating the international tourist industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
2. Distortion created by aggregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
3. Impact analysis of the tourist industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218

D. A general aggregation and disaggregation scheme for the creation of a new industry
not identified by ISIC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
1. Disaggregation scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
2. Aggregation scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

CHAPTER XI
INPUT-OUTPUT TABLES AND PRODUCTION ACCOUNTS IN CONSTANT PRICES

A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227

B. Double deflation method for benchmark years . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227

1. Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
2. Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
(a) Preparation of the use table in basic prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
(b) Deflation by types of products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
(c) Choice of index number formulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
3. Some important remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
4. Alternative procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239

C. Double deflation method for annual national accounts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239

1. Estimation of main aggregates in current prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
2. Estimation of the supply table in current prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
3. Estimation of the intermediate consumption in current prices . . . . . . . . . . . . . . . . . . . . . . . 240
4. Estimation of the use table in constant prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
5. Deflation process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
6. Special problem in the annual method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
7. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241

xii
CHAPTER XII
IMPACT ANALYSIS

A. Full-fledged impact analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243

1. Basic impact equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
2. Preparation of the initial impact vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
3. Calculation of total impacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
4. Endogenization of household income effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248

B. Other types of impact analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250

1. Multipliers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
2. Backward and forward linkages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251

C. The System of National Accounts (SNA) and Social Accounting Matrix (SAM)
as an extension of I/O model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252

CHAPTER XIII
AN ANALYTICAL APPROACH TO THE CALCULATION OF THE "GREEN GDP"

A. Adaptation of the I/O framework to the Green GDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256

B. Green GDP based on a simple I/O model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258

C. Green GDP based on a more realistic I/O model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259

D. Adaptation of the I/O model to study the economic impact of pollution abatement policies . . . . . . 261
1. Allocation of pollution abatement services to industries and final uses . . . . . . . . . . . . . . . . 262
2. Financing investments in abatement equipment by sectors . . . . . . . . . . . . . . . . . . . . . . . . . 264

xiii
 PART ONE

ANALYTICAL AND STATISTICAL FOUNDATION OF

INPUT-OUTPUT MODEL
 3

I. BASIC INPUT-OUTPUT SYMMETRIC MODEL (SIOT)

A. Historical background

1.1. Input-output (I/O) analysis as a theoretical framework and an applied economic tool in a market
economy was developed by Wassily Leontief with the construction of the first input-output tables for the
United States for the years 1919 and 1929 which were published in 1936. Since then, tables describing the
interrelationships among various producers of an economy have been constructed for over 90 countries. For
the development of input-output methodology and its application to important economic issues, Leontief was
honored with a Nobel prize in Economic Science in 1973. The integration of an input-output framework into
the system of national accounts was developed and published in 1968 by the United Nations as a System of
National Accounts, Studies in Methods. The integrated work earned Professor Richard Stone, a Nobel prize
winner in Economic Science in 1984 "for having made fundamental contributions to the development of the
systems of national accounts and hence greatly improved the basis for empirical economic analysis".

B. A simple input-output framework

1.2. The fundamental contribution of input-output in economics is the transformation of Francois

Quesnay's Tableau Economique - a descriptive device showing sales and purchases relationships between
different producers and consumers in an economy - into an analytical framework which facilitates economic
projections and analyses. It assumes that the inputs used in producing a product are related to the industry
output by a linear and fixed coefficient production function (at least in the short run). Under this assumption,
input and output relationships are transformed into technical relationships, with each column in an input-output
coefficient table representing a technique of production.

1.3. In a simple, refined form based on the open system1 originally designed by Professor Leontief, an
input-output coefficient table represents, in each of its columns, a technique of production by which only one
product is produced.2 One may understandably argue against the application of so simple an assumption as
the assumption of linear and fixed coefficients. However, the use of input-output economics can be justified
on the notion that a technique of production will not change much over a short period and, even if it does
change, the engineering data or statistical data collected by censuses on inputs utilized by an industry, which
represent a new technology in an input-output column, can be obtained to replace the old column. Because
input-output tables in theory can be in monetary units or in both physical and monetary units, the construction
of such tables may be implemented independently of national accounts. In fact many economists have
attempted to combine both physical and monetary units in an input-output table, especially with respect to

1
An open system does not internalize every transaction in the economy. It treats part of the economic transactions
in the economy, specifically final demand as exogenous.

2
Because of the assumption that one industry produces only one product, the inter-industry transaction table, which
is also called the table of intermediate consumption, must necessarily be not only square but also symmetric. The I/O
model based on this table will be called symmetric I/O model or SIOT.
4

energy and mineral products and therefore engineering data are quite useful to them. However, because input-
output tables elaborated as supply-and-use tables in monetary terms have been found to be powerful tools for
compiling production accounts in national accounts, they have been integrated into the United Nations System
of National Accounts since 1968. This enhances the accuracy of input-output tables through the process of
balancing supply and use of every commodity in a given economy as it is transacted in the market at different
prices to different users. Because of the latter reason, the Handbook will focus mainly on the construction of
an input-output table as part of a national accounting system, but the final product should satisfy the basic
requirements of an input-output model.

1.4. The following parts of this chapter will explain the basic input-output economic model and some of
its basic economic implications. For simplicity in presentation, the various elements of net final demand,
which include final consumption expenditures and gross capital formation of the household sector, the
government sector and the sector of non-profit institutions serving households and exports minus imports3 will
be considered as a single column vector and elements of value added, which are also referred to in economic
literature as primary inputs, will be considered as a single row vector. It is also assumed that producers do
not engage in any secondary production, i.e., they produce only their characteristic products or, in other words,
one type of producers produces only one type of products. Thus, henceforth, no distinction will be made
between industries and products.

1.5. An input-output table focuses on the interrelationships between industries in an economy with respect
to the production and uses4 of their products and the products imported from abroad. In a table form (see table
1.1) the economy is viewed with each industry listed across the top as a consuming sector and down the side
as a supplying sector.

Table 1.1. A highly simplified input-output accounting framework

Industries Net final demand Total output

Industries F Y X

Value added V
(Primary inputs)
Total input X

3
Because imports are netted out, hence the term "net final demand" is used.

4
It is important to note that purchases of goods and services by a producer should not be equated to usages or
consumption since on the one hand part of current purchases may be held as stocks for future uses, and on the other hand,
current consumption may exceed current purchases of input by the quantity or value of stocks which were bought in
previous periods and are brought out for use in the current production period. This distinction between purchases and
consumption is important for the balancing of input-output tables.
 5

1.6. Table 1.2 below shows a simplified set of accounts distinguishing three producers and showing the
input-output flow matrix describing their transactions. The values in the square box represent intermediate
consumption, i.e. uses of products as inputs in the production process.

Table 1.2. Input-output flow table and accounts

Industry A Industry B Industry C Final Total

demand output
Industry A 0 20 45 35 100
Industry B 30 0 30 140 200
Industry C 0 80 0 70 150

Value added 70 100 75

Total input 100 200 150

1.7. Input-output analysis became an economic tool when Leontief introduced an assumption of
fixed-coefficient linear production functions relating inputs used by an industry along each column to its output
flow, i.e., for one unit of every industry's output, a fixed amount of input of each kind is required. This fixed
relationship is introduced in table 1.3. The entries in each column of table 1.3 are obtained by dividing the
entries in the column by the total input of the consuming industry.

Table 1.3. Input-output coefficient table

Inputs per unit of output

Industry A Industry B Industry C

Industry A 0.00 0.10 0.30

Industry B 0.30 0.00 0.20
Industry C 0.00 0.40 0.00

Value added 0.70 0.50 0.50

1.8. In the above table, for example, one unit of output of industry B requires 0.10 unit of output of
industry A, 0.40 unit of output of industry C, and generates 0.50 unit of value added. Similarly, one unit of
output of industry C requires 0.30 unit of output of industry A, 0.20 unit of output of industry B and generates
0.50 unit of value added. Thus, in order to produce output X , X and X , the amount of product A (output
A B C
of industry A) required as intermediate input is equal to

(1.1) 0.00 XA + 0.10 XB + 0.30 XC

1.9. Equation 1.1 calculates the total amount of product A used as intermediate input in the production
process of an economy. If the remaining value of the same product left for net final demand, i.e. 35 in table
1.2 is further added to intermediate consumption, the total output of industry A is obtained in equation 1.2.

(1.2) 0.00 XA + 0.10 XB + 0.30 XC + 35 = 100

6

It is possible to check the equality property of equation 1.2 by replacing the values of X , X and X in table
A B C
1.2 by their actual values. The results are shown in equation 1.3.

(1.3) 0.00 x (100) + 0.10 x (200) + 0.30 x (150) + 35 = 100

The utilization of products B and C as intermediate inputs of production may be similarly calculated. In
general, the ratios shown in the box in table 1.2 could be written in more abstract terms, such as those in table
1.4, so that an input-output model may be formulated.

Table 1.4. Input-output coefficient table in more general terms

Industry 1 Industry 2 Industry 3 Net final demand

Industry 1 a11 a12 a13 Y1
Industry 2 a21 a22 a23 Y2
Industry 3 a31 a32 a33 Y3

Value added V1 V2 V3

The relationships in equations 1.1, 1.2, 1.3 using general terms of table 1.4 can be written as follows:

a11 X1 + a12 X2 + a13 X3 + Y1 = X1

(1.4) a21 X1 + a22 X2 + a23 X3 + Y2 = X2
a31 X1 + a32 X2 + a33 X3 + Y3 = X3

In matrix form, equation 1.4 can be written as follows:

a11 a12 a13 X1 Y1 X1

(1.5) a21 a22 a23 x X2 % Y2 ' X2

a31 a32 a33 X3 Y3 X3

1.10. In a more general form with n industry and n products, where a stands for input i (product of
ij
industry i) used in the production of one unit of output of industry j, systems of equations 1.4 and 1.5 can
be written as follows:

a11 X1 + a12 X2 + ..... + a1n Xn + Y1 = X1

(1.6) a21 X1 + a22 X2 + ..... + a2n Xn + Y2 = X2
. + . + ..... + .+ . = .
an1 X1 + an2 X2 + ..... + ann Xn + Yn = Xn

a11 a12 . a1n X1 Y1 X1

a21 a22 . a2n X2 Y2 X2 and in matrix form,
(1.7) x % '
. . . . . . .
a n1 a n2 . a nn Xn Yn Xn
 7

The computation of the coefficient matrix can be described in the following mathematical form:

Fij
a ij '
Xj

where F stands for an element of the flow table as described in a square box of table 1.1. Equation 1.7 is
ij
usually written in matrix form, as

(1.8) AX + Y = X

1.11. Relationship 1.8 is the basic input-output system of equations. Matrix A is called the input-output
coefficient matrix, vector X is the vector of output and vector Y is the vector of net final demand. The
dimension (size) of matrix A is constrained only by the statistical information on inputs and outputs
available to statisticians since some countries have constructed input-output tables of up to almost 500
industries.

C. The inverse matrix

1. Solution of an input-output model

1.12. Equations in the form of equation 1.8 are much more suitable to model-building or analysis. If the
values of the coefficients and of net final demand are known, then it is possible to solve this set of
simultaneous equations in order to find the level of output of various industries necessary to satisfy the
specified level of net final demand.

1.13. Mathematically, the vector of output X in the system of equation 1.8 can be solved as follows:

X - AX = Y
(1.9) (I - A)X = Y
X = (I - A)-1Y

where I stands for the identity matrix which is a square matrix where all the diagonal elements are equal to
-1
1 and all other elements are equal to zero. (I - A) is the Leontief inverse which can be calculated with
difficulty. At present, microcomputers can easily invert a matrix of a size greater than 200. The
availability of microcomputers has facilitated the application of input-output analysis in developing
countries.
-1
1.14. However, as shown in equation 1.9, (I - A) is just a big black box without much economic
meaning, and it is therefore important to explain the economic meaning of this inverse.

2. Economic interpretation of the inverse

8

1.15. The input structures represented by the A-matrix and discussed in previous sections of this chapter
show the type and amount of various inputs each industry requires in order to produce one unit of its
output but tell nothing about indirect effects. For example, the effect of the production of a motor vehicle
does not end with the steel, tyres and other components required. It generates a long chain of interaction in
the production processes since each of the products used as inputs needs to be produced and will, in turn,
require various inputs. The production of tyres, for instance, requires rubber, steel and cloths, etc. which,
in turn, require various products as inputs including the transport service provided by motor vehicles that
necessitates the production of motor vehicles in the first place. One cycle of input requirement requires
another cycle of inputs which in turn requires again another cycle. This chain of interactions goes into
infinity. However, the sum of all these chained reactions is determined from the value of the Leontief
inverse.

1.16. The chain of actions in a simplified form is illustrated in the following chart.

Direct effects Y Direct inputs Y Indirect inputs

Indirect effects

Direct effects initially results in the requirement for direct inputs which then require a chain of indirect
inputs. The sum of direct inputs and indirect inputs is normally called indirect effects.

1.17. In chain reactions in input-output analysis, the first exogenous shock is assumed to be initiated by
an exogenous increase in net final demand, like an increase in export demand, or an increase in fixed
capital formation. This assumption is made mainly for the sake of simplicity of exposition. Actually, the
first shock can happen anywhere. It can be an increase in domestic production of intermediate
consumption to replace imports, an increase in indirect taxes, a change in technology represented by
changes in input structures, etc.

1.18. In table 1.5, an initial increase in the net final demand of 1,000 units of product B is assumed.
 9

Table 1.5. Direct and indirect inputs

Initial Direct Indirect inputs Total

increase in input output
net final
demand Round Round Round Round ... Round
1 2 3 4 n
Formula F AF A2 F A3 F A4 F ... An F
A 0 100 120 11 16.8 ... ... 257
B 1,000 0 110 36 12.1 ... ... 1,171
C 0 400 0 44 14.4 ... ... 468

1.19. The chain reactions generated by an increase in net final demand include a string of outputs. In
the first round, it is the incremental output to meet the increase in net final demand. In the second round, it
is the incremental output to meet the input requirement of production to meet the increase in net final
demand. In the third round, it is the incremental output to meet the input requirement of the incremental
output of the second round. The number of rounds goes on to infinity. Since the coefficient matrix A
describes the input requirement of any increase in output, the chain reactions can be written as follows:

Exogenous shock Y F = F

First round F Y AxF = AF

Second round AF Y A x AF = A2 F

Third round A2 F Y A x A2 F = A3 F

... ... ... ... ...

nth round An-1 F Y A x An-1 F = An F

TOTAL
IMPACT
Y (I+A+..+An )F

Total impact = gross outputs generated

2 3 n
= F + AF + A F + A F + ... + A F
2 3 n
= (I + A + A + A + ... + A ) F

1.20. Table 1.5 includes the values of each round in the chain reactions to an
0 exogenous shock on net final demand. The values of these rounds can be
calculated as follows:
F ' 1000
0
10

0 .10 .30 0 100

AF ' .30 0 .20 x 1000 ' 0
0 .40 0 0 400

0 .10 .30 100 120

A 2F ' .30 0 .20 x 0 ' 110
0 .40 0 400 0

Mathematically,
2 3 n -1
(1.10)5 I + A + A + A + ... + A = (I - A)

as n approaches infinity (or, more formally, as n 6 4). However, it is not necessary to use the iterative
approach used in equation 1.10 to calculate the Leontief inverse as there are other approaches that can
calculate it directly. The Leontief inverse of the I/O coefficients in the example of table 1.3 is shown in
table 1.6 below.
-1
Table 1.6. The inverse matrix (I-A)

Industry A Industry B Industry C

Industry A 1.077 0.257 0.375

Industry B 0.351 1.171 0.340
Industry C 0.141 0.468 1.136

-1
1.21. The inverse matrix (I-A) is fundamental to input-output analysis as it shows the full impact of an
exogenous increase in net final demand on all industries. With such a matrix it is possible to unravel the
technological interdependence of the productive system and to trace the generation of output demand from
final consumption which is part of net final demand throughout the system. It is then possible to calculate
what output levels would be required to meet various postulated levels of net final demand and
consequently how output levels would be required to change to meet postulated changes in net final
demand.

5
This relation holds under certain conditions on the structure of matrix A as discussed in the appendix to this chapter,
para. 1.60.
 11

D. Measurement unit

1.22. In previous sections, the units used to measure inputs and outputs have not been discussed.
Nevertheless, it is easy to detect that the flow matrix of table 1.2 and the I/O coefficients derived from it in
table 1.3 are all based on monetary values. Henceforth, in an input-output table, the following postulates
are maintained:

(a) Total input is equal to total output in each producing unit;

(b) Each I/O coefficient is smaller than 1.0; and

(c) As a result of (a) the sum of I/O coefficients plus value added coefficients in each column of
the I/O coefficient table is equal to 1.0.

1.23. In the inverse matrix describing total direct and indirect input requirements, the diagonal elements
in the Leontief inverse shown in table 1.6 are at least equal to 1.0. This means that to produce one
additional unit for delivery to net final demand, it is necessary to increase output by at least one unit.

1.24. An input-output model, however, does not have to be measured in monetary values. In fact, each
input and output can be measured in different physical units - steel in tons, petroleum in liters or gallons,
services in money, etc. If so, each I/O coefficient would no longer be smaller than 1.0 and the sum of the
coefficients along each column cannot be meaningfully calculated, since it is not possible to add apples to
oranges.

1.25. As commonly done in most countries, an input-output system is compiled in monetary values,
sometimes but rarely supplemented with additional tables of physical units. The main advantage of a
monetary value table is that errors can be detected by the requirement that the column sum and the row
sum of each producer in the flow table must be equal (of course only if each producer is producing only
one product).

E. Prices and costs

1. Prices as costs of production

1.26. It was noted earlier that a column of an input-output matrix together with value added accounted
for all the expenditures of a producer. It follows that the price of each product can be built up from the
prices of inputs that are used to produce that product.

1.27. As can be seen from matrix A in equation 1.7, in order to produce one unit of product 1, one needs
the set of input (a11 a21 ... ai1 ... an1)6. Then, if p i is the price per unit of product i, the cost of intermediate
inputs to produce a unit of product 1 can be written as follows:

(1.11) p a 1 + p a 1 + ... + p a 1 + ... + p a

1 1 2 2 i i n n1

6
aij is now a "real" technical coefficient.
12

1.28. The difference between the unit price of product 1 and the cost of intermediate inputs for the
production of a unit of product 1 is equal to the value added per unit of product 1.

(1.12) p - (p a 1 + p a 1 + ... + p a 1 + ... + p an1) = v

1 1 1 2 2 1 i n 1
By switching elements of 1.12 around, a simplified system of price equations follows:

p = p a 1 + p a 1 + ... + p a 1 + ... + p a 1 + v
1 1 1 2 2 i i n n 1
p = p a 2 + p a 2 + ... + p a 2 + ... + p a 2 + v
2 1 1 2 2 i i n n 2
(1.13) .. .. .. .. ..
.. .. .. .. ..
p = p a n + p a n + ... + p a n + ... + p a n + v
n 1 1 2 2 i i n n n

1.29. Solving for price is easier using matrix methods. The price equations are transformed into:

p1 a11 a21 . an1 p1 v1

p2 a12 a22 . an2 p2 v2
(1.14) ' x %
. . . . . . .
pn a1n a2n . a nn pn vn

1.30. The block of "a" coefficients in the above equation 1.14 is the coefficient matrix "A" with its rows
and columns interchanged, i.e., it is matrix A that is transposed and can be written as A'. As p is defined
as the price vector and v as the value added vector, the above set of equations can then be simply written
as:

(1.15) p = A'p + v

which, when solved for p gives

-1
(1.16) p = (I - A') v

1.31. It should be noted that the inverse matrix in equation 1.16 is similar to the Leontief inverse
described in the previous section. In fact, it can be proven that the inverse in 1.16 is the transpose of the
Leontief inverse. (Proof can be found in most textbooks on linear algebra).

(I&A ))&1 ' ([I&A]&1))

1.32. As one can see below:

1.077 0.257 0.375

&1
(I&A) ' 0.351 1.171 0.340
0.141 0.468 1.136
 13

then

1.077 0.351 0.141

&1 )
((I&A) ) ' 0.257 1.171 0.468
0.375 0.340 1.136

each column of the matrix on the left hand becomes a row of the matrix on the right hand, with the column
indexing becoming row indexing.

2. Price analysis in input-output

1.33. In a simplified input-output model with constant coefficients, price analysis is shown in equation 1.16
which allows the calculation of the effects on price levels induced by changes in value added, i.e., given that
v is known, p can be calculated.

1.34. However, in order to clearly understand price analysis in input-output methodology, particularly when
most input-output tables are in monetary values, it is useful to show price analysis when input-output analyses
are constructed in physical units.

1.35. Equation 1.16 shows that v is the vector of value added in monetary terms per unit of physical output
if A is the input-output coefficient matrix constructed in physical output. However, if A is measured in
monetary terms, v is the vector of value added per monetary unit of output (for example per one dollar of
output). Then every element of the price vector p is equal to 1.0 (see table 1.7). This simply means that the
selling price of 1 dollar worth of output is 1 dollar. For the value of v, see table 1.3.

Table 1.7. Effects of value added on prices using I/O table in value terms
-1
p = (I-A') v

1 1.077 0.351 0.141 .70

1 ' 0.257 1.171 0.468 x .50
1 0.375 0.340 1.136 .50

1.36. The price of every product in vector p in table 1.7 is equal to 1 when no change is made in coefficient
A or in the vector of value added v. This does not mean that such a calculation is of no useful application.
In fact, it is quite useful to calculate changes in prices as a result of changes in value added.

1.37. Table 1.8 shows a useful application of price analysis. It attempts to answer a question of the type
"what will be the effect on the product prices if value added per unit of product B increases by 10 per cent."
14

Table 1.8. Effects of value added on prices using I/O table

in monetary terms
-1
pn = (I-A' ) vn

1.017 1.077 0.351 0.141 .70

1.058 ' 0.257 1.171 0.468 x .55
1.018 0.375 0.340 1.136 .50

Changes in relative prices are equal to:

n
pi n
' p i since pi ' 1
pi

1.38. With the new value of v (denoted by vn) for product B at .55 in table 1.8 (that is 10 per cent higher
than the old v value in table 1.7), the price of product B increases by almost 6 per cent while prices of other
products increase by almost 2 per cent.

3. Some rules in price analyses

1.39. The following rules can be derived in analyses of changes in relative prices due to technical changes
as reflected by changes in the I/O coefficients or value added:

(a) Changes in relative prices, given changes in value added or technical coefficients, will be the
same whether the I/O table is constructed in physical units or monetary units;

(b) Changes in value added or coefficients can be directly introduced into the I/O table in
monetary units without any further adjustment as long as the prices of the base period where
each price is equal to 1 is used for comparison; thus, the sum of the coefficients in each
column, including the coefficients with new values need not be equal to 1.

1.40. The example in table 1.9 (see below) will demonstrate that the relative price changes, as shown by
pn in table 1.8, will be the same whether the I/O coefficients table is in monetary terms or in physical units.
From the same example, if the new price vector is used to update the I/O coefficient matrix, by the relationship:

(1.17) p̂ A n p̂&1

then the sum of each column of the new updated I/O coefficient will be equal to 1. The following paragraphs
will explain the rules mentioned above through an example in table 1.9. The I/O table in constant prices
discussed in chapter XI is the table obtained by applying the relationship in 1.17

1.41. The physical tables shown on the left-hand side of table 1.9 present flow and coefficient tables
expressed in physical units; the value tables on the right-hand side are flow and coefficient tables in monetary
terms, similar to what has been shown in tables 1.2 and 1.3.
 15

1.42. In the flow table of physical units, value added and price per unit of physical output can be used to
derive the flow table and other information in monetary terms on the opposite side. These are shown in lines
1-6. For example, the first row of the flow table in monetary terms on the right-hand side is calculated by
multiplying the first row of the flow table in physical units on the left-hand side, which refers to the
intermediate consumption of the first product, by the unit price of the first product 0.20 shown as the first
element in line 6. Coefficient tables and value added coefficients are calculated on the basis of information
provided in lines 1-6 and are shown in lines 7-10. Unit prices for the table in monetary terms on the right-hand
side of line 6 are all equal to 1.0, as explained in para. 1.35.

1.43. The second set of information relating to changes in value added is shown in lines 11-12. New values
of value added per unit of physical output are shown on the left-hand side of line 11. Here, only value added
per physical unit of the second producer is assumed to change, increasing it 10 times, from 5.0 (line 10) to 50.
This increase may result from a substantial one in taxes on products due to, for instance, a desire by the
Government to drastically reduce the production and consumption of that product. New value added per unit
value of output of the second producer on the right-hand side of line 11 can be calculated by dividing value
added per unit of physical output on the left-hand side of line 11 by the price per unit of physical output on
the left-hand side of line 6. It should be recognized here that this calculation is based on the old price system.
In the table in monetary terms, the new value added coefficients on the basis of the old price system can be
calculated directly. In our example, if the new value added coefficient of producer 2 is increased 10 times, it
should be equal to 0.50 x 10 = 5.0.

1.44. Using the formula and procedure in table 1.8 above, the new price per unit of physical output based
on new value added coefficients is shown in line 12 while changes in relative prices, which are calculated by
dividing values in line 12 by corresponding values in line 6, are shown in line 13. One could conclude from
this exercise that changes in prices, given changes in value added coefficients, are the same whether one uses
physical or value I/O tables for analysis.

1.45. It is interesting to note, however, that when changes are introduced in the coefficient tables compiled
in monetary terms, the sum of the new coefficients in each column does not have to be equal to 1.0. For
example, as can be seen from column 2 of the coefficient table on the right-hand side, the sum of intermediate
consumption in line 7 to 9 and the new value added coefficient in line 11 is not equal to 1.0. In fact, the value
added coefficient in value terms alone is equal to 5. One may introduce changes in technical coefficients and
consequently in input-output coefficients in monetary terms and one would also obtain the same results without
having to reduce the column sum of the new matrix terms to 1.0. In this case, the new I/O coefficient table is
still measured in the old price system. The conversion of the new I/O coefficient table to the new price system
will bring back the standard rule that the sum of each column I/O coefficient table must be equal to 1.0. The
new I/O coefficient matrix in the new price system is obtained by using the following formula:

p̂ n A n (p̂ n)&1

where p̂ n is the diagonal matrix with values of vector p in the diagonal. The new value added coefficient
in the new price system is obtained by dividing each element of vn with the corresponding value in pn. An and
pn are shown in the lower part of table 1.9 from line 14 to line 17 To summarize then: When changes in
technology are introduced into coefficients of an input-output table in monetary terms in the same price system,
the sum of each column of the new table does not need to add up to 1.0. However, given these changes, a new
price system is established and if the input-output coefficients are recalculated on the basis of this new price
system, the sum of each column of the new table should add up to 1.0.
16

1.46. Furthermore, the introduction of an arbitrarily large change in value added per unit of physical output
and the reflection of that change on product prices on the basis of adding up costs were purely for illustrative
purposes. The basis for such an example may be considered economically unsound because the reaction to
an increase in taxes may result in production cut-backs or in changes in demand which cannot be easily be
formulated in a simple input-output model.
 17

Table 1.9. Analysis of price changes using physical and

value I/O tables
Physical tables Lines Value tables

Flow table Flow table

0 100 225 (1) 0 20 45

3 0 3 (2) 30 0 30

0 80 0 (3) 0 80 0

Value 70 100 75 (4) 70 100 75 Value added in

added in monetary terms
monetary
terms

Output in 500 20 150 (5) 100 200 150 Output in

physical monetary
units units

Price per 0.20 10 1.0 (6) 1 1 1 Unit price index

unit of
output

Coefficient table Coefficient table

0 5 1.50 (7) 0.00 0.10 0.30

0.006 0 0.02 (8) 0.30 0.00 0.20

0 4 0 (9) 0.00 0.40 0.00

Old value 0.14 5.0 0.5 (10) 0.70 0.50 0.50 Old value added
added per per unit value
unit of of output
physical
output

New value 0.14 50 0.5 (11) 0.70 5.0 0.50 New value
added per added per unit
unit of value of output
physical
output

New unit 0.516 62.69 2.528 (12) 2.580 6.269 2.528 New unit price
price of index
output

Changes in 2.580 6.269 2.528 (13) 2.580 6.269 2.528 Changes in

relative relative prices
prices

New coefficient table in new price system

(14) 0.000 0.041 0.306 Inter-

(15) 0.729 0.000 0.496 mediate

(16) 0.000 0.161 0.000 consumption

(17) 0.271 0.798 0.198 Value added

18

Appendix

MATHEMATICAL BACKGROUND

1.47. Following are some of the fundamental definitions and operations in matrix algebra. Readers are
advised to consult textbooks on advanced econometrics if they wish to learn more about matrix algebra.

A. Fundamental matrix definitions

1.48. Matrix: In a general form, a matrix is a rectangular grid containing elements in m rows and n columns.
It is said to have an (m x n) dimension. Below is an example of a matrix with m=3 and n=2:

4 5
A ' 2 1
6 3

1.49. Column vector: A column vector is a matrix with only one column and more than one row. Below
is an example of a column vector:

4
B '
2

1.50. Row vector: A row vector is a matrix with only one row and more than one colum. Below is an
example of a row vector:

C ' 4 5 6

1.51. Scalar: A scalar is a real number, i.e. a matrix with one row and one column: 4 for instance is a scalar.

1.52. Square matrix: In a square matrix, the number of rows is equal to the number of columns, i.e. m=n.
Below is an example of a square matrix:

4 5 6
D ' 2 1 3
6 3 8

1.53. Diagonal matrix: A diagonal matrix is a square matrix with all off-diagonal elements equal to zero.
A diagonal matrix is normally denoted with a hat. Below is an example of a diagonal matrix:
 19

4 0 0
Ê ' 0 1 0
0 0 8

1.54. Identity matrix: An identity matrix is a diagonal matrix with all diagonal elements equal to 1. Below
is an example of an identity matrix:

1 0 0
I ' 0 1 0
0 0 1

1.55. Transpose: A transpose of a matrix is formed by transforming each of the columns of the original
matrix into a row or vice versa. A transpose of A is normally written as A'. Below is the transpose of matrix
A:

4 2 6
A) '
5 1 3

B. Fundamental matrix operations

1.56. Addition: Matrices may be added together if they have the same dimension. For example, if A1 is
added to B1, then each element of the new matrix C1 identified by a row number and a column number (the
row number is always identified first followed by the column number) is the sum of the element of matrix A1
and the element of matrix B1 identified in the same position as the element in the resulting matrix. For
example element C12 = A112 + B112. Below is an example for adding the two matrices:

4 3 2 1 2 1 4%1 3%2 2%1

% '
2 2 1 0 0 2 2%0 2%0 1%2

5 5 3
'
2 2 3

1.57. Subtraction: Similarly to addition, one matrix can be subtracted from the other if they have the same
dimension. The process of subtraction is similar to the process of addition. Below is an example of subtraction:

4 3 2 1 2 1 4&1 3&2 2&1

& '
2 2 1 0 0 2 2&0 2&0 1&2

3 1 1
'
2 2 &1
20

1.58. Multiplication of a row vector and a column vector: In order for a row vector to be multiplied with
a column vector or vice versa, the number of elements in the two vectors must be equal. The result will be a
scalar which is the sum of the products of corresponding elements. Below is an example of a multiplication:

1
4 3 2 x 2 ' (4x1)%(3x2)%(2x0) ' 10
0

1.59. Multiplication: Two matrices may be multiplied to one another only if the second dimension of the
matrix on the left-hand side is equal to the first dimension of the matrix on the right-hand side. Thus:

K x M = N
(m x n) (n x k) (m x k)

The second dimension of K, i.e n is equal to the first dimension of M, i.e n. In multiplication, to obtain the
elements of the first row of N, the first row of K is multiplied respectively by the columns of matrix M
following the procedure in para. 1.58. The elements of other rows of N are also obtained similarly. Below is
an example of the product of two matrices:

1 0
4 3 2 10 11
x 2 3 '
2 2 1 6 7
0 1

In the example above, K has a 2x3 dimension, M has a 3x2 dimension, N must have a 2x2 dimension.

C. Solution of a linear equation system

1.60. Consider the following relationships:

X1 + 3X2 = 8
2X1 - X2 = 10

then

1 3 X1 8
B ' , X ' , Y '
2 &1 X2 10

The system can now be written as BX = Y and the solution of X is X = B-1Y. B-1 is called the inverse. The
necessary and sufficient conditions for B to be invertible is that B is a square matrix and that no row or column
in B is proportional respectively to other rows and columns, or a linear combination of other rows and
columns, i.e. all rows and columns are linearly independent. The system of equations may have a solution only
 21

if B is invertible. In a simple system where B is a scalar, X and Y are one-element variables. The solution is
quite simple. Let bx = y, then x = (1/b)y. If b = 2, y = 10 then x = 5. 1/a is similar in nature to the inverse
of matrix B. If a = 0 then x is indeterminate, i.e. it has no solution. No mathematical technique for matrix
inversion is presented because it is tedious and not economically revealing. Readers can find it in advanced
econometric textbooks.

1.61. The simple input-output economic model in which one industry produces only one commodity is a
linear equation system. Then B= (I-A)-1 which is called the Leontief inverse. This Leontief inverse is always
non-negative when:

-A is measured in value terms;

-aij, any element of A is non-negative and smaller than 1, which under normal economic conditions,
is always satisfied since the value of any input used is smaller than the value of output.
22
 23

II. THE SYSTEM OF NATIONAL ACCOUNTS (SNA) FRAMEWORK

OF SUPPLY AND USE TABLES (SUT): OVERVIEW

2.14. The input-output (I/O) model built around a symmetric I/O coefficient matrix and described in chapter
I can be derived from the SNA framework of supply and use tables (SUT) which is an essential part of the
SNA integrated national accounting structure. In this chapter, the SNA SUT, and the links between SUT and
institutional accounts of the SNA will be discussed. The appendixes will discuss the statistical units and
classification schemes based on which data for SUT are collected. Methods used in deriving the symmetric
I/O coefficient table (SIOT) from the SUT are discussed in chapter IV.

2.15. The SNA provides a comprehensive framework in which basic statistical data on transactions among
micro-producing units, i.e., the establishments, may be presented with minimum manipulation of statistical
data. Establishments as statistical units will be discussed in paragraph 2.54 below. Statistical data are
realistically presented in the basic supply and use framework of the SNA in the following senses:

-Any producing unit may engage in more than one activity producing more than one type of product;

-Goods and services as outputs are as far as possible valued at the prices at which they first entered
the market, i.e. basic prices or at equivalent market prices; they are only valued at costs when no
equivalent market prices are available;

-Goods and services as intermediate or final products are valued at the prices which users have to pay
for them.

2.16. Tables 2.1, 2.2 (see paragraph 2.24 below) describe fully the SNA SUT. They are the supply-and-use
tables that are closely linked. The supply table which is referred to in the SNA of 1968 as the commodity table
describes the sources of supply of products to the economy. The use table which was called the industry table
in the 1968 SNA describes where the products and other primary factors are used, as well as the derivation of
value added as the difference between output and intermediate consumption of goods and services in
production. The basic principle in deriving these two tables is that the total supply of a product must equal
the total use of the product.

A. Basic concepts used in SUT

2.17. Before the supply and use tables are described, many concepts used in the SNA SUT are reviewed
below.

2.18. Industry: "An industry consists of a group of establishments engaged in the same, or similar, kinds
of activity. At the most detailed level of classification, an industry consists of all the establishments falling
within a single Class of ISIC and which are therefore all engaged on the same activity as defined in the ISIC.
At higher levels of aggregation corresponding to the groups, divisions and, ultimately, sections of the ISIC,
24

industries consist of groups of establishments engaged in similar types of activities1. An establishment

“combines both the kind-of-activity dimension and the locality dimension”. An establishment is defined as
an enterprise, or part of an enterprise, that is situated in a single location and in which only a single (non-
ancillary) productive activity is carried out or in which the principal productive activity accounts for most of
the value added2. The ISIC is the United Nations International Standard Industrial Classification of All
Economic Activities3. The basis of these classifications is the establishment, i.e. the statistical unit which will
be elaborated on in part D of this chapter and in chapter V. The term "producer" is used interchangeably with
the term "industry". Thus a "producer" does not refer to the owner of an enterprise or to the enterprise itself
but to a group of similar economic activities classified under the same classification unit. Countries and groups
of countries may develop their own industrial classification to meet their specific requirements, but they should
be able to link with ISIC. European Union countries uses the general Industrial Classification of Economic
Activities within the European Communities (NACE, Rev. 1) for their European System of National Accounts
(ESA 95), which is broadly consistent with the System of National Accounts of the United Nations (1993
SNA) as regards the definitions, accounting rules and classifications but incorporates certain differences,
particularly in its presentation, which is more in line with its use within the European Union. NACE is
harmonized with ISIC.

2.19. Product: "Products" are used interchangeably with "goods and services" and are classified according
to the United Nations Central Product Classification (CPC), (ST/ESA/STAT/SER.M/77 Version 1.0) (see
appendix B). Similar to industries, countries and groups of countries may develop their own product
classification to meet their specific requirements, but they should be able to link with ISIC. European Union
countries use the Classification of Product by Activity (CPA) as the standard for their European System of
National Accounts, which is harmonized with the United Nations Central Product Classification.

2.20. Industry output: Industry output is the total value of all products produced by an industry including
both primary and secondary products.

2.21. Product output: Product output is the total value of a product produced by all industries, i.e. all
resident producers in the economy.

2.22. Market, own final use and other non-market producers: Industries in the SNA are grouped into
three broad categories: Market, own final use and other non-market producers. "A market producer is an
establishment or enterprise all or most of whose output is marketed. It is perfectly possible for market
producers, both small unincorporated enterprises and large corporations, to have some non-market output in
the form of production for own final consumption or gross fixed capital formation. Own-account producers
consist of establishments engaged in gross fixed capital formation for the enterprises of which they form part
or unincorporated enterprises owned by households all or most of whose output is intended for final
consumption or gross fixed capital formation by those households: for example, owner-occupiers or
subsistence farmers who sell none, or only a small fraction, of their output. Other non-market producers consist
of establishments owned by government units or NPISHs [non-profit institutions serving households] that
supply goods or services free, or at prices that are not economically significant, to households or the

1
System of National Accounts, 1993 (United Nations publication, Sale s No.E.94. XVII.4), para. 5.40.

2
Ibid , paragraph 5.21.

3
United Nations Publications, Sales [Link].11.
 25

community as a whole. These producers may also have some sales of secondary market output whose prices
are intended to cover their costs or earn a surplus: for example, sales of reproductions by non-market
museums.”4 Products are also similarly grouped into three broad categories: market, for own final use and
other non-market. The classification of activities and products into the three categories is not required to obtain
an I/O table SIOT as described in chapter I but it will provide useful information in estimating new I/O
coefficients if there are changes in the shares of the categories. This is because in an I/O model based on an
SIOT table, a column representing a production technology used in producing a product is normally a weighted
average of different technologies used in producing that product. To be more useful to I/O analysis, market
producers should be classified into two groups: unincorporated and own-use producers of household activities
which may use more primitive technology; and corporate enterprises which are large and may use more
advanced technology. This latter distinction is also useful for the purpose of identifying value added generated
by the household sector.

2.23. Valuation of outputs: "Goods and services produced for sale on the market at economically
significant prices may be valued either at basic prices or at producer's prices. The preferred method of
valuation is at basic prices, especially when a system of VAT, or similar deductible tax, is in operation,
although producer's prices may be used when valuation at basic prices is not feasible.”5 “The basic price is
the amount receivable by the producer from the purchaser for a unit of a good or service produced as output
minus any tax payable, and plus any subsidy receivable, on that unit as a consequence of its production or sale.
It excludes any transport charges invoiced separately by the producer; the producer's price is the amount
receivable by the producer from the purchaser for a unit of a good or service produced as output minus any
VAT, or similar deductible tax, invoiced to the purchaser. It excludes any transport charges invoiced separately
by the producer.”6

2.11 Value added: "Value added is the balancing item in the production account for an institutional unit
or sector, or establishment or industry. It measures the value created by production and may be calculated
either before or after deducting the consumption of fixed capital on the fixed assets used....Gross value added
is defined as the value of output less the value of intermediate consumption; net value added is defined as the
value of output less the values of both intermediate consumption and consumption of fixed capital.”7 Value
added at basic prices of an industry is the difference between the industry output at basic prices and the
intermediate consumption of the industry at purchasers' prices.

2.12. Operating surplus is a balancing item, which is equal to value added minus compensation of
employees, minus taxes less subsidies on production and imports. Operating surplus that includes
consumption of fixed capital is called gross operating surplus; without consumption of capital, it is called net
operating surplus. Operating surplus is calculated residually given value added and compensation of
employees, other taxes less subsidies on production and consumption of fixed capital. Value added at market
prices (either at producers' prices or purchasers' prices) for the total economy is the sum of value added at basic
prices plus taxes less subsidies on products.

4
System of National Accounts 1993, para.. 6.52.

5
Ibid, para. 6.218.

6
Ibid, para. 6.222.

7
Ibid, para. 6.222.
26

2.13. Compensation of employees: "Compensation of employees is defined as the total remuneration, in

cash or in kind, payable by an enterprise to an employee in return for work done by the latter during the
accounting period.... No compensation of employees is payable in respect of unpaid work undertaken
voluntarily, including the work done by members of a household within an unincorporated enterprise owned
by the same household. Compensation of employees does not include any taxes payable by the employer on
the wage and salary bill - for example, a payroll tax. Such taxes are treated as taxes on production in the same
way as taxes on buildings, land or other assets used in production.”8

2.14. Taxes on production and imports: Taxes on production and imports include taxes on products and
other taxes on production.

- Taxes on products: Taxes on products are “payable on goods and services when they are produced,
delivered, sold, transferred or otherwise disposed of by their producers; they include taxes and duties
on imports that become payable when goods enter the economic territory by crossing the frontier or
when services are delivered to resident units by non-resident units; when outputs are valued at basic
prices, taxes on domestically produced products are not recorded in the accounts of the System as
being payable by their producers.”9 Taxes on products can consist of: value added type taxes (VAT),
taxes and duties on imports excluding VAT, export taxes, taxes on products except taxes listed above.
These taxes are called commodity taxes in the 1968 SNA. Subsidies on products are defined in a
similar way.

- Other taxes on production: Other taxes on production consist “mainly of taxes on the ownership
or use of land, buildings or other assets used in production or on the labour employed, or
compensation of employees paid. Taxes on the personal use of vehicles, etc. by households are
recorded under current taxes on income, wealth, etc.”10. Examples of other taxes on production are
taxes payable by producers for business licenses, payroll taxes, stamp duties, etc. These taxes are not
proportional to the value of goods and services produced. In the 1968 SNA, these taxes are called
other indirect taxes.

2.15. Subsidies on production and imports: "Subsidies are current unrequited payments that government
units, including non-resident government units, make to enterprises on the basis of the levels of their
production activities or the quantities or values of the goods or services which they produce, sell or imports."11.
"Subsidies are not payable to final consumers, and current transfers that governments make directly to
households as consumers are treated as social benefits. Subsidies also do not include grants that governments
may make to enterprises in order to finance their capital formation, or compensate them for damage to their

8
Ibid, para. 7.21.

9
Ibid, para. 7.49.

10
Ibid., para. 7.49.

11
Ibid., para. 7.71.
 27

capital assets, such grants being treated as capital transfers.”12. Subsidies on production and imports are
divided into two distinct items:

- Subsidies on products: "A subsidy on a product is a subsidy payable per unit of a good or
service.”13

- Other subsidies on production: These “consist of subsidies except subsidies on products which
resident enterprises may receive as a consequence of engaging in production.”14 Examples are
subsidies on payroll or workforce, subsidies to reduce pollution.

2.16. Gross mixed income: "Mixed income contains an unknown element of remuneration for work done
by the owner of the enterprise, or other members of the same household, as well as the [operating] surplus
accruing from production.”15 Mixed income is the term reserved for the balancing item in the generation of
income account for "unincorporated enterprises owned by members of households either individually or in
partnership with others in which the owners, or other members of their households, may work without
receiving a wage or salary. Owners of such enterprises must be self-employed: those with paid employees
are employers, while those without paid employees are own-account workers. In a few cases it may be possible
to estimate the wage or salary element implicitly included within mixed income, but there is usually not enough
information about the number of hours worked or appropriate rates of remuneration for values to be imputed
systematically.”16 The concept of mixed income is necessary because, practically, satisfactory estimates of
compensation of employees for owners are not possible. Dwelling services for own consumption produced
by owner-occupiers do no generate mixed incomes since “there is no labour input into the production of the
services of owner-occupied dwellings so that any surplus arising is operating surplus.”17

2.17. Consumption of fixed capital: "Consumption of fixed capital is a cost of production. It may be
defined in general terms as the decline, during the course of the accounting period, in the current value of the
stock of fixed assets owned and used by a producer as a result of physical deterioration, normal obsolescence
or normal accidental damage. It excludes the value of fixed assets destroyed by acts of war or exceptional
events such as major natural disasters which occur very infrequently.”18 Consumption of fixed capital is
therefore not the same as depreciation allowed for tax purposes. Consumption of fixed capital is commonly
calculated by the perpetual inventory method (PIM).

B. Structure of SUT

12
Ibid., para. 7.72.

13
Ibid,, para. 7.73.

14
Ibid., para. 7.79.

15
Ibid., para. 7.85.
16
Ibid., para. 7.85.

17
Ibid., para. 4.150.

18
Ibid., para. 6.179.
28

2.18. The SUT in the revised SNA are presented as tables 2.1 and 2.2. A more elaborate framework with
numerical examples can be found in the System of National Accounts 1993 (see footnote 1). These tables are
shown in general form so that they can be used later in the discussion of SUT and production accounts in
constant prices; this is the main reason for showing a column of price indexes in table 2.1(column 10) which
normally should not be there. The same tables but with numerical examples can be found in tables 3.1 and 3.2
in chapter III on valuation. Readers are advised to read tables 2.1 and 2.2 together with tables 3.1 and 3.2 in
chapter III. The structure of tables 2.1 and 2.2 also differs slightly from table 15.1 on page 350 of the System
of National Accounts where taxes minus subsidies on products, trade and transport margins and total product
supply at purchasers' prices are shown on the left hand side of the supply and use tables.
2.19. In the SUT, every activity or industry is classified by the way its outputs are used (market, for own
final use or other non-market) and its aggregate industry outputs are similarly classified. The design of tables
2.1 and 2.2 is a simplified version of the SUT in SNA 1993, which does not take into account the identification
of industries by types of uses of the outputs they produce so as to assist the explanation of the double deflation
method in chapter XI . Activities may be classified into three groupings. In tables 2.1 and 2.2, goods, market
services and other non-market services. The first two groups produce products for sale so they have market
prices or equivalent . These two groups also include producers for own final uses. The last group representing
government services and services provided by non-profit institutions serving households has neither market
prices nor equivalent. Products are similarly classified.

2.20. The SUT framework has two tables: the supply table and the use table that are closely linked together.
The supply table shows both the value of different products at basic prices produced in every industry and the
total supply of every product in both basic prices and purchasers' prices. The use table shows both the cost
of production in every industry, and the use of every product at purchasers' prices in the economy. In both
tables, products are presented as rows better to illustrate the balancing of their supply and use in the SUT of
the SNA. The supply of every product must be equal to the use of that product when measured in the same
price, and the output of an industry must be equal to its cost of production: these two principles are used in
balancing the supply and use tables.

1. The supply table

2.21. In the supply table 2.1 (see also table 3.1), goods and services produced in the economy are measured
at basic prices. Basic prices do not include trade and transport margins and taxes on products. Values in basic
prices of products produced are shown in columns 1-3 and rows 1-3. The first row shows the supply of goods
by different industries from column 1-3. The total supply of goods at basic prices is obtained in column 6 after
imports c.i.f. are added to the supply domestically produced at basic prices. So:

SB1 = X11 + X12 + X13 + M1

2.22. Imports c.i.f. which exclude import duties are treated as equivalent to basic prices. The total supply
of goods at purchasers' prices is obtained by adding to the total supply of goods at basic prices their trade and
transport margins, and taxes minus subsidies on products including imports.

Thus: SP1 = SB1 + TM + TX1

 Table 2.1. The supply table

Industries with output at basic prices

Imports c.i.f. C.i.f./f.o.b Total product Trade and Taxes minus Total product Price
(Total f.o.b) . supply transport subsides supply indexes*
: Goods Market Other adjust- at basic prices margins on at purchasers'
services non-market ment products prices
services

Products: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Goods X11 X12 X13 M1 SB1=(1+.+5) TM TX1 SP1=(6+.+8) P1

Market services X21 X22 X23 M2 -ADJ SB2=(1+.+5) -TM TX2 SP2=(6+.+8) P2
Other non-market services X31 X32 X33 SB3=(1+.+5) TX3 SP3=(6+.+8)

C.i.f./f.o.b. adjustment -ADJ +ADJ 0

Purchases of residents abroad R R

Total industry output at basic I1=X11+ I2=X12+ I3=X13+

prices X21+X31 X22+X32 X23+X33

Other column total M 0 SB 0 TX SP

* Price indexes are irrelevant for this chapter but will be used to describe the double deflation method in chapter XI.
 Table 2.2 The use table at purchasers' prices

Industries Exports. Household final Government final Gross capital Total product
Intermediate consumption f.o.b. expenditure* expenditure formation** uses

Goods Market Other non-market

services services

(1) (2) (3) (4) (5) (6) (7) (8)

Goods U11 U12 U13 E1 HC1 K1 U1=(1+.+7)

Products Market services U21 U22 U23 E2 HC2 K2 U2=(1+.+7)
Other non-market services HC3 GC U3=(1+.+7)

Purchases of residents abroad R R

Purchases of non-residents at home NR -NR 0

Gross value added at basic prices V1 V2 V3

Compensation of employees W1 W2 W3
Other taxes on production OT1 OT2 OT3
Consumption of fixed capital CC1 CC2 CC3
Operating surplus/gross mixed income OS1 OS2

Total industry output at basic prices I1 I2 I3

Other column total E HC GC K U

Price indexes P3

* In order to reduce the size of the table, household consumption also includes final consumption of non-profit institutions serving households (NPISHs).
** Gross capital formation includes gross fixed capital formation, change in inventories and acquisition less disposal of valuables,
 31

2.23. The supply of market services and non-market services is similarly treated. To obtain the value of
output produced by each industry, we only need to sum the values of different products produced by the
industry as shown in the corresponding column. The value of the industry producing goods at basic prices is
equal to:

I1 = X11 + X21 + X31

2.24. The supply table needs one adjustment that can be explained immediately: the c.i.f./f.o.b. adjustment.
Other adjustments are taken in both the supply and the use tables or can only be understood in the context of
both tables and will be explained later.

The c.i.f./f.o.b. adjustment

2.25. As we know M1, the value of imported goods c.i.f. includes the value of:

- Imported goods f.o.b;

- Transport services rendered by both resident and non-resident transporters;
- Insurance services rendered by both resident and non-resident insurers.

2.26. The two latter services when rendered by non-residents are however already included in M2 which
shows the total value of imports of market services. In addition, the value of transport and insurance services
rendered by residents is domestic output and thus should not be treated as imports. If no adjustment is
introduced, imports are obviously higher than they should be by the total value of transport and insurance
services rendered by both residents and non-residents. One must be careful in the treatment of the value of
transport and insurance services on imports rendered by residents. If these services are bought and paid for
by non-resident exporters to resident carriers and insurers, their value will be recorded as export of services
in the balance of payments, and therefore must be excluded from export of services. To sum up, SUT has to
satisfy two conditions:

- Each imported product must be valued c.i.f. in the supply table since it is equivalent to the
basic value of the same domestic goods;
- The total value of imports must be valued f.o.b. since this is the true value of imports as
explained above.

2.27. To adjust for the above problem, a column of adjustment is introduced and an adjustment value (ADJ)
is introduced separately for transport services and insurance services though in table 2.1 the two commodities
are aggregated together in the group of market services. This adjustment value should be as follows:

ADJ = Value of transport services or insurance services (or both) rendered by both resident and non-
resident producers

By doing this the total value of imports is measured f.o.b., i.e.:

Imports f.o.b. = M1 c.i.f. + M2 - ADJ

The row of c.i.f./f.o.b. adjustment is introduced so that its total row value is equal to 0, like the total value of
the corresponding column.
32

2.28. In the 1968 SNA, the total value of imports is valued c.i.f. so ADJ has to be included in exports to
keep the balance of trade accurate. However, that treatment obviously distorts the value of both imports and
exports.

2. The use table

2.29. The use table shows in a column the cost of production in the corresponding industry. All
intermediate costs are measured in purchasers' prices. Thus U11 to U23 are in purchaser's prices. The
purchaser’s price is the price that the purchaser has to actually pay for the product unit. If the taxes paid on
the purchased products can be deducted by the purchaser when used as inputs in production, the deductible
value added tax (VAT) has to be excluded from the purchasers' price. This applies only to VAT countries.
For every industry, the total industry output at basic prices, i.e from I1 to I3 has to be calculated from the
supply table. Given this industry output and intermediate consumption obtained in the use table, it is possible
to measure value added as a residual. Thus for the goods industry,

V1 = I1 - (U11+ U21)

and then given other components of value added, operating surplus is measured residually.

2.30. Every product is shown as consumed in production, exports, household final expenditure, government
final expenditure and gross capital formation. The sum along every row is equal to the total use of the product
at purchasers' prices. So:

U1 = U11 + U12 + U13 + E1 + HC1 + K1

and then U1 in the use table must equal SP1 in the supply table. This is true for every U and SP.

2.31. The use of the product other non-market services as intermediate consumption by business is also
possible (for example, payment for some government services) but it is very small and assumed to be zero here
just for the sake of highlighting the presentation.

2.32. GC in column 6 of table 2.2 stands for government final consumption expenditure. For the purpose
of deriving detailed inflation indexes, government final expenditure GC should be split into two categories:

- The collective expenditure that benefits society at large; and

- The individual expenditure that benefits directly the individuals in the society Households
may pay part of it. This part which is normally called user charges has to be included in HC3
of column 5 and not included in GC in column 6.

HC3 in column 5 stands for final consumption of non-market services of non-profit institutions serving
households (which is also equal to its output). Thus HC3 includes both the part that is paid by households and
the part that is provided free by the Government or the non-profit institutions.

2.33. Gross capital formation (column 7) may be broken down by industries and/or institutional sectors:
by industries to calculate gross fixed assets by industry; by institutional sector for the purpose of integrated
institutional accounts. In the SNA, gross fixed capital formation is recommended to be classified by industry
only. Changes in inventories are normally prepared for the economy as a whole since data for changes in
 33

inventories by industry are more difficult to collect. Acquisition less disposal of valuables cannot be classified
by industry since they are owned by enterprises and do not relate to production.

2.34. In the supply and use tables there are three more adjustments to be explained:

(a) The adjustment for trade and transport margins

2.35. As seen in the use table, trade services included in the market services grouping in row 2 show only
those that are consumed directly by producers and final users. They do not include trade and transport margins
which are instead included in the value of the goods at purchasers' prices shown in the first row. Therefore,
trade and transport margins are not included in U2, the total use of market services in the use table (column
8). Consequently, in the supply table, trade and transport margins should be deducted from the total supply
of market services in row 2. This is done by entering trade and transport margins (TM) as a negative number
in column 7, row 2 in order to balance the supply and use of trade and transport services at purchasers' prices.

(b) The adjustment for purchases of residents abroad

2.36. Purchases of residents abroad are treated as both imports and household final expenditure by the SNA,
assuming that these purchases are mostly by household residents. Thus a value of R has to be entered in the
import column of the supply table and also entered in the column of household final expenditure in the use
table. The European System of National Accounts (ESA) refines the SNA treatment further and recommends
that these purchases be classified by product and two categories must be distinguished: R1, expenditure by
business travellers which are treated as intermediate consumption and R2, expenditure by households which
are treated as household final consumption. In the ESA, adjustments, if necessary, are introduced directly into
the import column 4 of the supply table and the household final expenditure (column 5) of the use table. The
ESA treatment will require expensive surveys to obtain detailed information on purchases by households and
business.

(c) The adjustment for purchases of non-residents at home

2.37. Purchases of non-residents at home should be treated as exports and if they are included in the column
of household final expenditure through the process of balancing for which this column is normally treated as
a residual, these purchases have to be deducted from household final expenditure. Thus a value of NR is
entered in the export column (column 4) and -NR is entered in the column of household final expenditure
(column 5). Similarly to the treatment of purchases abroad by residents, the European System of National
Accounts (ESA) recommends the classification of purchases by non-residents by products with adjustments
made directly on the column of household final expenditure.

3. The advantage of using basic prices in SUT

2.38. In SUT, one may use either basic prices or producer prices to measure industry outputs. However, as
will be shown below, it is much better to use basic prices. In the SNA supply table 2.1 taxes and subsidies on
products including import duties and trade margins are shown for every commodity. The link between taxes
paid and subsidies received and the values of products used in the economy can clearly be seen. In the tables
that have industry outputs measured at producer prices, as is currently done by many countries, there are two
alternative treatments. The first one is to include in producer prices taxes less subsidies on products only that
are collected directly by the producers for the Government. But in so doing, most of the taxes on products
collected through wholesaling and retailing would appear in the wholesale and retail trade industries. This
34

method shows the link between the output of each industry (which is the sum of possibly many products
produced by the industry) and product taxes paid, but the direct link between an individual product and the
taxes paid on it is not shown. The other alternative is to itemize taxes and subsidies paid by each industry by
product, i.e. to obtain a matrix of taxes on industry output classified by product and by industry. This method
which shows the direct link between taxes and products produced by each industry requires much more
information than does the SUT.

4. Balances between value added and final demand in the use table

2.39. Undoubtedly, SUT is an extremely useful device to arrange basic statistics not only for the derivation
of the symmetric I/O table (SIOT) that will be used for economic analysis as presented in chapter I, but also
for the compilation of value added by industry, final demand by product and by institutional sector in both
current and constant prices in a systematic way. In the following paragraphs the basic link between value added
and final demand will be further explored.

2.40. The gross domestic product (GDP) is the net contribution of economic activities to incomes (i.e. value
added) which are then used for final demand, therefore total incomes must equal total net final demand. The
relationship of value added and final demand is shown in table 2.3 below.

2.41. In the I/O framework, the GDP can be measured as sum of value added or of primary incomes or as
net final demand.

(a) GDP as sum of value added (the production approach)

2.42. GDP = Total industry output at basic prices

GDP = Total industry output at basic prices - Total industry intermediate consumption in
purchasers' prices + Taxes less subsidies on products.

In table 2.1. and table 2.2,

GDP = (I1 - U11 - U21) + (I2 - U12 - U22) + (I3 - U13 - U13) + TX = V1 + V2 + V3 + TX

In the supply and use tables 3.1 and 3.2 of chapter III, value added at basic prices can be found in row 9 of
the use table 3.2 and taxes less subsidies on products can be obtained in row 7, columns 9 and 10 of the supply
table 3.1.

Value added at basic prices = 108 + 49 + 39 = 196

Taxes less subsidies on products = 3 + 17 = 20

GDP = 196 + 20 = 216

2.43. This approach allows the calculation of value added at basic prices originating from each industry by
subtracting the intermediate consumption of each industry at purchasers' prices from its output at basic prices.
Information on industry output and intermediate consumption is available in the use table. Information on
taxes less subsidies on products is available in the supply table when deriving total supply at purchasers' prices.
 35

(b) GDP as sum of primary incomes (the income approach)

2.44. "GDP is also equal to the sum of primary incomes distributed by resident producer units.”19 Values
added at basic prices are calculated as the sum of compensation of employees, other taxes less subsidies on
production, gross mixed incomes, and gross operating surplus. The gross operating surplus of each
establishment or industry cannot be directly estimated because the operating surplus of each industry is, by
definition, calculated as a residual. However, the gross operating surplus of incorporated enterprises may be
estimated directly. It is equal to:

(1) Additions to retained earnings

(2) Plus Depreciation and depletion
(3) Plus Bad debts (not yet written off)
(4) Plus Net property incomes payable including:
Net rents on non-produced assets
Net interest payable
Dividends
(5) Plus Net current transfers payable
(6) Plus Net capital transfers payable
(7) Minus Gains (net of loss) on sales on fixed assets and securities

The net concepts in items 2-6 refer to receivables less payables. They are considered costs in business
accounting but part of gross operating surplus in national accounts. Therefore, they need to be added to
additions to retained earnings (item 1) in order to obtain gross operating surplus. Item 7 is part of incomes in
business accounting, but it is only capital gains due to changes in prices in national accounts and therefore
must be subtracted.

19
Ibid., para. 2.176
36

Table 2.3 Relationships in national accounts aggregates

Gross Domestic Product as net contribution to Gross Domestic Product as net final demand
incomes

Gross value added at basic prices Household final consumption expenditure

Compensation of employees Final consumption expenditure of non-profit institutions
Other taxes less subsidies on production serving households (individual)
Consumption of fixed capital Government final consumption expenditure
Operating surplus/Mixed incomes Collective
Taxes less subsidies on products Individual
Gross capital formation
Gross fixed capital formation
Changes in inventories
Acquisitions less disposal of valuables
Exports f.o.b.
Minus imports f.o.b.

2.45. Many countries have adopted this approach to measure GDP without using the full information in
SUT. This means that supplies of goods and services, product by product, need not be balanced with uses.
Only a balance of overall production incomes and final demand is obtained as can be seen in table 2.3. The
approach can provide value added for aggregated industries in a summarized way. However, many income
items listed from (1) to (7) such as additions to retained earnings, interest, etc., cannot be split into different
industries. Because of that, the approach does not provide value added by detailed industries and full
information on production accounts. In addition, the income approach does not work for small household
activities whose owners either do not keep full business accounts or are not able to separate income from
production from property income or current transfers. In this case, the production approach is more
appropriate. In conclusion, the income approach is not a pure approach and it is restricted to the measurement
of value added by institutional sectors and total GDP only. Without the supply and use tables, if discrepancies
arise, it will be very difficult to detect their origin and to reduce them. In contrast, using the income approach
jointly with a supply-and-use table would allow the compilation of components of incomes (compensation of
employees, other taxes less subsidies, operating surplus) by industry and the link between value added by
industry with value added by institutional sector discussed in section C.

(c) GDP as net final demand (the expenditure approach)

2.46. "GDP is also equal to the sum of the final uses of goods and services (all uses except intermediate
consumption) measured in purchasers' prices, less the value of imports of goods and services.”20 Final uses
or final demand are equal to: household final expenditure + final expenditure of non-profit institutions serving
households, the general government sector + gross capital formation + exports f.o.b. When final demand is

20
Ibid, para. 2.173.
 37

net of imports, it is called net final demand. Final demand is not the terminology used in the SNA but it is
convenient to refer to an analysis of future demand. For accounting purposes, demand and uses are used
interchangeably. Net final demand is a concept necessary for input-output analysis using the Leontief inverse.
It was introduced in chapter I and will be used frequently later in the Handbook.

In tables 2.1 and 2.2,

GDP = HC + GC + K + E - M

In the supply and use tables 3.1 and 3.2 of chapter III, total imports are obtained from row 7 and column 5 of
the supply table and other components of final demand are obtained from row 7, columns 5-8 of the use table.

GDP = 153 + 10 + 40 + 38 - 25 = 216

(d) The commodity flow approach

2.47. The SNA SUT framework combines all three approaches in estimating national accounts statistics by
balancing the supply and use of every product using the use and supply tables and in this way it has a
consistent way to cross-check information at a very detailed level. This SUT approach is more popularly called
the commodity flow approach.

5. Final consumption expenditure versus actual final consumption

2.48. A new concept of actual final consumption has been introduced in the revised SNA that is very useful
in studying the behaviour of households and in comparing their actual consumption among nations. The
relationship between the concept of final consumption expenditure and actual final consumption is presented
in table 2.4. In table 2.3, final expenditure was broken down by institutional sectors that actually spent for
different purposes. The Government spends for collective purposes such as public administration, defense,
security, general health improvement, etc. that benefit the society as a whole but not specific individuals.
Besides expenditure for collective benefit, the government also spends to benefit specific individuals for
instance health care, education, food aid, etc. The non-profit institutions serving households are defined by the
SNA to spend for the benefit of individuals only. Final consumption expenditure to benefit individuals of the
general government sector and non-profit institutions serving households can be added to household final
consumption expenditure to derive actual individual final consumption. The latter concept measures the actual
final consumption of households whether paid by themselves or by other institutions. Full elaboration of the
concept and compilation of final consumption expenditure and actual final consumption can be found in
chapter VII.
38

2.4 Relationship between two concepts of final consumption

Final consumption expenditure Actual final consumption

Households final consumtion Actual final consumption of households (or actual individual
expenditure consumption)
Government final consumption Households final consumption expenditure
expenditure
Collective Individual government final consumption expenditure
Individual Final consumption expenditure of NPISHs
Non-profit institutions serving Actual final consumption of general government (or actual
households' final consumption collective consumption)
expenditure
Collective government final consumption expenditure

C. Link between SUT and production accounts of institutional sectors in the SNA

2.49. National accounting studies not only value added generated by industries but also value added and
incomes generated and distributed among institutional sectors (see section D of this chapter for definition).
For example, a non-financial enterprise that manages or "owns" an activity classified as an industry in SUT
would have to pay compensation to its workers and taxes to the Government. Even the operating surplus
generated by the industry in question and by other industries that the enterprise manages must also be
distributed to other institutional sectors: interests on its loans to enterprises in the financial sector, dividends
to other enterprises or to households that own shares of the enterprise, etc. In addition, the same enterprise
may also receive income from its ownership of shares in other enterprises or of deposits at financial
institutions. It is thus not possible to trace the redistribution process on the basis of the industries but only
of the enterprises that manage or "own" them. Distribution and redistribution of incomes belong to another
part of the structure of national accounts, namely the distribution and use of income accounts. To facilitate
the integration of production accounts classified by industry to the distribution and use-of-income accounts
and other sequence of accounts in the SNA, production accounts by industry must be cross-classified by
institutional sectors. The cross-classification scheme is presented in table 2.5. Only aggregate values of outputs
and intermediate consumption of industries or establishments are needed. Information on establishments is
first to be classified acording their institutional sectors. To obtain industry outputs, intermediate consumption
and value added by institutional sectors, one simply adds up the corresponding information on specific
institutional sectors. For example, in table 2.5, the value added by the non-financial sector is the sum of value
added shown in columns 1-3. This scheme is implementable only if every establishment covered in production
censuses is identified by the type of institutional sector to which the enterprise that owns the establishment
belongs. Practically, production censuses should be designed to let enterprises fill in requested information for
all the establishments they own or to let the establishments fill in information under the supervision of the
enterprises.

2.50. In the above scheme, industries must be classified into one of five institutional sectors, or sub-sectors
if each institutional sector is divided further. In the scheme in table 2.5
 39

- The non-financial and financial corporations sectors include (i) all government and privately-
owned corporations and quasi-corporations that produce goods and services principally for
sale on the market but may have activities that produce goods and services for their own final
uses and other non-market uses and (ii) all private non-profit institutions serving corporations
and quasi-corporations;

- The household sector includes all unincorporated enterprises that produce for both the market
and their own final uses and household activities that perform mainly for their own final uses;

- The general government sector includes all central, state or local government units performing
other non-market activities, all social security funds, all non-profit institutions controlled and
financed by government units and all unincorporated government enterprises producing
mainly other non-market goods and services;

- The non-profit institutions’ sector serving households includes only private non-profit
institutions that produce mainly non-market goods and services to households.

2.51. In the table 2.5 scheme, gross fixed capital formation, opening stocks of produced fixed assets, and
closing produced fixed assets by industry may also be cross-classified by institutional sector to be used in other
SNA accounts and in the dynamic analysis of the economy. Closing produced fixed assets are equal to
opening produced fixed assets plus changes in assets. Changes in assets include:

- Changes in volume of produced fixed assets due to transactions which are actually gross fixed
capital formation;

- Changes in volume of produced fixed assets not due to transactions (such as unanticipated
destruction or disappearances); and

- Changes in produced fixed assets due to changes in prices.

The cross-classification scheme may also be applied to non-produced assets that are used in production.

2.52. The cross-classification matrix, if relationships are assumed not to change in the short run, can be used
to link the effects of production on industries to those on institutional sectors.

2.53. It should be pointed out that gross capital formation and fixed produced assets are classified by
ownership; so if owned by government agents and classified as producers of government services but
benefiting mainly other producers, they may need to be reclassified wherever appropriate for production
analysis. For example, roads are classified as the assets of producers of government services but in fact they
benefit directly producers of transport services.
 Table 2.5

Cross-classification of value added by industry and institutional sectors

Non-financial corporations sector Financial corporations sector General government sector Household sector (classified Non-profit institutions serving
(classified by ISIC) (classified by ISIC) (classified by ISIC) by ISIC) households (classified by ISIC)

Total Ind. ... Ind. Total . . . Total . . . Total . . . Total Ind. 1 ... Ind.n
1 n

(1) (2) (3)

Industry output at basic prices

Intermediate consumption at
purchasers' prices

Value added at basic prices

Compensation of
employees

Other taxes less subsidies

on production

Consumption of fixed
capital

Operating surplus/Mixed
incomes

Opening stocks of produced

fixed assets

Gross fixed capital formation

Other changes in volume

Revaluation

Closing stocks of produced fixed

assets
 41

D. The statistical units and classifications schemes for data collection

2.54. For the purpose of compiling the SNA SUT as shown in tables 2.1 and 2.2, it is important to decide
on the statistical units and on the schemes to classify them. Table 2.6 shows the elements of SNA and the
classification schemes they need for classification. In addition, items in the first two columns may also be
cross-classified with the classifications of the last column as discussed previously.

Table 2.6. Uses of classification schemes

Industrial classification Central product classification Institutional

(ISIC) (CPC) classification

Economic activities Intermediate consumption Enterprises (smallest

legal unit)
Industries Final demand Government units
Industry output Final consumption expenditure Non-profit institutions
serving households
Value added Gross capital formation Households
Exports
Imports
Product output

1. Statistical units and classification scheme for an activity or industry

2.55. The statistical unit for activities or industries recommended by the SNA is the establishment. An
operational definition of the establishment unit is a production unit consisting of either "an enterprise, or a part
of an enterprise, that is situated in a single location and in which only a single (non-ancillary) productive
activity ... accounts for most of the value added.”21 (SNA, para. 5.21). Under this operational definition, an
establishment may engage "in one or more secondary activities, [which] should be on a small scale compared
with the principal activity. If a secondary activity within an enterprise is as important, or nearly as important,
as the principal activity, then that activity should be treated as taking place within a separate establishment
from that in which the principal activity takes places. The definition of an establishment does not permit an
ancillary activity to constitute an establishment on its own." (SNA, para. 5.22). If an enterprise has more than
one establishment, normally identified with a separate location or book of manufacturing business accounts,
data manipulation is needed to allocate the managing and other costs shared by all establishments in the
enterprise. In many cases, the headquarters and warehouses may be in separate location, but not classified
separately as establishments by the SNA; they are treated only as ancillary activities.

2.56. The categories into which data on activities or industries may be classified are in the United Nations'
publication entitled International Standard Industrial Classification. (ST/ESA/STAT/SER.M/41 Rev.3)
These categories are used to classify establishments into various kinds of economic activities. At present, the

21
Ibid, para. 5.21.
42

ISIC contains 17 major sections (with alphabet coding), 60 divisions (with two-digit coding), 169 groups (with
three-digit coding) and 291 classes (with four-digit coding). Thus with ISIC it is possible to classify activities
into 291 industries. National classification standards may be elaborated from the standards recommended by
the United Nations. If a country prefers to have more details, it may further break down a certain class of
activities into various sub-classes, and on the other hand, if it requires less information it may aggregate many
classes into one.

2.57. The main purpose of the ISIC is to provide a hierarchical set of categories of economic activities which
can be utilized when dissecting production statistics according to kinds of activities. The definition of ISIC
categories is strongly linked with the way the economic process is organized, with each unit as homogeneous
as possible. For this reason, the groups and classes of the ISIC are in some cases too detailed to be used in
classifying enterprises or similar institutional units by kind of economic activity since a significant number of
enterprises will own establishments which engage in a range of activities spanning more than one ISIC class.
Thus, establishments are recommended as the statistical units used in identifying economic activities.

2.58. In SUT, "industry" refers to the kind of activity into which all units which engage primarily in that
kind of activity are aggregated.

2.59. Appendix A to this chapter gives the full list of divisions (two-digit level) of the ISIC classification
of activities. The detailed definition of each activity at four-digit level can be found in the United Nations
publication referred to in paragraph 2.56 above.

2.60. Appendix B which gives the classification of products is discussed below in paragraphs 2.61-2.62.
Appendix C gives the Classification of the Functions of Government (COFOG) which is important to split
expenditures by producers of government services into various kinds of activities and to identify the items that
can be classified as part of individual final consumption.

2. Statistical units and classification scheme for products

2.61. The statistical unit for products is a unit of homogeneous goods and services and the Central Product
Classification version 1.022 is recommended for the classification of products which are either outputs of
domestic production activities or imports from non-resident sources.

2.62. The product dimension of the SNA input-output framework may be based on Central Product
Classification (CPC) covering both goods and services. It is exhaustive and all products are mutually
exclusive. In principle, CPC is meant to be used for all kinds of statistics which need product detail. It was
developed primarily to include categories for everything that can be the object of a domestic and international
transaction. This means that in CPC, not only are all products which are an output of an economic activity
represented but also the purchase or sale of land are covered or similar transactions as well as those that arise
from legal contracts (such as patents, licenses and copyrights), though some of these assets are not regarded
as goods and services in the SNA.

2.63. Appendix B includes only the aggregated list of CPC by major division (with two-digit coding) with
correspondence with ISIC, Rev. 3. The detailed classification, however, consists of 10 sections (one-digit
coding), 69 divisions (two-digit coding), 291 groups (three-digit coding), 1,036 classes (four-digit coding) and

22
Central Product Classification, version 1.0, Series M, No. 77, United Nations, 1997.
 43

1,787 sub-classes (five-digit coding). The full classification can be found in the United Nations publication
referred to above. In theory, the classification coding can accommodate 65,610 categories. Thus, the CPC
allows for expanding classification into a more detailed one.

3. Correspondences between ISIC and CPC

2.64. In obtaining the symmetrical input-output table (SIOT) of either product-by-product or industry-by-
industry tables from the supply-and-use tables of SUT, the classification of economic activities must be the
same as the classification of products. Thus a correspondence between ISIC and CPC is needed. Theoretically,
these two classifications need not match. The CPC is a system of categories covering both goods and services
in which homogeneity within categories is maximized whereas the ISIC covers industrial activities of similar
types. The principles of classification used by the CPC are as follows:

(a) For transportable goods, categories of products should be based on the physical properties and
the intrinsic nature of the products, i.e. the raw materials of which they are made, their stage
of production, the use they are intended for, the prices at which they are sold, whether or not
they can be stored, etc.;

(b) Individual goods and services as far as possible should contain only goods and services which
are produced by a single industry.

2.65. With the first principle as the main criterion, different products may come out of the same industry.
For example, meat and hides are both produced by a slaughterhouse but they are put into entirely separate CPC
categories: unprocessed hides are considered animal materials and are classified in section 0 "Agriculture,
forestry and fishery products", whereas meat is classified in section 2 among food products. Similar problems
also arise in agriculture, mining, manufacturing, etc. A farm may produce farm products and also provide
farming services. A particular industry may produce a specific product and provide repair, maintenance
services or manufacturing services on a fee or contract basis. These goods and services are classified under
different parts of the CPC. These services are given a separate division (86). The CPC provides a
correspondence between itself and ISIC at the four-digit level but the correspondence is only a general
guideline because there can be no fixed correspondence at an aggregate level. For example, common salt (Na
Cl) can be produced in mining - a mining product, and also in food production (sea salt) - a food product. At
the most detailed level, an approximate correspondence is possible given that the most common production
processes used in a particular country are taken into account. For the member countries of the European Union,
Classification of Products according to Activities (CPA) and the detailed Classification of Products for
Community Surveys on Manufacturing Industries (PRODCOM) have been used as national classifications
since 1993 in order to build a better correspondence between product and activity classifications.
44

4. Statistical units for institutional sectors

2.66. An institutional unit that is used frequently in this chapter can be a household, or a legal, social entity.
It is defined as "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."23 The SNA provides five broad
categories of institutional sectors to classify institutional units: the non-financial corporations sector, the
financial corporations sector, the household sector, the non-profit-institutions-serving- households sector and
the general government sector.

2.67. In the area of business, an institutional unit is normally a legal entity that is collectively owned by
shareholders with the authority to appoint directors who are responsible for its general management, can enter
into contracts; receive and dispose of its income, maintain an independent, complete set of accounting
records, including profit and loss accounts and balance sheets; at the same time, the unit has its own rights,
privileges and liabilities as distinct from its owners. All corporate enterprises or corporations are classified into
either the non-financial sector or the financial sector. Quasi-corporate enterprises are unincorporated but
operate as if they are de facto corporations to their owners and therefore keep complete sets of accounts. They
are treated as corporate enterprises in the SNA. Cooperatives and limited liability partnerships are also treated
by the SNA as corporations. Non-profit institutions serving business interests are classified in the institutional
sector they serve.

2.68. Private non-profit institutional units serving households are also legal entities which are organized like
profit-oriented corporations. Their aims are however to produce mainly non-market goods and services to serve
households. The non-profit institutional sector serving households excludes the private non-profit institutional
units serving businesses and the units controlled and financed by the Government.

2.69. In the government sphere, a government institutional unit is a unique kind of legal entity established
by political processes with legislative, judicial or executive authority over other institutional units within a
given region. Its principal function is to assume responsibility for the provision of other non-market goods and
services to the community or to individual households. It has the authority to raise funds by collecting taxes
or compulsory transfers from other institutional units. By this definition, all corporations which produce
mainly market goods and services are classified into either the non-financial or the financial corporations
sectors. The general government sector includes only government units providing other non-market services
and government unincorporated enterprises (i.e. without full sets of accounts) owned by government units that
may produce both market and non-market goods and services. An example of a government unincorporated
enterprise is a unit within a governmental branch producing government documents. The classification of the
functions of government (COFOG) and the relationship between ISIC and COFOG are helpful in classifying
government activities both in terms of establishment-type units and institutional units and in the preparation
of two types of government final consumption: individual and collective final consumption. It should be noted
that COFOG is being revised to take into account changes in the SNA 1993.

23
System of National Accounts 1993, para. 4.2.
 45

Appendix A

INDUSTRY CLASSIFICATION

INTERNATIONAL STANDARD OF INDUSTRIAL CLASSIFICATION

OF ALL ECONOMIC ACTIVITIES

LIST OF TABULATION CATEGORIES AND DIVISIONS

(BROAD STRUCTURE)

_______________________________________________________________________________________________
Tabulation
categories Division Description

A AGRICULTURE, HUNTING AND FORESTRY

01 Agriculture, hunting and related service activities

02 Forestry, logging and related service activities

B FISHING

05 Fishing, operation of fish hatcheries and fish farms; service activities incidental to fishing

C MINING AND QUARRYING

10 Mining of coal and lignite; extraction of peat

11 Extraction of crude petroleum and natural gas; service activities incidental to oil and gas
extraction excluding surveying
12 Mining of uranium and thorium ores
13 Mining of metal ores
14 Other mining and quarrying

D MANUFACTURING

15 Manufacture of food products and beverages

16 Manufacture of tobacco products
17 Manufacture of textiles
18 Manufacture of wearing apparel; dressing and dyeing of fur
19 Tanning and dressing of leather; manufacture of luggage, handbags, saddlery, harness and
footwear
20 Manufacture of wood and of products of wood and cork, except furniture; manufacture of
articles of straw and plaiting materials
21 Manufacture of paper and paper products
22 Publishing, printing and reproduction of recorded media
23 Manufacture of coke, refined petroleum products and nuclear fuel
24 Manufacture of chemicals and chemical products
25 Manufacture of rubber and plastics products
26 Manufacture of other non-metallic mineral products
27 Manufacture of basic metals
28 Manufacture of fabricated metal products, except machinery and equipment
29 Manufacture of machinery and equipment n.e.c.
30 Manufacture of office, accounting and computing machinery
46

31 Manufacture of electrical machinery and apparatus n.e.c.

32 Manufacture of radio, television and communication equipment and apparatus
33 Manufacture of medical, precision and optical instruments, watches and clocks
34 Manufacture of motor vehicles, trailers and semi-trailers
35 Manufacture of other transport equipment
36 Manufacture of furniture; manufacturing n.e.c.
37 Recycling

E ELECTRICITY, GAS AND WATER SUPPLY

40 Electricity, gas, steam and hot water supply

41 Collection, purification and distribution of water

F CONSTRUCTION

45 Construction

G WHOLESALE AND RETAIL TRADE; REPAIR OF MOTOR VEHICLES, MOTORCYCLES

AND PERSONAL AND HOUSEHOLD GOODS

50 Sale, maintenance and repair of motor vehicles and motorcycles; retail sale of automotive
fuel
51 Wholesale trade and commission trade, except of motor vehicles and motorcycles
52 Retail trade, except of motor vehicles and motorcycles; repair of personal and household
goods

H HOTELS AND RESTAURANTS

55 Hotels and restaurants

I TRANSPORT, STORAGE AND COMMUNICATIONS

60 Land transport; transport via pipelines

61 Water transport
62 Air transport
63 Supporting and auxiliary transport activities; activities of travel agencies
64 Post and telecommunications

J FINANCIAL INTERMEDIATION

65 Financial intermediation, except insurance and pension funding

66 Insurance and pension funding, except compulsory social security
67 Activities auxiliary to financial intermediation

K REAL ESTATE, RENTING AND BUSINESS ACTIVITIES

70 Real estate activities

71 Renting of machinery and equipment without operator and of personal and household goods
72 Computer and related activities
73 Research and development
74 Other business activities

L PUBLIC ADMINISTRATION AND DEFENCE; COMPULSORY SOCIAL SECURITY

 47

75 Public administration and defence; compulsory social security

M EDUCATION

80 Education

N HEALTH AND SOCIAL WORK

85 Health and social work

O OTHER COMMUNITY, SOCIAL AND PERSONAL SERVICE ACTIVITIES

90 Sewage and refuse disposal, sanitation and similar activities

91 Activities of membership organizations n.e.c.
92 Recreational, cultural and sporting activities
93 Other service activities

P PRIVATE HOUSEHOLDS WITH EMPLOYED PERSONS

95 Private households with employed persons

Q EXTRA-TERRITORIAL ORGANIZATIONS AND BODIES

99 Extra-territorial organizations and bodies

48

Appendix B

CENTRAL PRODUCT CLASSIFICATION, VERSION 1.0

THE LIST OF SECTIONS AND DIVISIONS
OF CENTRAL PRODUCT CLASSIFICATION (CPC)

_______________________________________________________________________________________________
Tabulation
categories Division Description

0 AGRICULTURE, FORESTRY AND FISHERY PRODUCTS

01 Products of agriculture, horticulture and market gardening

02 Live animals and animal products
03 Forestry and logging products
04 Fish and other fishing products

1 ORES AND MINERALS; ELECTRICITY, GAS AND WATER

11 Coal and lignite; peat

12 Crude petroleum and natural gas
13 Uranium and thorium ores
14 Metal ores
15 Stone, sand and clay
16 Other minerals
17 Electricity, town gas, steam and hot water
18 Water

2 FOOD PRODUCTS, BEVERAGES AND TOBACCO; TEXTILES, APPAREL AND LEATHER

PRODUCTS

21 Meat, fish, fruit, vegetables, oils and fats

22 Dairy products
23 Grain mill products, starches and starch products; other food products
24 Beverages
25 Tobacco products
26 Yarn and thread; woven and tufted textile fabrics
27 Textile articles other than apparel
28 Knitted or crocheted fabrics; wearing apparel
29 Leather and leather products; footwear

3 OTHER TRANSPORTABLE GOODS, EXCEPT METAL PRODUCTS, MACHINERY AND

EQUIPMENT

31 Products of wood, cork, straw and plaiting materials

32 Pulp, paper and paper products; printed matter and related articles
33 Coke oven products; refined petroleum products; nuclear fuel
34 Basic chemicals
35 Other chemical products; man-made fibers
36 Rubber and plastic products
37 Glass and glass products and other non-metallic products n.e.c.
 49

38 Furniture; other transportable goods n.e.c.

39 Wastes or scraps

4 METAL PRODUCTS, MACHINERY AND EQUIPMENT

41 Basic metals
42 Fabricated metal products, except machinery and equipment
43 General purpose machinery
44 Special purpose machinery
45 Office, accounting and computing machinery
46 Electrical machinery and apparatus
47 Radio, television and communication equipment and apparatus
48 Medical appliances, precision and optical instruments, watches and clocks
49 Transport equipment

5 INTANGIBLE ASSETS; LAND; CONSTRUCTION; CONSTRUCTION SERVICES

51 Intangible assets
52 Land
53 Constructions
54 Construction services

6 DISTRIBUTIVE TRADE SERVICES; LODGING; FOOD AND BEVERAGE SERVING SERVICES;

TRANSPORT SERVICES; AND UTILITIES SERVICES

61 Wholesale trade services

62 Retail trade services
63 Lodging; food and beverage serving services
64 Land transport services
65 Water transport services
66 Air transport services
67 Supporting and auxiliary transport services
68 Postal and courier services
69 Electricity distribution services; gas and water distribution services through mains

7 FINANCIAL AND RELATED SERVICES; REAL ESTATE SERVICES; AND RENTAL AND
LEASING SERVICES

71 Financial intermediation services, insurance and auxiliary services

72 Real estate services
73 Leasing or rental services without operator

8 BUSINESS AND PRODUCTION SERVICES

81 Research and development services

82 Professional, scientific and technical services
83 Other professional, scientific and technical services
84 Telecommunications services; information retrieval and supply services
85 Support services
86 Production services, on a fee or contract basis
87 Maintenance and repair services

9 COMMUNITY, SOCIAL AND PERSONAL SERVICES

50

91 Public administration and other services to the community as a whole; compulsory social
security services
92 Education services
93 Health and social services
94 Sewage and refuse disposal, sanitation and other environmental protection services
95 Services of membership organizations
96 Recreational, cultural and sporting services
97 Other services
98 Domestic services
99 Services provided by extraterritorial organizations and bodies
 51

Appendix C

THE CLASSIFICATION OF THE FUNCTIONS OF GOVERNMENT (COFOG)24

_______________________________________________________________________________________________
Tabulation
categories Division Description

01 GENERAL PUBLIC SERVICES

01.1 Executive and legislative organs, financial and fiscal affairs, external affairs other than
foreign aid
01.2 Foreign economic aid
01.3 General services
01.4 Basic research
01.5 R&D general public services
01.6 General public services n.e.c.
01.7 Public debt transactions
01.8 Transfers of a general character between different levels of government

02 DEFENCE

02.1 Military defence

02.2 Civil defence
02.3 Foreign military aid
02.4 R&D defence
02.5 Defence n.e.c.

03 PUBLIC ORDER AND SAFETY AFFAIRS

03.1 Police services

03.2 Fire protection services
03.3 Law courts
03.4 Prisons
03.5 R&D public order and safety
03.6 Public order and safety n.e.c.

04 ECONOMIC AFFAIRS

04.1 General economic and commercial affairs

04.2 Agriculture, forestry, fishing and hunting
04.3 Fuel and energy
04.4 Mining, manufacturing and construction
04.5 Transport
04.6 Communication
04.7 Other sectors
04.8 R&D economic affairs
04.9 Economic affairs n.e.c.

24
The provisional COFOG was issued by the United Nations as part of the Draft Classifications of
Expenditure According to Purpose, ST/ESA/STAT/SER.M/84, 24 November 1998.
52

05 ENVIRONMENTAL PROTECTION

05.1 Waste management

05.2 Waste water management
05.3 Pollution abatement
05.4 Protection of biodiversity and landscape
05.5 R&D environmental protection
05.6 Environmental protection n.e.c.

06 HOUSING AND COMMUNITY AMENITIES

06.1 Housing development

06.2 Community development
06.3 Water supply
06.4 Street lighting
06.5 R&D housing and community amenities
06.6 Housing and community amenities n.e.c.

07 HEALTH

07.1 Prescribed medical products, equipment and appliances

07.2 Out-patient services
07.3 Hospital services
07.4 Public health services
07.5 R&D health
07.6 Health n.e.c.

08 RECREATIONAL, CULTURE AND RELIGION

08.1 Recreational services

08.2 Cultural services
08.3 Broadcasting and publishing services
08.4 Religious and other community services
08.5 R&D recreation, culture and religion
08.6 Recreation, culture and religion n.e.c.

09 EDUCATION

09.1 Pre-primary and primary education

09.2 Secondary education
09.3 Tertiary education
09.4 Education not definable by level
09.5 Subsidiary services to education
09.6 R&D education
09.7 Education n.e.c.
 53

10 SOCIAL PROTECTION

10.1 Sickness and disability

10.2 Old age
10.3 Survivors
10.4 Family and children
10.5 Unemployment
10.6 Housing
10.7 Social exclusion n.e.c.
10.8 R&D social protection
10.9 Social protection n.e.c.
54
 55

III. VALUATION IN I/O TABLE

3.1. The symmetric I/O model (SIOT) is assumed to measure production relationships on the basis of the
technical structure of the economy. In order to reflect that basic assumption, model builders should try not only
to maximize the homogeneity of establishments classified into the same ISIC, but also to use the same valuation
for both goods and services supplied and used that eliminates the effects of government policies or costs of
transactions on technical relationships. This chapter focuses on how to measure SIOT in the same price system.

A. Concepts of valuation

3.2. Goods and services may be valued in various ways. Following are the three ways in which they may be
measured in the SNA:

Purchasers’ price: "The purchasers’ price is the amount paid by the purchaser, excluding any deductible VAT
[value added taxes] or similar deductible tax, in order to take delivery of a unit of a good or service at the time
and place required by the purchaser. The purchasers’ price of a good includes any transport charges paid
separately by the purchaser to take delivery at the required time and place."1 It should be noted that in business
accounts, “freight-in costs” are normally separated from the purchased value of goods if these costs are paid
separately.

Producers' price: "The producer’s price is the amount receivable by the producer from the purchaser for a unit
of a good or service produced as output minus any VAT, or similar deductible tax, invoiced to the purchaser.
It excludes any transport charges invoiced separately by the producer."2

Basic price: "The basic price is the amount receivable by the producer from the purchaser for a unit of a good
or service produced as output, minus any tax payable, and plus any subsidy receivable, on that unit as a
consequence of its production or sale. It excludes any transport charges invoiced separately by the producer."3

3.3. The relationship between different types of prices can be given as follows:

1
Ibid., para. 6.215.

2
Ibid., para. 6.205.

3
Ibid.
56

Purchasers’ price
minus Trade and transport margins
equals Producers’ prices
minus Taxes on products (sales tax, manufacturers' excise tax,
and/or non-deductible VAT)
plus Subsidies on products
equals Basic price

3.4. Trade and transport margins are the difference between the purchasers’ price and the producers’ price
of a product. Because of the differences in trade and transport margins and taxes less subsidies on a given
product, a product can be sold at different purchasers’ prices. More concretely, purchasers’ prices and producer’s
prices vary non-uniformly for the same product for the following reasons:

(a) Trade margins vary from one transaction to another depending on whether the goods are
purchased directly from producers, wholesalers or retailers; they also vary by "class" of retail
establishments (at high prices in establishments with higher-quality services and at lower prices
at discount establishments with lower-quality services);

(b) Transport margins vary by mode of transport and the distance the goods have to be shipped;

(c) Policies on taxes on products are normally based on the purposes for which products are used
(e.g., no taxes are imposed, or taxes are fully or partly deductible, on products used for
production or export purposes);

(d) Who records the transaction: the seller may record it without transport cost while the buyer may
record it with transport cost.

3.5. For all these reasons, basic prices are more homogeneous than producer’s prices which in turn are more
homogeneous than purchasers’ prices.

3.6. Since the same goods and services can be measured differently in the market and since homogeneity is
one of the most important underlying assumptions of input-output economics, the SNA recommends products
to be measured as homogeneously or uniformly as possible in the SIOT, preferably in basic prices. When that
is not possible, they can be valued at producers' prices instead. In the case of services, purchasers’ and producers’
prices are the same since the services are sold directly by producers to consumers. Basic prices of services are
of course different from producers' prices since product taxes on services also do exist.

3.7. Because taxes are normally levied on products at the point at which they are sold by wholesalers and
retailers for the convenience of government tax collectors, a large part of taxes on products is included in the trade
margins. Consequently, some countries, when measuring output at producers’ prices, have treated these taxes
as taxes on the trade margins. This treatment is convenient but not very interesting since it is not possible to
obtain directly taxes collected on each kind of commodity. The SNA recommends linking taxes on products with
product output value as shown in the supply table 3.l. Information on taxes on each group of product as well
as wholesale and retail trade margins can be obtained from the surveys of wholesale and retail trade.

3.8. Transport margins are treated as part of trade margins when transport costs are paid for by trade
establishments and therefore are treated as their intermediate consumption. Where transport costs are paid directly
by purchasers, they are treated as services consumed directly by them.
 57

3.9. In the SNA supply and use tables (SUT) or in SIOT, it is important to note that trade margins can be
presented in one row (representing trade activities as a whole), two rows (representing separate wholesale and
retail activities), or many rows (each row representing a trading activity, either wholesale or retail, in one
particular kind of product). The columns also have to be split accordingly. There is an advantage to doing this
as each product may require a different trade margin, especially when products are weighty and have to be
transported over great distances. The detailed breakdown of trade margins is however rarely done due to the need
for detailed information and the large size of the resulting input-output tables.

B. Valuation and the tax system

3.10. The problem of valuation in the supply and use tables (SUT) is to organize data consistently using the
same price system and finally to arrive at tables in basic prices. Depending on the kind of tax system used,
valuation will be approached slightly differently. In non-VAT countries, it is easily to convert all elements of
SUT into one price system: either producers’ values or basic values. In VAT countries, the conversion of SUT
into one price system using basic values is as simple as the non-VAT case. However, the conversion of SUT
from mixed to producers’ prices is a lot more complicated, particularly with regard to the net treatment of
producers’ prices. Thus it is important to summarize again the definition of non-VAT and VAT before the
treatment of valuation in SUT is explained.

1. Non-VAT taxes

3.11. A non-VAT system is one of sales taxes on products which are normally assessed at the final stages of
transactions on final purchasers and, unlike VAT, not deductible. These taxes are mainly imposed on goods and
services for final consumption. They are usually not imposed on goods and services for intermediate consumption,
capital formation and sometimes exports. Non-deductible taxes on products at the factory gate usually called
manufacturers' excise taxes should be treated similarly.

2. Value added taxes (VAT)

3.12. Value added taxes are imposed on every stage of transaction but are usually deductible - although a non-
deductible part may occur - when they are used for intermediate consumption, capital formation or exports.
However, VAT are not usually charged on exports.

3.13. When taxes are deductible, a producer deducts those he paid on the intermediate and capital goods he
purchased, from the VAT he invoices to the purchasers of his products before transferring the difference to the
Government.

3.14. In a VAT country, some form of sales taxes such as excise tax on liquor, gasolines, cigarettes, etc. may
also be applied. These taxes are not deductible. Further discussion of the VAT system is left for part E of this
chapter.
58

C. The derivation of the supply and use tables (SUT) at basic prices, VAT and non-VAT

3.15. With SUT fully discussed in chapter II, their presentation will not be repeated here. The emphasis in
this part is on the derivation of the supply and use tables at basic prices universally applicable to either non-VAT,
VAT or mixed systems. Compilers need only read part C and ignore part E if they decide to derive SUT or SIOT
in basic prices which are recommended by this Handbook.

3.16. However, a few comments are needed on the presentation of supply table 3.1 and use table 3.2:

- The total supply of every product at purchasers’ prices in column 11, table 3.1, is equal to its
total use at purchasers' prices of the same product in column 9 of table 3.2.

- The industry outputs at basic prices in row 6 of table 3.1 are the same as shown in row 11 of
table 3.2.

- Uses in table 3.2 are at purchasers' prices excluding deductible VAT but including all other non-
deductible taxes on products so as to make the costs in the use table represent actual costs to
producers or consumers (see exhibit B, para. 3.28 below).

- Gross value added at basic prices in row 9 of table 3.2 (i.e. 196) is the sum of compensation of
employees, other taxes less subsidies on production, consumption of fixed capital and mixed
incomes.

- Total gross value added or GDP (row 8) is equal to gross value added at basic prices plus total
taxes less subsidies on production and imports. Thus, total gross value added = GDP = 196 +
20 = 216.

- Total final demand is equal to total final household final expenditures + total government final
expenditures + gross capital formation = 153 + 50 + 10 = 203.

- The GDP should also be equal to total net final demand. Thus, total net final demand = total
final household final expenditures (which also include final expenditures of NPISHs in the
example) + total government final expenditures + gross capital formation + total exports f.o.b. -
total imports f.o.b. = 153 + 10 + 40 + 38 - 25 = 216.

- In the supply table, table 3.1, product 2 includes both goods and all services except trade and
transport services. Thus in the column of c.i.f./f.o.b. adjustment, -1 is entered in row 2 as the
value of insurance services on imported goods rendered by residents and non-residents.

- In table 3.1 the total value of row 3 (column 11) is the transport services that are directly
purchased by producers and final consumers because trade and transport margins (-60) are
subtracted from the total supply of trade and transport services. These directly purchased
transport services are shown in the use table, table 3.2, as purchased by the household sector,
mainly passenger transportation but possibly also for goods transportation. However, they can
also be purchased directly by industries and other sectors.
 59

- In use table 3.2, row 4 does not appear so that a row of taxes less subsidies on products can be
inserted at basic prices while keeping the same row index numbers for all elements in tables 3.2,
3.4 and, 3.5.

3.17. The derivation of the use table at basic prices is fully shown in table 3.3. Shown there is the breakdown
of the value of every use at purchasers' prices, in rows 1 to 3 of table 3.2, into three parts: value at basic prices,
trade and transport margins and taxes less subsidies on products. In table 3.3, the values at basic prices are shown
in rows 1 to 3 in the same position as in table 3.2, with their trade and transport margins in rows 1' and 2', and
the taxes less subsidies paid on these products in rows 1" to 3".4 For example, the value of product 1 at
purchasers' prices used by households as final goods which is 100 in table 3.2 is split into:

- 71, the basic value shown in table 3.3 in row 1, column 6;

- 20, transport and trade margins shown in row 1', column 6 and;

- 9, taxes less subsidies shown in row 1", column 6.

3.18. Total uses at purchasers' prices in row 7 of table 3.3 are the same as those shown in row 7 of table 3.2.
The final result is shown in table 3.4 in which all product uses are measured at basic prices. Row 3 in table 3.4
is the sum of rows 3, 1' and 2' in table 3.3 and row 4 of table 3.4 is the sum of rows 1", 2" and 3" in table 3.3.

3.19. The final use and supply tables in basic prices can be combined to produce the symmetric I/O table
(SIOT). Table 3.5 shows imports as part of the use table which is the presentation that is needed to carry out I/O
modelling. In this table, imports are entered as negative values. Because of this, column 11 shows total use of
domestic products only. These are the same values of total domestic production as in column 4 of table 3.1.

3.20. To conclude the discussion on the derivation of SUT at basic prices, it should be pointed out that the
presentation in table 3.3 is only for the purpose of illustration. In practice, a table of trade and transport margins
and a table of taxes less subsidies on products are separately compiled using the same structure as the use table
in purchasers' prices. The values in these tables are then deducted from the corresponding values in the table at
purchasers' prices 3.2. Then, the row of column sums of the table of trade and transport margins (rows 1' and 2'
of table 3.3) is added back to the row of trade and transport services. Similarly, an extra row of taxes less
subsidies is created in table 3.4 which takes the values of the row of the column sums of the table of taxes less
subsidies shown in row 1" to row 3" in table 3.3.

D. The derivation of SUT at producers’ prices

3.21. The discussion below presents only the method of deriving SUT without VAT and the implications of
using this system. No discussion on the net system of SUT is presented because, as discussed in part E, this
system serves no purpose unless SUT assumes that every industry produces only one product, which means that
the system is symmetric. If that is the case, the derivation can be easily obtained on the basis of the previous
discussion of VAT.

3.22. The tables presenting SUT at producers’ prices are 3.6-3.10. These tables are structurally similar to tables
3.1-3.5 of the system at basic prices. They also contain values that are consistent with those in tables 3.1-3.5.

4
Rows with signs, e.g.1', 1", are split from row 1.
60

The derivation of the use table at producers’ prices is similar to the derivation of the use table at basic prices
discussed previously so it will not be repeated here. However, there are some important differences that need to
be pointed out:

- Unlike output of industries at basic prices in table 3.1, output of industries at producers’ prices
in table 3.6 includes taxes less subsidies on products. These taxes are, of course, not deductible.

- Since output supplies are measured in producers’ prices, imports must be too. The equivalent
value of imports in producers’ prices is the ex-customs value which is the sum of the value of
imports c.i.f. and import duties shown in supply table 3.6. Similarly, trade and transport margins
must include sales taxes less subsidies on goods and services sold by wholesalers and retailers
including those assessed on their margins.

- In table 3.7, taxes less subsidies on products payable by producers are included in row 9 and
columns 1-3. As can be seen in this example, a large amount of taxes less subsidies on products
is paid by wholesalers and retailers shown in row 9, column 3. This amount includes sales taxes
on goods sold by them and taxes assessed on their own outputs, i.e. trade and transport margins.
The taxes less subsidies on products shown in row 9 lose the direct link to specific products. At
row 9 and column 4, the total import duties are entered so that row 9 contains all taxes less
subsidies on production and imports.

- Gross industry value added at basic prices is calculated as residuals by subtracting total
intermediate uses at purchasers' prices and taxes less subsidies on products from industry output
at producers’ prices. If one wants to get gross industry value added at producers’ prices, one can
deduct total intermediate uses at purchasers' prices from industry output at producers' prices, or
simply the sum of gross value added at basic prices and taxes less subsidies on products. Gross
value added at producers’ prices is not shown in table 3.7 but it is equal to 213.

- The total gross value added or GDP is equal to the total gross value added at basic prices plus
the total taxes less subsidies on production and imports, i.e. GDP = 216 = 196 + 20. The GDP
can also be obtained by adding the total gross value added at producers’ prices to import duties,
i.e. GDP = 213 + 3.

- The GDP should also be equal to total net final demand (see table 3.7). Total net final demand
= total final household final expenditures + total government final expenditures + gross capital
formation + total exports f.o.b. - total imports f.o.b. = 153 + 10 + 40 + 38 - 25 = 216.

- In table 3.8, the derivation of the use table at producers’ prices is similar to that of table 3.3, but
here only a table of trade and transport margins is needed. Trade and transport margins in table
3.8 differ from those in table 3.3 because they include taxes assessed on the trade and transport
margins and taxes that are collected on products sold by wholesalers and retailers and then
transferred to the Government.

- Table 3.9 is the final use table at producers’ prices. Table 3.10 is the form of the use table which
after being transformed to become a product-by-product table is the I/O symmetric model that
is used for modelling. In this table, imports are entered in negative values. Because of this,
column 11 shows the total use of domestic products only. These values are the same as those
 61

in column 4 of table 3.6. Column 12 entitled total economy is moved to the extreme right so that
it is easier to show total taxes less subsidies on imports.

3.23. With the system measured at producers’ prices, it is much harder to analyse the impact of tax policy since
there is no direct linkage between the output of a product and the tax rates. The system measured at basic prices
is better in this respect.

E. The supply and use tables with VAT

1. Definitions

3.24. The derivation of SUT in basic prices from SUT in mixed prices with VAT was discussed in part C of
this chapter. The discussion here focuses only on the derivation of SUT with VAT in producers’ prices. To that
end, it is important to mention the following SNA definitions on VAT terminology:

Invoiced VAT: This VAT is payable on the sales of a producer and it is shown separately on the
invoice from the producer to the purchaser;

Deductible VAT: This VAT is payable on purchases of goods or services intended for intermediate
consumption, gross fixed capital formation or for resale which a producer is permitted to deduct from
his own VAT liability to the Government in respect of VAT invoiced to his customers;

Non-deductible VAT: This VAT is payable by a purchaser and is not deductible from his own VAT
liability;

Net VAT on product is the net total of taxes the Government collects on a product. It is equal to the
VAT invoiced on the product output produced by various industries minus the deductible VAT on the
same product when it is used either as intermediate input, for capital formation, or for export by all
purchasers. Thus the net VAT on a product is the aggregate concept that is derived for a particular
product.

2. Producers’ prices, gross and net treatment of VAT and basic prices

3.25. The following example will explain the difference between gross and net treatment of VAT when one
producer produces one product, i.e. SUT has to be symmetric. The example also assumes:

(a) No trade and transport margins and no non-VAT taxes on products so that producers' prices will
equal purchasers' prices;

(b) Invoiced VAT is equal to 10% of values at purchasers' prices.

In the example, a producer sold product X for $200 (see exhibit A.1) and spent $50 on intermediate goods and
services to produce the product. When selling, the producer invoiced $20 of VAT on the product he sold to his
customers, deducted $5 of VAT invoiced on the goods and services he purchased for his production operation
and turned over $15 (= 20 - 5) to the Government: $15 is payable VAT. Thus the actual cost of intermediate
inputs, i.e. intermediate consumption valued net of deductible VAT, is only $45 (50 - 5). The amount of $200
is called producer’s price, gross treatment. The basic price which is equal to the producers’ price -- gross
treatment less invoiced VAT on output sold is $180 (= 200 - 20).
62

3.26. The production account of the producer in the example with VAT can be presented in producers’ price-
gross treatment, producers’ price-net treatment and basic price.

3.27. The producers’ price-gross treatment is shown in exhibit A.1 and in the form of the use table in exhibit
A.2. This gross treatment, though simple, is impractical because business accounts normally report sales
excluding invoiced VAT, i.e. at basic prices, and also goods and services used in production are recorded
excluding deductible VAT. The gross system was tried in some countries, Norway for example, but had to be
abandoned partly for the reason mentioned and partly to conform with the convention set by the international
community (such as the European Union) for the sake of international comparability.

Exhibit A.1. Producers’ prices - gross treatment

Sale value (output at producers’ price - gross treatment 200

Less Intermediate consumption -50
Equal Value added at producers’ price - gross treatment 150

Equal Value added at basic price 135

Plus Payable VAT 15
Equal invoiced VAT on output sold 20
Less deductible VAT on intermediate consumption 5

3.28. The basic value of the product sold is equal to the output at producer’s prices-gross treatment less
invoiced VAT. It is $180 = 200 - 20. The components that make up the output at basic prices are shown in
exhibit B.1 and the presentation of the production and use of product X at basic prices in the use table is shown
in exhibit B.2. In the use table, the deductible VAT must be excluded from intermediate consumption and final
uses. Finally, in order to equate supply and demand, a column of negative net VAT on product must be added.
The net VAT on product (see definition in para. 3.24) is equal to the total VAT invoiced on the product less the
VAT paid on product but deductible ($16=20-(1+3)).

3.29. The system of producers’ prices-net treatment is similar to the case of output measured at basic prices
shown in exhibit B.2. Intermediate consumption of goods and services and final uses are valued at purchasers'
prices excluding the deductible VAT, which means that only the non-deductible VAT is included. However,
there is one important difference between the valuation at basic prices and the valuation at producers’ prices-net
treatment. In the former, production and consumption can be looked at from the points of view of both producers
and consumers. In the latter, production is looked at from the point of view of producers, i.e. output at basic
prices, but consumption is looked at from the point of view of consumers. In exhibit C.2 total of l96 is the total
cost to consumer but in exhibit B.2, 180, the total output at basic prices, is the total value received by producers.
The addition of 16, net VAT on product, to output at basic prices is needed to balance total supply and total use.
The totals in the column and the corresponding row now reflect the point of view of consumers. It is also
important to remind readers that in the exhibit C.2 example, only one type of product is assumed to be produced
by a producer.

Exhibit A.2. The use table in producers’ price - gross treatment

Industry X Other Final Gross capital Others Total

industries consumption formation
 63

At purchasers' prices including invoiced VAT

Product X 10 (1)* 160 (16)* 30 (3)* 0 200

Other products 40 (4)*

Value added at basic prices 135

Invoiced VAT 20
Deductible VAT on inputs -5

Output at producers’ price- 200

gross treatment

* Parenthesized numbers are invoiced VAT which is assumed to be with 10% of purchased values. In this example, only
the invoiced VAT, on intermediate consumption (1 for product X) and capital formation (3 for product X) is deductible.

Exhibit B.1. Basic prices

Intermediate consumption at purchasers' prices excluding deductible VAT 45
Gross value added at basic prices 135

Output at basic price 180

Equal Output at producers’ price - gross treatment (sale value) 200

less VAT invoiced on the output sold -20

3.30. In exhibit C.2, which describes the recent practice in many European countries with VAT, the net system
is used but with symmetric I/O tables (SIOT) in which one industry is assumed to produce only one product. If
a producer produces many products as described in SUT, the net VAT on them can only be obtained after
allocating the net VAT for each one. The value of net VAT on products obtained in this way does not even serve
the purpose of balancing the supply-and-use tables as in the case of the symmetric I/O table because the output
of an industry need not be equal to the value of the principal product it produces. Thus, in SUT with VAT, the
system in basic prices should be adopted. The European System of Accounts 1995 (ESA) (Eurostat, Brussels,
Luxembourg, 1996) has decided to use only basic prices for valuation.
64

Exhibit B.2. The use table in basic prices

Industry X Other Final Gross capital Others Net VAT Total

industries consumption formation on
products

At purchasers' prices At purchasers' prices excluding

excluding deductible VAT deductible VAT

Product X 9 160 27 0 -16 180

Other products 36

Value added at basic prices 135

Output at basic prices 180

Exhibit C.1. Producers’ price, net treatment in the use table

Intermediate consumption at purchasers' prices excluding deductible VAT 45
Gross value added at basic price 135

Output at basic price 180

Net VAT on product* 16

Total, net treatment 196

*20-4 (with 4 the deductible VAT on product X).

Exhibit C.2. The use table at producers’ price - net treatment

Industry X Other Final Gross Others Total

industries consumption capital
formation

At purchasers' prices At purchasers' prices excluding

excluding deductible VAT deductible VAT

Product X 9 160 27 0 196

Other products 36

Value added at basic prices 135

Net VAT on products 16

Total 196
 Table 3.1. Supply of products at basic prices

Output of industries Total economy Imports C.i.f./ Total product supply Trade and Taxes less subsidies on products Total supply at
c.i.f. f.o.b. adj. at basic prices transport purchasers' prices
(total margins
Import duties Other taxes less
f.o.b.)
subsidies on
products

(1) (2) (3) (4)=(1)+.+(+3) (5) (6) (7)= (4)+.+(6) (8) (9) (10) (11)= (7)+.+(10)

(1) Product 1 156 24 0 180 15 195 33 2 11 241

(2) Product 2 9 80 0 89 8 -1 96 27 1 6 130

(3) Transport services directly purchased 0 0 62 62 1 -1 62 -60 0 0 2

(4) C.i.f./f.o.b. adjustment -2 2 0 0

(5) Direct purchases abroad by residents 3 3 3

(6) Total industry output at basic prices 165 104 62 331

(1)+.+(5)

(7) Other column total 25 0 356 0 3 17 376

 Table 3.2. Use of products at purchasers' prices

Intermediate consumption Total Exports Household final Government final Gross capital Total use of products
of industries economy f.o.b. expenditures* expenditures formation at purchasers' prices

(1) (2) (3) (4) (5) (6) (7) (8) (9)=(1)+.+(8)

(1) Product 1 25 35 13 28 100 40 241

(2) Product 2 32 20 10 9 49 10 0 130

(3) Transport services directly purchased 0 2 0 2

(5) Direct purchases abroad by residents 3 3

(6) Direct purchases at home by non-residents 1 -1 0

(7) Total uses at purchasers' prices (11)+.+(6) 57 55 23 38 153 10 40 376

(8) Total gross value added/GDP 216

(9) Gross value added at basic prices (11)-(7) 108 49 39 196

(10) Taxes less subsidies on production and imports 20

(11) Industry output at basic prices 165 104 62 331

Other relationships in the supply and use tables are shown below with Uij standing for entry in row i and column j in this table and Sij in the supply table 3.1:

GDP = U8.4 = U9.4 + U10.4

Taxes less subsidies on production and imports = U10.4 = S7.9 + S7.10.

* Including final consumption expenditures of NPISHs.

 Table 3.3. Derivation of use of products at basic prices

Intermediate consumption Total Exports Household final Government final Gross capital Total use of products at
of industries economy f.o.b. expenditures* expenditures formation basic prices

Product at basic prices (1) (2) (3) (4) (5) (6) (7) (8) (9)=(1)+.+(8)

(1) Product 1 19 28 10 27 71 40 195

(2) Product 2 29 18 8 8 23 10 96

(3) Transport services directly purchased 2 0 2

(1') Trade and transport margins on product 1 5 5 2 1 20 0 33

(2') Trade and transport margins on product 2 2 2 1 1 21 0 27

(1") Taxes less subsidies on product 1 1 2 1 0 9 0 13

(2") Taxes less subsidies on product 2 1 0 1 0 5 0 7

(3") Taxes less subsidies on product 3 0 0

(5) Direct purchases abroad by residents 3 3

(6) Direct purchases at home by non-residents 1 -1 0

(7) Total uses at purchasers' prices (1)+.+(6) 57 55 23 38 153 10 40 376

(8) Total gross value added/GDP 216

(9) Gross value added at basic prices (11) - (7) 108 49 39 196

(10) Taxes less subsidies on production and imports 20

(11) Total industry output at basic prices 165 104 62 331

Other relationships in the supply and use tables are shown below with Uij standing for entry in row i and column j in this table and Sij in the supply table 3.1:

GDP = U8.4 = U9.4 + U10.4

Taxes less subsidies on production and imports = U10.4 = S7.9 + S7.10.

*Including final consumption expenditures of NPISHs.

 Table 3.4. Use of products at basic prices

Intermediate consumption of Total Exports f.o.b Household final Government final Gross capital Total use of
industries economy expenditures* expenditures formation products at basic
prices

(1) (2) (3) (4) (5) (6) (7) (8) (9)=(1)+.+(8)

(1) Product 1 19 28 10 27 71 40 195

(2) Product 2 29 18 8 8 23 10 96

(3) Trade and transport services** 7 7 3 2 43 0 62

(4) Taxes less subsidies on products 2 2 2 0 14 0 20

(5) Direct purchases abroad by residents 3 3

(6) Direct purchases at home by non-residents 1 -1 0

(7) Total uses at purchasers' prices (1)+.+(6) 57 55 23 38 153 10 40 376

(8) Total gross value added/GDP 216

(9) Gross value added at basic prices (11) - (7) 108 49 39 196

(10) Taxes less subsidies on production and imports 20

(11) Total industry output at basic prices 165 104 62 331

Other relationships in the supply and use tables are shown below with Uij standing for entry in row i and column j in this table and Sij in the supply table 3.1:

GDP = U8.4 = U9.4 + U10.4

Taxes less subsidies on production and imports = U10.4 = S7.9 + S7.10.

* Including final consumption expenditures of NPISHs.

** Trade and transport services are equal to trade and transport margins plus transport services directly purchased.
 Table 3.5. Use of products at basic prices: the conventional form of I/O model

Intermediate consumption Total Household final Government Gross capital Exports Imports c.i.f./f.o.b. Total
of industries economy expenditures* final formation f.o.b. c.i.f. adj. use of domestic
expenditures (total products at basic
f.o.b.) prices

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)=(1)+.+(10)

(1) Product 1 19 28 10 71 40 27 -15 180

(2) Product 2 29 18 8 23 10 8 -8 1 89

(3) Trade and transport services** 7 7 3 43 0 2 -1 1 62

(4) Taxes less subsidies on products 2 2 2 14 0 0 20

(5) Direct purchases abroad by residents 3 -3 0

(6) Direct purchases at home by non-residents -1 1 0

(7) C.i.f./f.o.b. adjustment 2 -2 0

(8) Total uses at purchasers' prices (1)+.+(7) 57 55 23 153 10 40 38 -25 0 351

(9) Total gross value added/GDP 273

(10) Gross value added at basic prices (12) - (8) 127 77 49 253

(11) Transport services directly purchased 20

(12) Total industry output at basic prices 165 104 62 331

Other relationships in the supply and use tables are shown below with Uij standing for entry in row i and column j in this table and Sij in the supply table 3.1:

GDP = U8.4 = U9.4 + U10.4

Taxes less subsidies on production and imports = U10.4 = S7.9 + S7.10.

*Including final consumption expenditures of NPISHs.

**Trade and transport services are equal to trade and transport margins plus transport services directly purchased.
 Table 3.6. Supply of products at producers’ prices

Output of industries Total economy Imports at ex-customs value Total product Trade and Total product supply at
supply at producer transport margins purchasers' prices
prices including sales
Imports c.i.f./ Import
taxes
c.i.f. f.o.b. duties
(total f.o.b.) adj.

(1) (2) (3) (4)=(1)+.+(3) (5) (6) (7) (8)= (4)+.+(7) (9) (10)= (8)+(9)

(1) Product 1 161 25 0 186 15 2 203 38 241

(2) Product 2 9 82 0 91 8 -1 1 99 31 130

(3) Transport services directly purchased 0 0 71 71 1 -1 0 71 -69 2

(4) c.i.f./f.o.b. adjustment -2 2 0 0

(5) Direct purchases abroad by residents 3 3 3

(6) Total industry output at producer prices (1)+.+(5) 170 107 71 348

(7) Other column total 25 0 3 376 0 376

 Table 3.7. Use of products at purchasers' prices

Intermediate consumption Import Total Exports Household Government Gross capital Total use of
of industries duties economy f.o.b. final final formation products at
expenditures expenditures purchasers' prices

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)=(1)+.+(9)

(1) Product 1 25 35 13 28 100 40 241

(2) Product 2 32 20 10 9 49 10 0 130

(3) Transport services directly purchased 0 2 0 2

(4) Direct purchases abroad by residents 3 3

(5) Direct purchases at home by non-residents 1 -1 0

(6) Total uses at purchasers' prices (1)+.+(5) 57 55 23 38 153 10 40 376

(7) Total gross value added/GDP 216

(8) Gross value added at basic prices (10)-(6)-(9) 108 49 -32 196

(9) Taxes less subsidies on production and imports 5 3 9 3 20

(10) Industry output at producer prices 170 107 71 348

Other relationships in the supply and use tables are shown below with Uij standing for entry in row i and column j in this table and Sij in the supply table (3.6).

GDP = U7.5 = U8.5 + U9.5

Import duties = U9.4 = S7.7
Taxes less subsidies on production and imports = U9.5 = U9.1 +...+ U9.4.
 Table 3.8. Derivation of the use table at producers’ prices

Intermediate consumption of Import Total Exports Household final Government Gross capital Total product
industries duties economy f.o.b. expenditures final formation uses at producer
expenditures prices

Product at producers’ prices (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)=(1)+.+(9)

(1) Product 1 20 30 11 27 75 40 203

(2) Product 2 30 18 9 8 24 10 99

(3) Transport services directly purchased 2 0 2

(1') Trade and transport margins on product 1 5 5 2 1 25 0 38

(2') Trade and transport margins on product 2 2 2 1 1 25 0 31

(4) Direct purchases abroad by residents 3 3

(5) Direct purchases at home by non-residents 1 -1 0

(6) Total uses at purchasers' prices (1)+.+(5) 57 55 23 38 153 10 40 376

(7) Total gross value added/GDP 216

(8) Gross value added at basic prices (10)-(6)-(9) 108 49 39 196

(9) Taxes less subsidies on production and imports 5 3 9 3 20

(10) Total industry output at producers' prices 170 107 71 348

Other relationships in the supply and use tables are shown below with Uij standing for entry in row i and column j in this table and Sij in the supply table (3.6).

GDP = U7.5 = U8.5 + U9.5

Import duties = U9.4 = S7.7
Taxes less subsidies on production and imports = U9.5 = U9.1 +...+ U9.4.
 Table 3.9. The use table at producers’ prices

Intermediate consumption Import Total Exports Household final Government final Gross capital Total product uses
of industries duties economy f.o.b expenditures expenditures formation at producer prices

Product at producers' prices (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)=(1)+.+(9)

(1) Product 1 20 30 11 27 75 40 203

(2) Product 2 30 18 9 8 24 10 99

(3) Trade and transport services* 7 7 3 2 52 0 203

(4) Direct purchases abroad by residents 3 3

(5) Direct purchases at home by non-residents 1 -1 0

(6) Total uses at purchasers' prices (1)+.+(5) 57 55 23 38 153 10 40 376

(7) Total gross value added/GDP 216

(8) Gross value added at basic prices (10)-(6)-(9) 108 49 39 196

(9) Taxes less subsidies on production and imports 5 3 9 3 20

(10) Total industry output at producer prices 170 107 71 348

Other relationships in the supply and use tables are shown below with Uij standing for entry in row i and column j in this table and Sij in the supply table (3.6).

GDP = U7.5 = U8.5 + U9.5

Import duties = U9.4 = S7.7
Taxes less subsidies on production and imports = U9.5 = U9.1 +...+ U9.4.

*Trade and transport services are equal to trade and transport margins plus transport services directly purchased.
 Table 3.10. The use table at producers’ prices: the conventional form of I/O model

Intermediate consumption Exports Imports at ex-custom value Household Government Gross capital Total use of Total
of industries f.o.b. final final formation domestic economy
expenditures expenditures products at
Imports cif Import c.i.f./f.o.b. producers’
(total f.o.b.) duties adj. prices

Product at producers’ prices (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)=(1)+.+(10) (12)

(1) Product 1 20 30 11 27 -15 -2 75 40 186

(2) Product 2 30 18 9 8 -8 -1 1 24 10 91

(3) Trade and transport services* 7 7 3 2 -1 0 1 52 0 71

(4) C.i.f./f.o.b. adjustment 2 -2 0

(5) Direct purchases abroad by residents -3 3 0

(6) Direct purchases at home by non-residents 1 -1 0

(7) Total uses at purchasers' prices (1)+.+(6) 57 55 23 38 -25 -3 0 153 10 40 348

(8) Total gross value added/GDP 216

(9) Gross value added at basic prices 108 49 39 196

(11)-(7)-(10)

(10) Taxes less subsidies on production and 5 3 9 3 20

imports

(11) Total industry output at producer prices 170 107 71 348

Other relationships in the supply and use tables are shown below with Uij standing for entry in row i and column j in this table.

Taxes less subsidies on production and imports = U10.12 = U10.1 + U10.2 + U10.3 + U10.6
Total gross value added/GDP = U8.12 = U9.12 + U10.12.

*Trade and transport services are equal to trade and transport margins plus transport services directly purchased.
 75

IV. CONVERTING SUPPLY AND USE TABLES INTO A SYMMETRIC

I/O TABLE: TREATMENT OF SECONDARY PRODUCTS

A. Introduction

4.1. The present chapter takes as a point of departure the supply and use tables that define the SNA input-
output framework that was reviewed in chapter II. It describes what methods could be used to convert the
SNA I/O framework of supply and use tables to a symmetric I/O table in which each industry produces only
one product, what conditions should be met to facilitate the conversion and/or its use in analysis and what
statistical issues must be faced in the conversion.

4.2. Theoretically, an I/O coefficient table may be compiled directly, using engineering information,
without first creating an input-output table in monetary values. To do so, one may ask detailed technical
information on all production processes operating in an economy. If a product is produced by more than one
process, an average input structure of those production processes would be created by weighing input columns
with output values. As weights change, the average input structures would have to be frequently updated.
With additional information on gross output, value added and final demand, it is possible to recreate a flow
table using the I/O coefficient matrix and output prices and thus check the accuracy and compatibility of the
engineering information with the data on value added and final demand, in the final balancing of the table.
This construction of I/O tables in physical terms is not simple, since enterprises may be reluctant to reveal
technical details of their production processes. Thus technical data may have to be collected through special
surveys, independently of censuses, and this may be very expensive. This is why the compilation of I/O tables
has relied mainly on data provided by censuses of establishments and special surveys. This information could
be supplemented by engineering-type data but this is rarely done. In view of these difficulties and the existing
practices, the present chapter will not deal with I/O tables in physical units.

4.3. The ideal situation for a symmetric I/O table is to have data that describe the input structure of every
type of activity producing a single product in the economy. With such an ideal situation, the I/O table is almost
symmetric and homogeneity in production function in the table is guaranteed, except for the cases of by-
products or joint products that are linked technologically in a production activity. If homogeneity is not
guaranteed, a distortion in analysis; particularly when total effects are calculated by the use of the Leontief
inverse may result. For example, in order to assess the total effects of an increase in demand of electricity on
the economy, reasonable estimates may not be obtainable if an I/O table only includes one industry producing
electricity. This is because electricity is produced by different processes: gas, coal, oil, water power or
windmill, etc., whereby each process requires a different set of inputs and consequently generates different
impacts. The column of inputs used for electricity production in this case may possibly be an aggregate of
different processes of electric generation. The assessment of the impact of an increase in electricity demand
using the I/O model is meaningful only if that increase is met by the same mix of electric generation
technologies that are shown as market shares of the technologies at the time the table is prepared. If the
technology that is used to meet the increase in demand for electricity is known in advance, first the inputs
needed for this particular type of technology have to be calculated, and then the indirect effects of the increased
inputs have to be estimated. But even here, the question needs to be asked whether or not the inputs used to
produce indirect demand of electricity are truly represented by the columns of the I/O table. Further
exploration of the use of I/O table for analysis will be the focus in later chapters on applications. The above
76

simply serves to emphasize the importance of the homogeneity property in input-output economics, i.e., each
industry in an input-output table should be as homogeneous as possible when used as a tool for analysis. To
improve the homogeneity property in analysis, it is important to maintain original data that are used in the
construction of an average technology in order to facilitate the construction of a different average technology
when needed. The present availability of computer power at low cost makes it feasible to maintain a huge
database that can be recombined to construct an I/O for a specific need.

4.4. In contrast to the ideal situation, the SNA use and supply tables as an integrated framework for
production statistics are designed to serve as the best statistical tool to compile national accounts aggregates
and provide information for the compilation of the symmetric I/O table. However, SUT, if taken only as a tool
for balancing supply and uses of products in the compilation of national accounts aggregates or institutional
sector accounts, would not need to apply rigorously the homogeneity assumption, which certainly would
require the identification of product production technologies. To extend this argument even further, SUT as
a tool for balancing supply and use products may even be based on enterprises as statistical units. In this case,
the derivation of the symmetric I/O table from SUT is purely a mathematical approximation. Even if an
establishment is used as the statistical unit, the resulting supply matrix still contains many secondary products,
though the derivation of the symmetric table is less an exercise in pure mathematical approximation. In order
to eliminate the need to either carry out supplementary surveys or live with the problems of secondary products
and resolve them by mechanical methods, I/O statisticians should engage in the planning of data collection by
censuses. The strategy is, as much as possible, to define operationally a purer establishment, linking each
product produced to an establishment. It is possible to design production censuses by asking business
accountants using the cost accounting approach to allocate costs to a specific type of product produced by an
establishment. Cost accounting is a common practice used by large enterprises in their pricing and general
management policy. Allocation keys are used to assign costs to every kind of product produced by the
enterprises. Some of the common allocation keys are work hours (for example, those spent by the personnel
of the maintenance department in other departments are used to allocate maintenance costs), surface area for
allocating rental, kilometres driven, percentage of administrative work done in each establishment to allocate
central administration costs, etc.1 In doing this, most secondary products will be eliminated from SUT.
Mathematical methods which are basically mechanical will be used only as a last resort. The approach
proposed above is feasible given the availability of inexpensive means of data processing at present. The
database should make it possible to link a product technology to a specific establishment which, in turn, is
identified with a specific enterprise.

4.5. Given the discussions in paragraph 4.4, the current chapter examines first the statistical approach for
eliminating secondary products before mathematical methods are discussed. However, whether statistical or
mathematical methods are usable depends on the types of secondary products produced.

1
See M. Francis Rousse and Vu Quang Viet, “Use of Business Accounts to Determine the Cost of a Particular Product
in a Multi-product Establishment”, chapter II of the Handbook of National Accounting: Links between Business
Accounting and National Accounting (ST/ESA/STAT/SER.F/76).
 77

B. Types of secondary products

1. Secondary products resulting from statistical practices

4.6. Secondary products may result from less refined statistical practices that do not adhere strictly to the
SNA definition of an establishment (see chapter II, para. 2.55). Such practices may originate from a statistical
policy to present production data as realistically as possible, and adopt the enterprise as the statistical unit.

4.7. If enterprises are used as statistical units and are multi-establishment, producing more than one
product, the issue of secondary products arises, i.e. in the supply table more than one product will be recorded
for the same industry.

4.8. Another example is when vertically integrated enterprises (or establishments) are not broken down
into separate establishments, as recommended by the SNA.

4.9. Also the case may occur of a horizontally integrated enterprise operating at one location and producing
different products for sale on the market. In this case, though, it may be obvious that each product is produced
through a different technology, but production data may be collected in the aggregate, treating the enterprise
as one establishment only.

4.10. Even when following the SNA recommendations, secondary products would still appear as long as
activities producing them are not identified as separate establishments, though the implementation of the SNA
recommendations will greatly reduce the scope of secondary products in input-output tables that are usually
resolved mechanically later.

2. Secondary products resulting from production technology

4.11. By-products exist because certain technologies produce more than one product simultaneously, and
similar products may also be produced elsewhere by other, quite different techniques of production. Three
types may be distinguished:

(a) Exclusive by-products are products that are not produced separately anywhere, e.g.,
molasses linked to the production of sugar, new scrap in metal industries.

(b) Ordinary by-products are products that are technologically linked to the production of other
products but are also produced separately elsewhere as main products. An example of this
type is hydrogen produced as a by-product in petroleum refining establishments but also
produced separately by other establishments in the chemical industry.

(c) Joint products are products that are more loosely linked technologically than ordinary
by-products; the common costs shared by joint products are more significant in value than is
the case for ordinary by-products. One example is milk and meat in the livestock industry
which may be produced on a scale that depends on the demand for each product and the ratio
of the two products may be varied in response to changing conditions of demand. One of the
joint products may be produced separately elsewhere. Joint products cannot be easily
distinguished from by-products.
78

C. Partitioning enterprises into units of homogeneous production

4.12. It is the enterprise unit that is able to provide statistics on productive activities, on incomes, profits and
losses, and its assets and liabilities. But as an enterprise may be involved in more than one activity, e.g. an
automobile company might produce airplanes, tanks and ammunition, trading activity or even computer
services for sale, they should not be used as the statistical unit for production accounts of which input-output
tables are a part because production technology of the enterprise is not homogeneous. In this case, the SNA
recommends that the enterprise be partitioned into separate establishments, each of which engages in only one
productive activity. An industry in an I/O table is then defined as a group of establishments engaged in the
same kind of productive activities.

4.13. Partitioning enterprises into units of homogeneous production is an important task in the collection
of production data and, as a consequence, the compilation of the symmetric I/O table. The best way to get
inputs and outputs from the units of homogeneous production in an enterprise is to ask it to do so with clear
instructions from data collectors so that there will be no need to use mechanical methods to separate them
afterwards. Below are the SNA rules recommended for partitioning enterprises into units of homogenous
production.

1. General SNA rules for creating a separate establishment

4.14. An establishment defined as a production unit consists of either an enterprise or a part of one, which
engages in a single kind of productive activity at a single location. To be an establishment, the unit must be
able to provide the outputs produced and inputs used, the number of people employed, the fixed assets and
changes in inventories used, consumed and put in place during the accounting period.2 With respect to
outputs, they need not be for sale and can be measured by imputation with the use of market prices for the
same type of products on sale or at cost as in the case of producers of government and non-profit services.

4.15. If an enterprise uses a single production process at a single location and produces only one product,
it is considered as one establishment. If the enterprise has many establishments, separate units of homogeneous
production should be created for each establishment. The SNA definition of establishments allows for
secondary activity producing secondary products.3 The operational definition is, therefore, inexact.
Furthermore, for I/O compilers of the symmetric table, secondary outputs and associated inputs must somehow
be separated. As a consequence, the I/O table would be much more reliable if the separation were done at the
source by business accountant who should be requested, whenever possible, to allocate costs to every product
the establishment produces. Cost allocation for the sake of determining product prices is not alien to business
accountants.

4.16. Most of the inputs used by each establishment and the outputs it produces can be itemized by the
enterprise accountants. However there are other costs that are shared by all establishments within the
enterprise. These costs are generated by ancillary activities, such as processing and communication of
information, transportation, storage, purchasing, sales promotion, cleaning, maintenance, security which are
needed to support the main productive activities of the enterprise. These activities cannot stand alone; they
are usually services produced for internal intermediate consumption and normally can be found in almost all

2
See System of National Accounts 1993, para. 5.26.

3
Ibid, para. 5.21.
 79

enterprises of similar type and therefore should be integrated with main activities by allocating the costs of
these activities to each main activity (more on allocation later). They should not be separated as establishments
but the inputs they use should be allocated to the establishments for which they provide services.

4.17. Certain types of activities, however, should never be treated as ancillary but should always be treated
as productive activities and separated as establishments. They include:

(a) The production of goods for further processing within the same vertically integrated enterprise
e.g. the production of cloth which is then manufactured into clothing, the extraction of crude
oil which is then refined into petroleum, etc.;

(b) Own-account production of goods and own-account construction by households,

Governments or enterprises;

(c) Mineral exploration in order to discover new deposits whether undertaken on own account
by enterprises or by other enterprises, which are hired to do the job.

2. Separation of own-account construction and vertically integrated activities

4.18. The separation of secondary activities from main activities, for example the separation of own-account
construction from other activities or of vertically integrated activities, as is the case of an integrated petroleum
industry combining both crude oil extraction and petroleum refining, is normally done by valuing outputs at
cost. In order to implement this separation in an input-output table, all costs pertaining to the activity have to
be separated out. The operating surplus of the enterprise as a whole has to be distributed proportionally to each
activity on the basis of value added minus operating surplus. Total costs plus operating surplus make up output
values. Despite the practical difficulties involved in partitioning vertical integrated enterprises into
establishments, the SNA recommends that when a vertical enterprise spans two or more headings at the first
level of breakdown of the ISIC (see chapter II, para. 2.56 of this Handbook) at least one establishment must
be distinguished within each heading.

3. Cost allocation of ancillary activities to establishments

4.19. When using establishments as the statistical units for production accounts, one has to deal with the issue
of cost allocation of ancillary activities in multi-establishment enterprises. A difficult accounting problem
arises within a multi-establishment enterprise when certain ancillary activities are carried out centrally for the
benefit of all individual establishments which make up the enterprise. They may include activities such as
marketing, sales, purchasing, accounting, data processing, computing, transportation, maintenance, and so on.
The central unit, which is often also the head office, is likely to have its own separate location away from the
establishments it serves. The costs of the ancillary activities are recommended by the SNA to be allocated over
all the establishments for which they provide services. If there are cost accounting records showing how much
of the central cost is attributable to the separate establishments, they should be allocated on this basis. In many
countries, cost accounting is normally prepared in large enterprises. If the necessary information is not
available, the cost of each type of input consumed by the central unit should be distributed over the
establishments served in proportion to the value of the output of each establishment less the value of its
intermediate costs excluding the costs of the ancillary activities themselves. If this method also proved to be
too difficult to apply, the cost of each central input may be distributed simply in proportion to the value of the
outputs of the establishments served. The approach proposed here in fact is not very different from that adopted
in enterprise cost accounting.
80

4. Ancillary activities with sales

4.20. In addition to the above case which is clear-cut, there are areas that are not so clear-cut. In fact, there
are enterprises that also sell part of the ancillary services to other enterprises. For instance, a computerized
database, an accounting system or storage facilities may first be developed for internal use but later on attract
customers. Should these activities continue to be treated as ancillary?

4.21. The SNA recommends that as ancillary units sell half or more of their outputs, they should be treated
as separate establishments. The outputs used internally will be treated as services consumed with imputed
values from the newly created establishment. When an ancillary unit is not classified as a separate
establishment, though a part of its output is sold on the market in addition to the outputs of the main activity,
secondary production shows up in the supply table. This secondary production reflects the output of the
ancillary unit that is sold. Of course, if enterprises are used as statistical units for production accounts, more
secondary products would appear in the supply table of the input-output framework. Thus, with the
recommendation of establishments as statistical units, the SNA attempts to reduce the secondary product
problem. At this point, it is important to note that the decision in the SNA to treat market output of ancillary
activities as secondary products actually leaves the difficult question of dealing with secondary products to
input-output statisticians and economists when they want to obtain symmetric input-output tables. The general
practice is to either sweep it under the rug by aggregation or to use mathematical methods on the basis of some
assumptions to reclassify the secondary outputs and their associated inputs. But this reclassification is nothing
other than creating a separate establishment of the market output of ancillary units.

4.22. The separation of ancillary units when at least half of their outputs are sold to other enterprises could
be complicated. If the entire output of an ancillary unit is separated as the output of a separate establishment,
several statistical difficulties will be encountered. For instance, how should its output be valued? Should the
entire output be valued at the price the output is sold to the market? Furthermore an additional operating
surplus needs to be imputed for the non-market portion of output of the ancillary activity. Secondly, if the
value of the non-market portion is based on costs plus an imputed operating surplus which is a portion of the
operating surplus of the enterprise itself, the SNA recommendation on valuation to treat each product as a
different product if it has a different price, is ignored.4 Therefore the best approach when compiling input-
output tables is to create only separate establishments for the market ancillary output while leaving the non-
market output with the main activity of the establishment. Though this is not recommended by the SNA, it is
better to remove secondary products by statistical rather than mathematical methods in the process of obtaining
the symmetric I/O table. The allocation of costs is similar to the procedure described above. In this case,
outputs of the same ancillary unit will be treated as different products since they are sold at different prices.

4
"In economic theory it is generally assumed that whenever a difference in price is found between two goods and
services which appear to be physically identical there must be some other factor, such as location, timing, conditions of
sale, etc., which is introducing a difference in quality...In most cases, therefore, differences in prices at the same moment
of time must be taken as prima facie evidence that the goods or services concerned, represent different qualities of the
same general kind of good or service." (SNA, para. 16.110).
 81

5. The redefinition method

4.23. The following example shows how the redefinition method operates. Let us take the example given
in table 4.1(a) below where product 3 is, in fact, produced elsewhere but due to reported data, that activity has
been aggregated as part of industry 1. Let us assume that we can find additional information on the input
structure of that activity as follows:

Product 1 1
Product 2 2
Product 3 0
Value added 3
Total output 5

Then the additional information can be inserted as column 3 of table 4.1(a) and deducted from column 1.5 This
method is similar to the mathematical method which is based on the commodity technology assumption, which
will be discussed later except that compilers may be directly involved in deciding which independent inputs
are to be assigned to the secondary products.

4.24. Given that data are collected on an establishment basis and consist of secondary products which are not
by-products or joint products resulting from production processes, it is still beneficial to collect input data
independently for the secondary products of the same kind and use this information to estimate the inputs
associated with the secondary products and transfer them out. For example, bicycle tyres may be a secondary
product of both the bicycle industry and the automobile tyre production, while no bicycle tyre single-activity
enterprises exist. The result is that in SUT no industry produces bicycle tyre. It is possible to use the
mathematical method based on the commodity technology assumption to transfer out secondary outputs and
their associated inputs, but the method may produce negative results. In this case, it is better to obtain
independent information on bicycle tyre production and use it to estimate inputs of the secondary products and
then transfer them out. In so doing, I/O compilers may be able to pass judgement on the quality of the
information. This is the essence of the redefinition method though it requires additional information.
Construction, which covers all kinds of activities with very different input structures, is also a secondary
product of many industries. It is important to identify it by specific kind and apply the redefinition method.
If the establishment is classified by activity at a detailed level in SUT, there would be more independent
information on the inputs of a specific product that is not tainted by the existence of other secondary products.

D. Treatment of secondary products resulting from production technology6

4.25. In order to separate by-products as defined previously from the main products of an industry, they
should be clearly identified in the product classifications that are used. Up to now there are no satisfactory
solutions for the separation of by-products and joint products in input-output tables, and only second best
solutions are offered here. Some of these solutions may be quite acceptable if the tables are desegregated and

5
This method is the same as the one suggested by P.J.A. Konijin in The Make and Use of Commodities by Industries,
on the Compilation of Input-Output Data from the National Accounts, Enschede, The Netherlands, Faculteit
Bestuurskunde, Twente University, 1994.

6
This is partly based on “Practices in Input-Output Table Compilation” by Vu Quang Viet in Regional Science and
Urban Economics, The Netherlands Vol. 24, No.1,1994.
82

detailed enough, but this is rarely the case. On the other hand, the issue may be exaggerated somewhat because
by-products and joint products are not pervasive in any economy, and generally low in values as compared to
principal products, and therefore second-best solutions are acceptable.

4.26. There are three types of methods in use to separate outputs and inputs of by-products and joint
products. These are the so-called negative transfer method developed by Richard Stone in which only output
is transferred, the aggregation or positive transfer method in which output and input of by-products remain
included with the main product of the industry in which they are produced, and a third method in which both
output and inputs are transferred to the by-products and joint products characteristically belong.

1. The negative transfer method

4.27. The negative transfer method developed by Stone is generally used for exclusive by-products. It is
illustrated with fictitious data presented in table 4.1(a) and the solution is given in table 4.1(b).

4.28. In the example of table 4.1(b) product 3 is the by-product of the first industry. It is entered as negative
input, so that the total output of the first industry is reduced from 115 to 110; the latter is equal to the output
of product 1. As the by-product is used by the second industry, it is recorded there as intermediate input. As
a result of this manner of recording, the output of the third industry representing the by-product is zero, and
the input column of the industry of by-products contains only zeros.

Table 4.1(a). An example of secondary products

Use table Intermediate consumption Final demand Product output

Product 1 10 20 80 110
Product 2 30 10 60 100
Product 3 5 0 0

Value added 75 65

Industry output 115 100

Supply table
Product 1 110 110
Product 2 100 100
Product 3 5 5
Industry output 115 100
 83

Table 4.1(b) The negative transfer method for exclusive by-products

Intermediate consumption Final demand Product output

Product 1 10 20 0 80 110
Product 2 30 10 0 60 100
Product 3 -5 5 0 0 0

Value added 75 65 0

Product output 110 100 0

4.29. The negative transfer method has implications for I/O analysis, which can be explained with the
following I/O coefficient table and Leontief inverse that are derived from table 4.1(b).

The I/O coefficient table resulting from negative transfer method

.0909 .2000 .0000

.2727 .1000 .0000
&.0454 .0500 .0000

The Leontief inverse

1.1786 .2619 .0000

.3571 1.1905 .0000
&.0357 .0476 .1000

4.30. Assume that final demand of the second product increases from 60 to 100 while the demand of the
first product remains the same. Since the by-product is required as input by the second industry, more
by-product will need to be produced to satisfy the increase in final demand of the second product. At the same
time, an additional amount of by-product (shown as negative inputs) would be required by an increase in the
output of the first product. There is no mechanism in the model that will generate a net output of by-products
equal to zero, as was the case in table 4.1(b) corresponding to a benchmark year. In the example, by
multiplying the Leontief inverse by the new final demand vector of (80 100 0), a total gross output of (120.5
147.6 1.9) would be needed. This implies a 1.9 net increase of by-products which can either be met by
additional imports of the by-product or increasing the production of the first product beyond what is needed
for final demand.

4.31. If alternatively it is assumed that final demand of the first product increases more than the second
product, for example (100 60 0), the required output vector will be (133.57 107.14 -.714). In this case, the
value of the by-product is not zero but negative. This is obviously the amount of the by-product produced by
the first producer but left unused.

4.32. When using the negative transfer method for ordinary by-products that are also produced elsewhere,
the input column of the by-product must be aggregated with the inputs of identical products that are produced
84

as principal products elsewhere. Assuming that the secondary product is also produced by the second industry,
the results are shown in table 4.1(c). The negative transfer method applied in this case will result in similar
distortions as were mentioned for exclusive by-products. In this case, changes may occur in the proportion
of market shares between the by-product and the product identical to it which was produced elsewhere and in
the relative weights between the output of the industry that produces the by-product and the outputs of
industries that use the by-products as inputs.

Table 4.1(c) The negative transfer method for by-products when by-products are aggregated with
similar products produced elsewhere

Intermediate consumption Final demand Product output

Product 1 10 20 80 110
Product 2 30-5 10+5 60 100

Value added 75 65

Product output 110 100

4.33. Mathematically, the I/O coefficient matrix Acc derived from the negative transfer method when by-
products are aggregated with the same products produced elsewhere is as follows:

4 V̂&1
Acc ' (U & V)
where V̂ denotes diagonalization by elimination
of the off&diagonal elements of matrix V;
V4 denotes off&diagonalization by elimination of
the diagonal elements of matrix V
and U the intermediate matrix of the use table.

4.34. The above examples show that when using the negative transfer method much care should be taken
in interpreting the final data after application of the Leontief inverse. They also show that this method will only
be workable as a basis for analysis, as long as the industry of by-products is not aggregated with other
industries. By identifying by-products and the corresponding industry separately, it will be possible to analyse
changes in market shares more explicitly.

2. The aggregation or positive transfer method

4.35. The positive transfer method is practiced by some I/O compilers for exclusive by-products (see
definition in para. 4.11). Following this method, exclusive by-products are treated as if they were primary
products of the industry where they are produced. This treatment is illustrated in table 4.2.
 85

Table 4.2 Treatment of exclusive by-products using the positive transfer method

Intermediate consumption Final demand industry output

Product 1 10 20+5 80 115
Product 2 30 10 60 100

Value added 75 65

Industry output 115 100

4.36. In this case, output of the first industry also includes the by-product, and thus is 115 instead of 100.
The third industry disappears in table 4.2 as compared to table 4.1(b) . The by-product is sold by the first
industry to the second industry as though it were the principal product; this results in an increase of inputs of
product 1 into industry 2 from 20 to 25, as compared to table 4.1. Again this method will only provide realistic
outputs in I/O analysis if the proportion of product 1 to product 2 remains constant over time. If this method
is used, it will not be possible to detect structural changes in the use of the by-product, as was the case with
the negative transfer method, as the by-product is not separately identified. The method has a major defect,
i.e. the input structure of the industry that produces these products as primary products is distorted by the
inclusion of the fictitious purchase. As a result, whenever the demand for the product increases, so will the
production of the industry that produces this product as by-products, which certainly is not true. To treat
exclusive by-products, the negative transfer method is preferable.

3. Transfer of outputs and inputs

4.37. For ordinary by-products and joint products, compilers may apply the redefinition method for
transferring out both outputs and inputs, if separate input information on similar products which are produced
elsewhere is available. This method may be arbitrary since the assumption of similarity is wrong but until now
no other method seems to be available to deal with ordinary by-products properly.

4.38. In the case of ordinary by-products there is an advantage in applying this procedure especially when
the product in question is largely produced separately. The use of this procedure could be rightly criticized on
the ground that by-products should not be separated from the principal product, since no additional inputs are
required, as was the case in the examples presented in table 4.1(a). In reality, however, this is rarely the case,
because at least some inputs specific to secondary products are required. Hydrogen as a by-product of
petroleum refining needs at least to be bottled before delivery to the market. The more specific costs are needed
for processing a by-product, the more reasonable the argument for allocating costs to the by-product.
Furthermore, producers of the principal and by-products, when making production decisions, would always
compare the revenues of the by-product with the corresponding costs.

4.39. There is no alternative to the separation of joint products into separate establishments by cost
allocation. Since there already are costs that are specific to each product, only the costs that are common to all
products need to be allocated. The allocation is similar to the allocation of ancillary costs or the redefinition
method.
86

E. Transferring secondary outputs and associated inputs by mathematical methods

1. General

4.40. The recommended solution for separating secondary products from the sources by using more
information which was discussed previously in paras. 4.4. and 4.24 is also based on the findings that until now
no mathematical methods are able to satisfactorily derive the traditional symmetric input-output coefficient
table that is the core of input-output economics, from both conceptual and practical points of view. These
methods may still be needed as a last resort given that all possible information has been fully utilized. They
will now be reviewed. There are two methods that are used for combining the make and use matrices7 to derive
the traditional input-output table. Before these methods are reviewed, some explanation is needed about the
traditional symmetric input-output table.

4.41. It is commonly known that the supply and use tables do not have to be square. In fact, the number of
products may be more than the number of industries (producers) and vice versa. In these cases, both the use
and make tables are rectangular. However, a symmetric or square input-output matrix is required for input-
output analysis, as only a square matrix can be inverted to obtain the Leontief inverse. A symmetric table can
be a product-by-product or an industry-by-industry matrix. Both are mentioned in the SNA. In the first
version a column in the intermediate matrix represents a product technology and a row represents the
distribution of a product to intermediate inputs and as final use. In the industry-by-industry version a column
represents an industry technology containing all inputs required by that industry, and a row represents the
distribution of the industry output (which also contains secondary products) to all industries and to final
consumers. The second type of I/O table is much less useful than the first one because an industry might
represent a group of establishments, part of which may be artificially created by mathematical methods, and
therefore does not reflect any "realistic" picture of the economy. It would be more useful if each industry
represented a group of enterprises instead. One should also be careful to use the inverse of the industry-by-
industry matrix when a significant time lag is involved because linear and fixed technical assumptions cannot
be applied here. The inversion of this matrix should be done indirectly through the inversion of the product-
by-product matrix so that whenever there are changes in market shares, a different industry-by-industry matrix
is obtained, even though the product-by-product matrix remains the same.

4.42. There are basically two methods to combine the use and supply matrices mathematically to generate
the traditional symmetric input-output matrix. These methods are based on either the industry technology
assumption or the commodity technology assumption.

4.43. The industry technology assumption assumes that inputs are consumed in the same proportions by
every product produced by a given industry, which means that principal and secondary products are all
produced using the same technology, i.e. the same input structure. This assumption has been used by many
countries on the basis of a recommendation made by the 1968 SNA mainly for two attractive reasons: first,
the method always generates positive symmetric input-output tables; second, it is also applicable to the case
of rectangular input-output tables. The method has since been found to break the fundamental economic rule
that products with different prices at a given moment must reflect different costs or different technology.
Therefore it cannot be considered acceptable.

7
The make matrix refers to the part of the supply table which describes domestic production. The use matrix refers
to the part of the use table which describes intermediate consumption.
 87

4.44. The commodity technology assumption assumes that the input structure of the technology that
produces a given product is the same no matter where it is produced. This assumption though economically
more reasonable than the industry technology assumption is not widely used because it tends to generate
negative symmetric input-output tables and requires the make and intermediate matrices of the use table to
be squared. In order to get rid of these negative coefficients, which are mostly small, additional methods have
been used to adjust data in order to produce positive symmetric tables. The more prevalent methods are (i)
setting all negative values to zero and using the RAS technique (which will be discussed in chapter IX) to
balance the table and (ii) optimization such as minimization of variances under constraints to generate positive
values. However, the latter is also questioned on other grounds such as an economic justification for a specific
form of the objective function. Economically, the commodity technology assumption makes more sense than
the industry technology assumption.

4.45. The two mathematical methods will be discussed in the following sections, based on the following
notations and definitions.

Notations

m number of products
n number of industries
Umxn the intermediate matrix of the use table (product by industry)

Bmxn the use coefficient matrix (product by industry)

Mmxn the make matrix (product by industry),

part of the supply matrix describing domestic production
Dsubnxm: the market share matrix (industry by product)
gn the vector of industry output

qm the vector of product output

ĝ the diagonal matrix of industry output

q̂ the diagonal matrix of product output

4.46. A matrix
with a " ^ " it is a diagonal matrix version of the vector with the same notation; all off-diagonal elements are
zeros. For example:

g1 0 0
ĝ ' 0 g2 0
0 0 g3

Basic definitions:
(4.1) B ' Uĝ &1
(4.2) D ' M )q̂ &1
88

4.47. The numerical examples shown in table 4.3 below will be used to clarify the derivations of Leontief
input-output coefficients on the basis of the make and use matrix.

2. The industry technology assumption

4.48. On the basis of this assumption, a product j can be produced by various industries k; each industry k
needs bik of input i per unit of industry product j, where bik, i = i ... n represents the industry technology of an
industry k; and each industry k has only a part of the market of product j. This market share of industry k in
the production of product j has a notation dkj. So all inputs i needed to produce one unit of product j by
different producers can be written as follows:

(4.3) aI,ij ' j b ik d kj

n

k'1

4.49. Thus, formula 4.3 shows that input i required for one unit of product j is a weighted average of the
input structures of the producers where product j is produced; the weights are the market shares of each
producer in the production of product j. In matrix form, equation 4.3 is written as:

(4.4) AI,cc = BD

where i refers to the industry technology and cc refers to the order of matrix A which is product by product.

4.50. As B is a product-by-industry matrix, D is an industry-by-product matrix, and the matrix AI,cc is a

product-by-product matrix. Thus AI,cc is the I/O coefficient matrix that describes products directly required to
produce other products.

4.51. The following example in table 4.3 which is the same example as in tables 3.1 and 3.5 of chapter III
will show how all various matrices are calculated:
 89

Table 4.3
An example of the use matrix and make matrix

U use matrix M make matrix

Industry Industry q

19 28 10 156 24 0 180
Product 29 18 8 Product 9 80 0 89
7 7 3 0 0 62 62

TX 2 2 2
VA 108 49 39

g 165 104 62 165 104 62

Industry outputs g = 165, 104, 62

Product outputs q = 180, 89, 62
Taxes less subsidies on products (on uses) TX = 2, 2, 2
Value added VA = 108, 49, 39

B = U•-1 D = MNq-1

Industry Product

19/165 28/104 10/62 156/180 9/89 0/62

Product 29/165 18/104 8/62 Industry 24/180 80/89 0/62
7/165 7/104 3/62 0/180 0/89 62/62

TX 2/165 2/104 2/104

VA 108/165 49/104 39/62

4.52. B normally refers to the product coefficient matrix only, i.e. the upper 3 rows and 3 columns. In
equation 4.4, if B also includes the rows of taxes and value added, one can obtain the rows of taxes and value
added by the production of each product. This can be observed in the resulting numerical example of AI,cc
shown below. Unless otherwise specified later, B refers to a product-by-industry matrix, excluding the rows
of taxes and value added. With the values of B and D given above,

.1357 .2536 .1613

.1754 .1733 .1290
AI,cc ' .0457 .0648 .0484
.0131 .0185 .0323
.6301 .4897 .6290

4.53. The basic equation 4.5 can be used to derive different types of symmetric I/O models:
90

- Product-by-product I/O model: this model calculates the impact of a final demand of
products on the production of products,

- Industry-by-industry I/O model: this model calculate the impact of a final demand of
industry outputs on the outputs of industries.

- Industry-by-product I/O model: this model calculates the impact of a final demand
of products on the outputs of industries,

4.54. In order to show these different models that can be derived from the industry technology assumption,
it is necessary to describe first the basic relationships in the use and supply tables in basic prices as shown in
tables 3.1 and 3.5 of chapter III.

4.55. The basic relationship in the SNA use table 3.5 is:

(4.5) q = Bg + YC

where

YC: final demand of products which is the vector of 3 rows from 1-3 summing over
columns 4 to 10 of table 3.5 of chapter III.

(4.6) g = Dq

(4.7) YI = DYC

(a) Product-by-product I/O model

4.56. Substituting equation 4.6 for g in equation 4.5 gives

(4.8) q = BDq + YC

then

(4.9) (I-BD)q = YC

4.57. Equation 4.9 can be used both to calculate the impact of YC on q and rearrange the I/O table 3.5 in
chapter III.

4.58. Using the product-by-product matrix AI,cc = BD shown in both equations 4.9 and 4.4, one is able to
calculate the flow of products, taxes and value added required for the product-by-product I/O table. Table 4.4
below presents the symmetric I/O table based on a product-by-product classification. It is the result of merging
the supply table 3.1 and the use table 3.5 in chapter III.

(b) Industry-by-industry I/O model

4.59. By multiplying both sides of equation 4.5 by D and then replacing DYC and Dq by the relationships
in equations 4.6 and 4.7, one obtains:
 91

(4.10) Dq = DBg + D YC

(4.11) g = DBg + YI

(4.12) (I-DB)g = YI

4.60. This model can be used both to calculate the impact of a final demand for industry outputs on industry
outputs and to rearrange use table 3.5 in chapter III. The I/O model based on equation 4.12 and tables 3.1 and
3.5 in chapter III is shown in table 4.5. This model is however of almost no interest to analysts since final
demand is rarely in terms of industry outputs. In this model, imports continue to be classified by products.

(c) Industry-by-product I/O model

4.61. Substituting DYC for YI in equation 4.12 gives

(4.13) (I - DB)g = DYC

or

(4.14) g = (I - DB)-1 DYC

4.62. This model is a variation of the model shown in paragraph 4.59. It can be used to calculate the
impact of a final demand for products, YC, on industry outputs. However, it cannot be used to arrange the
flow I/O table in a meaningful manner because final demand is now classified by products and outputs
classified by industries.

3. The commodity technology assumption

4.63. The basic premise of the commodity technology assumption is that a given product uses the same
input structure irrespective of the industry where it is being produced. This can be written as

(4.15) u ij ' j a ik mjk

n

k'1

where uij is input i required by industry j, m jk is product k produced by industry j and a ik is the input i required
to produce one unit of product k. Since an industry produces a number of products and each product requires
a different set of inputs, the amount of inputs required by industry j will be the sum of the inputs i required for
each of its products mjk. Equation 4.15 can be written in the following matrix form:

(4.16) U = AC,cc M
 Table 4.4. I/O symmetric table at basic prices, product by product
industry technology assumption

Intermediate consumption Total Household Government Gross Exports Imports c.i.f./f.o.b. Total
of industries economy final final capital f.o.b. c.i.f. adjustment use of domestic
expenditures expenditures formation (total products at basic
f.o.b.) prices

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)=(1)+.+(10)

(1) Product 1 24 23 10 71 40 27 -15 180

(2) Product 2 32 15 8 23 10 8 -8 1 89

(3) Trade and transport services 8 6 3 43 0 2 -1 1 62

(4) Taxes less subsidies on products 2 2 2 14 0 0 20

(5) Direct purchases abroad by residents 3 -3 0

(6) Direct purchases at home by non-residents -1 1 0

(7) C.i.f./f.o.b. adjustment +2 -2 0

(8) Total uses at purchasers' prices (1)+.+(7) 66 46 23 153 10 40 38 -25 0 351

(9) Total gross value added/GDP 216

(10) Gross value added at basic prices 114 43 39 196

(12) - (8)

(11) Taxes less subsidies on 20

production and imports

(12) Total industry output at basic prices 180 89 62 331

 Table 4.5. I/O symmetric table at basic prices, industry by industry
industry technology assumption

Intermediate Total Household Government Gross Exports Imports c.i.f./f.o.b. Total

consumption economy final final capital f.o.b. c.i.f. adj. use of domestic
of industries expenditures expenditures formation (total products at basic
f.o.b.) prices

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)=(1)+.+(10)

(1) Industry 1 19 26 10 65 1 35 24 -15 165

(2) Industry 2 29 20 8 29 9 5 11 -8 1 104
(3) Industry 3 7 7 3 43 0 2 -1 1 62
(4) Taxes less subsidies on products 2 2 2 14 0 0 20

(5) Direct purchases abroad by residents 3 -3 0

(6) Direct purchases at home by non- -1 1 0
residents

(7) C.i.f./f.o.b. adjustment +2 -2 0

(8) Total uses at purchasers' prices 57 55 23 153 10 40 38 -25 0 351

(1)+.+(7)

(9) Total gross value added/GDP 199

(10) Gross value added at basic 108 49 22 179
prices (12) - (8)
(11) Taxes less subsidies on 20
production and imports

(12) Total industry output at basic prices 165 104 45 314

 Table 4.6. I/O symmetric table at basic prices, product by product
commodity technology assumption

Intermediate Total Household Government Gross Exports Imports c.i.f/f.o.b. Total

consumption economy final final capital f.o.b. c.i.f. adjustment use of domestic
of industries expenditures expenditures formation (total products at basic
f.o.b.) prices

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)=(1)+.+(10)

(1) Product 1 19 28 10 71 40 27 -15 180

(2) Product 2 32 15 8 23 10 8 -8 1 89

(3) Trade and transport services 7 7 3 43 0 2 -1 1 62

(4) Taxes less subsidies on products 2 2 2 14 0 0 20

(5) Direct purchases abroad by residents 3 -3 0

(6) Direct purchases at home by non-residents -1 1 0

(7) C.i.f./f.o.b. adjustment +2 -2 0

(8) Total uses at purchasers' prices (1)+.+(7) 60 52 23 153 10 40 38 -25 0 351

(9) Total gross value added/GDP 216

(10) Gross value added at basic prices 120 37 39 196

(12) - (8)

(11) Taxes less subsidies on production 20

and imports

(12) Total industry output at basic prices 180 89 62 331

 95

then

(4.17) AC,cc = U M-1

where C refers to the commodity technology assumption, cc refers to the order of matrix A which is
product-by-product. As U is a product-by-industry matrix and M is a product-by-industry matrix then A
must be a product-by-product matrix. The relationship in 4.17 implies a strong restrictiveness of the
commodity technology assumption, i.e. M is invertible only if M is square or the number of industries must
equal the number of products. This mathematical requirement is unrealistic since the number of industries
needs not equal the number of products unless statisticians make it so by aggregation.

4.64. AC,cc can also be proved to be the same as BC-1 where:

(4.18) C ' Mĝ &1

The proof is as follows: insert (g-1g), which is an identity matrix, into the middle of the right-hand side of
equation (4.17), then:

(4.19) AC,cc ' U(ĝ &1ĝ)M &1

' (Uĝ &1)(ĝ M &1)
' BC &1
where C ' ĝ M &1
&1

4.65. Again the relation in equation 4.17 can be used to link directly product outputs and a final demand
for products. Table 4.6 shows the resulting product-by-product symmetric table merging the supply table
3.1 and the use table 3.5 in chapter III.

4.66. In order to relate industry outputs to a final demand for products, one can replace q in equation 4.5
(para.4.55 above) with Cg (this relation is based on the commodity technology assumption8 and obtain the
following relationship:

(4.20) g = C-1Bg + C-1 YC or

(4.21) g = (I - C-1B)-1C-1 YC

Equation 4.21 is an alternative to equation 4.14 in paragraph 4.61.

4.67. As can be observed from equation 4.17, the inversion of matrix M requires that the number of
products equal the number of industries. Thus a rectangular matrix is not applicable when the commodity
technology assumption is used. This requirement, which is not as restrictive as it sounds, merely means
that for any secondary product to be transferred out, there must be a producer that also produces this
product elsewhere so that its inputs can be used to remove the inputs of the secondary products. Other
products produced only as by-products have to be treated by aggregation as though they were part of the
primary products.

8
q = Cg may be obtained as follows: g = •i = (C-1M)i = C-1(Mi) = C-1q.
96

4.68. It is interesting to see the implication of equation 4.17 in the case of a two-industry economy with
the first industry producing a secondary product:

u11 u12 m11 0

where U ' and M '
u21 u22 m21 m22

u12
u11 & m21
m22 u12
m11 m22
(4.22) Ac '
u22
u21 & m21
m22 u22
m11 m22

For industry 2, there are the usual coefficients a12 = u12/m22 and a22 = u22/m22 but for industry 1 different
ones apply, i.e.

(u11 & a12m21)

(4.23) a11 '
m11

and

(u21 & a22m21)

(4.24) a21 '
m11

4.69. The derived technical coefficients in 4.23 and 4.24 are net input over net output of an industry
where net output is total industry output minus secondary products, and net input of an industry is equal to
total input minus the input required by the secondary products produced by that industry. More
importantly, from equations 4.23 and 4.24 one can see that as the inputs required for secondary products
are removed from total input, the derived technical coefficient can be negative if one of the following
occurs:

(i) There is over-specification of the secondary products, i.e. the output of the secondary
product in the make matrix (the supply table), in our example, product 2 produced in
industry 1, is misclassified;

(ii) The secondary product is not exactly the same as the product produced as a primary
product elsewhere; it requires less inputs than assumed;

(iii) There are errors in data.

 97

4.70. The commodity technology assumption may generate negative coefficients but economically these
can be explained by one of the reasons cited in paragraph 4.69. More often, a given input is not required by
an industry, but if secondary products require that input, the removal of secondary outputs and associated
inputs will certainly generate negative value for the technical coefficient aij. The solution to the problem of
negative coefficients is to recheck data themselves. Significant secondary products and their associated
inputs must be transferred by using the redefinition method on the basis of the information provided by
establishments producing only these kinds of secondary products or collected by special surveys. There is a
need to emphasize again the original suggestion that, as much as possible, data should be processed with
the application of separation rules and cost allocation from the sources.

4.71. In cases where negative coefficients are very small in comparison to other coefficients in the same
columns, practitioners may set them to zero and balance the tables by the RAS method (see chapter IX).

4.72. The commodity technology has been practiced in a very few countries of which Germany is one.
Germany applies the commodity technology assumption and then makes corrections for the coefficients
that are negative9 by transforming or obtaining a specific make matrix M to apply to the industry in
question (this implies that the reason for the negative coefficients is reason (i) mentioned in paragraph
4.69. However, changes in matrix M would automatically change matrix A since A is calculated on the
basis of U and M). The procedure can be explained by writing the flow intermediate matrix as follows:

F ' AC,cc q̂
' UM &1q̂
' UT &1
or Fk ' UkT &1
where T ' q̂ &1M

T is the transformation, which is nothing more than the matrix of row coefficients of the matrix of product
output. This means that a different Tk would be used for a different product k. Basically the transformation
relies on available information to change manually the matrix Mk that would be used for the specific
product k so that all ak. would be non-negative and to keep all the value Uk. to be the same as in the kth row
of matrix U. These relationships to be satisfied are shown in the table below:

Industries Row total

ak1M11 . . ak1M1m = ak1 q1
Products . . . . .
ak1M11 . . ak1M1m = ak1 q1
Column total Uk1 . . Ukm

4.73. The United States uses the redefinition method to separate secondary products for which
additional data can be found with the rest of the secondary products treated by the industry technology

9
Carsten Stahmer, “Transformation Matrices in Input-Output Compilation” in Input-Output Modeling, edited by
Smyshlyaev, Berlin, Springer, 1985.
98

assumption. Knonijn (see footnote 5) has tried to apply the redefinition method to the data for the
Netherlands, with the rest of the secondary products treated by the commodity technology assumption.

4.74. Austria uses the hybrid technology assumption, i.e. some secondary products are treated by the
commodity technology assumption10 and some by the industry technology assumption. Rainer and
Richter11 argued that in the case of joint products, secondary output is produced by the typical technology
of the respective industry and not by any commodity-specific technology. At least three cases can be
found: (a) chemicals of different kinds produced by refineries, (b) pipeline services provided by the
industry extracting natural gas, (c) production of electricity by the industry providing long distance heating.

4. Evaluation of the two assumptions

4.75. There are strong reasons to support the commodity technology assumption. The first one is that it
is economically plausible. The second one is that the assumption fulfills all the criteria mentioned below12:

(a) Material balance: Total output is equal to total intermediate consumption plus final
demand;
(b) Financial balance: For every industry, the price equation will hold when applied to
revenues and costs of producers;
(c) Scale invariance: The derived symmetric coefficient matrix is also invariant to a scaling
factor, i.e. if inputs and output of an industry in the original use and make matrices are
increased by the same proportion, it is possible to derive the same coefficient matrix;
(d) Price invariance: If a new price base is applied to the data, the same derived symmetric
input-output coefficient matrix A is obtained;

More explicit discussion of these criteria will be found in the appendix to this chapter.

4.76. On the other hand, the industry technology assumption has two distinct advantages:

(a) Matrix AI is always positive because matrices B and D are always positive.

10
The mathematics for combining the commodity technology assumption and the industry assumption can be found
in the annex to chapter III of A System of National Accounts (ST/STAT/SER.F/2/REV.3, Sales No. E.69, XVII,3, 1968)
or T. Gigantes, “The Representation of Technology in Input-Output System” in A.P. Carter and A. Brody, Contributions
to Input-Output Analysis, Amsterdam, North Holland, 1970. Since the I/O model is linear and additive, the application
of hybrid technology can be implemented sequentially using the appropriate formulas discussed in this chapter. M is
split into M1 and M2, each representing a given type of secondary products produced by a given technology. When M1
is used, the other type of secondary products is treated as if it were primary.

11
Nobert Rainer and Josef Richter, “Some Aspects of the Analytical Use of Descriptive Make and Absorption
Tables”, Economic Systems Research, Vol.4, No.2, 1992.

12
These criteria are formulated in a paper by Pieter K. Jansen and Thijs ten Raa, “The Choice of Model in the
Construction of Input-Output Coefficients Matrices”, International Economic Review, Vol. 31, No. 1, Feb. 1990. The
original paper in which problems with the technology assumption were first pointed out was, “An Alternative Approach
of Negatives in Input-Output Analysis”, Review of Economics and Statistics, No. 66 (1984) by T. ten Raa, D
Chakraborty and J. A. Small.
 99

(b) B and D can be rectangular and the product matrix AI is always square.

4.77. However, there are some fundamental problems with the industry technology assumption that may
lead economists to reject it out of hand. The assumption that different products are produced with the same
input structures of the industry in which they are produced is economically nonsensical. The reason is that
if any two products having the same costs because they are produced by the same technology in a market
where prices are always set equal to costs (this is the assumption in input-output economics and in a
perfect competitive market), they cannot have different prices. But this is exactly what is required by this
assumption which cannot be met except by coincidence. Besides that, the industry technology assumption
requires that market shares be constant over time, which is unrealistic. The interaction between this
assumption and the market shares has led to violations of three important criteria listed in para. 4.74,
except the material balance criteria.

4.78. It is suggested that the negative transfer method without aggregation be used to treat exclusive by-
products and the redefinition method be used in combination with the commodity technology assumption
method to treat ordinary secondary products. The redefinition method should be utilized as far as
additional information permits and especially for products that generate large negatives. Small negatives
may be set to zeros and differences in column and row sums are settled by table balancing. If the industry
technology assumption is used to calculate the symmetric I/O matrix, it is suggested that the redefinition
method be used to transfer secondary products of significant value by fully utilizing the basic information
available before it is applied. Some compilers13 have argued that the question of whether an industry
technology is valid or not is an empirical one and therefore cannot be excluded from theoretical
considerations alone.

4.79. The redefinition method is necessary when the number of products is larger than the number of
industries and when one wants to apply the commodity technology assumption which requires that the
number of products equal the number of industries. Another reason for use of this method is the treatment
of the use of own output within an establishment for intermediate consumption in the 1993 SNA. Let us
assume that dairy establishments produce both cheese and milk. The production of cheese requires milk as
an important input but because milk is produced and consumed within the same establishment, milk is not
treated as input by the SNA, only materials and services that are used to produce the milk that is then used
to produce cheese are shown as inputs. In this case, the dairy industry will produce cheese as secondary
output. If cheese is also produced elsewhere as a primary product, then the use of the commodity
technology assumption to separate cheese and milk products in the dairy establishments will generate
negative input of milk for the productions of milk. In this special case, even the redefinition will also
produce a negative input. The only way to prevent this is to treat milk explicitly as an input in the
production process of the dairy establishment. In this way, output of milk in the input-output framework
will be higher than in SUT.

Appendix

DESIRABLE PROPERTIES OF SYMMETRIC INPUT-OUTPUT COEFFICIENT MATRICES

13
Denmark Statistik, Commodity Flow Systems and Construction of Input-Output Tables in Denmark, 1986.
100

4.80. This appendix discusses the criteria formulated by Jansen and ten Raa14 to make judgements on
desirable properties of the symmetric input-output coefficient matrix A derived by using various assumptions
on secondary products. Their paper also shows that only the commodity technology assumption fulfils all their
proposed criteria. Readers can see complete proofs in their paper; this appendix however will only present the
criteria and show with examples that the industry technology assumption does not fulfil most of them.
According to their proposal, any derived input-output coefficient matrix should satisfy the following balances:

1. Material balance

4.81. The material balance requires that total product output should match total input requirements.

(4.25) q = A(U,M) q + y

where q is the vector of product output, y is the vector of final demand and A(U,M) is the input-output
coefficient matrix that is derived from the make and use matrices U and M on the basis of a certain assumption.

4.82. Both the commodity technology and the industry technology assumptions satisfy this balance, but
other methods such as the transfer method of output as negative input (the Richard Stone method) and transfer
of output as positive input do not.

2. Financial balance

4.83. The financial balance requires that the value of a product be equal to its costs.

(4.26) pN = pN A(U,M) + vN

where p is the product price vector and v is the vector of value added by product. The product prices calculated
in 4.26 must also hold when applied to the balance of revenues and costs of producer k. It is important to
remind readers that all elements of p' will be 1 (see chapter I for explanation). Producer k in producing a share
mjk of product j will receive a revenue pj mjk . The costs of this share will be the element jth of the following
vector (pN A + vN) multiplied by mjk. When all revenues are summed over all products j produced by k, these
revenues must also be equal to their costs:

(4.27) pN M.k = pN A(U,M) M.k + vN M.k

where M.k stands for column k of matrix M which shows all outputs produced by producer k. In equation 4.27,
the left side shows the revenues earned by producer k; the first term of the right side shows intermediate input
costs and the second term shows value added of producer k. This value added by producer k can also be
calculated straight from the U and M matrices as vNM.k = eNM.k - eNU.k where e stands for the vector in which
all elements are equal to 1 and U.k stands for column k of matrix U. This last calculation is simply the
difference between the sum of outputs produced by producer k and its intermediate costs. Substituting the
new value of vNM.k into 4.27, the financial balance for producer k is obtained:

14
See footnote 12 for references.
 101

(4.28) eN A(U,M) M.k = eN U.k

4.84. Thus in order to satisfy the financial balance, the derived matrix A has to satisfy equation 4.28. The
balance for all producers is written as follows:

(4.29) eN A(U,M) M = eN U

4.85. It is interesting to know that only the commodity technology assumption satisfies this balance.

4.86. Using the examples listed at the end of the appendix, the applicability of the industry technology
assumptions can be tested:

AI = BD

1 1 1 1
0 1
4 2 4 8
AI ' x '
1 1 1 1 1
0
2 2 2 2 2

The left side of 4.29 becomes

1 1
4 8 1 0 11 5
1 1 x x '
1 1 1 1 8 8
2 2

The right side of 4.28 is

1
0
2 3 1
1 1 x '
1 2 2
1
2

4.87. This example shows that the financial balance is not fulfilled by the industry technology assumption.
Readers could test the commodity technology as an exercise. Values of the A matrix derived from the
commodity technology assumption and other relevant values are given in the example at the end of the
appendix.

3. Scale invariance

4.88. This property requires that the derived matrix A remain unchanged when the outputs of a producer
and its associated inputs are increased by the same proportion, i.e.

A(Us,sM) = A(U,M)
102

where diagonal matrix s stands for the scales.

4.89. This criterion is not fulfilled by the industry technology assumption either, but it is by the commodity
technology assumption.

Let us take the case in which

2 0
s '
0 1

This means that the values of outputs and associated inputs of the first producer are assumed to double and
those of the second producer remain unchanged.

4.90. When this scale matrix is applied to the make-and-use matrices, the A matrix derived on the basis of
the technology would not remain the same. The new A matrix would be

1 1
4 6
AI '
1 1
2 2

which is not the same as the value of A shown in 4.86.

4. Price invariance

4.91. This property requires that when the use and make matrices are rebased on a different price base, the
derived A matrix must remain the same, i.e.

A(pU, Mp) = pA(U,M)p-1

where p is the diagonal matrix in which each element corresponds to the new price index, for example:

2 0
p '
0 1

4.92. This criterion is also met by the commodity technology assumption but not by the industry technology
assumption. For the industry technology assumption, the derived AI matrix is
different:
1 1
3 6
AI '
1 1
3 12
103
 PART TWO

COMPILATION OF SNA SUPPLY AND USE TABLES

 107

V. COMPILATION OF PRODUCTION ACCOUNTS OF INDUSTRIES

A. Introduction

5.1. Outputs, intermediate inputs (or intermediate consumption) and value added for each industry or kind
of economic activity are normally compiled together because conceptually they are linked together as parts of
the production function of an establishment or activity, and also because the main sources of information for
most nonfinancial activities are normally the same, particularly for the benchmark years with more detailed
information from the censuses. Annually, information comes from annual surveys, tax returns, business
accounts, and other secondary sources such as trade and producers' associations, etc. Outputs, total
intermediate inputs and total value added calculated residually, as a rule serve as total controls for the further
detailing of the components of inputs of goods and services used in production. For these reasons, they will
be discussed together in this chapter. To link production account by industry or establishment to production
account by institutional sector, it is necessary to link every establishment to the institutional sector it originates
from (see table 2.5 in chapter II).

5.2. Inputs can be observed from two points of view. From the production point of view, i.e. along the
columns of the use table, information for inputs comes from business accounts and production surveys. From
the marketing or distribution point of view, i.e. along the rows of the use table, information for inputs comes
from marketing boards, agencies that track the supply and uses of special products. Many products such as
grains, petroleum, other minerals, electricity, microcomputers, etc. are tracked from the distribution point of
view. These pieces of information are broad in scope, covering group of industries, and therefore need to be
broken down and reconciled with information from the production point of view. They are normally valued
in either basic or producers’ prices, in contrast to production inputs that are valued in purchasers' prices. In
this case, they must be transformed into purchasers' prices for reconciliation. General aspects of outputs,
intermediate inputs, and value added from the production point of view are discussed below before a more
detailed discussion of the compilation of specific activities or industries. Chapter VII will discuss final demand
from the distribution point of view. However, it is important to recognize that the compilation of gross capital
formation, a component of final demand, should be carried out simultaneously with the compilation of
production outputs because not only are the same sources of information used but gross capital formation needs
to be classified by industry so as to provide information to estimate capital stocks by industry if their time
series is of adequate length and available.

5.3. The discussion of SNA concepts in this chapter may be more extensive than needed by national
accountants. However, this is justified in two ways: (a) it frees input-output compilers from going through the
whole 1993 SNA unless it is really needed; (b) it illustrates more clearly SNA concepts that were newly
introduced in the 1993 SNA.
108

B. General remarks on compilation of production accounts

1. Outputs

5.4. Compilation of industry output and product output is an important task in the construction of the input-
output framework, i.e. the use and supply tables. Though the number of industries and products in the final
supply and use tables may be no larger than 100, the initial number of industries for which outputs are prepared
and the number of goods and services (or products) are normally much larger and should be as large as
information allows. The detailed elaboration allows compilers to use more reliable information from censuses
and secondary information as well as "expert" knowledge. With detailed product outputs, which some
countries extend to a few thousand products, the use of the commodity flow approach (which will be discussed
in chapter VIII) will also make it much easier to distribute outputs to various intermediate and final users. For
instance, in a highly desegregated form in countries like the United States, Canada and those of Western
Europe where coffee beans are entirely supplied from abroad, these imported products are utilized as inputs
to either the industry that makes roasted coffee or to households final consumption. However, if coffees are
aggregated with other agricultural products into one industry, we will not be able to use the above-mentioned
"expert" knowledge. More detailed elaboration also allows compilers a better selection of more appropriate
proxies on the basis of "expert" opinions for input structures of the industries under consideration. For
instance, within a given industry in which there are a few sub-sectors and out of which only one sub-sector is
significant in output value in comparison with outputs of other sub-sectors, we may try our best to obtain
information on the input structure of the most significant sub-sector and be satisfied with some rough
approximation for other insignificant sub-sectors if we do not have financial resources to survey input
structures of all sub-sectors. The distribution of product output in conjunction with input structure surveyed
will fill up the supply and use shown below in table 8.1 of the commodity flow method in chapter VIII which
becomes basic information for the SNA input-output framework.

2. Intermediate consumption

5.5. Intermediate consumption includes all non-durable goods and services with an expected life of under
one year which are used up in the process of production by industries. Small tools of low value, though
durable, should also be treated as intermediate goods.

5.6. In input-output tables, we are looking for input structures that best describe technical relationships in
production techniques. It is therefore best always to classify activities in as much detail as possible and, if
aggregation is necessary, only activities with similar input structures should be aggregated. The main reason
is that if an industry in an input-output table is an aggregate of many activities, each with a different set of
inputs, then when the shares of outputs of each type of activities change, the aggregate input structure of the
industry changes, even though the input structures of the component activities remain unchanged.

5.7. In production, we should distinguish between uses and purchases. An establishment may purchase
materials and semi-finished products for use as inputs but what it purchases does not necessarily match the
inputs actually used in production. Part of the purchases may remain in inventories or inputs may be withdrawn
from inventories if stock of necessary inputs is available for withdrawals. In input-output statistics, information
on uses is primary data to estimate input coefficients. Information on sales, such as from marketing boards or
associations of producers, must be used with care and adjusted to accommodate changes in stocks. From
business accounts that are made public as legally required, it is possible to obtain the total cost of goods and
services used after adjusting for changes in inventories. Detailed information on goods and services used as
inputs can only be obtained with the cooperation of business accountants of enterprises. From their cost
 109

accounting, it is possible to assign manufacturing costs to each kind of products they produced. General
administrative costs may be allocated in proportion to product outputs.

5.8. Statistics on outputs and inputs for non-financial activities are collected by many countries through
censuses for the benchmark years and supplemented annually by annual surveys. Censuses have more complete
and detailed information on outputs and inputs and components of value added such as compensation of
employees, capital consumption (or depreciation) and other taxes on production. Annual surveys normally
cover only total output, total input and components of value added. Censuses and particularly industrial
censuses normally follow the recommendations in International Recommendations for Industrial Statistics.1
However, intermediate consumption in input-output tables and national accounts is broader in concept than
the same definition used in the publication mentioned above. The latter concept includes only agricultural and
industrial goods and services. Industrial services include only payments for contract, commission, repair and
maintenance work and freight charges. Thus in order to arrive at the SNA concept of intermediate
consumption, it is necessary to add the cost of non-industrial services such as bank and financial charges,
patent and license user fees, insurance charges, storage or warehousing charges, advertising, legal, accounting,
consulting services, printing costs, cost of travelling, entertainment, meetings, motor-vehicle running expenses,
cleaning costs, postal, telephone and telegraph charges, and elements of labour cost that cannot be considered
as compensation of employees such as vocational training, work clothes, workers' transport, cost of cultural,
recreational services to employees, etc. Censuses are normally carried out for mineral and manufacturing
industries, electric, gas and water establishments, construction establishments, wholesale and retail trades,
service industries, government departments, etc. Since information from censuses of industrial activities is
not comprehensive, it must be supplemented by additional special surveys and technical knowledge from
experts on specific fields.

5.9. Another source of information is business accounts of enterprises. Inputs reported in business
accounts that are in the public domain are rarely detailed enough for the purpose of input-output compilation.
Even if a more detailed business account is obtained, it contains only broad categories of inputs since only they
are of interest to business managers. For example, the category of office supplies is reported instead of the
detailed information such as paper, clips, pens, etc. In these cases, it is necessary to make separate
supplementary surveys to break down these aggregate items. A useful approach is to assume a common content
of goods or services within a given generic group of goods or services. For example, office supply expenses
may be such a group, which can be assumed to contain the same proportion of content of papers and paper
products, pens, pencils, diskettes, etc. Normally, it is not possible to survey all establishments due to cost
constraints so that only selective surveys of a limited number of establishments which are believed to be
representative of the broad group of industries under study have to be carried out because it is reasonable to
assume that the details of office supplies may be the same by type of activities such as manufacturing, trade
services, banking, business, government services, etc. Other broad groups of expenses that can be approached
in a similar way are business travel expenses, expenditures on office furniture, utility expenses, maintenance
of fixed assets, purchases of buildings (the last one must be split into the cost of the real estate and other
service costs such as legal services and taxes).

5.10. Another problem with business accounts, particularly of manufacturing and trade establishments, is
that only purchased materials, fuels and supplies to be matched with sales are reported after deducting changes
in stocks of materials, fuels and supplies. Compilers must adjust for changes in stocks for each kind of goods
purchased in order to arrive at inputs used to match outputs produced. Unless establishments are asked to do

1
United Nations, ST/ESA/STAT/SER.M/48/Rev.1. Sales No. [Link].8.
110

the adjustments themselves, some rough allocation rules may have to be used to assign changes in stocks to
various inputs. The rule may be as simple as allocating changes in stocks in proportion to values of input. The
balancing of the use and supply tables will contribute to the correction of errors later.

5.11. Inputs may also be used to check the outputs of the industries that use them. One example is cement
which is mainly connected with the construction industry. The use of cement is a useful indicator to check the
output of the construction industry, which is also an important part of capital formation.

5.12. Because small establishments with few employees are usually not covered by industrial statistics,
supplementary surveys on their inputs are needed; otherwise the input structure of large establishments may
be applied to them.

3. Value added at basic prices

5.13. The SNA recommends the separation of value added at basic prices into six categories (a to f below)
which have been defined in chapter II:

(a) Compensation of employees (including members of the armed forces) which includes:

(1) Wages and salaries in cash which include regular payments of wages and salaries,
payments by results and piecework; special allowances for working overtime, at
night, on weekends, away from home or in hazardous circumstances, expatriation
allowances for working abroad; supplementary allowances for housing or travel to
and from work; wages or salaries payable to employees away from work for short
periods, e.g. on holiday or as a result of a temporary halt in production; ad hoc
bonuses; commissions, gratuities and tips;

(2) Wages and salaries in kind which include the following most common types of goods
and services provided without charges or at reduced prices to employees and their
families: remuneration of products such as free meals and drinks including those
consumed when travelling on business; free housing; free uniforms or special
clothing that can be worn outside of the workplace; services of vehicles or other
durable provided for the personal use of employees; goods and services produced as
outputs; sports, recreation or holiday facilities; transportation to and from work, free
parking; creches for the children of employees, the value of interest forgone by
employers when they grant loans at reduced rates;

(3) Employers' actual social contributions which include their contribution to social
security funds, insurance enterprises or other institutional units responsible for the
administration and management of social insurance schemes;

(4) Employers' imputed social contributions which include social benefits employers
provide directly to employees, former employees, or dependents out of their own
resources without involving a social scheme or institution such as children's, spousal,
family, education or other allowances in respect of dependents, payments to workers
absent from work because of illness, injury or maternity leave, severance payments
and unfunded pension payments.
 111

(b) Other taxes on production which include:

(1) Taxes on payroll or work force;

(2) Recurrent taxes on land, buildings and other structures;

(3) Business and professional licenses;

(4) Taxes on the use of fixed assets or other activities;

(5) Stamp taxes;

(6) Taxes on pollution;

(7) Taxes on international transactions.

minus

(c) Other subsidies on production which include:

(1) Subsidies on payroll and workforce;

(2) Subsidies to reduce pollution.

(d) Consumption of fixed capital;

(e) Operating surplus; and

(f) Mixed incomes.

5.14. To arrive at the GDP, taxes less subsidies on products must be added to the total value added at basic
prices. Major components of taxes on products and subsidies on products are listed below:

(a) Taxes on products include:

(1) Value added type taxes (VAT);

(2) Taxes and duties on imports excluding VAT which include import duties, general
sales taxes, excise duties, taxes on specific services, profits of import monopolies,
implicit taxes on imports resulting from multiple exchange rates;

(3) Export taxes which include export duties, profits of export monopolies, implicit taxes
on exports resulting from multiple exchange rates;

(4) Other taxes on products which include general sales or turnover taxes, taxes on
specific services, taxes on financial and capital transactions, profits of fiscal
monopolies.

(b) Subsidies on products include:

112

(1) Import subsidies which include direct subsidies on imports, losses of government
trading organizations involved in imports, implicit subsidies on imports resulting
from multiple exchange rates;

(2) Export subsidies which include direct subsidies on exports, losses of government
trading organizations involved in exports, implicit subsidies on exports resulting from
multiple exchange rates;

(3) Other subsidies on products which include subsidies on products used domestically,
losses of government trading organizations involved in domestic trades, subsidies to
public corporations and quasi-corporations.

5.15. The total value added at basic prices of industries should be calculated first to serve as total control.
The value added at basic prices of an industry is equal to its output at basic prices minus its intermediate
consumption. Sources of information for calculating value added are censuses of production. However, as
pointed out previously, census value added is not the same as national account value added.2 Census value
added in an industry is equal to output minus cost of materials and industrial services consumed. Thus the costs
of other services such as accounting, advertising and other overhead costs which are paid at the company level
and not allocated to the establishment level are included in census value added. National account value added
is normally smaller than census value added. Table 5.1 shows the value of national account value added as a
percent of census value added for the United States. The difference between census value added and national
account value added may be much smaller in other countries. But, in order to obtain national account value
added for the use table, it is necessary to calculate intermediate consumption by adding other service costs to
the cost of materials and industrial services consumed.

Table 5.1. National account value added and census value added in the United States3

National account value added as percent of census value added

1982 1987

Manufacturing 77 73

Nondurable goods 74 71

Durable goods 79 75

5.16. Without integrating data collection for both enterprises and establishments, it is not easy to allocate
systematically non-industrial services paid by enterprises to their constituent establishments. In many countries,
statistics on establishments are collected by censuses independently of enterprise statistics, so that the
allocation of non-industrial services to manufacturing establishments is only an approximation. If statistics on
establishments are collected together with statistics of parent companies, it is possible to ask for non-industrial
services paid and consumed by the companies and then to distribute them to component establishments. It will

2
Ibid. For the definition of census value added and its difference from the concept of value added in national account.

3
United States Department of Commerce, Bureau of the Census Bureau, 1987 Census of Manufactures, General
Summary.
 113

be recalled that national account value added may be calculated only on a consolidated company-industry basis
and then converted to an establishment-industry basis. Value added on a company-industry basis is calculated
either as the sum of income payments (e.g. wages, profits, and interest) and other cost (e.g. depreciation and
other production taxes) or as the difference between output (value of shipment plus change in inventories) and
cost of all materials and services consumed. Data on components of value added on a company-industry basis
can be obtained from annual surveys and income tax returns. The difference between company-industry
classification and establishment-industry classification can be used as one of the factors for allocating non-
industrial services to different I/O industries. Additional surveying of non-industrial service uses may also be
necessary for the allocation purposes if enterprise (e.g. company) statistics are not available.

5.17. Taxes and subsidies: Data on taxes on products and other taxes on production are available from
government sources. Other taxes on production which are classified by kind of activity of the establishments
that pay the taxes are available in production censuses such as the census of manufacturing. Some other taxes
on production may be paid by the enterprise instead of its component establishments. These taxes must be
allocated. Only other taxes on production enter into the calculation of value added of a single establishment-
industry in the use table. Since output is valued at basic prices, taxes on products are to be allocated by kind
of product in the supply table. Turnover taxes, manufacturers' excise taxes, which may vary by kind of product,
are normally paid and reported by establishments so that it is possible to estimate taxes paid by kind of
product. Import and export duties can be easily identified and estimated by kind of products. Other product
taxes (VAT, sales taxes) which are collected by types of producers and mostly by types of wholesalers and
retailers are not easily identifiable with products on which taxes are assessed, and it may be necessary to use
detailed statistics and tax rates on products sold in wholesale and retail trade to estimate product taxes by kind
of product. Since taxes on products are by definition proportional to either values or quantities of products,
given that taxes on a product are identified, these taxes are then simply distributed proportionally to products
on the basis of either value or quantity as appropriate. Subsidies on products should be distributed in a similar
way. It should be emphasized again that SUT provides an excellent framework to estimate and allocate taxes
on products.

5.18. Compensation of employees: Compensation of employees is often available from both production
censuses and annual surveys, frequent labour-force or employment surveys, financial reports of Government
and non-profit organizations. In some cases, only employment and hours worked are available so that wages
must be estimated separately through household surveys, which also provide statistics on payments in kind.
Data on compensation of employees should be supplemented by data from tax and social security reports, and
from reports of pension funds, and insurance companies. Social security records often provide the best current
information on wages and salaries paid since contributions are usually proportional to wages.

5.19. Consumption of fixed capital: “The value of a fixed asset to its owner at any point of time is
determined by the present value of the future rentals (i.e., the sum of the discounted values of the stream of
future rentals) that can be expected over its remaining service life. Consumption of fixed capital is therefore
measured by the decrease, between the beginning and the end of the current accounting period, in the present
value of the remaining sequence of rentals.”4 “The calculation of consumption of fixed capital is a forward
looking measure that is determined by future, and not past events....Unlike depreciation as usually calculated
in business accounts, consumption of fixed capital is not, at least in principle, a method of allocating the costs

4
System of National Accounts 1993, para. 6.182.
114

of past expenditures on fixed assets over subsequent accounting periods.”5 Because it is not easy to estimate
the stream of future values, the most commonly used method to estimate consumption of fixed capital is to
estimate it together with gross capital stocks by the perpetual inventory method (PIM). Gross capital stocks
were built up by “gross fixed capital formation undertaken in previous years [but] have survived to the current
period. Average service lives, or survival functions, based on observations or technical studies may be applied
to past investments for this purpose. Fixed assets purchased at different prices in the past have then to be
revalued at the prices of the current period.”6 “The value of the capital consumption on a fixed asset may be
estimated by applying either the linear or geometric depreciation formula to the actual or estimated or current
purchaser's price of a new asset of the same type.”7, 8 To estimate consumption of fixed capital, it is necessary
to know for every industry its stock of capital assets by type, the historical values of these assets, their dates
of purchase and their expected useful life which can be collected through benchmark surveys. The
replacement cost can be estimated on the basis of price movements on capital goods. With benchmark surveys
and information on annual capital formation and price movements for capital goods, consumption of fixed
capital can be estimated annually. The perpetual inventory method requires data on capital formation by type
over 30-40 or more years which may not be available in many developing countries. Though the SNA does
not recommend the use of business depreciation in place of consumption of fixed capital, in case of difficulty
in estimating consumption of fixed capital, it suggests that "if data on depreciation are used, they must, at the
very least, be adjusted from historic costs to current prices."9 Unlike the 1968 SNA, the revised SNA also
recommends the calculation of consumption of fixed capital for roads, bridges and railway tracks. Depletion
of natural assets such as mineral deposits, reserves of oil, gas, etc. is not covered by consumption of fixed
capital; it is treated as "other changes in volume of assets."10

5.20. Gross operating surplus: This by definition is calculated as residual. It is the difference between
industry output and the sum of intermediate input, compensation of employees and net taxes on products and
production. If consumption of fixed capital is known, then the operating surplus is also calculated residually
as the difference between gross operating surplus and consumption of fixed capital. Consumption of fixed
capital has to be estimated by the perpetual inventory method. For many developing countries, the calculation
of consumption of fixed capital may be difficult because data are not available. If such a calculation is not
possible, the gross operating surplus which is the sum of the operating surplus and consumption of fixed
capital should be calculated as a residual.

4. Statistical sources

5
Ibid., para. 6.183.

6
Ibid., para. 6.189.

7
Ibid., para. 6.198.

8
For the estimation of fixed capital stock and consumption of fixed capital, see Vu Quang Viet, “Measurement of
Fixed Capital Stock and Consumption of Fixed Capital” in the Handbook on Linking Business Accounts to National
Accounts for the Non-financial Corporation Sector, United Nations, ST/ESA/STAT/SER.F/76, or the handbook on The
Measurement of Capital by Michael Ward, OECD, 1976; Stock of Fixed Assets in Industry in the Community Member
States: Toward Greater Comparability, Studies of National Accounts No. 2, Eurostat, 1983.

9
System of National Accounts 1993, para. 6.184.

10
Ibid., paras. 12.29 and 12.30.
 115

5.21. Major sources for input-output and production statistics for non-financial activities come from various
censuses:

- Census of mineral industries;

- Census of agriculture;

- Census of manufactures;

- Annual surveys of enterprises or establishments;

- Census of trade;

- Census of services;

- Census of transportation;

- Census of construction industries;

- Census of Governments and/or annual administrative records and accounts of government

bodies, either central or local;

- Survey of household expenditures;

- Survey of business capital formation;

- Foreign trade statistics;

- Balance of payments.

5.22. All these censuses and annual surveys are supplemented by annual statistics from various government
departments and private agencies, professional associations, business associations on specific areas, financial
reports of major corporations, banks, insurance companies, pension funds, etc. Taxes and incomes, besides
being provided by censuses, can be found in documents of tax authorities and national budgets. The survey
of household incomes and expenditures is useful for the estimation of personal consumption expenditures as
well as outputs and incomes from own-account production that are not covered by any other sources. For more
information on other sources on outputs and inputs, readers are advised to consult the United Nation
publication, Handbook of National Accounting: Accounting for Production: Sources and Methods.11 Sources
of information and specific methods of compilation for all industries cannot be adequately covered in this
publication because of the substantial richness of the subject matter.

5.23. For financial activities, information comes mainly from the central bank which is set up to regulate
and monitor financial activities. In many countries, insurance companies are also under its supervision, unless

11
United Nations, Sales No. [Link].11, ST/ESA/STAT/SER.F/39. The publication, though obsolete in respect
of recent changes in the 1993 SNA, is still a valuable source of information on production accounts for compilers.
116

an independent agency is set up to regulate them. General information on inputs may be obtained directly from
public business accounts of financial companies. This information must be supplemented by limited surveys
in order to detail some input components reported in business accounts. The number of financial companies
is relatively small for many countries, so the use of all their business accounts for compilation is possible.

C. SNA production concepts reviewed

1. Production boundary

5.24. In the SNA, output is measured only for production activities within its production boundary. Thus
it is necessary to define the production boundary of the SNA and the output to be measured within it.

5.25. "Economic production may be defined as an activity carried out under the control and responsibility
of an institutional unit that uses inputs of labour, capital, and goods and services to produce outputs of goods
and services... A purely natural process without any human involvement or direction is not production in an
economic sense."12 However, not all human activities are included in the production boundary by the SNA
and the following will clarify the borderline cases:

(a) All activities by households that produce goods, whether they are sold in the market or used
for own final consumption or own capital formation, or whether they are legal or illegal, are
included in the production boundary. Household activities producing goods include the
production of agricultural products, gathering of uncultivated fruit, wood cutting and
firewood collecting, hunting, fishing, mining, water supplying, storing of agricultural goods,
processing of agricultural products,13 making of cloth, leather, footwear, pottery, utensils,
furniture, etc. Goods produced by households for intermediate consumption and excluded
from production boundary are, for example, own-produced seeds used in agricultural
cultivation.

(b) All "do-it-yourself" activities by households that produce services for own use such as taking
care of children and other members of the family, cooking, cleaning, home repairs, repairs of
personal objects including own vehicles, etc. are excluded from the production boundary.
However, this rule of exclusion does not apply to the own-account production of housing
services.

(c) Domestic services by paid domestic servants are within the production boundary and
classified in the unincorporated (or household) sector in which compensation of employees
is the only cost of production. These services are for own final use.

(d) Only activities of the establishments that produce goods and services that are intended for
exchange with, or sold to, other establishments, even within the same enterprise, are treated
as belonging to the production boundary. Excluded are goods and services produced for own
intermediate consumption within the same establishments.

12
System of National Accounts 1993, para. 6.15.

13
Storing and processing of agricultural goods are treated within the production boundary as an extension of goods-
producing activities.
 117

(e) Own-capital formation of any sector must always be included in the production boundary.

2. Scope of output

5.26. The output of an establishment or institutional unit is the result of a production activity within the
production boundary defined by the SNA and may either be a market product or own final use or other
non-market product. A producer may be classified as market, own final use or other non-market producer
depending on whether his output is mainly for the market, his own final use or other non-market use.

5.27. Market output: Market output is output that is sold or otherwise disposed of on the market or
intended to be disposed of on the market at prices that are economically significant. Market output should
consist of the total value of:

(a) Goods and services sold;

(b) Goods and services bartered;

(c) Goods and services used for payments in kind, including compensation in kind;

(d) Goods and services supplied by one establishment to another belonging to the same market
enterprise, to be used as intermediate inputs;

(e) Changes in inventories of finished goods and work-in-progress intended for one or other of
the above uses.

5.28. Own final use: Own final use output is output that is retained for own final use by the owners of
enterprises, either as own final consumption by unincorporated enterprises (incorporated enterprises by
definition do not incur final consumption) or own gross capital formation. Examples of own final consumption
output are agricultural goods and services, and owner-occupied housing which are provided and used by
members of the same households. Domestic services provided by paid servants are included. Own capital
formation includes own-constructed housing, machine tools produced for their own use by engineering
enterprises, etc. Own final use output should consist of the total value of:

(a) Goods and services produced by household enterprises and consumed by the same
households;

(b) The fixed assets produced by an establishment that are retained within the same enterprises
for use in future production (own-account gross capital formation, for instance the computer
hardware and software that are produced by, and used within, the establishment (products that
are retained within the establishment and used as intermediate inputs are not considered
output by the SNA));

(c) Changes in inventories of finished goods and work-in-progress intended for one or another
of the above uses.
118

5.29. Other non-market: Other non-market output includes the total value of the following goods and
individual or collective services produced by non-profit institutions serving households (NHISHs) or
Government:

(a) Goods and services supplied free, or at prices that are not economically significant, to other
institutional units, either individually or collectively;

(b) Goods and services supplied by one establishment to another belonging to the same non-
market producer, to be used as intermediate inputs;

(c) Changes in inventories of finished goods and work-in-progress intended for one or another
of the above uses.

5.30. Ancillary activities such as administration, purchasing, sales, communication, accounting, data
processing, etc., supporting main activities are not classified as separate establishments and therefore their
costs must be allocated to the production costs of the establishments within the enterprise (see also chapter IV
for allocation methods). If ancillary activities involve own-account capital formation such as construction, the
value of own-account capital formation should be allocated to every establishment in the enterprise using the
same allocation method used in allocating costs of ancillary activities. In table 5.5 following paragraph 5.52,
own-account capital output of ancillary activities has not been allocated. But if it was, then gross establishment
outputs would be greater than those shown in column 10. In case ancillary activities sell part of their products -
for example, accounting or data processing services - to other enterprises, this market output should also be
allocated to every establishment of the enterprise to be consistent with the SNA valuation of output. If the
market output of ancillary activities makes up more than 50% of their costs, a separate establishment must be
created for the outputs that are sold.

5.31. Establishment output is equal to the sum of products produced for sale, exchanged with other
establishments, put in inventories or used for own capital formation. Industry output is equal to the sum of
outputs of all establishments belonging to the industry.

3. Output and input valuation

5.32. Market output is measured at market prices. The output for own final use should also be measured
at comparable market prices; for instance agricultural, fishing and forestry products can be measured in prices
at farm or communal markets. When comparable prices are not available, output for own final use may be
measured at production costs. Other non-market outputs are measured at production costs.

5.33. However, market output, except output of financial intermediaries, is often measured indirectly as the
sum of sales of produced products, other uses (including items b-d in paragraph 5.27 above and a-b in
paragraph 5.28 and changes in the stock of products produced.

(1) Output = Sales net of discounts, returns, VAT and sales taxes + other uses + changes
in inventories14

14
Changes in inventories = Ending inventories - beginning inventories.
 119

5.34. The above formula is the conventional one to calculate output because businesses report data in terms
of value of sales and changes in inventories. However, the SNA definition of changes in inventories is
different from the practice in the business world. Thus, changes in inventories of business accounts cannot,
in principle, be used for SNA changes in inventories except when the prices of goods stored are stable (see
appendix A below for a detailed description of various methods used in inventory valuation and output
calculation according to the SNA). In practice, "so many different methods are liable to be used in business
accounts, it is impossible to suggest algorithms, or rules of thumb, which would be generally applicable for
the purposes of transforming data on inventory changes in business accounts to the data required by the
System. Each case has to be treated individually, depending upon the precise way in which the business
accounts have been drawn up."15

5.35. Outputs of other non-market goods and services are measured at production costs which include the
following items:

(a) The value at market prices of the goods and services consumed as intermediate
inputs;

(b) The value of the compensation of employees payable;

(c) The value of consumption of fixed capital;

(d) The value of other taxes on production (if any) less other subsidies on production (if
any).

5.36. Inputs or intermediate consumption must also be valued at market prices and specifically at
purchasers' prices in the use table. They are not information that can be collected directly from business
accounts. From business accounts, inputs of materials (or goods) must be derived by the following formula:

(2) Materials used in production = Materials purchased net of discounts and returns - changes in
inventories

The valuation of inventories is important in the calculation of both materials used and outputs.

15
Ibid., para. 6.71.
120

4. Work-in-progress

5.37. Work-in-progress (sometimes called work put in place) consists of inventories of goods held by an
establishment or enterprise which have been produced as output but require further processing and are not yet
in a form in which they can be suitably delivered or sold to purchasers. Work-in-progress is recorded under
the general heading of changes in inventories and usually happens to products that take more than one year
or accounting period to complete. Examples are construction of buildings, dams, ships, other large capital
goods, and production of some agricultural products. Output of work-in-progress in an accounting period, like
output of non-market goods and services, is to be measured by the costs actually incurred in its production
during that accounting period.

5.38. However, at the accounting period in which the project is completed and the sale takes place, the
reduction in work-in-progress must be recorded. It is equal to the value of work-in-progress accumulated in
previous periods valued at the prices or costs prevailing at the moment the sale occurred.

Table 5.2 (a). Calculation of output of work-in-progress

Basic data

Period 1 Period 2 Period 3 Period 4 Year

Sales 0 0 0 2,500

Intermediate consumption 50 83 102 220

Capital consumption 15 22 33 60
Compensation of employees 60 95 130 260

Total production costs 125 200 265 540

Table 5.2 (b). Calculation of output of work-in-progress

At the prices and wage rates of period 4

Period 1 Period 2 Period 3 Period 4 Total

Sales 0 0 0 2,500 2,500

Intermediate consumption 250 260 210 220 940

Capital consumption 75 69 66 60 270
Other taxes less subsidies on production 0 0 0 0 0
Compensation of employees 310 290 255 260 1,115

Total production costs 635 619 531 540 2,325

 121

Table 5.2 (c). Calculation of output of work-in-progress

At the prices and wage rates of period 4

Period 1 Period 2 Period 3 Period 4 Total

Sales 0 0 0 2,500 2,500

Addition to inventories 125 200 265 540 1,130

Withdrawals from inventories 0 0 0 -2,325 -2,325
Change in inventories 125 200 265 -1,785 -1,195
Output 125 200 265 715 1,305
Holding gains realized on work-in- 510 419 266 0 1,195
progress in period 4 16

5.39. The work-in-progress should also include the value of the addition to it since it was stored, which is
due not to the effects of general price inflation but to the improvement in the physical quality of the good over
time (such as wines) or to seasonal factors affecting supply and demand for the good, such as preserved
summer vegetables sold in winter. When there is a reduction in value not due to general price deflation,
reduction in the work-in-progress must also be introduced. So in table 5.2(c), period 4, output (715) is equal
to sales ( 2,500 ) + addition to inventories (540) - withdrawals from inventories (-2,325). In general,

Output = Finished products sold or bartered;

Plus Addition to inventories;
Less Withdrawals from inventories;
Plus Finished products used by their producer for his own final use.

5.40. Tables 5.2(a), (b) and (c)17 show how output of work-in-progress should be calculated. Table 5.2(a)
shows the basic data, table 5.2(b) shows the production costs revalued at the prices of the period 4. The
general price indexes used for period 1 to period 4 are consecutively equal to 20, 32, 50, 100. Table 5.2(c)
shows how work-in-progress should be treated in the SNA. In the last table, the value of inventories withdrawn
in period 4 (-2,325) when the final product is sold, is equal to the sum of production costs incurred by work-in-
progress during the production process revalued to the prices and wages of period 4 shown in table 5.2(b) i.e.
-(635+619+531+540).

5. The treatment of second-hand goods and old scraps in input-output table

5.41. The output of establishments involving the transaction of second-hand goods is measured in a fashion
similar to the treatment of wholesalers and retailers, i.e. it is equal to the difference between revenues received
from selling the goods and the cost of restocking these goods. The value of second-hand goods is entered as
a positive value to the industry that buys them and as a negative value to the industry that sells them. If capital

16
Calculated by the increase in production costs due to the application of prices and wages of period 4 to the basic
data. For example, for period 1, holding gains are 510 = 635 - 125.

17
The examples are taken from Peter Hill, Handbook on Inflation Accounting.
122

goods are sold by enterprises, a negative value must enter the column "gross capital formation" of enterprises.
If current goods are sold, a negative value may enter either the column change in inventories or the column
final household consumption. The treatment of goods bought is treated similarly but in an opposite way.
Following are examples that will be used to demonstrate the different treatments of second-hand goods
depending on the nature of those goods and where they are utilized.

Revenue for selling the second-hand good 60

Cost of restocking the same good 40
Trade margin 20

(a) Case 1: The second-hand goods are sold by the household sector for household final consumption
uses. The treatment is shown in table 5.3. Only the output of trade margins is created, the sale of second-hand
goods cancels their purchase.

Table 5.3. Second-hand goods sold for final consumption

Intermediate consumption Household final Output

consumption

Industries Trade margins Second-hand goods

Industries

Trade margins 20 20

Second-hand goods +40-40 0

Total 20 20 20

(b) Case 2: The second-hand goods are used as intermediate inputs. The treatment is similar to the
first case but a positive value of 40 is entered in row 3, column 1, as intermediate inputs and a negative value
is entered in the household final consumption, row 3, if the goods are sold by the household sector or entered
as negative value in the column "change in inventories" (not shown in table 5.3) if the goods are sold by
enterprises. Also in the first column and second row, 20 is entered as intermediate consumption of industries.

(c) Case 3: The second-hand goods are capital goods that are sold from one sector (enterprises,
households or Government) to the same or other sectors. Gross capital formation of the purchasing sector
increases by the value of trade margins. The same treatment applies in selling old homes but in this case, real
estate services instead of trade services are provided and include legal services, banking services and taxes
involved in the transaction (one may also break down real estate services into separate flows and enter them
separately in the column of "gross capital formation"). In case second-hand goods are capital goods sold by
enterprises or government to households (i.e. automobiles or other consumable durable goods such as
refrigerators, air-conditioners, etc.), there is a reduction in gross capital formation of the sellers and an increase
in household final consumption.

Table 5.4. Second-hand goods as capital goods

Intermediate consumption Gross capital Output

formation
 123

Industries Trade margins Second-hand goods

Industries

Trade margins 20 20

Second-hand goods +40-40 0

Total 20 20 20

5.42. Old scraps (e.g. waste and scrap and materials for recycling such as paper, glass, cans, bottles) can be
treated the same way as second-hand goods in case 2 since old scraps may often be used as intermediate inputs
into the production of new products. The margins for old scrap collecting units are classified under version
1.0 of the United Nations Central Product Classification (ST/ESA/STAT/SER.M/77) as part of retail trade
services. Besides trade margins that need to be calculated, the value of old scraps used for intermediate
consumption may be balanced by a negative change in inventories (if sold by business) and negative final
consumption (if sold by households and Governments). Also as in case 2, trade margins are entered as
intermediate consumption of industries that recycle scraps.

5.43. New scraps such as metal scraps produced as secondary products of the steel-making process are
treated as secondary products and not as second-hand goods as discussed above. New scraps can be treated
as a separate product in the supply table (or make matrix). However, given that they are not independently
produced anywhere else, they later can probably be better aggregated with the main product as, for example,
steel scraps being aggregated with steel.

6. Operating leasing versus financial leasing

5.44. Operating leasing refers to the renting out of machinery or equipment for varying periods of time but
not for the entire expected service life of the equipment during which the owner, or lessor, is responsible for
maintaining and repairing it as part of the service to the lessee and may also have to replace the equipment in
the event of a serious or prolonged breakdown. Rentals paid by the lessee to the lessor are treated as the output
of leasing companies similar to the renting of dwellings.

5.45. “In contrast to operating leasing, financial leasing is not itself a process of production. It is an
alternative to lending as a method of financing the acquisition of machinery and equipment. A financial lease
is a contract between a lessor and a lessee whereby the lessor purchases machinery or equipment that is put
at the disposal of the lessee and the lessee contracts to pay rentals which enable the lessor, over the period of
the contract, to recover all, or virtually all, of his costs including interest. Financial leases may be distinguished
by the fact that all the risks and rewards of ownership are, de facto, transferred from the legal owner of the
good, the lessor, to the user of the good, the lessee. In order to capture the economic reality of such
arrangements, a change of ownership from the lessor to the lessee is deemed to take place, even though legally
the leased good remains the property of the lessor, at least until the termination of the lease when the legal
ownership is usually transferred to the lessee. The lessor is treated as making a loan to the lessee which enables
the latter to finance the acquisition of the equipment. The rentals are then treated as covering repayments of
the loan and interest payments.”18 For financial leases, the 1993 SNA treats them as a purchase with a loan
from the lessor. Rentals are divided into two parts: the repayment of principal and interest payment. Both of

18
System of National Accounts 1993, para. 6.118.
124

these payments would not appear as costs in production accounts. In fact, the payment of interest is part of
value added of the lessee and the repayment of principal is part of his financial account. There are some
“incidental services ...provided by the lessor in the process of arranging the lease,... but the value of these
services is very small compared with the total rentals paid.”19 However, the SNA does not specify any method
to estimate the output of these incidental services. We may try to estimate implicit service charges on interests
which are treated as the output of the lessor and also as intermediate consumption by the lessee. The estimation
is not easy since an enterprise may have both operating leasing and financial leasing and the interests payable
and receivable which relate directly to financial leasing are not easy to separate out except with detailed
information from financial leasing companies.

5.46. In practice, national account compilers have to rely on business accounts and therefore can only apply
the SNA principle if business accountants apply the principle of treating financial leasing differently from
operating leasing. In fact, many countries have applied business accounting principles on leasing like those
advocated by the SNA.

7. Treatment of entertainment, literary and artistic originals and copyrights

5.47. “The production of books, recordings, films, software, tapes, disks, etc. is a two-stage process of which
the first stage is the production of the original and the second stage the production and use of copied of the
original. The output of the first stage is the original itself over which legal or de factor ownership can be
established by copyright, patent or secrecy.”20 The output of the first stage is treated as own-account gross
capital formation, an intangible fixed asset, of the originator if the asset is not sold. If the original is sold, its
output is measured by the price paid. If it is not sold, its value is measured by production costs plus a mark-up.
The value of the mark-up is the discounted value of the future expected receipts from licensing its use. This
creates uncertainty in assigning a value to the original. If we are uncertain about future value of the original,
the mark-up may be given no value.

5.48. The second stage is the use or reproduction of originals. “The owner of the asset may use it directly
or to produce copies in subsequent periods. Consumption of fixed capital is recorded in respect of the use of
the asset in the same way as for any other fixed asset used in production.”21 “The owner may also license other
producers to make use of the original in production.... In these cases, the owner is treated as providing services
to the licensees that are recorded as part of their intermediate consumption. The payments made by the licenses
may be described in various ways, such as fees, commissions or royalties, but however they are described they
are treated as payments for services rendered by the owner.”22

5.49. There is a special category of originals and patents, which is called scientific originals and patents,
the production of which is treated by the SNA as outside the production boundary. Scientific originals and
patents such as inventions, new drugs, new processes, etc. must be understood in the context of expenditures
on research and development (R&D) which may yield benefits long after being undertaken and are considered

19
Ibid., para. 6.119.

20
Ibid., para. 6.143.

21
Ibid., para. 6.145.

22
Ibid., para. 6.146.
 125

by most economists as gross capital formation. However, the 1993 SNA treats R&D only as output that is
consumed by the unit that produces it as intermediate consumption. In practice, scientific originals may be
produced which must be assets from an economic point of view but they are not recognized by the SNA.
There is no category of scientific originals" under intangible fixed assets, though patents may be taken out
which establish legal ownership over non-existent assets. The SNA tries to get around this by recognizing a
category of asset called patented entities" which are part of non-produced intangible assets. Payments of
royalties to patent owners for the use of these patented entities are treated as property income like interest or
dividends. They are not part of the production boundary and therefore not an output.

D. Allocation of production activities of institutional units to establishments

1. Production activities of corporate and quasi-corporate enterprises

5.50. Output of an enterprise is usually compiled according to the procedure shown in table 5.5 below.

5.51. In table 5.5, an example of the statistics for a corporate or quasi-corporate enterprise is given to show
what outputs should be accounted for. In this example, the enterprise consists of two establishments, a research
and development activity and ancillary activities serving the two establishments. These two establishments are
classified as market producers since most of their output is intended for sale at economically significant prices.
Two additional establishments are imputed: one to research and development activities and the other to own-
account capital formation. Own-capital formation is assumed to be carried out separately by establishments
1 and 2 including ancillary units (see para. 4.15) with research and development treated as carried out outside
these two establishments. Imputed outputs of R&D and own-account capital formation are not counted as part
of the outputs of establishment 1 and 2. R&D activities may be classified as for the market, for own use or
for other non-market use depending on whether a major part of the services are sold in the market, internally
used or provided as free government and non-profit services. A non-market part of the output of R&D is then
treated as intermediate consumption and not as gross capital formation. In this example, own-account capital
formation is shown as a single establishment, but in actual compilation, this establishment may have to be
subdivided depending on the number of capital products the enterprise produces. Normally, only own-account
construction is produced. The producers identified as establishments may simultaneously produce market
output, output for final use and possibly other non-market output, as defined below. Table 5.5 helps classify
outputs into separate categories. Each product produced by an establishment may be used for different
purposes and therefore must be classified accordingly.

5.52. Surveys of establishments are the conventional method used to collect production statistics. A separate
establishment is normally identified by a physical production location belonging to an enterprise. If all
activities happen in one location, the enterprise is normally treated as an establishment. Conceptually, each
product produced, assuming there are no by-products, is best identified as an establishment. Surveys may be
designed to ask for outputs as well as associated inputs used in producing each output separately. However,
costs are not always linked directly to a specific establishment producing a specific product within an
enterprise. There are general costs at the enterprise level that do not apply specifically to an establishment, such
as general administration, insurance, interests receivable and payable. There should be guidelines to instruct
respondents how to allocate these costs to various establishments. A rule may be as simple as allocation on the
basis of sales or value added.23 This type of rule is not alien to business accountants as they have to apply cost

23
To see how to link business accounting of non-financial enterprises to the SNA, see Vu Quang Viet, “Compilation
of National Accounts from Business Accounts: Non-financial Corporations” in Handbook of National Accounting: Links
accounting methods to prepare production costs associated with each product in order to price their products
in the market. This is also a practice of the Malaysian Department of Statistics in its establishment surveys.
In Malaysia, even bank charges, interest and dividend payments and receipts, and insurance premiums are
assigned to separate establishments. This information will be useful in calculating bank and insurance service
charges as intermediate inputs of establishments. This topic will be covered later in this chapter. The allocation
of costs to separate establishments and to products, if feasible, would avoid the mechanical application of
methods in treating secondary products.

2. Production of government units and non-profit institutions serving households

5.53. Outputs or products produced by governmental agencies to serve the society at large or by non-profit
institutions serving households supplied without charge or at prices which are not economically significant (see
table 5.6) are normally the first three items of the following list, although in many countries the last items are
also included as parts of output of government services:24

(a) Market goods and services that are sold at economically significant prices and normally
supplied by non-profit schools, colleges, universities, clinics, hospitals, etc. including
publications sold by government offices or reproductions sold by non-market museums (the
part that is sold is treated as market and the residual is treated as non-market);

(b) Non-market goods and services such as health and educational services that are provided free
or almost free;

(c) Other non-market services that are collective services such as public administration and
defense, etc.;

(d) Research and development that must be treated as a separate establishment;

(e) Own-account construction that also must be treated as a separate establishment.

between Business Accounting and National Accounting, (ST/ESA/STAT/SER.F/76).

24
In I/O analysis, it is preferable that R&D and own-account capital formation, particularly construction be separated
from pure government services. To do this, associated inputs must also be estimated and separated out.
 127

Table 5.5. Production statistics of a non-financial or financial enterprise with secondary products

Products of specified establishments, characteristic and secondary Establishment

and ancillary activities of an enterprise output

Product 1 Product 2 Product 3 Prod. 4

Research and Own-account

Development capital formation

Market output at basic prices Non-market Other non- Total . Total

market

(1) (2)* (3) (4)* (5)* (6) . (7) (8)* (9)* (10)*

Output Other Total Own final use Other non- (3)+.+(5) (6)+(7)+(8)
sold uses (1)+(2) (capital goods) market

Establishment 1 17 17 2 1 20 8 (4) 28

Establishment 2 9 3 12 12 4 (2) 16

Research & Development 2 2

establishment

Ancillary activities x x x x x x x (2)

Own-account capital formation (8) 8

Total enterprise 26 3 29 2 1 32 12 2 8 54

*Columns: (2) Other uses include products that are bartered with other enterprises, used in other establishments within the same enterprise and delivered to employees as remuneration.
(4) Own final use includes only products (i.e durable goods) that are used as capital goods. In the case of unincorporated enterprises which belong to the household sector, these products are classified as final
consumption expenditures.
(5) Other non-market includes output that is distributed free or sold at not economically significant prices to non-profit institutions serving households.
(8) Costs of R&D should be separated from other costs of the establishment that generates R&D and imputed as the output of a separate imputed establishment. This output is then imputed as intermediate consumption
of R&D of the establishment that generates R&D.
(9) Own-account capital formation such as construction, development of software and other originals which can be used for more than a year should also be separated from other costs of the establishment that produces
them and each activity is imputed as the output of a separate imputed establishment. This output is then imputed as gross capital formation of the establishment that produces it. In fact, there is no such output as
own-account capital formation; column 9 is an aggregate of many capital products for the sake of abbreviation. Ancillary activities may also produce capital products. In this case, these capital products have to
be allocated to other establishments. The values are entered into parentheses to show that they do not enter as establishment output in column 10. Own final use in column 4, which shows the products, such as
microcomputers, produced but internally used as capital goods should be later included with own-account capital formation.
(10) For the grand total for the enterprise, 8 has to be included.

x Means that nothing should be included.

 Table 5.6. Production statistics of a government or non-profit unit serving households
enterprise with secondary products

Products of specified establishments, characteristic and Establishment

secondary and ancillary activities of an enterprise output

Other non-market Non-market Market Research and Own-account

output output output at Development construction
measured at costs measured at basic prices
costs
Output sold

(1) (2) (3) (4)* (5)* (6)

Establishment 1 200 (14) 200

Establishment 2 40 17 (3) 57

Research & Development 10 10

establishment

Ancillary activities (2)

x 0

Own-account construction 19

Total enterprise 200 40 17 10 (19) 286

*Columns: (4) Research and Development is treated like own-account constructions (see next note) though in the sample, R&D is separately provided.
(5) Own-account construction must be treated as separate establishments and not as part of the outputs of establishments 1 and 2. They are therefore parenthesized and treated as output of
a separate establishment. Output of own-account construction is measured at costs. Here costs must be transferred out of the costs of establishments 1 and 2.
 129

5.54. The outputs of government units and NPISHs are normally measured by costs as previously discussed.
These outputs net of sales make up either collective or individual final consumption, so it is important that
they be prepared in conjunction with the compilation of final consumption which will be discussed in detail
in chapter VII where an example is given of how output of government services and government final
consumption are compiled from data on government budget.

3. Production activities of unincorporated enterprises of households

5.55. Production within the household institutional sector takes place in enterprises that are directly owned
and controlled by members of a household or shared with other households. These production units are called
unincorporated enterprises. In the input-output framework, activities in an unincorporated enterprise are
classified into different ISIC categories depending on the nature of its activities and thus may be separated into
different establishment units similar to those shown in table 5.5. Members of households who work as
employees for corporations, quasi-corporations, government or non-profit institutions serving households
contribute to the production activities outside the household sector.

5.56. Similar to the corporate sector, each activity (or establishment) can be classified into market or own-
use producers and their output can also be classified accordingly.

5.57. Household market producers include all unincorporated enterprises that produce goods and services
for sale. The value of the products for sale must be at least 50% of their total output. Table 5.6 can also be
used to calculate output of the household market producers.

5.58. Household non-market producers include:

(a) Subsistence farmers, fishermen, woodcutters, etc.;

(b) Households engaged in the construction of dwellings for their own use or in major structural
improvement or extension of their dwellings;

(c) Households engaged in communal construction projects such as bridges, roads, schools, water
supplies, irrigation works, etc.;

(d) Households engaged in the production of other goods for their own consumption such as
cloth, clothing, furniture, other household goods, fuel materials for cooking and heating,
foodstuffs (other than meal preparation), etc.;

(e) Imputed services of owner-occupied housing;

(f) Domestic services produced by paid employees.

5.59. For communal construction projects by household non-market producers, when no comparable market
prices are available to value them, their values as previously discussed can be measured at production costs.
Since a major part of the work is done by unpaid voluntary labour, wage rates paid for similar kinds of work
on local labour market should be used to estimate labour costs.
130

E. Compilation of production accounts of selected industries

5.60. Elaboration of methods, contents of output and inputs, and sources of information for most industries,
particularly non-financial industries can be obtained from the United Nations publication, Handbook of
National Accounting: Accounting for Production: Sources and Methods mentioned previously. This part will
mainly elaborate on the calculation of outputs only and only sometimes on inputs of the industries deemed to
be conceptually complicated or whose definitions in the 1993 SNA are not the same as the 1968 SNA.

1. Agriculture, forestry and fishery

5.61. The SNA recommends that, in order to have a concept of production that is consistent with other
economic activities, the managed growth of crops, standing timber or other trees and fish and livestock reared
for food, under the control of institutional units, have to be treated as a continuous process of production whose
output consists of work-in-progress -- that is, output that is not yet sufficiently processed to be marketed or
used to produce other goods and services. “Assume the process of production takes several periods (months,
quarters, or years, as the case may be) to complete. The value of the output produced in each period can then
be measured as work-in-progress by distributing the value of the finished agricultural products (harvested
crops, slaughtered animals, etc.) in proportion to the costs incurred each period.”25 The measurement of output
of work-in-progress in each period can be seen in paragraphs 5.37-5.40 of this chapter. The only difference
in application here is that work-in-progress of agriculture, forestry and fishing is not a production cost in each
period but the value of finished goods allocated on the basis of production cost.

5.62. The output of agriculture, forestry and fishery is normally not estimated by using sale values but is
based on the quantities and prices that are estimated separately. For crops, for instance, quantity is estimated
by sampled cultivated areas and their corresponding yields, losses and seeds.26 One point that needs
clarification is the definition of agricultural output in the 1993 SNA which excludes from output and inputs
the output that is used as intermediate consumption within the establishment. This means that the products that
are produced by a household and then used as seeds will be excluded from agricultural output and inputs.
Only seeds that are obtained from the market are included. Conceptually, value added of agricultural activities
does not change whether non-marketed seeds are included or excluded. Thus, if I/O practitioners wish to
include seeds in both output and inputs in order to see clearly the inputs required, the method should be
footnoted so that no misunderstanding may occur. Agricultural statistics, however, use concepts of output that
are different from that of the SNA. The main reasons given are: (i) the SNA concept does not provide the
information needed to prepare a seed replacement plan; (ii) it is difficult to gather data required for
intermediate uses of harvested output within the same establishments. The relationship of the different
concepts27 is shown below:

25
System of National Accounts 1993, para. 6.96.

26
For detailed instructions on how to compile the output of agriculture, forestry and fishing, see A System of Economic
Accounts for Food and Agriculture, FAO, 1996.

27
Ibid., page 75.
 131

Harvested output

less Losses (wastage) on producing farm between harvest and the use or storage of the
output. In such cases output covers both the principal product and by-products

equals Total output

less Intermediate uses of harvested output, principal product or by-products within the
producing farm (establishment) itself

equals Output (SNA)

2. Wholesalers and retailers

5.63. The output of wholesalers and retailers is "measured by the total value of the trade margins realized
on the goods they purchase for resale. A trade margin is defined as the difference between the actual or
imputed price realized on a good purchased for resale and the price that would have to be paid by the
distributor to replace the good at the time it is sold or otherwise disposed of. The margins realized on some
goods may be negative if their prices have to be marked down. They must be negative on goods that are never
sold because they go to waste or are stolen.”28 The reason the output of wholesalers and retailers is measured
by trade margins is because their productive activity is construed to be the provision of services in transporting
and storing goods, displaying the goods in an informative and attractive way and then packaging them
according to the requirement of an individual customer. The goods themselves are not regarded as being
significantly transformed in this process and are therefore not treated as intermediate input into the distributor's
own process of production.

5.64. The output of wholesalers and retailers would include delivery costs if it is included in the cost of the
goods sold. If the delivery cost is billed separately, it will be treated separately as output of transport services.
In the latter case, the trader performs two separate services: trade and transport.

5.65. The output of wholesale and retail trade, or trade margins, is calculated by the SNA by the following
formula:

(3) Trade margins = Sales - value of goods purchased for resale + changes in inventories

5.66. Table 5.7 illustrates how trade margins are calculated given three methods of valuing inventories. Its
structure is similar to that of table 5.10 in appendix A, except that production is replaced by purchases for
resale as a reason for inventories to increase. Another difference is that goods produced are valued at the prices
at which they may be sold in the market. Thus, the unit cost for production and sales are the same in table
5.10. In table 5.7, additions to inventories are valued at purchasers’ prices of goods bought for resale at the
time goods are entered into inventories, and withdrawals from inventories are valued at purchasers’ prices of
good bought for resale at the time goods are withdrawn. By this treatment, trade margins may be calculated
by subtracting cost for purchases from sales at the prices of the accounting period. They reflect trade mark-up
at the time of selling.

28
System of National Accounts 1993, para. 6.110.
132

Table 5.7. Example describing the calculation of trade margins given different methods of inventory
valuation

SNA LIFO FIFO

Period 1 2 Year 1 2 Year 1 2 Year

1. Beginning inventories

Unit 0 60 0 0 50 0 0 50 0

Value 0 300 0 0 250 0 0 250 0

2. Purchases

Unit 60 5 65 60 5 65 60 5 65

Unit cost 5 7 5 7 5 7

Value 300 35 335 300 35 335 300 35 335

3. Addition to inventories

Unit 60 5 65 60 5 65 60 5 65

Unit cost 5 7 5 7 5 7

Value 300 35 335 300 35 335 300 35 335

4. Sales

Unit 10 40 50 10 40 50 10 40 50

Unit cost 6 8 6 8 0 6 8

Value 60 320 380 60 320 380 60 320 380

5. Withdrawal from inventories

Unit -10 -40 -50 -10 -40 -50 -10 -40 -50

29 30 31
Unit cost 5 7 5 7&5 5 5&7

Value -50 -280 -330 -50 -210 -260 -50 -200 -250

6. Changes in inventories

Unit 50 -35 15 50 -35 15 50 -35 15

Value 250 -245 5 250 -175 75 250 -165 85

7. Ending inventories

Unit 50 15 15 50 15 15 50 15 15

Value 250 5 5 250 75 75 250 85 85

8. Output calculated as sales - purchases for resales + changes in inventories

Value 10 40 50 10 110 120 10 120 130

29
LIFO withdraws the last 5 units of incoming inventory first, values them at the unit cost of 7 (i.e. last in, first out).
The other 35 units withdrawn are valued at the unit cost of 5.

30
FIFO values the 10 units withdrawn at the unit cost of 5 (i.e. first in, first out).

31
FIFO withdraws the 40 units that enter inventories first. They are valued at the unit cost of 5.
 133

5.67. The trade margin calculated by the SNA method for period 1 is 10, which is equal to the unit sold (10)
multiplied by the market mark-up prevailing at the time the goods are sold (6 - 5 = 1), and for period 2 it is 40,
with a quantity sold of 40 and the market mark-up of 1 (8 - 7). This valuation rule in the SNA applies even
when prices change within the same period. Thus, to approximate trade margins by the SNA method, one may
take the product of quantity sold and average mark-up during the accounting period. As long as there is a
mark-up, the minimum value of output of trade margins is zero even when no sale takes place.

5.68. The estimates of outputs by the LIFO and FIFO methods are higher than the SNA’s because the former
also contain capital gains on the units sold while they are in inventories because the inventories withdrawn are
not all valued at the prevailing replacement prices. The significant differences in outputs estimated by the three
methods shown in table 5.7 are due to the large ratio of inventories over sales and the large change in prices
given in the example. In countries with inflation under control and a normal ratio of inventories,32 the
differences should not be large.

3. Transport services

5.69. Transport services move people and goods, with the latter called transport margins. Transport margins
can appear in three different forms: (i) margins paid as part of the price of goods, i.e. no bills on transport
services are issued to consumers, (ii) margins paid as separate bill to traders who will arrange for the transport,
and (iii) margins provided by purchasers either as own-account transport or as hire from the transport industry.
There are two methods of treating transport services, and depending on which method is adopted in the input-
output framework, the output of the sector could be different.

5.70. The first method aimed at presenting data as they are actually transacted is recommended by the SNA
as follows:

(a) Transport services provided by producers, wholesalers, retailers or third parties will be treated
separately as transport margins only if they are billed separately from the values of goods to
their customers. If bills are issued separately, customers buy not only the goods, but also
transport services from producers, wholesalers, retailers or third parties. These transport
services are then treated as secondary product of the trade enterprise, producers or third
parties. However, if they are not billed separately, but included in purchasers' prices of the
goods bought, they become part of the trade margins or values of the goods (in the case of
producers) and the transport costs become their intermediate consumption;

(b) Own-account transport services consumed within the establishments which, by SNA rule, are
treated as own-account goods [Link] not be treated as part of the output of the
establishments but as their operating costs, and therefore do not form part of the transport
outputs.

5.71. The second method adopted by some input-output practitioners who want to treat transportation as a
functional concept would always remove transport services billed or not billed to customers from trade margins
and include them into the output of the transport industry. They would also apply the same method to own-
account transportation. The second method always produces higher output value of transport services than the

32
In the United States, inventory-sales ratios for most goods were below 2, except for a few durable goods. In 1995
and 1996, the average ratio for manufacturing goods was below 1.4.
134

first one. Outputs of own-account transport services, if based on the second method, are usually estimated by
simple rule, such as by multiplying the total number of trucks with the estimated average output per truck.
When doing this, inputs associated with own-account transport must also be removed from the mother activity
and these costs are replaced by the consumption of transport services equal to the sum of values removed. To
be fully consistent with the SNA, the second method should not be used in the SNA input-output framework.
It could be introduced as a satellite study of the transportation industry like a study of a tourist industry that
will be discussed in chapter X. In the SNA, the transport industry contains only services that are billed
separately or consumed outside of the establishments.

5.72. The complications in deciding output of transport services involving mobile equipment such as ships,
aircraft, railways, etc. that operate outside the economic territory in which the units are resident will be
discussed in chapter VI on treatment of imports.

4. Owner-occupied housing services and government-subsidized housing

5.73. The SNA assumes that owner-occupied dwellings provide housing services for their owners. These
services are part of the output of the unincorporated enterprises sector. The output of these services can be
approximated by market rents for the same kind of housing, taking into account factors such as location and
amenities as well as the size and quality of the dwelling itself. In countries where no market rents are
available, the output of the services of owner-occupied housing can be measured at least as the value of
depreciation of the dwelling measured at current market value or current replacement cost. Remember that
do-it-yourself repairs are not within the SNA production boundary and thus do not enter as part of the housing
services. Expenses paid for housing repairs are treated as household final consumption.

5.74. To be compatible with the treatment of owner-occupied housing, the output of government subsidized
housing should also be measured at equivalent market rents. The difference between market rents and
subsidized rents is treated as social benefits in kind provided by Governments, which is part of the individual
final consumption of government.33 Assuming the market rents are 100, but renters pay only 80, then output
should be 100 of which 80 goes to household final consumption and 20 goes to individual final consumption
of government (see definitions of final consumption in chapter VII). If the subsidized housing is government-
owned and the output estimated by equivalent market rents is higher than the subsidized rent, then the
difference between output estimate and the subsidized rent should also be imputed as social benefits in kind.

5. The output of financial intermediaries and its treatment in the input-output framework

5.75. Many financial intermediaries like money-market funds, trusts, and stock brokers, charge their
customers for their services. However, others such as banks and the like do so for only a very few services,
while major ones rendered to their customers in channelling funds from depositors and savers to borrowers
are not charged explicitly. Charges for these services are implicit in the difference between interest received
from borrowers and interest paid to depositors. As a consequence, the implicit service charges of financial
intermediaries are measured by the difference between interests received and interests paid. Interests received

33
This is based on the interpretation of the Inter-Secretariat Working Group on National Account, “...that payments
by governments to market producers for goods and services should be treated as social benefits in kind and not subsidies
on products if those market producers provide those products directly and individually to households in the context of
social risks or needs and to which the households are legally entitled.” SNA News and Notes, No. 3, January 1996,
United Nations Statistics Division (UNSD).
 135

should exclude those received from lending own funds. The total output of financial intermediaries is equal
to the sum of explicit and implicit service charges. The output of the central bank is measured like that of other
financial institutions. However, if its gross output fluctuates widely, or is negative, a second-best approach of
measuring output by cost may be adopted.34 Below is a review of conceptual issues. Measurement issues are
discussed in appendix B since until now there are no internationally agreed standards for treating them.

Conceptual issues

5.76. In the 1968 SNA, implicit service charges are not distributed directly among the users of the services
provided by the financial intermediaries but are assumed to be consumed by a fictitious industry representing
all industries. Since this is a fictitious industry that is supposed to produce zero output, a negative value which
is equal to the value of implicit service charges has to be entered as the value added of the fictitious industry
in order to make the output of this section equal to zero. The treatment is aimed at avoiding allocating implicit
service charges to every consumers. It creates two problems:

(a) It is not possible to compute the input-output coefficients of the fictitious industry. As a
consequence, this fictitious industry has to be aggregated to the financial intermediary
industries;

(b) Given that all industries and the household sector as well as the Government use the services
of financial intermediaries, not imputing these implicit service charges to the intermediate
consumption of individual industries or to the final consumption of final consumers will
increase the value added of the individual consuming industry. Overall, imputing to the value
added of the fictitious industry a negative value which is equal to the total implicit service
charges reduces the GDP since the final consumption of bank services by the household and
government sectors is treated as intermediate consumption by industries.

5.77. Though not rejecting the 1968 method, the 1993 SNA recommends an alternative, i.e. the distribution
of implicit service charges to all consumers, including industries as intermediate input and final consumers as
final consumption. In this case, when measured as final expenditures, the GDP will increase in comparison
to the 1968 SNA treatment by the amount of implicit service charges allocated to final consumption. The
value added of each individual industry will be lower than before but the total value added, i.e. the GDP, will
be higher since less than a full amount of implicit service charges is allocated as input to industries.

5.78. Table 5.8 shows how banking service charges are treated in the old 1968 SNA and table 5.9 shows
how they may be treated in the 1993 SNA. The total explicit service charges are assumed to be 5 and the total
implicit service charges are assumed to be 15. In the 1968 SNA the full value of implicit service charges are
assumed to be consumed by the fictitious financial industry. In the 1993 SNA, 7 is allocated to be consumed
by industries and 8 allocated to final demand. The GDP in the 1968 SNA is equal to the total value added or
total final demand which is 59 while the GDP in the 1993 SNA is higher and equal to 67. The difference in
GDP is equal to the implicit banking services allocated to final demand, which is 8 in our example.

34
According to the decision by the Inter-Secretariat Working Group on National Accounts. Ibid.
136

Table 5.8. The treatment of imputed bank service charges in the 1968 SNA

Intermediate consumption Final demand Output

Other industries Banking industry Fictitious

industry

Others 35 6 59 100

Banking 5 15 0 20

Fictitious 0 0 0

Value added 60 14 -15

Output 100 20 0

Table 5.9. The treatment of imputed bank service charges in the 1993 SNA

Intermediate consumption Final demand Output

Other industries Banking industry

Others 35 6 59 100

Banking 5+7 8 20

Value added 53 14

Output 100 20

5.79. The distribution of explicit service charges, in principle, is easy as long as banks and other financial
intermediaries keep records of their revenues supplemented by records of expenditures by industries,
government and the household sector. The distribution of implicit service charges is quite complicated. The
SNA recommends the use of a proxy to represent pure interest rate, which is called the reference rate. The
estimation of the reference rate will be discussed later. Borrowers will have to pay a rate higher than the pure
interest rate. The difference between the rate borrowers pay and the pure interest is the implicit service charge.
Depositors also have to pay for service charges, and therefore they receive an interest payment less than the
pure interest. Pure interest may differ from one class of consumers to another. This is particularly true for
financial transactions in foreign financial instruments because of differences in inflation rates and other factors
such as government control measures among countries.

5.80. Overall, for an economy, interest payments by borrowers (ib) are equal to pure interest (i) plus implicit
service charges to borrowers (Sb); and interest receipts by depositors (id ) are equal to pure interest (i) minus
implicit service charges (Sd) to depositors, or:

ib = i + Sb
id = i - Sd

The difference between interest payments and interest receipts must be equal to total implicit service charges.

ib - id = (i + Sb) - (i - Sd) = Sb + Sd
 137

From the equations above, implicit outputs of financial intermediaries, which are indirectly measured and
abbreviated as FISIM (Financial intermediation services indirectly measured) equal ib - id. Service charges on
borrowers are Sb = ib - i and service charges on depositors are Sd = i - id. The amount of service charges on
borrowers and depositors is dependent on where the pure interest rate i is located between ib and id.

5.81. The pure interest rate (i), or reference rate, for borrowers and depositors in the domestic market can
be approximated by the average inter-bank lending rate which has the advantage of not incorporating risk
premiums and intermediation services. Given banks' interest receipts and payments, broken down by
institutional sector, and statistics for outstanding bank deposits and loans in the balance sheet of banks, broken
down by institutional sector, we can easily calculate imputed service charges consumed by each sector. The
charges involve the calculation of pure interest flows on the basis of the pure interest rate and sectoral bank
deposits and loans.

6. Production accounts of insurance enterprises

5.82. Activities of insurance companies are similar to those of financial intermediaries. Insurance companies
collect premiums from units that buy insurance against specified events, manage these funds and pay out
claims to the same units when the specified events occur. Besides receiving premiums, insurance companies
also receive property income from the investment of their technical reserves. Technical reserves are necessary
because there is a time lag between the time when premiums have to be paid, which is at the start of the
insurance period, and the time claims are paid out.

5.83. The output of insurance companies should be the value of services rendered by them for protection
against specified events.

(a) Life insurance

5.84. The value of the output of life insurance establishments is calculated as follows:

Formula 1

(a) Gross premiums earned

(b) + Net income from investment of the insurance technical reserves
(c) - Total claims due
(d) - Increase in the actuarial reserve that excludes allocation of capital gains and losses
to the reserves which are treated as revaluations in the balance sheets outside of the
input-output framework.

5.85. Item (b) includes:

(i) Property incomes receivable less property income payable

(ii) Rentals and other incomes

Item (ii) covers secondary products of insurance companies as most of them are normally involved in real
estate and some other business services. The output calculated by formula 1 should include the primary
product which is insurance services and the secondary products, i.e. rentals and other incomes.
138

5.86. Item (a) must relate only to the accounting period the insurance policy covers and must include all
premiums paid in the current or previous periods which cover the risks incurred during the current accounting
period and exclude late payment of premiums for previous periods and prepayment of premiums for the
coming periods. Item (c) must cover claims which become due for payment during the accounting period
though they might not have been paid yet. Net income from investment and capital gain or loss of the insurance
enterprise's own capital are excluded from items (b) and (d). Therefore the following relationships should be
observed when one calculates the insurance output:

Gross premiums earned = Premiums receivable - change in the reserve due to prepayment
Claims due = Claims payable + change in the reserve against outstanding claims.

In fact, there are three types of reserves: reserve due to prepayment, reserve against outstanding claims, and
the actuarial reserve against future claims. Because insurance companies may use different technical terms for
the three reserves it is important to understand the nature of each one. Since many insurance companies report
on premiums receivable and claims payable rather than premiums earned and claims due, the following
formula can be used to calculate life insurance output:

Formula 2

(a) Gross premiums receivable

(b) + Net income from investment of the insurance technical reserves
(c) - Total claims payable
(d) - (Increase in the reserve due to prepayment + increase in the reserve against
outstanding claims + increase in the actuarial reserve).

In calculating changes in reserves, it is important to exclude capital gains/losses from the value of total reserves
in the subsequent period. These capital gains/losses are treated as revaluations in the balance sheets outside
of the production framework. In business balance sheets, reserves are revalued mainly for investments in
financial assets and rarely for investments in fixed assets.

(b) Non-life insurance

5.87. Either formula 1 or 2 can also be used to calculate the output of non-life insurance, except that change
in the actuarial reserve is zero since non-life insurance companies do not have the actuarial reserve. However,
they should have the other two reserves: the reserve due to prepayment and the reserve against outstanding
claims.

5.88. Sometimes, there is a reserve that is not directed to any of the purposes of the three reserves mentioned
above. A change in this reserve, which may be called contingency reserve, should be considered part of the
operating surplus that has not been distributed as dividends.

5.89. On the basis of the experiences of some countries, it is likely that the insurance output and
consequently the insurance value added may fluctuate widely from one year to the next It is also possible that
value added is negative. These problems may happen even if insurance companies are healthy and growing.
The instability of output and value added makes the use of I/O analysis useless as the latter has to rely on the
stability of I/O coefficients. There are two solutions to this problem: (a) spread output and value added over
a number of years so that claims may be spread out, or (b) calculate output from costs. Solution (a) can only
cover the years before the year for which national accounts need to be prepared in order to release national
 139

accounts statistics on time. Even with that, the resulting output does not necessarily move along with cost of
production. Solution (b) clearly deviates from the SNA but may be needed for the use of the I/O coefficient
matrix in economic modelling.

7. Production accounts of privately funded pension funds

5.90. Pension funds for which output is calculated must be separate funds established for the purpose of
providing income on retirement for specific groups of employees. They are organized and run by private or
public employers or jointly by the employers and their employees. Pension funds that are not independent are
treated as part of the operation of the companies that run them and, payments would be treated as normal
compensation of employees. Pension funds should not include compulsory social security schemes imposed,
controlled or financed by the Government.

5.91. The output of privately funded pension funds is taken to be a service charge to the beneficiaries equal
to:

(a) Total actual contributions earned

(b) + Total imputed contribution supplements35
(c) - Benefits due
(d) - Increases (plus decreases) in pension reserves.

The new method is different from the 1968 SNA. In the 1968 version, since pension funds are assumed not
to be organized for profit-making purposes, their output is measured by the cost of operating them. This
includes intermediate costs, compensation of employees, other taxes on production and consumption of fixed
capital. If pension funds are also organized for profit making, the measurement of output by costs is not
accurate. However, the new method is more complicated in the sense that an increase in reserves, which is
basically an increase in the funds, must exclude capital gains/losses in the investments of the funds.

8. Production accounts of research and development

5.92. Since the 1993 SNA treats all expenses on research and development (R&D) incurred by enterprises
as part of their outputs they should be classified in separate establishments belonging to the enterprises. The
new ISIC also allows for this classification. However, as in the 1968 SNA, research and development output
for the enterprise’s own consumption is treated as intermediate consumption by the establishments within the
enterprises. The allocation of the intermediate consumption of research and development to other
establishments in the enterprise may be based on the proportion of value added or output of each establishment
as in the allocation of ancillary activities. The output of research and development services for its own use in
an enterprise may be measured by the total costs of production including the imputed rentals on its own
buildings. In the case of specialized commercial research laboratories and institutes, their output can be
measured by receipts of sales, commissions, fees, etc. After detailed inputs consumed by R&D are singled out,
it is important to deduct them from the inputs of the industries that consume them and in their places, only a
single value of expenditures on R&D is included as input.

35
These are equal to investment income from invested pension reserves.
140

9. Production accounts of mineral exploration

5.93. The output of mineral exploration is measured by the total expenditures on mineral exploration.
However, unlike research and development which are treated as intermediate consumption, mineral exploration
is treated as gross fixed capital formation.

10. Production accounts of producers of government services and of non-profit

organizations serving households (other non-market output)

5.94. Producers of government services and of non-profit organizations’ services serving households consist
of all agencies providing collective services such as public administration, regulation, the maintenance of law
and order, defence, non-profit services and also non-collective services that are supplied free or at nominal
prices to individual households. Examples of the latter that are mostly supplied on a non-market basis are
education, health and housing.

5.95. The output of all other non-market producers, whether providing collective services or individual
goods and services should be, according to the SNA, valued on the basis of their total costs of production and
not by prices of similar goods or services charged on the market, because of the difficulty of obtaining suitable
prices in practice. Thus, the outputs of producers of government services and of non-profit organizations are
measured by the sum of:

- Intermediate consumption
- Compensation of employees
- Other taxes less subsidies on production
- Consumption of fixed capital.

The net operating surplus is zero by convention. As the above value covers the whole of the output of a
non-market producer, receipts from any incidental sales of goods and services are ignored for the calculation
of output, while incidental sales determine which part of the output is consumed (incurred) in other sectors.

5.96. In an input-output table, producers of government services or of non-profit organizations’ services

serving households do not appear as only two sectors but as many separate sectors classified by distinctive
economic activities. For example, government services consist of general public administration as well as
education, health, electricity, sewage and refuse disposal, library, radio and television services, etc. These
government activities may be split and grouped with similar ones provided by the private sector and by
non-profit organizations serving households.
 141

Appendix A

VALUATION OF INVENTORIES IN OUTPUT CALCULATION

5.97. Business accounting does not concern itself with the concept of output. Its focus is on sales and the
cost of goods sold in order to measure net income. National accountants need to use business accounts in order
to derive the SNA concept of output. Output of non-financial activities is measured as follows:

Output = Sales net of discounts, returns, VAT and sales taxes + other uses + changes in inventories

Thus the valuation of inventories is quite important to the measurement of output, intermediate consumption,
gross capital formation and finally GDP in the SNA. Different values of inventories would produce different
output values. For this reason, this note will discuss in depth why business inventories should be revalued
according to the SNA concept.

5.98. In business accounting, many methods are used in valuing inventories at cost. The most frequently
used ones are: FIFO (first in, first out), LIFO (last in, first out), average cost method. While LIFO best matches
revenues and cost of goods sold, especially when there has been a prolonged period of inflation or deflation,
it is not best suited for measuring the current balance sheet: FIFO is better in this regard as the value of ending
inventories is closest to current market values. Table 5.10 shows the valuation methods of the SNA, LIFO and
FIFO using the perpetual inventory method of recording (PIM), instead of the periodic method of recording.
The PIM recording method is similar to the method recommended by the SNA and also adopted by some
business firms. Under PIM, sales and purchases of each item are recorded continuously. With the availability
of powerful micro-computers, PIM is increasingly adopted by business in replacing the periodic recording
method. From the illustration in table 5.10, it is possible to derive some basic ways to approximate the SNA
method in measuring non financial output. Table 5.10 divides the annual accounting period into sub-periods
in order to see the accuracy of the methods used in measuring output. In the table, products produced are
entered immediately into inventories and remain there until they are withdrawn for sale. Differences in
valuation methods lie in the valuation of inventories withdrawn.

5.99 The SNA method values addition to inventories at the basic prices prevailing at the time of entry and
withdrawal at the prices at which they are then sold. The SNA method, however, values ending inventories
(assets) at market prices (which are basic prices) at the time of valuation. This revaluation is not introduced
in table 5.10.

5.100. The LIFO method values incoming inventories similarly to the SNA but withdrawals at the cost of
the last items entered into inventories.

5.101. The FIFO method values incoming inventories as do the other methods but values withdrawals at
the cost of the first items acquired.

5.102. The output calculated by the SNA method is 335, which matches the actual output assumed in the
example. The output calculated by LIFO is 345 which is more than the actual output. The output by FIFO is
355 which is farther away from the actual output than LIFO. The difference, or error, is capital gains due to
inflation. When prices are declining, FIFO gives an output value that is closer to the actual one than LIFO.
Changes in inventories calculated according to the SNA include all capital gains (or loss) on the inventories
sold and therefore the output calculated by the SNA eliminates all capital gains from sales.
142

Table 5.10. Example describing output calculation of goods production activities using different
methods of inventory valuation

SNA LIFO FIFO

Period 1 2 Year 1 2 Year 1 2 Year

1. Beginning inventories

Unit 0 60 0 0 60 0 0 60 0

Value 0 300 0 0 300 0 0 300 0

2. Production

Unit 60 5 65 60 5 65 60 5 65

Unit cost 5 7 5 7 5 7

Value 300 35 355 300 35 335 300 35 335

3. Addition to inventories

Unit 60 5 65 60 5 65 60 5 65

Unit cost 5 7 5 7 5 7

Value 300 35 355 300 35 335 300 35 335

4. Sales

Unit 0 10 10 0 10 10 0 10 10

Unit cost 5 7 5 7 5 7

Value 0 70 70 0 70 70 0 70 70

5. Withdrawal from inventories

Unit 0 -10 -10 0 -10 -10 0 -10 -10

Unit cost 5 7 5 7&536 537 538

Value 0 -70 -70 0 -60 -60 0 -50 -50

6. Changes in inventories

Unit 60 -5 55 60 -5 55 60 -5 55

Value 300 -35 265 300 -25 275 300 -15 285

7. Ending inventories

Unit 60 55 55 60 55 55 60 55 55

Value 300 265 265 300 275 275 300 285 285

8. Output calculated as sales + changes in inventories

Unit 60 5 65 60 5 65 60 5 65

Value 300 35 335 300 45 345 300 55 355

36
LIFO withdraws the last 5 units of incoming inventory first and values them at the unit cost of 7 (i.e. last in, first
out). The other 5 units withdrawn are valued at the unit cost of 5.

37
FIFO values the 10 units withdrawn at the unit cost of 5 (i.e. first in, first out).

38
Idem.
 143

5.103. In terms of evaluating changes in inventories and ending inventories, which are used in the balance
sheet, the SNA values all inventories at current market prices. Therefore changes in inventories and the ending
inventories must be revalued at the unit cost of 7. The ending inventories, after revaluation, are equal to 385
(table 5.11), which is higher than the ending inventories in table 5.10 by 120. This difference is the capital
gain for goods remaining in inventories. The results in the table also show that, in case of inflation, FIFO
produced a value of ending inventories that is closer to market price than with LIFO.

5.104. There is no easy and accurate short-cut to convert LIFO and FIFO values of inventories to SNA values
without detailed information as shown in table 5.10. The revaluation is, in fact, better done by business
accountants than by national accountants at the last stage of data collection. It would be significant for national
accounting if business accountants agreed to value inventories as the SNA recommends, but it is highly
unlikely because banks and financial analysts always want to value inventories in the most conservative
manner, i.e. at either cost or market value, whichever is lower.

5.105. However, as long as prices do not change, whatever method used in valuation, output and inventories
would be the same. Errors will be great if prices increase or decrease rapidly.

5.106. “This suggests that even when prices are changing a good approximation to the PIM may be obtained
by taking the difference between the quantities of goods held in inventory at the beginning and the end of the
accounting period and valuing this difference at the average prices prevailing within the period. This method,
which may be described as the "quantity" measure, is widely used in practice and is sometimes mistakenly
considered to be the theoretically appropriate measure under all circumstances. The quantity measure will be
the same, or virtually the same, as the perpetual inventory method measure not only when prices are constant
but also when the quantities of goods held in inventory rise or fall at a steady pace throughout the period.
Conversely, the conditions under which the quantity measure may provide only a poor approximation to the
PIM are when prices are rising or falling and when inventory levels fluctuate within the accounting period.”39

5.107. Change in inventories as an approximation as suggested by the SNA can be split into two parts:
volume (or quantity) changes and nominal holding gains.

Volume changes in inventories at current market prices = VPC = (qt - qt-1) pt

Nominal holding gains = IVA = (pt - pt-1) qt-1

where q stands for quantity, p stands for price and t stands for time. The above equations are used to calculate
volume change in inventories at current prices and nominal holding gains which, given the information in table
5.10, are shown in table 5.11. When the PIM method of recording inventories is used, as shown in table 5.11,
the volume changes in inventories at current market prices will approximate SNA value of change in
inventories. In this example, they are exactly the same. Changes in inventories in the balance sheet (table 5.11,
row 4) are equal to volume change in inventories at current prices plus nominal holding gains at the end of
every period and of the year (rows 6 and 7).

39
System of National Accounts 1993, para. 6.68.
144

Table 5.11. The calculation of output, volume change and holding gains of inventories

Period 0 1 2 Year

1. Quantity of inventories 0 60 55 55
2. Prices 5 5 7
3. Balance sheet values 0 300 385 385
4. Change in inventories (SNA balance sheet 300 85 385
values)

5. Volume change in inventories 60 -5 55

6. Volume change in inventories at current 300 -35 265
prices (VPC)
7. Nominal holding gains (IVA) 0 120 120

8. Sales 0 70 70
9. Volume change in inventories at current 300 -35 265
prices (VPC)
10. Output 300 35 335

5.108. In practice, inventories of many products may not be recorded by using the PIM as shown in tables
5.10 and 5.11 with full information on quantities and prices, but valued only at the beginning and end of the
accounting period and an approximation of the method shown in table 5.12 may have to be used. Practices
may vary among countries in treating inventories. Below only the Canadian practice is given as an example.40
The practical calculation in Canada follows eight steps:

(1) B Book values as reported

(2) D Deflators
(3) K = B/D Constant dollar book values
(4) DK$ = Kt - Kt-1 Constant dollar value of volume change
(5) R Revaluers
(6) VPC = R*DK$ Current dollar value of volume change
(7) CB = Bt - Bt-1 Change in reported book values
(8) IVA = CB - VPC Nominal holding gains (which in Canada is called inventory
valuation adjustment).

5.109. On the basis of the turnover period, the average length of time that a good spends in inventory, an
appropriate price deflator is set which reflects the costs of goods entering inventory during the turnover period
of the reference period. For example, with a 2.3-month turnover period, the fourth quarter deflator would be
the sum of the December price index for November and 0.3 of October, divided by 2.3. For revaluers (R),
monthly reference price series are used reflecting the average costs of the inventories during the period;

40
This part is based on a paper by Kishori Lal, titled “Recording of Change in Inventories in the SNA and in the
Business Accounts - A Case Study of Canadian Practices” in the United Nations Handbook of National Accounting:
Links between Business Accounting and National Accounting, Series F, No. 76.
 145

quarterly revaluers are simple averages over the quarter. And VPC is the value of changes in inventories used
as part of gross capital formation. Annual changes in inventories, if calculated directly, in principle are equal
to the sum of quarterly changes. But in practice, they differ because the underlying principles of aggregation
are not met in an inflationary situation. There is no solution to this problem yet.

5.110. The same example in table 5.11 calculated according to the Canadian method is shown in table 5.12
below:

Table 5.12. The calculation of volume change and holding gains in inventories by the Canadian
method

Period 0 1 2 Year=sum of Year, calculated

2 periods given yearly data
Steps

B 0 300 385 385 385

D 1.0 1.0 1.4
K=B/D 0 300 275 275 275
DK=Kt-Kt-1 0 300 -25 275 275
R 1.0 1.0 1.4
VPC=R*DK 0 300 -35 265 33041
CB=Bt-Bt-1 0 300 85 385 385
IVA=CB-VPC 0 0 120 120 55

The last column shows that holding gains calculated by using the revaluer as an average price index of the year
distorts the true value of holding gains. The method is less distortional with a monthly or at least quarterly
calculation.

41
This is the simple average of price indexes in two periods (=(1+1.4)/2=1.2).
146

Appendix B

MEASURING OF OUTPUT OF FINANCIAL INTERMEDIARIES

5.111. The indirect measurement of financial intermediation services (FISIM), or implicit output of financial
intermediaries is not straightforward. Until now, there are no internationally accepted methods of measurement.
Below are a few alternative methods of measurement that are proposed or attempted by some countries. In
order to show how FISIM are calculated and allocated, an example shown in table 5.13, representing a
realistic set of data, will be used. Table 5.13 shows the balance sheet of all financial intermediaries including,
on one side, assets with the corresponding interest receivable from them and, on the other side, liabilities with
their corresponding interest payable. In this table, securities include bonds, bills, etc. Own funds should be
equal to net worth and shares, and can be calculated as the difference between total assets and other items other
than own funds on the liability side. In practice, as we can observe from the table, loans are not equal to
deposits because financial intermediaries raise funds not only from deposits but also their own funds or by
incurring liabilities such as selling securities. They earn revenue not only from making loans but also from
buying securities.

Table 5.13. The balance sheet of financial intermediaries and their interest incomes and expenses

Assets Liabilities

Stocks Interest Stocks Interest

receivable payable

A Fixed assets, shares and accounts 6,162 C Own funds (net worth and 3,741
receivable less payable, insurance shares)
technical reserves, shares and other
equity

B1 Securities other than shares 13,349 1,102 D1 Securities other than shares 10,935 568

B2 Loans 62,004 4,634 D2 Deposits 69,467 3,453

Interbank 9,288 402 Interbank 7,188 287

Residents 6,188 247 Residents 6,188 247

Non-residents 3,100 155 Non-residents 1,000 40

Others 52,716 4,232 Others 62,279 3,166

Residents 51,975 4,226 Residents 62,279 3,166

Non-residents 741 6 Non-residents 0 0

B3 Others (currency and deposits) 4,298 85 D3 Other liabilities 1,670 36

Total 85,813 5,821 85,813 4,057

 147

Canada

5.112. Canada42 calculates FISIM by taking all interests receivable (from either loans or securities) less all
interests payable (to depositors or other liability holders), less the interests generated by the part of own funds
which is lent out. The formulas it adopts are as follows:

FISIM = RAB - RLD - [C - (A+B3-D3)] x (RAB + RLD) / (B + D)

Where B = B1 + B2 + B3, D = D1 + D2 + D3
RA = Actual average interest rate on assets
RL = Actual average interest rate on liabilities

If [C-(A+B3-D3]) is positive, part of own funds is lent out

If [C-(A+B3-D3)] is negative, fixed and other similar assets are funded from non-owners' funds

5.113. In the formula of FISIM above the third item shows a deduction of the interest on the part of own
funds, C-(A+B3-D3) lent out, which is not supposed to generate FISIM. The interest rate on the own funds
lent out is estimated by the average interest rate on B and D.

RA = (interest on B)/B = 5,821/79,651 = .073

RL = (interest on D)/D = 4,057/82,072 = .049

FISIM = (.073 x 79,651) - (.049 x 82,072) - [3,741 - (6,162+4,298-1,670)] x (.073 x 79,651 + .049
x 82,072) / (79,651 + 82,072)
= 1,764 - (-308) = 2,072

5.114. Canada does not use a reference rate representing the pure interest rate to allocate FISIM to borrowers
and depositors. It allocates FISIM by the ratio between deposits and loans including securities other than
shares, i.e., with borrowers the ratio B/(B+D) is used and with depositors the ratio D/(B+D) is used. The
advantage of the Canadian approach is that FISIM allocated to borrowers and lenders are always positive.

5.115. For imports and exports of FISIM, the average domestic FISIM rate FISIM/((B+D) will be applied
to the total of deposit assets abroad plus the loan liabilities of the country.

42
Statistics Canada, “Calculating and Allocating FISIM: Canadian Approach”, a paper submitted to the Working
Party on National Accounts of Eurostat, Luxembourg, 26 June 1996.
148

European Union proposal

5.116. Though the final form of the European Union proposal43 is not certain at the time of this writing, it
seems that it will be close to the one described below. The European Union’s proposal calculates FISIM by
taking only interests receivable from loans less interests payable to depositors, and then allocates FISIM to
borrowers and depositors by reference rates. The separate calculations of FISIM on loans and deposits are as
follows:

FISIM on loans to residents = interest receivable on residents - (loan stocks to residents x internal
interbank reference rate)

FISIM on deposits with resident financial institutions = (deposit stocks by residents x internal
interbank reference rate) - interest payable by residents.

5.117. Interest reference rate applied only to loans to residents and deposits with resident financial
institutions is calculated by the ratio of interest receivable on interbank loans over stock of interbank loans.
In table 5.14 (see row 2) this internal reference rate is 247/6,188 = .04. Thus,

FISIM on loans to residents = (4,226+247) - ((51,975+6,188) x .04)) = 2,146

FISIM on deposits by residents = ((62,279+6,188) x .04)) - (3,166+247) = -675
Total FISIM on residents = 2,146 + (-675) = 1,471.

5.118. In comparison to the Canadian approach, the European Unions’s FISIM proposal is much lower
because it takes only interests on loans and deposits into account.

5.119. Exports and imports of FISIM are calculated for loans granted to non-residents and for deposits of
non-residents. The formulas used are similar to those used for residents, except that "internal reference rate"
is replaced by "external reference rate".

5.120. The external reference rate is calculated as the average rate weighted by the levels of stocks in “loans
to non-resident financial institutions (FIs)" and "deposits by non-resident financial institutions (FIs)”. In table
5.14, these are transactions with interbank non-residents.

External reference rate = (155+40)/(3,100+1,000) = .048. Thus,

FISIM on loans to non-residents = (155+6) - ((3,100+741) x .048) = -23

FISIM on deposits by non-residents = (1,000 x .048) - 40 = 8
Total exports of FISIM = -23 + 8 = -15.

5.121. Imports of FISIM will use the same formulas as exports with the same external reference rate, except
that one needs to use loans granted by non-resident FIs and deposits with non-resident FIs. This information
may be obtained from the balance of payments.

43
Eurostat, “Proposal for a Council Regulation on the Allocation of Financial Intermediation Services Indirectly
Measured (FISIM)”, a paper submitted to the Working Party on National Accounts of Eurostat, Luxembourg, 26 June
1996.
 149

5.122. The total FISIM will be the sum of FISIM on residents plus exports of FISIM = 1,471 + (-15) = 1,456.
This value is also shown in column (5) bottom row of table 5.14.

5.123. The problem with the European Union’s proposal is that the estimated pure interests payable and
receivable will not be equal because total loans are not the same as total deposits. From the calculation shown
above, the estimated pure interests receivable on loans are equal to interest receivable on loans (4,634) less
FISIM on loans to residents (2,146) shown in paragraph 5.117, less FISIM on loans to non-residents (-23)
shown in paragraph 5.120, and similarly the estimated pure interests payable by residents are 3,453 + (-675+8)
= 2,786. Even taking into account other interests receivable and payable shown in table 5.13, rows B1, B2,
D1 and D3, which are pure interests, total pure interests receivable on loans (=2,511+1,102+85=3,698) are
not equal to total pure interests payable by depositors (=2,786+568+36=3,390). To make the pure interests
on both sides equal, some adjustment is necessary. For instance, the difference can be allocated proportionally
to borrowers and depositors on the basis of their shares of estimated reference interests.

5.124. The use of a reference rate does not always yield positive service charges if it does not lie in between
the deposit rate and the lending rate. In the example above, depositors are "paid" by financial institutions to
use their services. The reason is that, while the lending average (4,226/51,975)= .0813) is higher than the
average deposit rate (3,166/62,279 = .051), the deposit rate is higher than the "internal reference rate" (=.04).
This problem happened in many countries, at some particular time.

Malaysia

5.125. The Malaysian method for calculating total FISIM is similar to the Canadian method except in the
allocation of FISIM to borrowers and depositors. The reason for taking in all interests receivable and payable,
except from own funds, is that many financial companies raise intermediated funds by deposits but invest them
in securities, not loans to generate income, and those securities should generate some output.

5.126. The definition of the references rate is similar to the European proposal but the estimation is different.
The reference rate is taken as a simple average interbank rate during the accounting period. It is based on the
type of loan commonly used, which for Malaysia was a three-month interbank rate. The reason for this
decision is that the internal calculation of a reference rate as proposed by Eurostat may be biased as interests
paid are incurred throughout the accounting period but interbank loan stocks reflect only the value at the end
of the accounting period, particularly when interbank loans are only a small amount of total loan stocks. In
1987, when the average interbank rate was very low (3%), the internally calculated rate was much higher (6%).
The interest rates paid by borrowers and received by depositors are, however, calculated internally as other
loans are larger in value and therefore do not have the bias problem faced by the internally calculated reference
rate. The internally calculated interest rates paid and received are calculated as follows:

ib = (4,226/51,975) = .08
id = (3,166/62,279) = .05
r = .04 (This rate in the example is not based on table 5.14 but supposed to be the simple average of
most commonly used interbank rate)

5.127. FISIM is, however, allocated only to depositors and borrowers as follows:
150

To borrowers: (ib-r)/(ib-id)
To depositors: (r-id)/(ib-id)

Where ib: actual average interest rate paid by borrowers

id: actual average interest rate received by depositors
r: reference rates
ib, id are calculated directly from a table similar to table 5.13.

Table 5.14. Allocation of FISIM to borrowers and depositors

Allocation of FISIM to borrowers

Stocks Interest Reference Reference interest Estimated

receivable rate (estimated pure interest) FISIM

(1) (2) (3) (4)=(1)x(4) (5)=(2)-(4)

Interbank 9,288 402 396 6

Residents 6,188 247 0.04 247 0
Non-residents 3,100 155 0.048 149 6

Others 52,716 4,232 2,115 2,117

Residents 51,975 4,226 0.04 2,079 2,147
Non-residents 741 6 .048 36 -30

Total 1 4,634 2,511 2,123

Allocation of FISIM to depositors

Stocks Interest Reference Reference interest Estimated

payable rate FISIM

(1) (2) (3) (4)=(1)x(3) (5)=(4)-(2)

Interbank 7,188 287 295 8

Residents 6,188 247 0.04 247 0

Non-residents 1,000 40 0.048 48 8

Others 62,279 3,166 2,491 -675

Residents 62,279 3,166 0.04 2,491 -675

Non-residents 0 0 0.048 0

Total 2 3,453 2,786 -667

Total FISIM = Total 1 + 1,456

Total 2

5.128. The two proportions are always added to 1.0. The reason for a negative proportion is that the reference
rate does not lie between the interest payable by borrowers and that receivable by depositors. This problem
is similar to that faced by the European proposal but this method will yield the equality of the total of estimated
pure interests receivable and the total of estimated pure interests payable (see the last row of table 5.15).
 151

Table 5.15. The balance sheet of financial intermediaries and their interest incomes and expenses
by the Malaysian method

Assets Liabilities

Interest FISIM Pure Interest FISIM Pure

receivable interest payable interest

A Fixed assets, shares C Own funds

and accounts (net worth
receivable less and shares)
payable, insurance
technical reserves

B1 Securities other than 1,102 0 1,102 D1 Securities 568 0 568

shares other than
shares

B2 Loans 4,634 2,756 1,878 D2 Deposits 3,453 -684 2,769

Interbank 402 0 402 Interbank 287 0 287
Others 4,232 2,756 1,476 Others 3,166 -684 2,482
B3 Others (currency and 85 0 85 D3 Other 36 -684 36
deposits) liabilities

Adjustment for 30844

interest on own
funds

Total 5,821 2,756 3,373 4,057 -684 3,373

Implication for I/O compilers

5.129. The calculation of FISIM as discussed above may not be fully relevant to I/O compilers as long as the
output of financial intermediaries is prepared by national accountants. However, it is important to understand
the process in order to allocate the output to intermediate and final consumption correctly. In as much as we
know stocks of loans and deposits by the institutional sectors (such as non-financial corporations, financial
corporations, government, households, rest of the world), we can easily allocate FISIM to these sectors.

5.130. The distribution of FISIM to each industry, i.e. each establishment, is a lot more difficult. Similar to
the problem of distributing head-office administration costs of an enterprise to its component establishments,
loans and deposits cannot be linked to a specific establishment. The decision of depositing or borrowing is
made by the enterprise and not by establishments. For this reason, only ad hoc methods can be used to
distribute bank service charges to establishments. Preferably, FISIM should be distributed to each
establishment on the basis of its share of capital stock, but this information may not be available, in which case
FISIM may be distributed on the basis of its share of the enterprise output. Another difficulty in distributing
FISIM to individual establishments is the data itself. Banks keep information on their loans to and deposits
from their customers but are unlikely to reveal it except in aggregate form that only serves the purpose of
distributing the information among institutional sectors, and broad groups of economic activities such as

44
308 is the value of the third item in formula (b).
152

agriculture, construction, manufacturing, services, etc. Then, survey of industries in relation to interest receipts
and payments will be needed to estimate service charges paid by industries.

5.131. For the household sector owning unincorporated establishments and owner-occupied housing, it is
important to divide it into the unincorporated sector and the household sector itself because for the former
service charges are intermediate inputs and for the latter they are final consumption services and therefore will
have a different impact on the GDP.

5.132. For bonds issued by banks, their whole return could be treated as pure interest, that is, no implicit
services are rendered in respect of bonds. For non-domestic lending and borrowing where rates are quite
different from domestic rates, the interbank rates may not be appropriate and a reference rate for each currency
must be found which could be the interbank rates in the country that issues the currency. Service charges on
lending abroad or on deposits from non-residents are exports of services and the charges on borrowing abroad
and on deposits abroad are imports of services.
 153

VI. TREATMENT OF IMPORTS

6.1. This chapter will deal with an in-depth treatment of imports in the I/O model and the use table in basic
values in particular. Valuation is further discussed and the use of input-output for the analysis of imports is
also briefly discussed. For a full discussion of impact analysis on imports, reference is made to chapter XII.

A. Compilation of imports

6.2. Imports are goods and services provided by non-residents to residents through sales, barter, gifts or
grants. Imports take effect when transfer of ownership does. To account for imports, it is important that
residents and non-residents and the time when transfer of ownership takes place be clearly identified. Readers
are advised to read the SNA for a full discussion of the above issues. It suffices to note here that:

(a) Residents include all institutional units (individuals, households, corporations, quasi-corporations,
unincorporated enterprises, government units, non-profit institutions serving households) having a
long-term economic interest in the economy or residing in it for more than one year, students studying
abroad, institutional units working or going abroad for less than one year, ships, aircraft, drilling rigs
and platforms owned by residents but operating in international waters or airspace;

(b) Non-residents include all institutional units that stay or operate for less than a year or do not have a
long-term economic interest in the economic territory of the economy. Non-residents include students
from foreign countries, embassies, consulates, military establishments and other entities of a foreign
Government that are considered as extraterritorial by the economy and international organizations.

6.3. Transportation needs special attention in the SNA as a carrier may operate outside the economic
territory in which the corporation is resident, either in international waters, in airspace, or in one or more other
economies. Below are some rules that the SNA applies to transportation:47

(a) A carrier that operates in international waters or airspace should be attributed to the economy of
residence of the operator; the same rule applies if the activity takes place in more than one country
during the course of, but for less than a year;

(b) A carrier who operates in other economies is a resident of the country in which the activity occurs, if
accounted for separately by the operator and so recognized by the tax and licensing authorities there;
otherwise the activity may be attributed to the country of residence of the operator;

(c) In the particular case of ships flying flags of convenience, the country of registry in most instances is
different from the country of residence of the operator but the shipping activity is to be attributed to
the country of residence of the operating unit; if the unit has branches for tax or other considerations,
their activities are attributed to their country of residence;

47
See System of National Accounts 1993, paras. 14.25-27.
154

(d) When a corporation is established by special legislation by two or more Governments acting jointly
and is registered in each of the countries concerned, there are two possible treatments: the first one
is to allocate all transactions to each of the countries in proportion to the amount of financial capital
which they have contributed; the second one is to treat the corporation as resident of the country in
which its headquarters are located and the premises of the corporation in the other countries as foreign
branches (direct foreign investment enterprises) which are residents of the countries where they are
located.

6.4. Goods that are imported and then re-exported or returned unprocessed are omitted from exports and
deducted from imports.

6.5. Direct purchases abroad by residents are included in imports of goods and services. These purchases
abroad by residents have fewer reasons to be identified by product when they are analysed by using input-
output analysis, so, the SNA's treatment is quite satisfactory for this. Residents are assumed to be households
only. If residents include business persons or government officials, their purchases abroad should be treated
as intermediate consumption of enterprises. A modification in the use table must be introduced to handle this
problem. These purchases may be treated as one value of non-competitive imports of either households,
government, enterprises or non-profit organizations, which will be clarified later in the chapter.

6.6. Imports of goods, in addition to regular ones, include:

(a) Imports of goods for processing;

(b) Imports of goods which are exported for processing abroad;
(c) Imports of repairs to fixed assets;
(d) Imports of goods under financial leasing;
(e) Imports of non-monetary gold used as valuables and industrial production;
(f) Transfers in kind from the rest of the world such as household goods and gifts transferred by migrants,
foreign countries, etc.;
(g) Imports of fuel and stores purchased abroad by ships and aircraft operated primarily in international
waters by resident enterprises, fish and salvage purchased abroad by national fishing vessels;
(h) Direct purchases abroad by residents;
(i) (-) Transport and insurance services on imports rendered by residents and non-residents.

6.7. Imports of services, in addition to regular ones, include:

(a) Transfers in kind from the rest of the world;

(b) Imports of insurance services;
(c) Imports of financial intermediation services indirectly measured;
(d) Transport and insurance services on imports rendered by non-residents.1

1
One must be careful in analysing and accounting for these items. The reason is that if an importer pays upon
delivery at a domestic port, costs of insurance and transport services on imports are included in the value of the goods
and would be paid by the sellers. In that case, payments for insurance and transport services to non-residents are not
registered as import of services while in the SNA, they should be treated as such. The payments by non-resident shippers
to resident insurers or carriers are registered as export of services while they will be treated as domestic production in
the SNA.
 155

6.8. Estimation: Data for imports of merchandise (goods) come mainly from foreign trade statistics. Data
for imports of services come mainly from the balance of payments. Merchandise trade statistics must be
adjusted to exclude goods in transit and goods that are returned. They must also be adjusted to include
transactions that are not covered by foreign trade statistics which are elaborated on in para. 6.6. Other sources
of data are: foreign exchange records, customs, maritime transport statistics, government accounts, accounts
of organizations involved in international business like post offices, telephone and telegraph, insurance
companies and special surveys on expenditures by tourists.

B. Treatment of imports in the use table

1. Valuation of imports

6.9. Imports c.i.f. are treated as equivalent to basic values by the SNA. They are the values of imported
products delivered at domestic ports before being subjected to import duties. Imports c.i.f. plus import duties
are treated as equivalent to producers' values. In the SNA, imports are shown as part of the supply table. In
input-output analysis, imports must be incorporated as part of the use table. To do this, the use table must be
in basic prices and imports must be entered as negative values and become part of net final demand. In this
chapter, imports are shown as part of the use table.

6.10. In the SNA, imports are all valued f.o.b., i.e. measured without the transnational costs of transport and
insurance services on imports to bring goods from the border of one country to that of another country. In
input-output tables, however, components of imports must be valued c.i.f. to make them equivalent to basic
values that are used to measure domestic products, but in order to be compatible with the SNA the total value
of imports must be measured f.o.b. through a c.i.f./f.o.b. adjustment.

2. Types of imports

6.11. Imports are usually classified into two types: competitive and non-competitive imports, the latter being
also called complementary imports. Competitive imports include imported products that are also being
produced by the domestic economy. Non-competitive imports include products that are either not producible
or not yet produced in the country. For example, products that can only be grown in tropical climate are treated
as non-competitive imports by countries with cold weather. This distinction is important for the modelling of
imports in input-output tables that will be discussed later.

(a) Non-competitive imports

6.12. The first table (table 6.1.a) shows the import of goods that are used as inputs by industry A. The same
example will be used to illustrate how non-competitive and competitive imports are treated in the use table in
basic prices. Table 6.1.b shows the treatment of non-competitive imports as a separate row in the use table.
It has no corresponding column since no equivalent products are produced domestically. One can create as
many rows as there are non-competitive imported products instead of only one row in order to show more
detail in the input-output table.

6.13. The example in table 6.1.a shows the change of value of the imported product from a foreign port to
the gate of industry A. The total cost to the producers of product A is 32, which is the purchasers' price paid
by industry A that uses the non-competitive imports. Non-competitive imports are valued c.i.f. at only 24 in
the use table to make them equivalent to basic values. The total value of imported goods must, however, be
valued f.o.b. in the SNA, so a column and a row of c.i.f./f.o.b. adjustment on imports must be introduced in
156

the use table. While the principle of adjustment was discussed in chapter II, the example in table 6.1.b is
intended to clarify all the detail of the treatment of imports.

6.14. The following will explain how the content of imports c.i.f. in part A of table 6.1. a is entered into the
final demand quadrant in table 6.1.b. The payment of transoceanic margins has already been included in the
c.i.f. value of imports. The part of transoceanic margins supplied by resident producers (.6 + .2) is treated as
domestic services, the other part supplied by non-resident producers (1 + .2) is treated as imports of services
in the column of imports of services (negative values). The transport and insurance margins are entered as part
of the column of c.i.f./f.o.b. adjustment (1.6 and .4) so that the differences between total margins and imported
margins become domestic margins. Domestic production of transport services serving imported goods is equal
to 2.6, which includes .6 as transport services on imports rendered by resident carriers and 2 as internal
transport margins to bring the goods to users (see table 6.1.b). The internal transport margins in the example
show the complete picture of the treatment of imports, but compilers of the use table need not worry about
the collection of data on internal transport margins in the industry quadrant because they would be filled up
when compilers prepare intermediate inputs of industries.

6.15. In table 6.1.b, a positive value of imports value c.i.f. (24) is entered as intermediate inputs to industry
A, and a negative value (-24) is entered in the column on imports of goods so that no output of the imported
product is produced in the country. In the column on imports of goods, an adjustment equal to transoceanic
margins (2) is added to make the total value of imports of goods equal to an f.o.b. value of -22. Duties (3) and
sales tax (1) on imports (table 6.1.a ) which are taxes on products are treated as part of the total cost of
production of industry A (table 6.1 b).

(b) Competitive imports

6.16. Table 6.1.c shows the treatment of competitive imports of product B used as input by industry A. It
uses the same example as table 6.1.a, the only difference being that the whole value of non-competitive imports
is now considered as the import of only one product - product B. The imported product is treated in the same
fashion as the domestic one. Product B in basic prices (24 i.e. imports c.i.f.) is entered as input to industry A,
regardless of whether it is imported or not, then imports of B are entered as negatives (-24) in the column of
imported goods so that no value equal to the value of import of product B is produced inside the economy.
Transoceanic margins, duty and other internal margins are treated as in the case of non-competitive imports.

6.17. With the above method of valuation of imports, it is important to remember that transoceanic margins
on imports including transport and insurance services provided by resident producers are part of final demand
and must be estimated exogenously in I/O modelling by using the I/O coefficient matrix.
 157

Table 6.1.a. An example of imports broken down by components

Imports of products by industry A

Cost element Amount

A Foreign port value (imports f.o.b.) 22.0

Transoceanic margins 2.0

Transport 1.6

by resident carriers .6

by non-resident carriers 1.0

Insurance .4

by resident carriers .2

by non-resident carriers .2

Imports c.i.f. 24.0

B Duty 3.0

Internal margins 5.0

Transport 2.0

Trade with sales tax 3.0

Trade margins 2.0

Sales tax 1.0

Purchasers' value 32.0

Table 6.1.b. Treatment of non-competitive imports in the use matrix

Non-competitive imports in use matrix in basic prices

INDUSTRY NET FINAL DEMAND PRODUCT

OUTPUT

PRODUCTS A Imports C.i.f./f.o.b. Final consumption,

adjust- gross capital
Goods Services ment formation, exports

Trade 2.0 2.0

Transport 2.0 -1.0 1.6 2.6

Insurance -.2 .4 .2

Non-competitive imports 24.0 -24.0 0.0

C.i.f./fob adjustment on imports 2.0 -2.0 0.0

Total uses at basic prices 28.0

Taxes on products 4.0

Total uses at purchasers' price 32.0 -22.0 -1.2 0.0

158

Table 6.1.c. Treatment of competitive imports in the use table

Competitive import of product B by industry A in the use matrix in basic prices

INDUSTRY NET FINAL DEMAND PRODUCT

OUTPUT

PRODUCTS A Imports C.i.f./f.o.b. Final

adjust- consumption,
ment gross capital
Goods Services formation,
exports

Product B 24.0 -24.0 0.0

Trade 2.0 2.0

Transport 2.0 -1.0 1.6 2.6

Insurance -.2 .4 .2

C.i.f./f.o.b. adjustment on 2.0 -2.0 0.0

imports

Total uses at basic price 28.0

Taxes on products 4.0

Total uses at purchasers' 32.0 -22.0 -1.2 0.0

price

3. Alternative presentation of imports in the use table

6.18. This section will discuss alternative ways to present imports in the use table so as to provide
information for economic analysis by using I/O models. With regard to non-competitive imports, there is no
alternative to the treatment mentioned above, i.e. they must be included in the use matrix as either one row,
each of its element representing all non-competitive imports used by an industry, or as a number of rows, each
row representing a non-competitive imported product.

6.19. However, there are two different ways to show competitive imports in the use table, i.e.
uses of domestic and imported goods and services shown together or separated.

6.20. The first way is to show all inputs consumed in the intermediate demand matrix and in final
expenditures regardless of origins of supply. This method is shown in table 6.2. The sum total of consumption
in each row is the supply of a product. The total product output produced by the economy is then obtained by
subtracting imports from total supply. The method has the advantage of showing the technical requirements
of every industry. In table 6.2, only a simple input-output structure with no secondary products is shown to
obtain the balance between input and output; the table also assumes no c.i.f./f.o.b. adjustment on imports to
simplify the presentation.

6.21. The other method (uses of domestic and imported goods and services separated) shows competitive
goods and services in the same manner as the treatment of non-competitive imports, that is imported inputs
are shown separately from domestic ones. This method is shown in table 6.4. For example, industry 1 uses 18
as the value of product 1, of which 8 is imported. Then, table 6.4 will show only the value of 10 of domestic
product. The value of imported 8 is moved down to the row of (all) imports. Product 2 is treated similarly. The
values of competitive imports are moved down to the rows of competitive imports. These rows are treated
similarly to the value added in an I/O model.
 159

Table 6.2. Input-output table with total input presented in the intermediate and final demand
matrices

Intermediate consumption Net final demand Product output

Products 1 2 Final demand Imports

1 18.0 36.0 80.0 -34.0 100.0

2 42.0 36.0 62.0 -38.0 102.0

Non-competitive 5.0 -5.0 0.0

imports

Value added 35.0 30.0

Industry output 100.0 102.0

Table 6.3. Matrix of competitive imports

Intermediate consumption Final demand Total competitive

imports by
Products 1 2 products

1 8.0 16.0 10.0 34.0

2 12.0 20.0 6.0 38.0

Competitive imports by 20.0 36.0 16.0

industry

Table 6.4. Input-output table in which domestic and imported resources are separated

Intermediate consumption of domestic Final demand Product output

products

Products 1 2

1 10.0 20.0 70.0 100.0

2 20.0 16.0 56.0 102.0

Non-competitive 5.0
imports

Import of product 1 8.0 16.0 10.0

Import of product 2 12.0 20.0 6.0

Value added 35.0 30.0

Industry output 100.0 102.0

6.22. The second method is used by many countries. It will provide information for the basic I/O model.

6.23. Imports as shown in matrix form in table 6.3 are usually not readily available. In fact, they are derived
from special surveys and assumptions. For many products, especially when labels of origins are required, users
160

are well aware if they are imported or domestically produced. Surveys will then serve the purpose of filling
the import matrix of table 6.3. For other products such as petroleum or gas etc., the users may not be able to
determine their origins. In such cases, we have to assume that the same import share for a particular product
is applied to every user. For example, if a certain percentage of a product is imported by the country in
question, that percentage of import can be assumed for every user of the product. Or if surveys on imports can
only identify import shares for a group of users, the rest of the imports can be distributed proportionally to the
rest of the users on the basis of their total uses. Exports are always from domestic production. So, before
applying proportional distributions for the estimation of import shares per user, it is helpful to set aside exports.
In order to use assumptions effectively based on the proportionality principle, product classification should
be as detailed as possible. In some countries, the number of products exceeds 2,000.

C. Analysis of imports in an input-output model

6.24. How an input-output model is used for the analysis of imports is dependent on many factors such as
the influence of changes in prices, exchange rates, incomes and so on but it also depends on how the input-
output table is constructed. This section will emphasize the latter problem more than other economic factors.
The analysis is, therefore, expected to be simple and provide answers mainly to simple questions of a
comparatively static nature.

6.25. Given that the input-output table is organized as recommended in table 6.2, the relationships in
production and final demand can be written in the following general formula:

(6.1) X = AX + Y

where

X = Vector of output
A = I/O coefficient matrix
Y = Vector of net final demand which is equal to final consumption expenditures, gross capital formation
and exports less imports.

6.26. Matrix A represents total input coefficients including input from both domestic and foreign sources.
The vector of final demand represents total final expenditures minus total imports including imports for both
intermediate and final expenditures. The model in table 6.1 provides a neat way to show the relationships in
production, consumption, gross capital formation, exports and imports in the economy, but it is not yet a model
that can be used to calculate, for example, the impact of a change in final demand on imports. The reason is
that in order to get the value of Y so as to find the value of X, it is necessary to know the total import of each
product because net final demand Y is equal to final demand less imports. But imports are, in fact, the
unknown entity that the model aims to estimate. Thus, equation 6.1 is just a balancing equation and as it
stands it cannot be used to calculate impacts on outputs or imports of a change in final demand. It can be used
for analysis given that one assumes a value for the vector of net exports, a component of net final demand.

6.27. Table 6.4 showing domestic input is the equation that can be directly used to calculate the impact on
outputs and imports of a change in final demand. The relationships in table 6.4 can be written as follows:

(6.2) X = AdX + Yd

6.28. In equation 6.2, Yd represents only final demand that is produced domestically. Ad , which is the
matrix derived from the intermediate flow matrix in table 6.3, also represents input coefficients that are
 161

domestically produced. It is possible to show Ad as including rows of imported products with corresponding
columns of zeros. The other alternative is to show it without these rows and corresponding columns. In the
latter case, the rows of imported products are treated in the same manner as the rows of value added. After
the vector of output is obtained, imports can be calculated by multiplying the matrix of import coefficients to
the output vector. Equation 6.2 can be derived mathematically as follows:

From equation 6.1, we can write

X = (Ad + Mid)X + (Yd + Ym - M)

where

Yd is the vector of domestically-produced final-demand products

Ym is the vector of imported final-demand products
M is the vector of imported products
Mid is the coefficient matrix of imported intermediate demand

thus,

M = Mid X + Ym

X = Ad X + Yd + (Mid X + Ym - M)

then,

X = Ad X + Yd

One can solve for X by using the formula:

(6.3) X = (I - Ad)-1Yd

6.29. Equation 6.3 needs all the information that is or can be assumed to be, available. Economically,
equations 6.2 and 6.3 show that as there is an increase in final demand, with part of this final demand lost to
imports. So only Yd, the part that is domestically produced will have an impact on domestic production. And
then, as intermediate input is required to satisfy this increase, only input that is domestically produced will
have an impact on other rounds of domestic production, which is represented by Ad, and the other part is
leaked out of the system. In order to use equation 6.3 for analysis, it is necessary to first estimate Yd, i.e. final
consumption of domestic goods and services, demand of domestic capital goods and exports. In other words,
it is necessary to estimate total final consumption expenditures, gross capital goods and also the imports of
final consumption goods and services, and of capital goods. Only imports of intermediate goods and services
are estimated by model 6.3.

6.30. It is also possible to use matrix A directly by introducing the import matrix into the system in a
different way as follows:

(6.4) Ad = RqA

adij = rij x aij for all i and j.

162

6.31. The symbol q stands for product multiplication in which each element of Ad is the product of the
corresponding elements in R and A, i.e. with the same row and column indexes. For example (using the input
coefficient derived from table 6.2):

.18 .35 .0
A ' .42 .35 .0
.05 .00 .0

.75 .75 .0
R ' .71 .44 .0
.00 .00 .0

6.32. Rij is the domestic ratio of the input-output coefficient aij, which is the ratio of input aij supplied by
domestic producers. In other words, it is equal to 1 minus the import ratio of coefficient aij.

From the example above

ad21 = r21 x a21

or .2982 = .42 x .71

6.33. This example of relationship 6.3 is based on the assumption that was discussed previously, i.e. the
import of each product used as input by each industry is proportional to the total input used. The matrix of
domestic ratios R assumes that for the first product, the same import proportion is applied to all users since
there is no information other than the ratio of import over total supply. For the second product, the import ratio
varies from one user to another; this information should obviously be based on surveys. In the third row and
column corresponding to non-competitive imports zeros are entered as noted previously.

6.34. With this method, new assumptions on imports can be introduced easily into matrix R, while the input-
output matrix A is assumed constant.

6.35. Matrix R allows different import ratios for each element of the A matrix. However, if the same
import ratio is assumed for all users of a product, then matrix R becomes diagonal, i.e. only the element in the
diagonal is positive, the elements off diagonal are all zeros. In this case, we will have:

A d ' R̂A
r1 0 . 0
0 r2 . 0
R̂ ' that is
. . . .
0 0 . rn
 163

This is the most commonly used assumption on imports.

6.36. From the calculation above, one can also calculate the import of intermediate goods and services by
type of product necessary to support a given increase in final demand.

6.37. Competitive imports are equal to:

(A - Ad)X + (Y - Yd)

or

MidX + Yd

6.38. The main advantage of the analysis presented in this chapter is its simplicity and the ability to
incorporate vast detailed data in the import matrix without resorting to sophisticated theories. When prices and
technology are stable, the modelling of imports presented above is quite useful and, in fact, has been used by
many input-output practitioners to calculate import requirements and domestic production. Of course,
sophisticated economic theories have been introduced into the analyses using input-output model but this
literature, which is vast, will be found elsewhere since the description of all analytical methods is beyond the
objective of this book.
 165

VII. COMPILATION OF FINAL DEMAND

A. Introduction

7.1. Final demand is part of the use of goods and services. In the use table, these values are shown along
the rows. This chapter will look into the conceptual framework and data sources for preparing final demand
as broadly discussed in chapter II. Expenditures on goods and services are either for intermediate consumption
by industries or for final demand ( or final uses). Intermediate consumption is the value of goods and services
that are used as inputs in production processes, as discussed in chapter V. Final demand includes final
consumption expenditures to satisfy individual needs or wants and the collective needs of members of the
community, gross capital formation and exports.

7.2. The term final demand is not mentioned in the SNA but is used here for the convenience of discussing
input-output economics. From the point of view of modelling, as presented in table 1.1 of chapter I and YC
in equation 4.8 of chapter IV, net final demand - final demand less imports - is required. It is the vector
summing:

(a) Final consumption expenditures to satisfy individual needs or wants and the collective needs of
members of the community;

(b) Gross capital formation;

(c) Exports f.o.b.;

(d) Minus imports f.o.b.

Net final demand can be observed in table 4.4 of chapter IV as the sum of columns 5-10. Net final demand
is equal to the GDP. In order to arrive at net final demand, components of final demand should all be in basic
values because imports are all in basic values. Thus values of final demand have already gone through a
process converting purchasers' prices into basic prices. The basic data that need to be collected first are final
demand in purchasers' prices. The definition of final demand is shown below with more details in table 7.1.

7.3. Final consumption expenditures are broken down into:

(a) Final consumption expenditures of the household sector including the value of goods and services that
are purchased, actually consumed and ultimately paid for by resident households;

(b) Final consumption expenditures by non-profit institutions serving households (NPISHs) which, for
simplicity’s sake, are assumed by the SNA to serve individual needs of resident households even
though some may serve collective needs, for example some research carried out by NPISHs;
 Table 7.1. Components of final demand at purchasers' prices

FINAL CONSUMPTION EXPENDITURES GROSS CAPITAL FORMATION EXPORTS FINAL

DEMAND

Households Non-profit General government Gross Changes Acquisition

institutions fixed in inven- less disposal
serving capital tories of valuables
households formation

Individual Individual Collective Individual (1)+.+(8)

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Product 1 100 32 6 2 28 168

Product 2 48 1 5 5 9 68
Product 3 2 0 2
Taxes less subsidies on products 0

C.i.f./f.o.b. adjustment. 0

Direct purchases abroad by 3 3

residents

Direct purchases at home by -1 1 0

non-residents

TOTAL 152 1 5 5 32 6 2 38 241

 167

(c) Final consumption expenditures by general government which may include the national Government
and state and local governments. These expenditures are divided into two parts: one part serving the
collective needs of members of the community and the other part serving individual needs of resident
households.

7.4. Gross capital formation consists of:

(a) Gross fixed capital formation;

(b) Changes in inventories of finished goods, goods for resale, materials and supplies and work-in-
progress;

(c) Acquisition, less disposal, of valuables: valuables are assets that are not used primarily for production,
do not deteriorate over time under normal conditions and are held as stores of values. Examples of
valuables are precious stones and metals, jewelry, works of arts such as paintings, sculptures, antiques,
etc. These items are not covered in the 1968 SNA.

Gross capital formation should also be identified by institutional sectors, i.e. by non-financial, financial, non-
profit institutions serving households (NPISHs), households and general government sectors. It should also
be classified by industries so as to facilitate the estimation of capital stocks by industries.

7.5. Final consumption expenditures and gross capital formation are all measured at purchasers' prices in
table 7.1. To convert them into basic prices, one should consult chapter III on valuation. However, whether
components of final demand are measured at purchasers' prices, producers’ prices or basic prices, the total
value of final consumption or gross capital consumption by institutional sectors is the same.

7.6. Exports are measured f.o.b. at either the detailed or aggregate level. Imports are measured c.i.f. at the
detail level but the total value of imports is valued f.o.b..

B. Final consumption expenditures and actual final consumption

1. Government final consumption

7.7. The final consumption expenditures of the general government sector are divided into individual final
consumption and collective final consumption expenditures while all NPISHs final consumption expenditures
are treated as individual final consumption. So, what is collective final consumption and what is individual
final consumption?

7.8. Individual final consumption expenditures of government include individual goods and services
that “are essentially ‘private’, as distinct from ‘public’ goods. They have the following characteristics:

(a) It must be possible to observe and record the acquisition of the good or service by an individual
household or member thereof and also the time at which it took place;

(b) The household must have agreed to the provision of the good or service and take whatever action is
necessary to make it possible, for example by attending a school or clinic;
168

(c) The good or service must be such that its acquisition by one household or person, or possibly by a
small, restricted group of persons, precludes its acquisition by other households or persons.”1

7.9. Included in individual goods and services are expenditures by general government for the following:

(a) Health services including public health;

(b) Recreation, culture and religion;

(c) Education;

(d) Social security and welfare services;

(e) Housing, refuse collection and sewerage services.

The above items are classified in division 14 of the Classification of Individual Consumption by Purpose
(COICOP) and cross-classified with divisions in the Classification of the Functions of Governments
(COFOG).2 Also included in individual goods and services are overhead expenses made by ministries in
connection with the administration and functioning of these institutions (see para.9.86 of the SNA). The final
consumption expenditures by general government for individual goods and services are also called social
transfers in kind by general government in the SNA. The social transfers in kind provided to individuals may
take three different forms whereby government:

(a) buys goods and services on the market and transfers them to individuals;
(b) partly or fully reimburses goods and services bought by individuals;
(c) fully operates and pays for the facilities that produce non-market goods and services (i.e. government
output) and provides them at prices that are not economically significant to individuals.

The first two items are called by the SNA social benefits in kind (SNA, paras. 8.101 to 8.104) since recipients
have little or no choice; they are not part of the output of government services because the government unit
"does not engage in any further processing of such goods or services."3 The third item is called transfers of
individual non-market goods and services (SNA, paras. 8.105 to 8.106) and is a part of the output of
government services.

7.10. The treatment of social transfers in kind means that a given quantity of goods consumed by individuals
or households will be split into two parts: the part funded by government is individual final consumption of
government and the part eventually paid by households is final consumption of households. Thus it is also
important to note that social benefit reimbursement by government is not treated as current transfers to

1
System of National Accounts 1993, para. 9.81.

2
International organizations have been working together to produce international standards for identifying items in
government expenditures that may be classified as individual consumption. A draft linking individual goods and services
and COFOG items has been published by the United Nations, see Draft Classification of Expenditure by Purpose on 24
November 1998. Compilers should consult with the final one when it is ready in order to produce internationally
compatible statistics.

3
Ibid., para. 9.79.
 169

households. Only government transfers in cash to households for which households have total freedom to
spend are treated as current transfers. These current transfers are not part of a production account.

7.11. Collective final consumption expenditures include only services (there are no collective goods) with
the following characteristics:

“(a) Collective services can be delivered simultaneously to every member of the community or of
particular sections of the community;

(b) The use of such services is usually passive and does not require the explicit agreement or active
participation of all the individuals concerned;

(c) The provision of a collective service to one individual does not reduce the amount available to others
in the same community or section of the community. There is no rivalry in acquisition.”4

7.12. Included in collective services are the provision of security and defence, public administration, public
research and development, maintenance and improvement of law and order, general administration including
the setting and enforcement of policies, standards and regulation of public health, education, etc. In other
words, they should include all current expenditures that are not treated as current transfers or individual final
consumption expenditures.

7.13. The above clarification is also shown in the following simple example:

4
Ibid., para. 9.83.
170

Table 7.2. An example of government revenues and expenditures

Government expenditures 183

Government capital expenditures 30

Purchases of capital goods 23

Own-account construction 7

Current expenditures for non-market activities 100

Collective 80

Individual 20

Social benefits in kind (reimbursements included) 10

Social benefits in cash 43

Others* 0

Government revenues 183

Sales to households 5

Taxes 150

Others* 0

Deficit financing 28

*”Others” denote property income, capital and current transfers and transactions in financial assets and/or liabilities. They normally take
values other than zero assumed here for simplicity’s sake.

From the example, output, final consumption and capital formation of the government sector are as follows:

Table 7.3. Output, final consumption and capital formation of government sector

Output produced by government activities 107

Output of government services Equal to current expenditures for non-market activities 100

Secondary output Own-account construction 7

Government final consumption 105

Individual final consumption of government Or social transfers in kind equal to current expenditures for 25
non-market activities for individuals (20), less sales (5), plus
social benefits in kind (10)

Collective final consumption of government Equal to current expenditures for non-market activities for 80
collective needs

Government capital formation 30

7.14. The output of government services is equal to current expenditures for non-market activities. In this
example, the output of non-market activities of general government is 100 which may contain both non-market
output such as education, health, defense, etc., and market output produced by secondary activities such as
sales of documents, museum reproductions, etc. Out of 100, 80 are collective non-market services which are
treated as collective final consumption expenditures of general government, and 20 are individual non-market
goods and services which are treated as individual final consumption expenditures of general government after
deducting a sale of 5 paid by households. Sales of outputs produced by government activities to be used as
 171

the intermediate consumption of industries do exist, but are assumed to be zero in the example for the sake of
simplicity. In addition to the above expenditures, government partly or fully finances through reimbursement
final expenditures of the household sector for market goods and services, which are 10 in the example. These
values, which are commonly called in economic literature consumption subsidies,5 must be treated in the 1993
SNA as part of individual final consumption expenditures of general government.

7.15. By the SNA definition, the output of producers of government services, less the value of government
sales of non-capital goods and services, plus social benefits in kind is equal to government final consumption
expenditures. Government sales include such items as receipts from sales of postcards and reproductions by
museums, firewood sold by the government forestry department, government publications, fees for medical
treatments and school fees, etc. Information on government final consumption expenditures comes from reports
on government budget and expenditures. Other government expenditures such as subsidies to industries to
reduce operation costs or costs of capital goods, interest payment, costs of capital good procurement,6 etc. lie
outside the framework for calculating government output and final consumption.

7.16. The balancing of the output of government services,7 government final consumption is shown in table
7.4.

Table 7.4. Treatment of general government final consumption expenditures*

Other Activities of General government’s final consumption expenditures Household final

activities general consumption
government Collective Individual

x x 10

Market goods and 0 80 20-5 5

services

General x x
government
services

Value added x 100 80 25 5

*x means that values are not specified in the example.

7.17. It may be worthwhile here to point out the following SNA treatments:

(a) Subsidies by general government to enterprises to reduce input costs, prices of imported or exported
goods and services are treated as subsidies on products in the SNA sense of the word and included as
part of the value added of the recipient sectors; they do not form part of the output of government
services;

5
In the SNA, only producers get subsidies from government, not consumers.

6
Own-account capital formation, which is quite important, should be part of government output.

7
Secondary output is not included here since it may be treated as output of a separate establishment. For the example
given, it should be included in the construction industry.
172

(b) Subsidies by general government to enterprises to reduce prices of capital goods to enterprises are
capital transfers and do not form part of the output of government services either;

(c) Subsidies effected by government trading activities: when government buys products and resells them
at a loss as a matter of deliberate economic or social policy, the difference between the purchase and
selling prices is treated as other subsidies on production to government trading organizations. The
output of trading organizations are part of the general government sector (SNA, para. 7.78);

(d) Subsidies to households by government are channelled indirectly to households through subsidizing
enterprises and treated as social benefits in kind. They should therefore be part of individual final
consumption of government.8

2. Final consumption expenditures of NPISHs

7.18. Final consumption expenditures of NPISHs are equal to the gross output of producers of NPISHs
services less sales plus social transfers in kind. If own-capital formation is part of the output, it has to be
deducted from it to obtain final consumption. In short, all output of NPISHs, except for the part that is sold,
is treated as individual final consumption of NPISHs. The treatment of NPISHs is very similar to that of
general government. However, unlike the government sector, there is no collective final consumption of
NPISHs. In most countries, statistics on NPISHs are very weak, sometimes almost nonexistent. Surveys need
to be carried out to cover their activities. Many activities of NPISHs are quite similar to government activities
such as health services, recreation, culture, education and welfare services. But the following activities should
also be included: political parties, civic associations, religion, professional and labour organizations, activities
of environmental protection organizations, and other activities such as trust funds and charitable organizations
that fund research and scientific studies, etc.

3. Household final consumption expenditures

7.19. Household final consumption expenditures should include:

(a) All purchases of consumer non-durable and durable goods except dwellings which are treated as
capital goods of unincorporated enterprises of the household sector producing housing services;

(b) Imputed purchases of consumer durables by financial leasing (i.e. households may pay in installment,
but the value of the goods bought through financial leasing must be calculated and imputed as
individual final consumption);

(c) Imputed gross rental for owner-occupied housing (see para. 5.73 in chapter V again for more detail);

(d) Own-account final consumption of goods by owners of unincorporated enterprises;

(e) Bartered consumer goods and services (net);

(f) Domestic services provided by domestic servants;

8
This is based on a further interpretation of the SNA by the Inter-Secretarial Working Group on National Accounts,
SNA New and Notes, issue 3, January 1996.
 173

(g) Goods and services in kind provided by enterprises as a form of compensation of employees (see also
paras. 5.15 and 5.27 in chapter V);

(h) Imputed financial intermediary (banking, insurance, pension, etc.) service charges;

(i) Purchases minus sales of second-hand goods by households except dwellings (see chapter V, section
C.5 on the treatment of second-hand goods for further detail);

(j) Purchases by residents abroad;

(k) (Minus) Purchases by non-residents at home.

Estimation and sources of final consumption expenditures

7.20. Final consumption expenditures should be estimated by the commodity flow approach (the method
will be discussed in the next chapter) which utilizes various sources, the most common and important being
information obtained in the preparation of industry outputs, agricultural and industrial censuses and annual
surveys, annual reports of industrial associations or societies, reports of public utilities, reports from the tourist
industry, trade and balance of payments statistics, household expenditures surveys, censuses on retail sales and
services, central and local government budgets.

7.21. Data on household expenditures collected in surveys of households are useful for making benchmark
and annual estimates of distribution of household consumption expenditures by broad categories (i.e. objects
of expenditure). To break down these broad expenditures into a more detailed product classification, and to
supplement the gap in the household expenditure surveys which are generally based on small samples, it is
necessary to resort to a census of retail sales or annual retail sale statistics. However, household surveys would
not give any estimate on imputed gross rental, or direct information on imputed service charges of banks,
casualty and life insurance, though interest payments and premium payments may be given, and these SNA-
defined expenditures must be independently estimated. The method has been discussed in section E.5 of
chapter V.

7.22. Information from household surveys has to be cross-checked with information from retail sales and
other sources. Retail sale statistics give a vast source of data on household consumption expenditures.
Household consumption of goods and services can be sometimes identified by the nature of the goods and
services. Care must be taken to exclude from retail sales purchases by government and enterprises and also
second-hand goods. Only trade margins of second-hand goods involving households are treated as household
expenditures -- see again chapter V. Other sources of information on household expenditures are from
censuses of agriculture for estimating own-account production, and from reports of particular establishments
such as public utilities (i.e. telephone companies, gas and electric companies, water and other sanitary
services), schools and hospitals, restaurants and hotels, and financial firms such as banks, stockbrokers, etc.

7.23. Because no source of information gives the most accurate and complete data, one has to use the supply
and use tables introduced in chapter II, also called the commodity flow technique - to evaluate the data. This
technique is discussed in detail in the next chapter.

7.24. One final point that needs to be clarified is that, in the SNA, the concept of final consumption
expenditures of enterprises does not exist so that expenditures by enterprises for the benefit of individuals who
are not their employees must be imputed as current transfers to the household sector which will use the
174

transferred income for its final consumption9 even though the household sector does not have to pay for it.
These transfers in kind (not the same as social transfers in kind which are made by general government or
NPISHs) must be imputed to household final consumption expenditures in column 1 of table 7.1( para. 7.3
above). Whatever goods and services an enterprise buys for the main benefit of its employees, these
expenditures would have to be treated as compensation of employees in terms of payments in kind even though
they may be recorded by enterprises in their books as intermediate consumption. These expenditures would
then be treated as a part of both value added and imputed final household consumption expenditures. Outlays
on goods and services which, in addition to being of some benefit to employees, mainly benefit the employers
for purposes of public relations or employees’ morale, should be treated as intermediate consumption.
Examples of these outlays are expenditures on sports tournaments, medical examinations, work-related training
or educational programmes. Also important are the cases of proprietors of unincorporated enterprises who may
infuse both expenditures for their business and their own household consumption. In the latter case an
apportioning of expenditures between final and intermediate consumption is needed. On the other hand, if
employees are required to provide certain tools, equipments or uniforms at their own expense, these
expenditures should be treated as intermediate consumption by enterprises and then the compensation of
employees must be accordingly adjusted downward.

4. Final consumption expenditures versus actual final consumption

7.25. The distinction between final consumption expenditures and actual final consumption is an extension
of the distinction between collective and individual final consumption in the SNA. Final consumption
expenditures of an institutional sector are the expenditures incurred freely by the sector. Actual final
consumption of an institutional sector refers to either the taking possession of goods and services or the
consuming of goods and services by the sector. Actual final consumption aims at capturing the consumption
of goods and services that are provided in kind by other sectors. The distinction serves international
comparisons of the actual levels of final consumption by households and provides two alternative concepts of
consumption for analysis. Specifically, the actual final consumption of households includes:

(a) Household final consumption expenditures (column 1 of table 7.1);

(b) Final consumption expenditures of NPISHs (column 2 of table 7.1); and

(c) Individual final consumption expenditures of general government (column 4 of table 7.1): These
expenditures make up social transfers in kind by the general government sector.

Items (b) and (c) above are also called social transfers in kind by the SNA. All of the final consumption of
NPISHs is treated as social transfers in kind by the SNA, so that the actual final consumption by the NPISHs
is zero. The actual final consumption of the general government sector is equal to its collective final
consumption expenditures.

C. Gross capital formation

7.26. Gross capital formation includes:

9
These current transfers appear in the secondary distribution of income account of the SNA and do not appear in the
input-output table.
 175

(a) Gross fixed capital formation;

(b) Changes in inventories;

(c) Acquisitions less disposal of valuables.

The following paragraphs will discuss in detail the components of gross capital formation.

1. Gross fixed capital formation

7.27. Gross fixed capital formation is measured by the total value of a producer's acquisition, less disposal,
of fixed assets during the accounting period plus certain additions to the value of non-produced assets realized
by productive activities of resident producers. Fixed assets are tangible or intangible assets which are outputs
of production processes. They must have a life span of not less than one year, except for small tools of
economically insignificant value. The threshold value under which an expense is considered insignificant is
normally decided by convention or by the business accounting rules of a country. The 1995 European System
of National Accounts (ESA) sets the threshold values for European Union members.

7.28. For institutional sectors, their gross fixed capital formation may include existing or second-hand goods,
but for the total economy as a whole, gross fixed capital formation includes only new expenses on fixed capital
formation and net second-hand capital goods imported from abroad. For second-hand capital goods that
already exist in the economy, a purchase by one institutional sector, say unincorporated enterprises of the
household sector, must be netted out by a sale of the same value by another institutional sector, say the non-
financial sector. The increase to gross capital formation is the transfer cost only. The second example in the
annex to chapter X of the SNA is repeated below to show how a transaction in second-hand capital assets is
treated in input-output tables. In this example, say the first owner who is classified in the non-financial sector
sold second-hand machinery to an unincorporated enterprise which is classified in the household sector. The
transaction costs involved are shown in table 7.5. The seller received from the buyer 750, but the actual value
he received after paying the transfer costs was only 670. The buyer had to pay more than the price he paid to
the seller, i.e. 880. Thus 880 is considered gross capital formation of the buyer and 670 is considered the
reduction in gross capital formation of the seller. In the process of transaction, an output of trade services of
110 was generated, and an additional value of 210 (880-670) including taxes was added to gross capital
formation in the total economy. Table 7.6 shows all the transactions in an abbreviated form of the use table.
Here, one may wonder why new capital formation was created for the seller when no real new asset had been
created nor any improvement made. The solution to this problem in the SNA is to write the value of 80
incurred to the seller in the balance sheet as a holding loss. This solution, however, is not covered in the
production account, i.e. the I/O framework.
176

Table 7.5. Transaction in second-hand capital goods

Seller (non-financial) Buyer (household)

Acquisition / disposal value (basic value) 750 750

Costs of transfer 80 130

Trade 80 30

Taxes 100
Value received/Value paid 670 880

Table 7.6. Treatment of transaction costs in second-hand capital goods

Intermediate consumption Gross capital formation

... Trade ... Non-financial sector Household sector

...

Machinery -750 750

Trade 80 30

...

Taxes on products 100

Value added

Output/Total 110 -670 880

7.29. As shown in table 7.6, sales of domestic second-hand capital goods have to be entered as negative
values to net out the purchase value of the same good. For the acquisition value to net out the disposal value
of the same good, both of them must be measured at basic prices, i.e., excluding transfer costs. Net second-
hand capital goods from abroad are netted out by net imports.

7.30. Dwellings including houseboats, mobile homes and caravans used as principal residences are treated
as capital goods of owners of unincorporated enterprises producing housing services. Other consumer durable
goods such as refrigerators, air-conditioners, etc. are treated as final consumption expenditures, not capital
goods. Included in gross fixed capital formation are:

(a) Acquisitions, less disposal, by enterprises of new and existing tangible fixed assets (dwellings, other
buildings and structures, machinery and equipment) and by households of dwellings including:

Purchases;
Barter transactions;
Own gross fixed capital formation;
Capital transfers in kind;

(b) Acquisitions, less disposal, of intangible fixed assets (mineral exploration, computer software,
entertainment, literary or artistic originals) including:
 177

Purchases;
Barter transactions;
Own gross fixed capital formation;
Capital transfers in kind;

(c) Transfer costs of existing tangible and intangible fixed assets: the transfer costs cover expenditures
for agents' or lawyers' fees, dealers' margins or commissions, stamp duties and so on. The value of
non-produced fixed assets (i.e. land, mineral deposits, invented patents) themselves should not be
netted out as capital formation for the economy as a whole because these assets are not an output of
the processes of production for the existing accounting period or, in the case of natural assets, they are
not an output of any production process;

(d) Major improvements to tangible non-produced assets including land and costs associated with the
transfers of ownership of non-produced assets;

(e) Expenses to transform existing capital goods into different kinds of capital goods such as major
renovations and extensions:

(i) Expenses to transform natural assets such as draining, reclamation, clearing and levelling of land
for purposes of agriculture and construction, clearing for planting of forests, etc.;

(ii) Changes in animal stocks which are not reared primarily for purposes of meat production; these
consist of draught animals, animals used for purposes of sport or entertainment, cattle used to
produce dairy products, or wool and breeding stocks;

(f) Acquisitions of produced fixed assets under financial leasing.

7.31. It is important to recognize that though (a) and (b) above are valued at purchasers' prices, the
separation of these values into two components (basic values and transfer costs) would allow the value of items
(a) and (b) to be measured at basic prices so that the buyer and seller of a capital good face the same price.
This is important for the balancing of gross capital formation from the points of view of both the seller and
buyer. Transfer costs which make the difference between the buyer's price and the seller's are included in (c).

7.32. Gross capital formation should exclude:

(a) Outlays by government on construction and durable equipments that can only be used for military
purposes (they are treated by SNA as intermediate consumption by the producers of government
services). However, other capital goods that can also be used for civilian purposes like military
airports, roads, troop carrying transport equipment like buses, aircraft, ships, etc. are treated as capital
goods);

(b) Outlays to acquire non-produced tangible assets such as land, mineral deposits or timber tracts, or non-
produced intangible assets such as patented entities, leases, goodwill, etc. (but the transfer costs that
arise when these goods change hands are included) ;

(c) Other changes in volume of tangible assets owing to new finds of subsoil assets and natural growth
of other natural assets or to losses of tangible assets;
178

(d) Outlays on fixed capital formation by non-residents like embassies, consulates, and other
extraterritorial bodies such as international organizations.

7.33. Gross fixed capital formation should be prepared separately by industries as well as for these sections:

(a) Financial sector;

(b) Non-financial sectors;

(c) Government sector;

(d) Household sector;

(e) Non-profit institutions serving households sector.

7.34. Estimation: Data on construction are available from construction surveys or can be estimated from
building permits preferably on completed works. Data on plants, structures, equipment and other capital
expenditures are from production surveys of industries. It is important that data on gross capital formation be
classified by type and by industry. Own-capital formation such as housing construction by households is very
important in developing countries. It may be covered in household income and expenditure surveys, or else
a special survey should be conducted to capture this important output. Capital formation of government and
non-profit institutions comes from government financial accounts supplemented by information from the
government agencies involved in various projects and from financial accounts of non-profit organizations. One
reasonable approach to estimate gross capital formation by product is first to set up a matrix of gross capital
formation by industries cross-classified by institutional sectors (similar to table 2.5) and by general types of
goods (dwellings, other buildings and structures, transport equipment, other machinery and equipment). The
next step is to expand the general types of goods into more detailed products using the commodity flow
method. The use of the commodity approach which will be presented in chapter VIII can also help identify
capital goods from domestic production, imports and exports by the nature of the goods, though it does not
always help identify owners.

2. Changes in inventories

7.35. Changes in inventories held by producers, general government and non-profit institutions serving
households are the second main component of gross capital formation. They cover the following:

(a) Inventories (or stocks) of raw materials and stocks of semi-processed or finished products purchased
by producers for use as input into their production process;

(b) Work-in-progress which refers to goods produced during the accounting period but in need of further
processing to be sellable on the market;

(c) Livestock raised for slaughter (but breeding stock, draught animals, dairy cattle and animals raised for
wool and hair clips should be treated as fixed assets);

(d) Inventories of finished products produced as output but unsold.

7.36. The classification of changes in inventories by type of goods is shown in table 7.7.
 179

7.37. To evaluate changes in inventories according to the SNA principles, and the practical experience in
calculating volume changes in inventories which exclude holding gains as required for table 7.7, see appendix
A of chapter V.

7.38. Estimation: Data on the value of inventories come from production censuses and surveys and also
special quarterly and annual surveys of inventories. Countries may also hold inventories of commodities such
as petroleum, grains, etc., that they deem important. Information on these inventories should also be used.
These data may have to be supplemented by data from tax returns, reports of marketing boards and balance
sheets of enterprises. Changes in inventories by products may be estimated by, first, setting up a matrix of
changes in inventories according to the classification in table 7.7 and by industry; the more detailed the latter
the better since it allows the identification of products. From production surveys, it is possible to identify
inventories of work-in-progress and finished products held by industries. The identification of wholesale and
retail trade establishments by general types of products traded allows the broad identification of the types of
products held in inventories. The next step is to expand the general types of goods into more detailed products
using the commodity flow method.

Table 7.7. Classification of changes in inventories

1. Changes in inventories, total

2. Goods producing industries

3. Materials and supplies
4. Work-in-progress
5. Livestock, except breeding stocks, dairy cattle, etc.
6. Finished goods
7. Wholesale and retail trade
8. Other, except government inventories
9. Government inventories

D. Exports

7.39. Exports of goods and services consist of sales, barter, gifts or grants of goods and services from
residents to non-residents (see para. 6.2 in chapter VI for definition of residents and non-residents). Exports
take effect when transfer of ownership from residents to non-residents does.

7.40. In the SNA, exports are all valued f.o.b. i.e. measured without the transnational costs of transport and
insurance services to bring goods from the border of one country to that of another country.

7.41. Goods that are imported and then re-exported or returned unprocessed are omitted from exports and
deducted from imports. Also included in exports are transport, communication and insurance services in
respect of merchandise imports, which are provided by domestic carriers and domestic insurers as exports of
services.
180

7.42. Direct purchases in the domestic market by non-residents are included in exports of goods and
services. Unlike national accounts, direct purchases in the domestic market by non-residents in the use table
have to be broken down by types of goods and services so that it is possible to know the value of every product
consumed by non-residents and residents. This is what the 1995 ESA recommends. However, if exports
cannot be broken down by product or if the details are not needed for economic analysis, as a purchase of a
product by residents or non-residents would have the same effect on the total economy, the SNA's treatment
would be acceptable. For countries with many non-residents (i.e. foreign armed forces, diplomatic and
international extraterritorial bodies, etc.) and a large tourist sector, the breakdown is quite important.

7.43. Exports of goods, in addition to regular goods, include:

(a) Exports of goods for processing;

(b) Exports of goods imported for processing;

(c) Exports of repairs to fixed assets;

(d) Exports of goods under financial leasing;

(e) Exports of non-monetary gold used as valuables and in industrial production;

(f) Transfers in kind to the rest of the world;

(g) Exports of fuel and stores, fish and salvage sold to non-resident ships and aircraft operated primarily
in international waters by resident enterprises;

(h) Direct purchases at home by non-residents.

7.44. Exports of services, in addition to regular services, include:

(a) Transfers in kind to the rest of the world;

(b) Exports of insurance services;

(c) Exports of financial intermediation services indirectly measured.

7.45. Estimation: Data for exports of merchandise (goods) come mainly from foreign trade statistics. Data
for exports of services come mainly from balance of payments. Merchandise trade statistics must be adjusted
to exclude goods in transit and goods that are returned. They must also be adjusted to include transactions that
are not covered by foreign trade statistics which are elaborated above. Other sources of data are foreign
exchange records, customs, government accounts, accounts of bodies involved in international business such
as the post office, telephone and telegraph, insurance companies and special surveys on expenditures by
tourists. Special surveys must be used to estimate and break down by kind of product direct purchases of non-
residents at home.
 183

VIII. COMMODITY FLOW METHOD AND TABLE BALANCING

A. Introduction

8.1. The commodity flow method is a term which is commonly known in national accounting and used by
many developed countries to estimate gross domestic product (GDP). It has been a formal part of the SNA
in the form of the use and supply tables of the input-output framework and is fully described in chapter II. This
chapter further details the practical aspects of completing the use and supply matrices of the input-output
framework by using various sources of information from censuses, surveys, company and industry reports, and
expert knowledge on either the input structure of an industry or the market consumption of a product.

B. Table construction

8.2. In constructing the use and supply tables of the input-output framework, one normally utilizes
information on the input uses as well as input structure of every industry in the tables that are provided by
censuses, government, business reports. When censuses of most economic activities are available, a benchmark
SUT may be compiled which may then serve as the basis for the compilation of annual tables for subsequent
years. Even though censuses are available, supplementary information is always needed and, in most cases,
special studies or surveys may be needed. For example, para. 5.9 in chapter V discussed the need for a special
survey on the commodity breakdown of office supplies that are always reported in aggregate in business
reports. In addition, other information on various uses of a product either from expert knowledge or from the
very nature of the product, information on product market or sales published by producers’ marketing
associations and information on the inter-linking within groups of industries are always needed. This chapter
will focus only on the general aspects of compilation and balancing since the preparation of output,
intermediate consumption, components of final demand and value added have been discussed in detail in
previous chapters. For the compilation of annual SUT on the basis the benchmark SUT, one should refer to
chapter IX.

1. Industry input

8.3. An important part of the work of input-output compilation is to fill in each column of the use table
with inputs used by the industry identified with the column. Surveying inputs used is the principal method
in estimating intermediate consumption because it is based on the theory of production function with constant
input coefficients. Because of this theory it is legitimate to use (i) input information on a few establishments
in an industry to generalize for the whole industry, (ii) engineering information to estimate input flows for the
establishments on which only their outputs are known, or (iii) information on different time vintages to
estimate the input coefficients for a certain industry in the period when the input-output tables are prepared.
The last technique sometimes is necessary when the implementation of production censuses is spread out over
a number of years due to financial and logistical constraints. The information on inputs is at best approximate
and therefore must be cross-checked systematically with information from other sources basically by balancing
the use and supply of each product (e.g. commodity) at the most detailed level. An SUT is recommended by
the SNA for the compilation of annual production accounts in an integrated and consistent manner.
 Table 8.1. Commodity flow method

Supply of goods and services Total supply Uses Total uses

at purchasers' at purchasers' prices
prices
Domestic Imported Trade & Taxes less Intermediate Final Gross capital Exports
output at products transport subsidies on consumption consumption formation**
basic c.i.f. margins products expenditures
prices *

(1) (2) (3) (4) (5)=(1)+.+(4) (6) (7) (8) (9) (10)=(6)+.+(9)

Products
(CPC)
(1)
(2)
...
(k)

* Final consumption expenditures should be split into final consumption expenditures of households, NPISHs and Governments.
** Gross capital formation should be split into three columns: gross fixed capital formation, changes in inventories and acquisition less disposal
of valuables.
 185

2. Commodity uses

8.4. The elaboration of industry uses of products provides a major method to fill in the use table along each
row. This method which is presented in more detail in the use and supply tables is summarized in table 8.1.
The aim of the method is to trace across each row of the use matrix the consumption of every good and service
by industries as intermediate consumption and by various institutional sectors as final demand. In order to do
this, the total supply of a product must first be compiled, by adding to the domestic output of the product, both
market, own final use and other non-market, the import of that product c.i.f., domestic trade and transport
margins and taxes on products in order to obtain the total supply of the product at purchasers' prices (see
column 5). This total supply is then used as the control total for the distribution of the product to various uses
along the same row from column 6 to column 9. The total uses of the commodity must be equal to the total
supply, e.g. column 10 must be equal to column 5.

8.5. The consumption of a product as intermediate inputs is determined by three factors: (i) input
coefficients, (ii) outputs of the industries that consume the product as inputs, (iii) output of the product itself.
Sources and methods for estimating components of final uses have been discussed in chapter VII. Given (i)
and (ii), it is possible to estimate intermediate uses of a product; and given final uses of the product, it is
possible to estimate its total uses. Total supply includes the product output and imports. The total use of a
product in excess of its total supply indicates that either domestic production is higher or imports are. Whatever
adjustment compilers want to introduce must be judged on the basis of the reliability of data sources or of our
own knowledge of the industry.

8.6. Commodity flow analysis along the rows requires the output of any establishment to be broken down
by detailed kind of goods and services, following a certain scheme of classification, for instance the United
Nations Central Product Classification (CPC), version 1.0 which has 1,787 separate classes of products. The
classification of products has to be much more detailed than the classification of establishments. At a
minimum, it is normally necessary to identify the product output of 200 to 300 items. In many countries, for
instance Norway and Denmark, the commodity flow method was applied to 2,000 commodities. Norway
recently reduced the number of commodities from 1,750 to 1,000 when the 1993 SNA and the 1995 ESA were
implemented.

8.7. The commodity flow approach gives a clear advantage in identifying many flows. For many products,
by their nature, it is possible to identify whether they are current or capital goods, and even where they are
used. For example, tractors can only be capital goods that are used in agricultural sectors. Even with current
goods and services, one may be able to identify whether they are used for intermediate or final consumption,
for instance haircut services should go to personal final consumption. The more detailed the way commodities
are classified, the better it is possible to use expert knowledge to supplement surveys in allocating products
to different uses. For instance, one may not be able to allocate coffee in general, but if it is broken down into
coffee beans and roasted coffee, then, one may allocate a part of the coffee beans to exports (which are known
most of the time), and the residual to industries which process coffee. The allocation is obviously much more
difficult if coffee is aggregated with tea, coconuts, etc. to form a sector of cash crops. Similarly, machinery
classified by type is easier to allocate (obviously to fixed capital formation) than if it is aggregated to fabricated
metal products some of which can be used for intermediate consumption.
186

3. Sales ratio

8.8. Another source of information that might be used to fill in the use matrix is the distribution of goods
and services to various sectors by sale ratios (i.e. sale of a product to a sector over the total output of that
product). This method should only be used as a last resort and always in combination with a careful checking
of the likelihood of consequent input coefficients. The reason is that (i) sales may not match uses as uses can
be satisfied by a withdrawal from inventories, (ii) the sale ratio of a product to an industry may drastically
change while the same product input coefficient of the industry remains unchanged. For example, a certain
industry may lose its market to a foreign producer, say by one half and as a consequence its output is halved.
Obviously the sale ratios of the output of this industry to other industries would be cut dramatically. But if
there is no change in the production technology of the industry that consumes the product, the input coefficient
of the industry should remain unchanged.

4. Flow and coefficient adjustment

8.9. When compilers decide to adjust a certain flow, obviously the decision will affect other flows that are
closely associated with it either as suppliers of input or as market consumers. To make the adjustment more
systematic, i.e. to take into account inter-industrial relationships, compilers should set up the most clearly
separable blocks of sectors and assign them to separate groups of input-output statisticians responsible for
balancing. A block should include industries that are closely related in terms of input requirements. For
example, cement would be definitely required in construction, therefore cement supplies can be used to cross-
check estimates of construction, particularly for the household sector in developing countries. The work can
be divided into the following blocks of sectors:

(a) Agriculture, fishing, manufacture of food products and beverages, hotels and restaurants;

(b) Metal and machinery industries, construction, auto repair shops;

(c) Forestry and industries producing wood and wood products, quarrying and non-metallic mineral
products, construction;

(d) Manufacture of textile and textile products, footwear;

(e) Chemical industries, including plastic products;

(f) Energy sectors;

(g) Service industries and producers of government services.

Within each block, there might be technical and behavioral changes that are more discernible by the same
experts. It is always important to adjust flows while taking into account the resulting input coefficients.

8.10. Given input-output coefficients of previous periods, one should also inquire why there are changes
in the coefficients, particularly big changes and whether they are justified. Reasons for coefficient changes
are numerous, but the most important ones are:

(a) Changes in classification where a group of establishments has been recently added to the industry or
a subgroup has been removed and aggregated in another industry;
 187

(b) Changes in the composition of the industry in terms of subgroup outputs;

(c) Changes in relative input prices such as the drastic changes in the price of petroleum and
consequently of energy as a whole in 1973 and 1979;

(d) Changes in technology: for example in Indonesia from 1980 to 1985, there were important changes
in the poultry industry as a result of commercialization of the sector with large increases in input
coefficients of feed, medicines, wages and salaries, while there was a reduction of operating surplus.

8.11. Finally, an important reason for large changes is error. By analysing changes, it is possible to
understand them in each industry and reduce error to a minimum. Thus it is always important to list in parallel
new coefficients and coefficients of past periods to help the analysis. To do this, it is important to use the same
schemes of classification for both industry and product.

C. Construction of margins

8.12. The construction of the import matrix together with the construction of the trade margin matrix and
that of transport margins are the weakest parts in terms of reliability in the construction of the use table. It is
clear that for some products users may not know whether they are domestic or imported. It is even much more
difficult to know the actual transport margins and trade margins of many products by specific users. These
margins are estimated more or less as averages given total margins of specific products and margins of some
individual flows in the use matrix. Margins are assigned to product flows on the basis of the proportionality
principle, i.e. for a given product, a higher flow is assigned proportionally higher margins and vice versa.
Detailed product classification should help in allocating margins.

1. Construction of import margins

8.13. Though imports and the construction of the import matrix are discussed in detail separately in chapter
VI, it is important to emphasize again that the detailed classification of products greatly simplifies the task of
distributing an imported product across each row in proportion to the use of the product by each industry and
by final demand. With very detailed classification, it is possible more fully to use expert knowledge on the
product, for example, to know where the imported product is not utilized as input, and, given actual figures
of imports for a number of industries or final demand, compilers only have to distribute the residual
proportionally to the sectors that are believed to use the imported product. The distribution of imports on the
basis of the proportionality principle supplemented by expert knowledge will certainly lead to a more accurate
compilation of the import matrix.

2. Construction of the trade margin matrix

8.14. Detailed product classification also helps in allocating trade margins. The derivation of trade margins
which include both wholesaling and retailing margins either as one sector or as two separate sectors also
follows the same principle as the distribution of imported products. The proportionality principle is used in
conjunction with knowledge of wholesale and retail trade margins. There are some rules of thumb that need
to be kept in mind, namely:

(a) Most retail margins will be allocated to households' final consumption expenditures, with the
exception of the procurement of office supplies, gasoline and probably some other goods that are also
obtained from retailers;
188

(b) Industry and government purchases are mostly from wholesalers;

(c) Exports do not involve retail trade, with the exception of some minor transactions.

8.15. For developing countries, it may be better as far as possible, to divide each product into six separate
groups:

(a) Products which are produced and marketed by major corporations in an organized and modern
fashion;

(b) Products which are produced by small manufacturers and craftsmen, and mainly used in local markets;

(c) Products which are produced by small producers like rice or wheat farmers but are procured and
marketed in an organized manner through wholesalers and retailers;

(d) Products which are produced for own consumption;

(e) Products which are produced for own capital formation;

(f) Products which are imported.

Obviously, each type of product will require more or less in the way of trade margins. Cases (d) and (e) may
require no trade margins, cases (a) and (c) may require higher trade margins than case (b). Finally, case (f)
may require the highest trade margin in developing countries because of import monopolies.

3. Construction of the transport margin matrix

8.16. Transport margins which relate to the movement of goods should be clearly distinguished from
transport services that are consumed directly as a means of personal movement. The construction of a transport
margin matrix is similar to the construction of a trade matrix and normally they are prepared together. But it
is important to note that in the 1993 SNA only transport costs which are billed separately to purchasers by
suppliers or paid to third parties by wholesalers and retailers are treated as transport margins, not as part of
trade margins (SNA, para. 6.112(c) or paras. 5.64 and 5.65 of chapter V) and will appear in the consumption
of transport services by industries. In this case, these known values assigned to a known destination should
be subtracted from the total output of transport services before the residual is distributed proportionally. Other
known transport costs that are paid by purchasers of goods directly to transport establishments because these
purchasers hire movers of the goods themselves can be obtained by surveys on input used. By SNA rule, own-
account transport of goods by purchasers will not appear as output of transport services.

D. Final balancing of the tables

8.17. The main focus in SUT balancing is on balancing the use and supply tables together. The balancing
may begin after all components shown in table 8.1 have been prepared. One should use all statistical and
engineering information to fill in the intermediate demand matrix of the use table. One should use information
from household expenditure surveys, surveys on retail sales capital formation, inventories, government
expenditures and foreign trade to fill in the final demand part of the use matrix. Only small residuals should
be mechanically distributed. Discrepancy between supply and use by commodity may be the result of under-
reporting, over-allocation, mis-classification or improper valuation and timing of production, intermediate
 189

consumption, final uses, gross capital formation, exports and imports. All of these problems must be
thoroughly investigated, which requires the re-examination of basic records and an intimate knowledge of the
industries and commodities concerned.

8.18. One of the tools for distributing the residuals in existing flows and to keep industry output (column
sum) and product output (row sum) fixed at predetermined values is the RAS method. The method will be
discussed in chapter IX but here it is sufficient to say that the basic objective of the method is to distribute the
differences between known product outputs and the sums of the rows of the present intermediate matrix
proportionally along the rows and then to distribute the differences between known industry outputs and the
sums of the columns proportionally along the columns. This process is carried out continuously until all
discrepancies are reduced to an acceptable minimum. This method is equivalent to minimizing differences
between the flows that have been obtained by using available information and the final calculated flows of the
use table given row and column sums. It is also possible to fix certain flows to known figures while using RAS
to distribute the discrepancies only in the rest of the flows. The latter method is called the modified RAS
method. Thus, in balancing the use table, it is important to divide the flows into two parts: those that are based
on sure information, and those that can be modified. If the use table is a benchmark where most of the
information comes from censuses, then one must keep the flows that one is sure of constant and distribute the
residual proportionally to the rest of the flows both horizontally and vertically. The RAS method is used mostly
to balance the intermediate demand matrix, given known industry output, product output, value added and final
demand. However, the RAS method can also be used for the entire use matrix, given additional information
on total values for each category of final demand, like total personal consumption expenditures, total gross
capital formation, total exports, total imports, etc.

8.19. Most of the methods used for balancing input-output tables are ad hoc, but they have to be based on
statistical information, and expert knowledge. Only as a last resort, when discrepancies are reduced to the
minimum possible, should an arbitrary method like the proportionality principle be used to allocate the
residuals. If one does not want to distribute flows arbitrarily, one may create a row and a corresponding
column to store discrepancies, though they may look odd in an input-output tables. However, the latter method
may still be preferable to the allocation of discrepancies to changes in inventories as some countries have
attempted.

8.20. In some countries, there are no clear-cut benchmark tables, and additional information is introduced
every year to arrive at new annual tables. Some use the modified RAS method to update the old table into the
new table by incorporating new information. Because of the nature of the method, it is possible to force the
new table to be close to the old one when even the old one is not based on the fullest possible information.
Thus, it is still preferable to compile a benchmark table using SUT on the basis of the fullest possible
information provided by censuses to check accuracy of the tables derived by other methods. Annual SUT
should also be compiled given the benchmark table and annual information because that is the most reliable
and integrated method to check the consistency of the data and components of national accounts.
 191

IX. UPDATING INPUT-OUTPUT TABLES: RAS METHODS

9.1. This chapter discusses two techniques--i.e. the simple RAS and modified RAS -- that are widely used
to update input-output tables on the basis of the benchmark tables compiled with detailed census and survey
data. Since the techniques are based on the stability over time of input-output coefficients, the chapter starts
with that topic by reviewing studies that confirm this stability (see section A below). The widespread use of
the techniques is explained because in most countries it takes two to five years after the end of the year before
the input-output tables for that year can be prepared and published. If the tables are to be used in current
analysis, it is important to check whether and to what extent the coefficients have changed since the year to
which they relate. By incorporating the most up-to-date information, the two RAS techniques are used to carry
out this check and update those coefficients that are clearly changed in view of the newly available
information. The RAS techniques are used not only to update entire I/O tables, but, as will be shown below,
to balance supply and use tables at the last stage, which was discussed in chapter VIII. The Lagrangian
multiplier approach has also been tried but it does not seem to provide better results. This approach will be
briefly discussed in paragraph 9.39 at the end of the chapter. It is important to emphasize from the beginning
that no purely mechanical method would be able to replace updating SUT on the basis of comprehensive data.
The best approach is always to use all available information to compile annual SUT and then use modified
RAS at the last stage of balancing. In this context, the current Canadian method is worth mentioning.1 It uses
the following steps:

- Real value added by industries of the current period, to calculate real industry output;
- Real industry output and the intermediate coefficient matrix of the base year, to estimate
intermediate consumption flows;
- Real industry output and the market share matrix, to estimate product output and the supply
table;
- Use price indexes, to transform the real supply table into the supply table and industry output
at current prices;
- Use price indexes, to transform the real intermediate consumption matrix into the matrix at
current prices;
- Compile components of final uses at current prices from available statistics;
- Estimate trade and transport margins by base-year margins, and reflate them to current prices;
- Estimate taxes on products using the benchmark method and then confront them with control
totals;.
- Balance SUT with supplementary data and expert knowledge.

A. Stability of I/O coefficients

9.2. Changes in the input coefficients in current values may be caused by three major reasons:

(a) Changes in technology;

1
Yusuf Siddiqi and Mehrzad Salem, “A Synthetic Approach to Projecting Input-Output Tables” in Economic Systems
Research, Volume 7, No. 4, 1995.
192

(b) Changes in relative prices;

(c) Imperfect data.

9.3. The most important changes, in technology, are beyond manipulation by statisticians. These resulted
for instance from the sudden and drastic changes in relative prices, due to increases in the price of petroleum
in 1973 and 1979. In current prices, the input coefficients showing the use of petroleum in various industries
increased significantly immediately after petroleum prices increased, while the increases abated and even
declined after a while when conservation (technical change) took effect. Another cause of changes in input-
output coefficients is inflation also causing changes in relative prices; the effect of these changes can be
eliminated by using input-output tables in constant prices (see chapter XI for a discussion on input-output
tables in constant prices). The changes due to imperfect data might be reduced --but can never be entirely
eliminated-- by the use of statistical classifications developed on the basis of the homogeneity principle. Using
such classifications would avoid adding together products with different technical coefficients, so that changes
in product-mix within a given classification are insignificant. If these conditions are met, changes in input-
output coefficients at constant prices can be treated as caused by technical changes.

9.4. Many studies have shown that input coefficients are not stable over time but that changes take place
fairly gradually. Vaccara2 of the United States showed that during the period 1947-1958, the average annual
rate of change in intermediate output requirements (ignoring signs) for a fixed bill of final demand was 2.3 per
cent. Over the 1958-1961 period, the change averaged 1.7% per year. Tilanus3 of the Netherlands showed that
the median coefficients for all observed cells indicated a tendency for coefficients to change by about 2 per
cent a year and he found that over half the coefficients of the time trends were statistically significant. The
fact that changes over time are fairly gradual makes it possible to extend the useful life of any input-output
table. The work of Sevaldson4 in Norway showed that 60% of coefficients had either a moderate or clear
(positive or negative) trend, and that trends which changed the coefficients as much as 1% a year were rare.
He also noted a considerable dispersion about the mean value of each coefficient, but when the values were
compared with a time trend the dispersion was reduced but still quite large. It was drastically reduced when
tables were aggregated from 64 industries to 14.

9.5. However, because of a still large dispersion about a linear trend, the method of projecting coefficients
by linear extrapolation does not provide good estimates of future input-output tables. Tilanus showed in the
study mentioned above that such extrapolations generated worse results than simply using the most recent
coefficients. When T.S. Barker5 performed the same type of exercise, projecting the 1963 intermediate
demand for Britain using the 1954 table as the benchmark, results showed that trend projection was worse than
using the most recent table of 1960. In all cases, the tables updated by simple RAS method, which will be
described later, performed better.

2
Beatrice N. Vaccara, “Changes over Time in Input-Output Coefficient for the United States” in Readings in Input-
Output Analysis, edited by Ira Sohn, New York, Oxford University Press, 1986.

3
C. [Link], Input-output Experiments : The Netherlands 1948-1961, Rotterdam University Press, 1966.

4
P. Sevaldson, “The Stability of Input-Output Coefficients” in Applications of Input-Output Analysis edited by A.P.
Carter and A. Brody, North Holland, 1969, also in Readings in Input-Output Analysis, edited by Ira Sohn, op. cit.

5
T.S. Barker, “Some Experiments in Projecting Intermediate Demand” in Estimating and Projecting Input-Output
Coefficients, edited by R.I.G. Allen and W.F. Gossling, England, Input-Output Publishing Company, 1975.
 193

9.6. Another method of projecting future input-output tables is the "best practice" firm method proposed
by William Miernyk6. Miernyk suggested the use of the coefficients of the best practice firms in each industry
as the future technology. He argued that these technologies of "best practice" firms in the present should
gradually become the norm of the future, five to six years ahead. Many objected to the arbitrariness in selecting
the length of time needed for the present best technology to mature. However, this objection can be met by
figuring out a predicted time schedule for a given technology to be adopted, industry by industry. Until now
no study has been carried out to test the accuracy of this proposal.

B. The mathematics of the RAS method

9.7. As changes in the coefficients are shown to be gradual, it might be possible to use a current input-
output table to project to the near future, with its calculated values adjusted for technological changes.
However, the tables used should be the most up-to-date ones. This raises the question whether there are
methods that are efficient in updating tables that are normally completed five years after the year they refer to.

9.8. The problem of updating, i.e. adjusting an intermediate consumption matrix to fit new column and row
sum constraints was first regarded as a statistical problem and tackled by statisticians working as early as
19407. New column sums refer to the differences between gross outputs and value added, and the new row
sums refer to the differences between gross outputs and net final demand. The new values are normally
provided by annual surveys. The method used was the so-called RAS method, a well-known technique to
update input-output tables, which was adapted from the work of Deming and Stephen to input-output tables
by Richard Stone in 19618 and remains the simplest and most widely used. Even though variations of this
technique have been developed, practically no other methods seem to perform better.

9.9. The RAS approach assumes that there exists an input-output table estimated from full data for a past
year and that row and column sums for the input-output table of the present year are available. The basis of
the RAS method suggested by Stone consists in finding a set of multipliers to adjust the rows of the existing
matrix and a set of multipliers to adjust the columns so that the cells in the adjusted matrix will sum up to the
required row and column totals relating to the later current year. In mathematical terms, if A0 is the coefficient
matrix corresponding to the benchmark input-output matrix F0 table and A1 is the updated matrix of input-
output coefficients corresponding to the estimated input-output matrix F1, then
(9.1) A1 ' r̂A0ŝ

6
William L. Miernyk, The Elements of Input-Output Analysis. New York, Random House, 1965. See also “Long-
Range Forecasting with a Regional Input-Output Model” of the same author in Readings in Input-Output Analysis, edited
by Ira Sohn, op. cit.

7
W. E. Deming and F.F. Stephen, “On a Least-Squares Adjustment of a Sampled Frequency Table When the Expected
Marginal Totals are Known” in Annals of Mathematical Statistics, 11, 1940.

8
Richard Stone, Input-Output and National Accounts, Paris, Organization for European Economic Cooperation, 1961.
Also Richard Stone and A. Brown, A Computable Model for Economic Growth, Vol. 1, London, Chapman and Hall,
1962. The full mathematical development of RAS should be read in Michael Bacharach, Bi-proportional Matrices and
Input-Output Change, Cambridge, U.K., Cambridge University Press, 1969. Various variations of the RAS technique
as optimization methods can be found in R.I.G. Allen and W.F. Gossling, Estimating and Projecting Input-Output
Coefficients, London, Input-Output Publishing Company, 1975.
194

where r and s are row and column multipliers and ^ stands for a diagonal matrix.

9.10. An elaboration of the above formula for two industries (1 and 2) illustrates the role of the multipliers
r and s.

r1 0 a11 a12 s1 0
A1 ' x x
0 r2 a21 a22 0 s2

r1a11s1 r1a12s2
A1 '
r2.a12s1 r2a22s2

The example shows that the row multipliers r1 or r2 are merely scaling factors to adjust elements of each
respective row equally so that the sum of the new row is equal to the known row total. In the above formula,
elements of the first row are multiplied by r1 and elements of the second row are multiplied by r2. Similarly,
a given column multiplier also acts as a scaling factor applied equally to every element of that column so that
the sum of the new column is equal to the known column total. Thus, all elements of the first column are
multiplied by s1 and all elements of the second column are multiplied by s2.

9.11. An economic interpretation of the RAS multipliers r and s is that input-output coefficients aij of the
input-output coefficient matrix A0 change over time due to two effects: (a) the effect of substitution, expressed
by r, measuring the extent to which product i has been replaced by, or used as a substitute for, other products
in industrial production; and (b) the effect of fabrication, expressed by s, measuring the extent to which
industry j has come to absorb a greater or smaller ratio of intermediate to total inputs in its production. The
method further assumes that each effect works uniformly, e.g., that product i is increasing or decreasing as
an input into all industries at the same rate and that any change in the ratio of intermediate to total inputs into
an industry has the same effect on all products used as inputs. The substitution multipliers which operate along
the rows are denoted as vector r and the fabrication multipliers operating on the columns as vector s. Each cell
in the benchmark matrix, A0, will be subject to these two effects. The assumption of uniform effects on inputs
across industries is, however, economically unrealistic.

9.12. In order to find r and s, one must introduce the input-output flow matrix to be estimated, together with
the known row and column totals, which will be called u* and v*. Let F1 be the input-output flow matrix of
the current year which is unknown, X19 the known output vector of the current year, and 1A is the new

9
If output by industry is not the same as product output because of the existence of secondary products, X stands for
the industry output vector.
 195

coefficient matrix to be estimated corresponding to F1. If X1 is converted into a diagonal matrix10, indicated
by the sign ^ over it , the following identity holds:

F1 ' A1X̂1
(9.2)
' (r̂ A0 ŝ)X̂1

Define u* to be the row total11 of intermediate inputs of matrix F1. Then

u ( ' F1i
' (r̂ A0 ŝ)X̂1 i

' (r̂ A0X̂1) ŝ i

(9.3)
' r̂ (A0X̂1) s

In the above definition, i is a column vector in which each element is equal to 1. Vector i is used to sum the
flow matrix across the rows to obtain the row sums of the flow matrix F1. In the second equation F1 is
replaced by the expressions in equations 9.1 and 9.2. Then, in the next equation the order of matrices ×X1 is
switched to X1×, which does not change the value of u*, because the × and X1 matrices are diagonal. And,
finally in the last equation, × i is replaced by the column vector s. If the column totals of F1 are defined as

v* = F1 ' i
and v'* = i'F1

then 9.3 can be rewritten as:

(9.4) v )( ' r ) (A0X̂1) ŝ

where r' is a row vector.

10
A diagonal matrix is a matrix with non-negative values on the diagonal and zeros everywhere else:

x1 0 . 0
0 x2 . 0
X̂ '
. . . 0
0 0 . xn

11
The subscripts 1, 2, ... in u and v (see table 9.1) refer to the year to which the input-output table is intended to be
updated and do not refer to the elements of these matrices.
196

9.13. The two equations 9.3 and 9.4 consist of the two unknown r and s vectors and known information on
the benchmark coefficient matrix Ao, the new row and column constraints u* and v*, and the new output levels
X1. Thus, if these equations are solved simultaneously the values of the r and s vectors will be found and then
the updated matrix A1 can be derived on the basis of equation 9.1. The mathematical properties of the RAS
method have been explored by Bacharach12 who shows that the iterative method will converge and produce
a unique solution, which does not depend on whether rows or columns are adjusted first. Further, experience
has shown that the method is not too demanding on computing time.

C. The iterative solution of the simple RAS method

9.14. The estimation process of obtaining F1 from F0 or alternatively from A1 from A0, in effect amounts
to nothing more than a proportional adjustment of the benchmark matrix elements successively along the rows
and columns until convergence is reached between the row and column sums of the new matrix found through
iteration, and the known row and column sums. Due to the nature of the RAS method, it is also called the "bi-
proportional adjustment" method.

9.15. How the RAS method functions will now be illustrated and later its properties will be outlined and
some refinements discussed. The simplified accounting matrix for the benchmark year 0 is given in table
9.1(a) and all known information for the current year 1, is presented in table 9.1(b). From (a) one can calculate
the coefficient matrix A0 as presented in (c). If initially it is assumed that there is no technical change, the
matrix for year 1 would be found by applying these coefficients to the new output levels, resulting in A0 X1
shown in (d). The row and column sums of this matrix, u1 and v1 , are then calculated and compared with the
known row and column sums for year 1, u* and v*. Then in table 9.1(e) - (g) the elements of X1 calculated in
(d) are adjusted to fit the new constraints by solving for the r and s multipliers. This effectively removes the
assumption of unchanged technical coefficients.

9.16. The first step in the iteration is to calculate the first set of row multipliers r1 at (d) and then to multiply
the values in each of the rows to the multiplier for that row so that the row totals meet the constraints as
reflected in (e). The column totals of the matrix at (e) are then compared with v*, the know column sums, and
a set of column multipliers are described which bring the column sums of the matrix to equal v* in (f). As a
result of this step in the iteration, the row totals of this matrix are no longer equal to u*. A second set of
multipliers, r2, must now be applied to the rows so that they sum again to u*. The column sums will then
require further adjustment and this process will continue until no further adjustment is necessary. The
resulting matrix after many iterations is shown at (g), together with the r and s multipliers. The latter are the
product of the multipliers used at each of the successive rounds of adjustment. Thus, the accumulated row
multiplier for the first row, 0.884, is r1 x r2 x r3 x ... x rn, which in the case of the example presented is equal
to 0.873 x 1.010 x 1.002 x 1.001 = 0.884.

12
M. O. L. Bacharach, Bi-proportional Matrices and Input-Output Change, Cambridge University Press, 1969.
 197

Table 9.1. Illustration of the RAS method

(a) I/O data for year 0

Product13 Final Total

demand output
A B C Total

A 50 100 0 150 50 200

Product B 30 50 20 100 200 300

C 20 50 30 100 100 200
Total 100 200 50 350 350 700

Value added 100 100 150 350

Total output 200 300 200 700

(b) Available I/O data for year 1

Product Total Final Total

u* demand output
A B C

A 160 40 200
Product B 150 250 400
C 120 180 300

Total 100 250 80 430 470 900

Value added 100 150 220 470

Total output 200 400 300 900

13
For simplicity, no distinction is made between products and industries in this example.
198

(c) Input-output coefficients for year 0 (A0)

A B C
A 0.250 0.333 0
B 0.150 0.167 0.100
C 0.100 0.167 0.150

(d) Year 0 coefficients applied to year 1 outputs (A0X1) and calculation of the first set of row
multipliers (r1)

A B C u1 u* r1=u*/u1

A 50 133.3 0 183.3 160 0.873

B 30 66.7 30 126.7 150 1.184
C 20 66.7 45 131.7 120 0.911

v1 100 266.7 75
v* 100 250 80

(e) Adjustment of matrix along rows based on the first set of row multipliers (r1) and calculation of
the first set of column multipliers (s1)

A B C u1=u*
A 43.6 116.4 0 160
B 35.5 79.0 35.5 150
C 18.2 60.8 40.1 120
v1 97.3 256.2 76.5
v* 100 250 80
s1=v*/v1 1.027 0.976 1.046
 199

(f) Adjustment of matrix down columns based on the first set of column multipliers (s1), and
calculation of the second set of row multipliers (r2)

A B C u2 u* r2=u*/u2
A 44.8 113.6 0 158.4 160 1.010
B 36.5 77.1 37.1 150.7 150 0.996
C 18.7 59.3 42.9 120.9 120 0.992
v2=v* 100 250 80

(g) Further adjustment of rows and columns, until both u =u* and v =v* in the final updated matrix

A B C u=u*/u

A 45.3 114.7 0 0.884

B 36.2 76.6 37.2 1.177
C 18.5 58.7 42.8 0.902

v=v* 100 250 80

s=v*/v 1.025 0.974 1.054

(h) Short-cut method

A B C u=u*/u

A 45.3 114.7 0 1.119

B 36.2 76.6 37.2 1.488
C 18.5 58.7 42.8 1.140

v=v* 100 250 80

s=v*/v 0.811 1.025 1.201

9.17. Each cell in the benchmark matrix (d) can be multiplied by its summary row and column multipliers
to yield the value of cells in the final matrix at (g). Thus the input of A into B which is 133.3 in the benchmark
matrix must be multiplied by its row multiplier (0.884) and its column multiplier (0.974) to give 114.7, which
is the value in the updated matrix at (g).

9.18. It can be seen in this example that product B with a row multiplier of 1.177 is replacing products A
and C as an input into intermediate demands. When reviewing the column multipliers it becomes apparent
that product B is subject to a downward fabrication effect, i.e., it is using fewer product inputs and thus more
primary inputs in its production process, perhaps because the production process has become more
200

complicated. On the other hand products A and C are using more intermediate products as inputs and fewer
primary inputs per unit of output, perhaps because their inputs are being purchased in a form which requires
less processing before being transformed into the finished product. This would be the case if a firm in the
motor industry began to purchase the engines for its vehicles from another supplier, whereas previously it had
purchased the various parts and made the engine itself.

9.19. If a particular cell of the benchmark matrix was zero, the RAS method would not change this. It would
remain zero, since the proportional scaling along both row and column will not affect a zero value. Also no
negative entries will appear in any cells in the final matrix as the allocation by the RAS method is based on
the base matrix that is non-negative.

9.20. An alternative short-cut to the method shown in tables 9.1(a)-(g) is to start with the benchmark flow
matrix and simply adjust this matrix to the new controls. This omits stages (c) and (d). The resulting matrix
in this case, shown in table 9.1(h), is identical to that obtained in (g) but the row and column multipliers are
different. Indeed, the same meaning cannot be attached to the multipliers which are covering changes in output
levels as well as the two effects operating on the coefficients. This so-called short-cut method is defined by
the following mathematical presentations:

(9.5) F1 ' r̂ F0 ŝ

that is
(9.6) A1X̂1 ' r̂ A0X̂1 ŝ

(9.7) A1X̂1 ' r̂ [A0X̂0(X̂1)&1 ] ŝ

9.21. Equation 9.7 is similar to the RAS equation 9.1, however, with output levels X0 and X1 additionally
included. If one is interested in the values of r and s as measures of substitution and fabrication effects, this
method is not suitable; they cannot be derived. If there is no interest in these effects and the RAS methods are
seen only as mechanical techniques of adjustment, the use of this short-cut method is recommended.

D. The modified RAS method

9.22. A modified version of the RAS method incorporates some firmly based information about some of
the cells in the matrix being estimated. This method is illustrated in table 9.2 below where it is assumed that
information is available on the input of B into A and hence there is no need for this cell to be involved in the
adjustment process. The value of this cell is subtracted from the required row and column constraints, u* and
v*, and this cell is set at zero in the benchmark matrix. This forms the starting point for the RAS adjustment
in (c). The RAS adjustment procedure is then carried out in the normal way as was explained above, and when
a solution has been reached at (d) the zero for the input of B into A is replaced by the known value 40.

Table 9.2. Illustration of modified RAS

 201

(a) I/O data for year 0

Product Total Final Total

demand output
A B C

A 50 100 0 150 50 200

Product B 30 50 20 100 200 300

C 20 50 30 100 100 200

Total 100 200 50 350 350 700

Value added 100 100 150 350

Total output 200 300 200 700

(b) Available I/O data for year 1

Product Total Final Total

u* demand output
A B C

A 160 40 200

Product B 40 150 250 400

C 120 180 300

Total 100 200 80 430 470 900

Value added 100 150 220 470

Total output 200 400 300 900

It is assumed that the input of B into A is 40, otherwise the table is the same as table 9.1(b).

(c) Year 0 coefficients applied to year 1 outputs

A B C u1 u* r1=u*/u1

A 50 133.3 0 183.3 160 0.873

B 66.7 30 96.7 110 1.138
C 20 66.7 45 131.7 120 0.911
v1 70 266.7 75
*
v 60 250 80

(d) Final updated matrix

202

The RAS adjustment continues as in table 9.1(d)-(g) and when a solution is obtained the exogenous cell
value of 40 is entered in the result.

A B C
A 42.7 114.3 0
B 40 73.7 36.3
C 17.3 59.0 43.7

E. Evaluation of the simple and modified RAS methods, based on a literature survey

9.23. It is important to know how accurate the RAS method of updating input-output tables can be and
various tests have been carried out to compare tables thus updated with tables subsequently estimated for that
year from full statistical data.

9.24. Paelinck and Waelbroeck14 compared an input-output table for 1959 estimated by RAS from the 1953
table with the actual 1959 table, and Schneider15 performed similar tests with United States tables over the
period 1947-1958. These tests showed that a matrix estimated by RAS from a past year gave a somewhat better
estimate of the later year than did the matrix of the past year, i.e., r A0 s was closer to A1 than A0. The number
of errors in individual cells greater than 1 per cent was reduced from 17 to 9 in the Belgian case and from 121
to 103 in the United States case, with a larger table. In a more recent study by R. G. Lynch16 using the 1963
and 1968 United Kingdom input-output tables, the improvement of the updated tables versus the tables of the
past years was also confirmed and it was shown that the root mean square error decreases with the level of
aggregation. However, in all these studies, the improvement in terms of mean square error is small unless the
level of aggregation is increased significantly, which reduces the usefulness of the input-output table for
detailed studies of industries.

9.25. Another recent study by P. Gretton and P. Cotterell of Australia17 showed that errors in the estimates
of large coefficients tend to be smaller than in those of smaller coefficients and that coefficients aggregated
from those estimated by RAS method using more desegregated tables have smaller errors than coefficients
estimated directly from aggregated tables.

9.26. The simple RAS method will normally fail to produce an accurate estimate of A, since the assumption
that the row and column effects work uniformly along rows and columns is not justified. This will be the case
if product groups are not homogeneous but consist of, say, one product which is replacing a product in another

14
J. Paelinck. and J. Waelbroeck, “Etude Empirique sur l'évolution de Coefficients Input-Output” in Economie
Appliquée, 1965.

15
J. Schneider, An Evaluation of Two Alternative Methods for Updating Input-Output Tables (B.A. Thesis, Harvard,
1965).

16
R.G. Lynch, “An Assessment of the RAS Method for Updating Input-Output Tables” in Readings in Input-Output
Analysis, edited by Ira Sohn, New York, Oxford University Press, 1986.

17
P. Gretton and P. Cotterell, The RAS Method for Compiling Input-Output Tables, Australian Bureau of Statistics,
Canberra, 1979.
 203

row and one product which is not replacing another or is being replaced by some other product. The net effect
as recorded by RAS may be, say, an upward substitution effect on the row. However, if the two products are
used in different proportions by consuming industries the substitution effect will not be uniform. A similar
problem arising from lack of homogeneity was observed in the Belgian tests where it was found that the cells
relating to fuel inputs were badly estimated by the simple RAS method. This was because coal, for instance,
is used as a raw material in energy-producing industries but as a source of power in all other industries but in
the latter case coal was being replaced by other fuels at a much faster speed than in the former case. This
markedly different substitution rate seriously affected the results of the RAS updating estimate.

9.27. The incorporation of exogenous data into the simple RAS method will tend to improve the accuracy
with which other cells can be estimated in the same rows and columns as the known cells. This does not mean
that theoretically the improvement is always achieved as shown in an example given by R.E. Miller and P.D.
Blair. The example is a small table of 3 by 3 in which one coefficient is assumed known. The overall accuracy
of updated coefficients using the modified RAS method actually declines when compared with the normal
RAS method.18 This observation is also confirmed by a test by Gretton and Cotterel of Australia mentioned
above when the inclusion of the coefficients of one more industry made overall errors worse. In general, most
tests on real data show an improvement in the use of the modified method over normal RAS. The improvement
is more significant if more exogenous data are incorporated.

9.28. In the Belgian tests, six of the cells which the simple RAS method had estimated were treated as if they
were known exogenously and the modified RAS method was used. The number of cells with errors of more
than 0.5% was reduced from 20 in the simple RAS method to 8 in the modified method. This conclusion is
also reported in Lynch's study mentioned above.

F. Conclusions for application of the RAS methods in country practices

9.29. From the above it may be concluded that when input-output coefficients only change gradually, the
RAS method and in particular the modified RAS which uses additional information that is available, provides
satisfactory results. Thus, in order to make optimum use of the limited statistical resources available in many
countries, it is acceptable to compile benchmark input-output tables based on a comprehensive database every
5-10 years and to compile annual tables integrated with the national accounts using the modified RAS updating
techniques described here. While the annual tables based on a modified RAS method require some data on
inputs, these requirements are nothing compared to the statistical requirements of benchmark input-output
tables. Some data on inputs may be available on an annual basis and should therefore be used. On the other
hand, the annual updating exercise of the input-output table should not involve the additional collection of
statistical information on a major scale.

9.30. It is difficult to generalize as to which cells, rows or columns are well recorded by normally available
statistics. In many countries fuel industries are well documented and information on the cells relating to fuel
inputs should be available. The same may be true of the major inputs (fertilizers and feed) into agriculture.
In some cases, it may be worthwhile to investigate a particular cell or cells to be incorporated into an updating
exercise. Alternatively, it may be possible to combine indices of price and volume movements with the
benchmark year value of a cell and obtain a measure of the value in the current year. Where there is no

18
Ronald E. Miller and Peter D. Blair, Input-Output Analysis: Foundations and Extensions. New York, Prentice-
Hall, 1985, pp.292-293.
204

domestic production of a particular product it may be possible to obtain an estimate of the input of that product
from international trade statistics, e.g. the input of crude oil into the petroleum refining industry.

9.31. In this context it should be noted that exogenous data on certain cells may lead to a greater
improvement in the accuracy of the updating methods than data on other cells. Tests on United Kingdom input-
output tables have shown that substantially improved results can be obtained when relatively small amounts
of additional information about preselected major coefficients were incorporated into the updating exercise19.
It seems useful, therefore, to perform some tests of updating exercises to find the most important cells in
particular national tables. When these have been identified, efforts can be made to find exogenous data on
those coefficients which will improve the accuracy of the updating exercise.

9.32. In developing countries, the modified RAS method could be used to cut the cost of frequent full-
fledged surveying. The key coefficients are often determined by large establishments over which the
Government has significant control. In addition, the number of such establishments is small, which allows less
costly supplementary surveys for the purpose of identifying changes in the key coefficients or key industries.20
In general, when using the modified RAS, it is important to divide the flows into two parts: those that are based
on sure information, and those that can be modified. If the use table is compiled for a benchmark year, where
most of the information comes from censuses, then only the flows that are based on census and similar data
sources are kept constant with the residual proportionally distributed in the rest of the flows both horizontally
and vertically.

9.33. In many countries it should be possible to extract from regularly available statistics some information
on inputs or trends in input coefficients (if clear trends can be established) since the year of the previous full
table. It should be possible to produce an input-output table by this method quite soon after the full national
accounts are produced. Australia is the country that has adopted the modified RAS method to compile its
annual input-output tables with significant annual supplementary basic data in order to speed up the availability
of more current input-output tables. According to Australian reports more than 28% in 1974-1975, 18% in
1977-1978, and 20% in 1978-1979 of all intermediate inputs (in terms of value) were estimated from basic
data.21

9.34. Such an updated table will obviously not be so accurate as a table estimated on full data (although even
these will suffer inevitably from errors of measurement, particularly of smaller inputs), but should be
sufficiently accurate to remove the need for a full table each year, particularly if some exogenous data on
coefficient can be incorporated. This is certainly likely to be true if the input-output table is used in any
analysis or model-building. If, on the other hand, the input-output table is used to check the consistency of
data in a statistical framework, there may be some need for an updated table. For instance, if a very high
multiplier is found for a particular product which does not fit in with knowledge of changes in industrial
techniques, this would suggest either that final demand has been underestimated or total output overestimated,

19
R.I.G. Allen, “The Effects on Intermediate Output Predictions of Incorporating Coefficients with Known Values
into the RAS Updating Method” in Aspects of Input-Output Analysis, edited by I.G. Stewart, Edinburgh University
Press.

20
This point was discussed in Victor Bulmer-Thomas, “Application of Input-Output Analysis for Less Developed
Countries (LDCs)” in Readings in Input-Output Analysis, edited by Ira Sohn, New York, Oxford University Press, 1986.

21
Australian National Accounts: Input-Output Tables, 1974-1975, 1977-1978, 1978-1979, Canberra.
 205

resulting in an overestimate of intermediate demand. Similarly, an unusual column multiplier might lead to
suspect over-estimation of primary inputs or total output of the industry.

9.35. It follows from the above observation that the application of the RAS method, either simple or
modified, will only produce an accurate estimate of the input-output table if the control totals are accurate.
This requires accurate estimates of industry and product outputs, primary inputs (net input) and final demand.
Any errors in this area of national accounting data are likely to be transmitted to the input-output table whose
row and column totals are derived as residuals from these other vectors.

9.36. It should be noted that when the short-cut RAS method is used, no economic significance can be
attached to the values of the row and column multipliers. This should not prevent the use of the modified RAS
method. It may be preferable under those circumstances to regard the RAS method simply as a statistical tool
which can be used to adjust two-way tables. The basic method can be applied to many tables other than input-
output tables. In some of these applications some meaning may be attached to the values of the multipliers
as in the simple input-output case but in other cases the values of particular multipliers may have no
significance. In this context it should be mentioned that the use of the RAS method is not necessarily limited
to balancing the intermediate demand matrix, but could also be used to balance the use matrix of a supply and
use table, and in the case of the modified version of RAS, not only additional data on intermediate uses could
be incorporated, but also additional information on total values for each category of final demand like total
household final consumption expenditures, total gross capital formation, total exports, total imports, etc. might
be used.

9.37. In some countries, there are no clear-cut benchmark tables; additional information is introduced every
year as new data become available, to arrive at new annual tables. These practices are found in many countries
and some use the modified RAS method to update the old table into the new one by incorporating new
information. Because of the nature of the method, however, one should realize that the data in the new table
have to be close to those in the previous table, even though the latter may be based to a large extent on less
reliable data. Therefore those practices still require that from time to time a benchmark table be compiled on
the basis of the fullest possible information provided by censuses to check the accuracy of the tables derived
by other methods. In Norway, a country with a long tradition of compiling the full SUT annually, the full SUT
benchmark framework is revised every five years, the annual SUT is an updated version of the benchmark on
the basis of as comprehensive data as are available, either by current values or by indices, specially where only
more aggregated data are available.22

9.38. Finally, it is important to realize that all the tests carried out until now were on individual cell accuracy
(i.e. partitive accuracy) rather than on overall accuracy of the estimated input-output table in practical
projection (i.e. holistic accuracy), for example tests on how estimated tables perform on predicting gross
outputs given that final demand is known. The reason is that tests will be rendered trivial by the nature of the
RAS method to equate the sum of estimated elements in the intermediate matrix and final demand (which is
known) to total outputs. However, it is still possible to carry out tests on the performance of an updated
coefficient matrix on the most recent set of final demand and gross outputs, e.g. to compare errors in gross
output of time t+1 generated by using the coefficient matrix at time t updated from time t-1 with errors in gross
output generated by using the more outdated real coefficient matrix based on surveyed data of time t-1.

G. Other alternative methods for updating coefficients

22
Communication from Mr. Erling Flottum, Statistics Norway.
206

9.39. Bacharach23 also demonstrated that the RAS method is similar to the minimization of an objective
function of the form:

Minimize j X̃ ij ln(X̃ij / X ij)

i,j
subject to row and column sum constraints
where X̃ is the estimated value of X.

The Lagrangean multiplier approach is a further extension of the above formula:

Minimizej
(x̃ij&xij)2
i,j Bij
where Bij are weights.

If one assumes that the larger elements are less subject to change than the smaller, then weights assigned are
larger, for example by using X2ij. A weight of infinity means that the element Xij is fixed. Similarly, some
elements may be more likely to change, a weight can be Xij. Elements may be negative unless they are
constrained to be non-negative. W.I. Morrison and R.G. Thurman24 experimented with this method and found
that the RAS method produced better results that the Lagrangian multiplier approach with simple weights and
without additional constraints. The RAS is also computationally much cheaper. Statistics Netherlands25 has
tested the Lagrangean multiplier approach and argued that it has some advantages over RAS. However, its
advantages over the modified RAS are not clear.

23
M.O.L. Bacharach, op cit.
24
[Link] and R.G. Thurman, “A Lagrangian Multiplier Approach to the Solution of a Special Constrained
Matrix Problem”, Journal of Regional Science, Vol. 20, No.3, 1980.

25
Harthon, R. and J. van Dalen, “On the Adjustment of Tables with Lagrange Multipliers”, National Accounts
Occasional Paper, No. NA-024, Statistics Netherlands, 1987.
 PART THREE

SOME APPLICATIONS OF INPUT-OUTPUT TABLES

AND MODEL
210

X. CREATING AN INDUSTRY NOT IDENTIFIED FROM ISIC

CLASSIFICATION: ANALYSIS OF THE INTERNATIONAL TOURIST
INDUSTRY

A. Introduction

10.1. This chapter was written to demonstrate an application of input-output modelling in studying a specific
industry. International tourism is selected because, firstly, it is an economic activity that attracts a lot of
attention in many countries but at the same time is not separately classified in either national accounts or input-
output tables and, secondly, it is not a very complex issue to deal with. The principles of impact analysis will
be discussed in chapter XII and this chapter will focus mainly on how to create the international tourist
industry in input-output tables and how to apply the principles discussed in chapter XII. It should be noted that
the World Tourism Organization may come up with a different way of creating the tourist industry. Its plan
is to aggregate establishments that produce mainly tourist services.

10.2. International tourism is not an industry that can be immediately identified in input-output tables or in
national accounts which are constructed on the basis of the United Nations International Industrial
Classification of All Economic Activities (ISIC). However, this industry can be formed by aggregating
different economic activities which serve it directly. Procedures for forming the industry will be discussed
in this chapter, together with methods to analyse some of the major roles of the industry in generating
employment and taxes or in creating demand for capital and imports in the economy. These analyses are very
simple but may be used as powerful tools to provide important information on the industry. The procedures
used in analysing international tourism can also be applied to study other industries such as energy and
information that do not appear explicitly in input-output tables and national accounts.

10.3. The chapter will first define the industry and then discuss technical issues relating to its analysis.

B. Definition and classification of the international tourist industry

1. Definition

10.4. The importance of international tourism as a contributor to GDP and as a source of employment is
being increasingly recognized. It is estimated that in 1984 receipts from world international tourism totalled
$100 billion, which is over 5 per cent of total world merchandise exports in the same year. International tourist
receipts amounted to 8 per cent of GNP and 36 per cent of foreign exchange earnings in 1982 in Austria, 12.3
per cent of GNP in Cyprus, 20 percent of employment in Barbados, and 48 percent of employment in the
Bahamas.1

10.5. The international tourist industry of a country may be defined as the sum of domestic activities that
directly support the consumption of goods and services of foreign tourists in the country. Therefore, definition

1
“World Tourism: an Economic Analysis” by S. I. Papadopoulos, Braford in Revue de Tourisme, No.1, 1987 and
UNDP/TCDC News, Vol. 3, p. 9.
 211

of the international tourist industry is dependent on the definition of international tourists, which in turn can
be based on a narrower definition of non-residents. That is, tourists should include:

(a) Persons visiting the country for less than one year, specifically for purposes of recreation
or holiday, medical care, religious observances, family affairs, participation in
international sport and cultural events, conferences and other meetings, study tours and
other student programmes, as well as persons in transit to another country;
(b) Foreign students remaining for more than one year;
(c) Crew members of foreign vessels and aircraft docked in the country or on lay-over;
(d) Foreign business travellers who are in a given country for less than one year;
(e) Employees of international bodies who are on a mission of less than one year;
(f) Nationals who are residents of other countries who come back for visits of less than one
year.

10.6. In comparison to the boundary of activities of non-residents in a country, only activities of diplomats
and extra-territorial bodies like members of the armed forces of other countries are excluded as not belonging
to tourists' activities.

10.7. In general, it is quite difficult to measure directly the domestic activities that support the consumption
of goods and services of foreign tourists in a country. When this happens, the activity may be measured in
terms of the share of goods or services that are consumed by foreign tourists. Those expenditures (e.g. on
meals, accommodations, purchases of gifts, international as well as local transportation), according to the SNA,
should be treated as the country's exports.

2. Classification

10.8. ISIC recommends the classification of parts of the following activities into the international tourist
industry:

Table 10.1. ISIC components of the international tourist industry

Class Activities

5510 Hotels; camping sites and other provision of short-stay

accommodation

5520 Restaurants, bars and canteens

6010 Transport via railways

6021 Other scheduled passenger land transport

6022 Other non-scheduled passenger land transport

6110 Sea and coastal water transport

6120 Inland water transport

212

6210 Scheduled air transport

6220 Non-scheduled air transport

6304 Activities of travel agencies and tour operators; tourist

assistance activities n.e.c.

7111 Renting of land transport equipment

7112 Renting of water transport equipment

7113 Renting of air transport equipment

9219 Other entertainment activities n.e.c.

9231 Library and archives activities

9232 Museum activities and preservation of historical sites and

buildings

9233 Botanical and zoological gardens and nature reserves

activities

9241 Sporting activities

9249 Other recreational activities

.. Other direct purchases.

 213

Table 10.2. Compilation worksheet for the international tourist industry

ISIC Purchasers' value Trade, transport Taxes on Imports Basic values

margins products
(1) (2) (3) (4) (5) (6)

(1) 5510

(2) 5520

(3) 6010

(4) 6021

(5) 6022

(6) 6110

(7) 6120

(8) 6210

(9) 6220

(10) 6304

(11) 7111

(12) 7112

(13) 7113

(14) 9219

(15) 9231

(16) 9232

(17) 9233

(18) 9241

(19) 9249

(20) Trade and *

transport
margins

(21) Taxes on **
products

(22) Other products

purchased by
tourists broken
down by ISIC

(23) Total * **

Steps to calculate the output of the international tourist industry:

(a) Break down tourist purchases into ISIC listed in column (1);
(b) Break down each detailed purchase at purchasers' price into the components listed in col. (3)-(6);
(c) Each component of the output of international tourism in col. (6) is equal to col. (2) minus (3) minus (4) minus (5);
(d) Move total trade and transport margins and taxes on products (i.e. sum of col. 3 and 4 to asterisked places in col. (6);
(e) The sum of col. (6) is the total output of international tourism and can be treated as the whole vector of col.(6) or one
single value shown at row (23) of col. (6).
214

10.9. In fact, it is only possible to survey tourist expenditures by products, not by economic activity. In this
case, using ISIC, one implicitly assumes that the products consumed by tourists are matched with the ISIC
classification in table 10.1. The expenditures by tourists on the outputs of these activities should either be
direct purchases by the tourists themselves or through their purchases of package tours from tourist agencies.
When expenditures are made by buying package tours, they have to be broken down into ISIC components
as listed in table 10.1. For example, an expenditure on air travel tickets should be split into the payment for
air tickets (ISIC 6210 or 6220) and the payment of fees to travel agents (6304).

10.10. Other direct purchases of goods and services by tourists like souvenirs which are directed exclusively
to tourists, and others such as hairdressing services, etc. listed as the last item in table 10.1 above are not shown
with a corresponding ISIC class number because the kinds of goods and services detailed in ISIC bought by
tourists may only be known by conducting a survey on their expenditures. These other direct expenditures are,
unfortunately, not included by the United Nations in its recommended ISIC categories of the international
tourist industry. However, to be complete these items should be included since without tourists these purchases
would not be made. All these expenditures are at purchasers' prices and have to be broken down into basic
values, trade and transport margins and taxes on products. In the case of tourism, except for the goods directly
purchased by tourists, most of the items listed as parts of the activities of the international tourist industry are
services and therefore do not have trade and transport margins. However, the margins are still shown in table
10.2 so as to make the table compatible with a more general approach.

10.11. After all expenditures by international tourists are classified by ISIC, the vector of expenditures is
broken down into those supplied by resident producers versus non-resident producers. Only those goods and
services that are domestically supplied will impact the domestic economy. For example, air transport may be
provided by non-resident carriers but services of domestic travel agencies and airport services represented by
airport user fees charged on carriers and tourists are part of exports and should impact the domestic economy.
Similarly, the goods that are bought directly by tourists might be imported but the margins on these goods that
are domestically supplied will impact the economy. The final vector of international tourist purchases of
domestic products in detailed ISIC should be the vector of output of the international tourist industry shown
in table 10.2. It is important to know that trade and transport margins on all goods and services consumed by
tourists should also be part of the industry output. When included, the output is at purchasers' values.

10.12. By the definition given above, the output of the international tourist industry consists of either direct
expenditures made by the tourists themselves or through tourist agencies that are supplied by resident
producers. The output of the industry, therefore, is not the same as the total tourist expenditures in the country.
In addition, the output of this industry is also the same as the final consumption by tourists of domestic
products, which is then treated as exports. The treatment of the total output of this industry as exports is a
special case since normally only part of the output of an industry is exported. This special characteristic has
to be taken into account in modelling the industry in an I/O table. Before creating an international tourist
industry in input-output table is discussed, a general scheme of creating a new industry by aggregating parts
of other industries in an existing input-output table must be first. This general scheme will require techniques
of disaggregation and aggregation of industries.

C. Modelling the international tourist industry in I/O table

10.13. The output of the international tourist industry in an input-output table as described in section B above
is a vector contained in column 6 of table 10.2 rather than a single value, which consists of parts of the outputs
of many industries. With regards to inputs, there should be an input structure corresponding to every detailed
component output of the international tourist industry contained in column 6 of table 10.2. It is possible to
 215

proceed with the work of creating the international tourist industry by, at first, separating every output and its
associated inputs which are part of the international tourist industry from other industries and then aggregating
these newly separated rows and columns to form the new industry. Fortunately, the international tourist
industry is much simpler in the sense that its output is all consumed by non-residents and therefore must be
treated as exports which are part of final demand. None of its output would be used as intermediate inputs in
other industries. Or in other words, no domestic activities would generate output of international tourism
(remember we focus only on tourism by non-resident tourists). To create the international tourist industry, one
important assumption has to be introduced, i.e. the input coefficients of the industry that are separated out are
similar to the input coefficients of the industry that remain. This may not be applicable to other industries
since final demand for the output of a particular industry normally generates only part of the total impact of
that industry, with other parts generated by final demand of the outputs of other industries which, in turn,
indirectly generates the output of that particular industry. To create an industry that is not independently
identified by an ISIC classification will be discussed in part D below.

1. Creating the international tourist industry

10.14. To analyse the international tourist industry, it is necessary to assume that the input coefficients of the
industry which are separated out are similar to those of the industry that remain. Thus for the international
tourist industry, only the inputs corresponding to every output which is a component of the industry are
separated out. Its total composite output is then treated as exports. An example in table 10.3 will be used to
demonstrate how the international tourist industry is created. The results are shown in tables 10.4 and 10.5.
In the example, it is assumed that 10% of the output of industry 1 and 10% of the output of industry 2 make
up the output of the international tourist industry.

10.15. The following steps will be used:

(a) The components of output that belong to the international tourist industry are separated;
(b) Associated inputs of the output components are separated;
(c) An aggregation of inputs of the components of the tourist industry is used to form the
inputs of the tourist industry;
(d) The final demand of the remaining industries from which outputs and inputs are extracted
to form the tourist industry is equal to its remaining outputs less intermediate
consumption.

10.16. Table 10.3 shows the original table. Table 10.4 shows the separation of industry 1 and industry 2, each
into two parts. From industry 1, 10 is the output of the tourist industry and from industry 2, 20 is also the
output of the tourist industry. Final demand of industry 1 that remains is equal to its output (90) less the
intermediate consumption of its output -- [90 - (9+1+13.5+1.5+30)=35]. The final demand of the part of
industry 2 that remains is calculated in a similar manner. Disaggregation of inputs is simply done by
extracting 10% from the inputs of the corresponding industries. Columns and rows 2 and 4 of table 10.4 show
the inputs and outputs of the tourist industry that are extracted. Industries 2 and 4 in table 10.4 are then
aggregated in table 10.5 into one industry (the third row and column). In both tables, the outputs of the
international tourist industry are entirely consumed as final demand.
216

Table 10.3. Matrix without explicit components of international tourist industry aggregated

Matrix f Final demand Total output

Industries 1 2 3

1 10 15 30 45 100
2 20 9 7.5 163.5 200
3 30 50 22.5 47.5 150

Value added 40 126 90

Output 100 200 150

Table 10.4. Matrix with components of international tourist industry disaggregated

Matrix f Final demand Total output

Old industries 1 2 3

New disaggregated 1 2 3 4 5
industries

Row 1 in table 1 9 1 13.5 1.5 30 35 90

11.3 is split
2 10 10

Row 2 in table 3 18 2 8.1 0.9 7.5 143.5 180

11.3 is split
4 20 20

5 27 3 45 5 22.5 47.5 150

Value added 36 4 113.4 12.6 90

Output 90 10 180 20 150

Table 10.5. Matrix with components of international tourist industry aggregated

Matrix f Final demand Total output

New industries 1 2 3 4

Industries of table 11.4 to be aggregated 1 3 2+4 5

1 9 13.5 2.5 39 35 90
2 18 8.1 2.9 7.5 143.5 180
3 30 30
4 27 45 8 22.5 47.5 150

Value added 36 113.4 16.6 90

Output 90 180 30 150

 217

10.17. The form in table 10.4 or table 10.5 can be used to analyse the international tourist industry. If the
industry is treated as made up of two separate industries, its final demand would consist of two non-zero
components (0 10 0 20 0). If it is treated as one industry, then its final demand would consist of only one non-
zero component (0 30 0 0).

10.18. The second method may be preferred if the original input-output table is not very detailed to the extent
of having each industry corresponding to a 4-digit ISIC, and if analysts want to see the international tourist
industry separately identified. Table 10.5 shows the inputs used in the new industry as well as the value added
generated.

2. Distortion created by aggregation

10.19. The aggregation of the components of the international tourist industry is mainly for the purpose of
presentation rather than for analysis because of the distortion it creates. One can see the distortion if one
applies impact analysis using equation 12.2 in chapter XII with the final demand vector of international
tourism shown below:

0
0
Y '
0
x

in which x is both the output and final demand of the international tourist industry. Given that the components
of x change relatively in the next period, the impacts would be distorted when the same output and input
structures of the base-year period are used. In this case, the formulation shown in table 10.4 is preferred.

3. Impact analysis of the tourist industry

10.20. The analysis of the tourist industry should follow the procedure described in chapter XII.

D. A general aggregation and disaggregation scheme for the creation of a new industry not
identified by ISIC

10.21. Many industries that are normally referred to in common language are not identified in ISIC, I/O tables
or national accounts. For instance, the energy industry includes many diverse activities such as burning of
organic substances (e.g. thatches, wood, etc.), mining and extraction of coal, crude oil, natural gas, uranium
and thorium ores, the production of coke, petroleum products and nuclear fuel, the production and
distribution of electricity, gas, steam and hot water. Another example is the pollution abatement industry
which consists of all abatement activities connected with every industry identified by ISIC. Other industries,
such as the port industry which is composed of the operation of ports and other services serving them such
as transport, financial and administrative services, etc., are made up of segments of many activities. One
important "industry" that one may want to look at as a whole in order to analyse its interrelationships with other
218

industries is the food and agriculture industry. According to the recommendation of FAO, the economic
accounts for food and agriculture consist of the following activities:2

(a) Crop and animal husbandry (ISIC division 01);

(b) Forestry, logging and related activities (ISIC division 02);
(c) Fishing (ISIC division 05);
(e) Food manufacturing production (ISIC division 15);3
(f) Environmental protection activities (desertification control, water resource protection,
marine pollution control, erosion control, flood prevention or control and natural reserves;
(g) Research and development relating to crops, livestock, forestry, fisheries and food
production (part of ISIC class 7310);
(h) Veterinary activities (ISIC class 8520);
(i) Technical and vocational and higher education relating to agriculture (part of ISIC classes
8022 and 8030);
(j) Agro-industries other than food products (part of ISIC divisions 16, 17, 19, 20);
(k) Production of inputs for the agricultural sector (ISIC class 2412 - manufacture of
fertilizers and pesticides);
(l) Production of machinery and equipment for the agricultural sector (ISIC class 2921 -
manufacture of agricultural and forestry machinery, ISIC class 2925 - manufacture of
machinery for food, forestry and tobacco processing, and part of ISIC class 3511 -
building of fishing boats and fish processing factory vessels);
(m) Development of infrastructure for rural areas (part of ISIC division 45).

Many economic activities cited above are only a part of an ISIC class and therefore must be separated before
they can be re-aggregated into one industry. As a consequence, the formation of such industries requires the
application of the procedures described in this part on disaggregation and aggregation.

10.22. In creating a new industry in an I/O table from existing information, parts of the new industry have
to be separated from the existing I/O table before they can be aggregated to form the new industry. The
separation technique will now be discussed before the aggregation technique.

1. Disaggregation scheme4

11.23. The general scheme is based on two assumptions:

(a) The input coefficients of the industry as separated out are similar to the input coefficients
of the industry that remain;

(b) The consumption by other industries of the output of the industry that is separated out is
proportional to the consumption pattern of the output of the original industry by other
industries.

2
FAO, A System of Economic Accounts for Food and Agriculture, Rome, 1996.

3
Food production is defined by FAO as including food manufacturing production and primary food production
included as part of ISIC divisions 01, 05.

4
See Alan M. Wolsky, “Disaggregating Input-Output Models” in Review of Economic Statistics, May 1984.
 219

10.24. Assume that there exist two matrices: matrix f is the original flow matrix and matrix a is the coefficient
matrix derived from the f matrix. Matrix f includes the row(s) of value added, matrix a includes only the
intermediate coefficients. Corresponding to these matrices are the vectors of industry output x and of final
demand y. Then, let's assume that the nth industry is split into the remaining industry and the part that is
separated out. The output of industry n is xn, the remaining industry has the output xn = w1 xn and the output
of the part that is separated out is xn+1 = w2 xn where w1 + w2 = 1.0. w1 is the share of output that remains to
be the output of industry n and w2 is the share of output of the newly separated part of the industry.

10.25. The newly disaggregated flow matrix F is obtained by:

- first, multiplying the nth column by w1 to obtain the column of the remaining industry and
then multiplying the same nth column by w2 to obtain the column of the newly separated
part;

- second, multiplying the nth row by w1 to obtain the row of the remaining industry and then
multiplying the same nth row by w2 to obtain the row of the newly separated part.

10.26. The new final demand matrix Y can be easily obtained by multiplying yn alternatively with w1 and w2.
Below f is the original flow matrix, F is the newly disaggregated flow matrix, and Y is the newly
disaggregated final demand matrix.

Table 10.6. The new disaggregated flow and final demand matrices

Ind.1 Ind. 2 Industry n Ind. m

Remain split

f11 f12 . . f1n w1 f1n w2 . . f1m y1

f21 f22 . . f2n w1 f2n w2 . . f2m y2
. . . . . . . . . .
. . . . . . . . . .
w1 fn1 w1 fn2 . . w1 fnn w1 w1 fnn w2 . . w1 fnm w1 yn
w2 fn1 w2 fn2 . . w2 fnn w1 w2 fnn w2 . . w2 fnm Y= w2 yn
F =
. . . . . . . . . .
. . . . . . . . . .
fm,1 fm,2 . . fm,n w1 fm,n w2 . . fm,m ym

10.27. In a more concise mathematical term, F can be written as:

F = SfS'

where S is a disaggregation matrix and S' is the transpose of S. The disaggregation matrix S, which is similar
to an aggregation matrix that will be discussed later, is formed in terms of a (m+1)xm matrix in which the two
220

rows of the new matrix F corresponding to the new nth and (n+1)th rows of matrix F have respectively the
values w1 and w2 on its nth column, the column which needs to be split. Other rows of S are assigned values
as follows: the first row has 1 at the first column and zero elsewhere, the second row has 1 at the second
column and zero elsewhere, etc.. The S matrix corresponding to F is shown in table 10.7.

Table 10.7. The disaggregation matrix

Ind. 1 Ind. 2 Industry n Ind. m

1 0 . . 0 . . 0
0 1 . . 0 . . 0
. . . . . . . .
. . . . . . . .

The 2 0 0 . . w1 . . 0
new
S= split up 0 0 . . w2 . . 0
rows

. . . . . . . .
. . . . . . . .
0 0 . . 0 . . 1

10.28. The newly diseggregated output is as follows:

X = (x1, x2, ..., w1 xn, w2 xn, ..., xm)

10.29. If the newly disaggregated coefficient matrix that is derived from the original matrix f is called A, then
A can be derived either by dividing each column of matrix F in table 10.6 by its corresponding industry output
or directly by using the following form:

Table 10.8. The new disaggregated coefficient matrix

a11 a12 . . a1n a1n . . a1m

a21 a22 . . a2n a2n . . a2m
. . . . . . . . .
. . . . . . . . .
w1 an1 w1 an2 . . w1 ann w1 ann . . w1 anm
A = w2 an1 w2 an2 . . w2 ann w2 ann . . w2 anm
. . . . . . . . .
. . . . . . . . .
am,1 am,2 . . am,n am,n . . am,m
 221

10.30. The following example of a matrix f will be used to show how disaggregation is carried out:

10 15 30 45
20 9 7.5 163.5
(10.1) f '
30 50 22.5 47.5
40 126 90 0

The last row of matrix f is the row of value added. The original vector of industry output and the original
vector of final demand are as follows

x = (100, 200, 150)

y = (45, 163.5, 47.5)

The coefficient matrix a has the following value:

0.10 0.075 0.200

(10.2) a ' 0.20 0.045 0.050
0.30 0.250 0.150

10.31. Now we separate:

- Industry 1 in matrices f and a into industries 1 and 2 in the new matrix F and A and

- Industry 2 in matrices f and a into industries 3 and 4 in the new matrices F and A,
assuming that output shares are the same for the two cases with w1 = 0.9 and w2 = 0.1.

10.32. The disaggregation matrix S for the case mentioned above can be written as follows:

0.9 0 0 0
0.1 0 0 0
0 0.9 0 0
(10.3) S '
0 0.1 0 0
0 0 0.1 0
0 0 0 1

10.33. In the S matrix with the order of 6 x 4, which is the matrix to disaggregate the rows of matrix f, we
see that the first two rows aim at splitting the first row of matrix f, the next two rows aim at splitting the second
row of matrix f, with the respective weights of 0.9 and 0.1. The last two rows of matrix S correspond to the
rows of the third industry and value added in matrix f and the last two columns correspond respectively to the
third industry and final demand in matrix f. The matrix S' with the order of 3 x 5, matrix S,' which is the
matrix to disaggregate the columns, should have one column and one row less than the transpose of matrix S
222

since there is no column corresponding to the row of value added in matrix f. In this example, S' is the
transpose of S but with the last row and the last column eliminated .

0.9 0.1 0 0 0
)
(10.4) S ' 0 0 0.9 0.1 0
0 0 0 0 1

10.34. The final F and A are shown below:

8.1 0.9 12.15 1.35 27

0.9 0.1 1.35 0.15 3
16.2 1.8 7.29 0.81 6.75
(10.5) F '
1.8 0.2 0.81 0.09 0.75
27 3 45 5 22.5
36 4 113.4 12.6 90

The final F matrix also includes value added in the last row.

.09 .09 .0675 .0675 .180

.01 .01 .0075 .0075 .020
(10.6) A ' .18 .18 .0405 .0405 .045
.02 .02 .0045 .0045 .005
.30 .30 .2500 .2500 .150

Equivalent X and Y are as follows:

X = (90, 10, 180, 20, 150)

Y = (40.5, 4.5, 147.15, 16.35, 47.5)

10.35. In matrix A, the input structures of industries 1 and 2 must be the same. Similarly, those of industries
3 and 4 must also be the same. Assumption (b) can be checked by dividing row 3 by row 4 in the final matrix
F.

2. Aggregation scheme

10.36. Suppose that industries 2 and 4 in the new matrices F and A are parts of the international tourist
industry. The creation of the new industry would require the aggregation of the above industries in matrices
F and A.

10.37. Call F and A the matrices to be aggregated and a and f the resulting aggregated matrices. In matrices
F and A, industry 2 must be aggregated to industry 4 to form industry 3 in matrix a, with w2 = 0.207 (=
 223

4.5/(4.5+17.25) and w4 = 0.793 (=17.25/(4.5+17.25)) which are output shares of the industries that are subject
to aggregation. The aggregation scheme of the flow and coefficient matrices are discussed separately below.

Aggregation scheme of the flow matrix

10.38. Aggregating the flow matrix is quite simple: one needs to add the rows to be aggregated together and
then the columns to be aggregated together or vice versa. Mathematically, the aggregation can be written
neatly by using an aggregation matrix. Call G the aggregation matrix. Say F is a 6 x 5 matrix, the same as the
matrix in equation 10.5, and assume that industries 2 and 4 are aggregated together and put into the new
industry in the third position of the new matrix f, then the aggregation matrix G can be written as follows:

1 0 0 0 0 0
0 0 1 0 0 0
(10.7) G ' 0 1 0 1 0 0
0 0 0 0 1 0
0 0 0 0 0 1

10.39. In matrix G, on the third row where the resulting new row is put, 1 is entered in the second and fourth
positions. This means that rows 2 and 4 of the original matrix F are aggregated and then put into the third row
of the new matrix f. Since rows l, 3, 5 and 6 are not aggregated, 1 is entered in the first, third, fifth and six
elements respectively of the first, second, fourth and sixth row of G. Row 2 is to be aggregated to row 4 and
therefore does not have a corresponding row in the aggregation matrix. The transpose of G, G’, is for
aggregating the corresponding columns in a similar manner. The new matrix f can then be written as follows:
f = GFG) 5.

10.40. The resulting flow and coefficient matrices, after aggregation, are shown in tables 10.9 and 10.10

5
If we need to aggregated only rows together we can apply the formula f = GF. If we need to aggregate only columns
together we can use the formula f = FG. Aggregation of rows or columns is useful in some economic analysis. For
instance, if one wants to know the use of electricity in manufacturing industries, one has to aggregate all columns of
manufacturing industries together.
224

Table 10.9. The new aggregated flow and final demand matrices

Ind. Ind. Aggregated Ind.

1 3 industry 5
F11 F13 F12 + F14 F15 Y1
F31 F33 F32 + F34 F35 Y3
f= F21 + F41 F23 + F43 (F22 + F24 ) + (F42 + F44 ) F25 + F35 y= Y2 + Y 4
F51 F53 F52 + F54 F55 Y5

Table 10.10. The newly aggregated coefficient matrix

Ind. Ind. Aggregated Ind.

1 3 industry 5

A11 A13 w1 A12 + w2 A14 F15

A31 A33 w1 A32 + w2 A34 F35
a= A21 + A41 A23 + A43 w1 (F22 + F24 )+ w2 (F42 + F44 ) F25 + F35
A51 A53 w1 F52 + w2 F54 F55

where the new vector of industry outputs is as follows:

x = (X1, X3, X2 + X4, X5 )

10.41. In general, when industries n and n+1 are aggregated, the following formula can be used to calculate
the new coefficients:

aij = Aij for i, j … n

anj = Anj + An+1,j for j … n

ain = w1 Ain + w2 Ai,n+1 for i … n

ann = w1(Ann + An+1,n) + w2 (An,n+1, +An+1,n+1)

10.42. Looking at column 3 of table 10.10 or the formulas above, one can see that the newly aggregated
industry has a new output which is equal to X2 + X 4, with the shares of output equal respectively to w 2 and w4 2.
The new coefficients are, as a result, the weighted average of the coefficients in columns 2 and 4 of the original
matrix A.

10.43. Applying the formulas in table 10.9 to the example matrix of 10.5 (para. 10.34), one can derive the
new matrix f with the third industry as the newly aggregated industry which is composed of industries 2 and
4 of the original matrix F in 10.1 (para. 10.30) as follows:
 225

8.1 12.15 2.25 27

16.2 7.29 2.61 6.75
(10.8) f ' 2.7 2.16 0.54 3.75
27 45 8 22.5
36 113.4 16.6 90

10.44. The last row of matrix f is the row of value added. The new matrix a corresponding to the f matrix
above is:

0.09 0.0675 0.180 0.075

0.18 0.0405 0.045 0.087
(10.9) a '
0.30 0.2500 0.150 0.267
0.03 0.0120 0.025 0.018

10.45. From matrix f in 10.8, one can observe in column 3 the inputs used for the production of the newly
aggregated industry and in row 3 the uses of the output of the new industry by other industries; 16.6 is the
contribution to value added or GDP by the new industry. The latter figure can easily be obtained without going
through the procedures of disaggregation and aggregation shown above, but the former figures can only be
obtained by using input-output analysis.
226

XI. INPUT-OUTPUT TABLES AND PRODUCTION ACCOUNTS IN

CONSTANT PRICES

A. Introduction

11.1. This chapter should be read as a continuation of chapter II. It discusses first the double deflation
method and the procedures for applying the method for the benchmark years when full information is available
and then on the basis of the relationships in the benchmark year and of incomplete annual information, it
discusses the application of the method to estimate annual quantity and price indices in an integrated manner.

B. Double deflation method for benchmark years

1. Definition

11.2. Before discussing the double deflation method, it is necessary to review the definition of GDP:

GDP = value added at basic prices + taxes minus subsidies on products

Value added at basic prices is defined in turn as:

Value added at basic prices = industry output at basic prices - intermediate consumption at
purchasers' prices

11.3. The double deflation method is aimed at deriving real GDP implicitly by deflating current industry
output, intermediate consumption and taxes minus subsidies on products. Though the double deflation method
may be applied as a short cut for each industry, it can only be consistently and systematically applied within
the context of the full SNA supply and use framework (SUT). By using the full SUT framework, it is possible
to derive implicit price indices for all aggregates in it such as: GDP, exports, imports, household final
expenditure, government final expenditure, gross capital formation (which may be disaggregated by industries
and/or by institutional sectors).

2. Procedures

11.4. The double deflation method when applied in the most comprehensive manner requires:
 Table 11.1. The use table at basic prices

Industries Exports Household Government Gross Total product Deflator

Intermediate consumption f.o.b. final final capital uses
expenditure expenditure formation

Goods Market Other non-market

services services

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Domestic Goods UB11 UB12 UB13 EB1 HCB1 KB1 UB1=(1+.+7) Basic

products Market services UB21 UB22 UB23 EB2 HCB2 KB2 UB2=(1+.+7) price

Other non-market services HCB3 GC UB3=(1+.+7) indices

Imported Goods M11 M12 M13 MHC1 MK1 M1=(1+.+7) Import

products Market services M21 M22 M33 MHC2 MK2 M2=(1+.+7) price indices

Other non-market services

Trade and Goods TM11 TM12 TM13 TME1 TMHC1 TMK1 TM1=(1+.+7) Base year trade
transport margin
margins

Taxes Goods TX11 TX12 TX13 TXE1 TXHC1 TXK1 TX1=(1+.+7)

on products Market services TX21 TX22 TX23 TXE2 TXHC2 TXK2 TX2=(1+.+7) Base year tax

Other non-market services TXHC3 TX3=(1+.+7) rates

Gross value added at basic prices V1 V2 V3

Compensation of employees W1 W2 W3

Other taxes less subsidies on production OTC1 OTC2 OT3

Consumption of fixed capital CC1 CC2 CC3

Operating surplus/Mixed incomes OS1 OS2

Total industry output I1 I2 I3

at basic prices

Other column total E HC GC K U

Price indices based on cost P3

228

- The supply and use tables of the benchmark or base year;

- The current supply and use tables at current prices;
- Basic price indices for all commodities;
- Import price indices;
- Changes in wage rates.

11.5. The procedures for applying the double deflation method can best be demonstrated using table 2.1 of
chapter II and table 11.11, both compiled with detailed transactions measured at basic prices. (There is an
alternative way of applying the double deflation method that does not require the conversion of the full system
to basic prices. It is discussed in para. 11.45.) There are two reasons for converting the system into basic
prices, namely:

- Price indices are more consistent when the price survey is carried out for basic prices,
which are not affected by changes in trade margins and in tax rates. Basic prices in many
countries are more popularly named "producers’ prices" or "wholesale prices" which are
the first prices offered in the process of transactions, excluding taxes on products;

- Provided there are differential changes in trade margins, import prices or taxes rates, the
separation of intermediate consumption at purchasers' prices into domestic components
at basic prices, imported components c.i.f., trade margins and taxes minus subsidies on
products would allow the uses of different price indices with different components and
thus the value added at constant prices would be more accurately derived.

(a) Preparation of the use table in basic prices

11.6. Table 11.1, the use table at basic prices is derived by splitting every transaction flow of commodities
at purchasers’ prices, for example U11 in table 2.2 of chapter II into four separate components: the domestic
component at basic prices (i.e. UB11) , the imported component c.i.f. (i.e. M11), the component of trade
margins2 (i.e. TM11) and the component of taxes on products (i.e. TX11). For example:

U11 = UB11 + M11 + TM11 + TX11

Thus along each row of table 11.1, the total uses at basic prices of every domestic product (i.e. UB1) must
equal the total domestic production of the product (i.e. X1 = X11+X12+X13, see table 2.1).

Application of proportionality for trade margins

11.7. Normally the ratios of wholesale trade margin, retail trade margin and taxes on products for each
commodity use are not available. The common procedure used in deriving these ratios is to assume that they
are the same for each kind of use of the commodity whether it is domestic or imported goods and/or services.
Thus, the ratio of trade margins for goods 1 is equal to TM1/SP1; the ratio of taxes on product 1 is equal to

1
The presentation here benefits from Bent Thage’s Commodity Flow Systems and Construction of Input-Output
Tables in Denmark, Danmarks Statistik, 1986.

2
Trade margins may be broken down separately into wholesale and retail trade margins but for simplicity of
presentation, they are shown aggregated in this chapter.
 229

TX1/SP1 with all information available in the supply table 2.1 of chapter II. With the above assumption, it is
possible to derive for example:

TM11/U11 = TM12/U12 = TM13/U13 = TMHC1/HC1 = TMK1/K1 = TM1/SP1 and

TX11/U11 = TX12/U12 = TX13/U13 = TXHC1/HC1 = TXK1/K1 = TX1/SP1

11.8. The census or survey of distributive trades and services provides the wholesale and retail trade ratios.
The ratios when available and applied for a detailed list of commodities would enhance the accuracy of the
method. Thus, it should be applied at the most detailed level of commodities available. In some countries such
as Denmark and Norway the method is applied at the level of over 1,000 commodities. Obviously, when the
use and supply tables are aggregated to a smaller number of commodity groups, the ratios of trade margins will
vary by user due to the variation of detailed commodities used in each grouping. The application of the
proportionality assumption at a more aggregated level will lose a lot of information available.

Splitting uses into domestic and import shares

11.9. The splitting of uses into domestic and imported goods is important for the purpose of deflation since
prices of domestic goods and imports may change differently, and thus would affect the real value of value
added differently, particularly at the industry level. The splitting of uses which are already known into
domestic and imported goods requires estimating imports for each element of uses in the use table. Since the
total import of each commodity is known, splitting is simply the problem of allocating this total to its users.
The allocation of an imported commodity may be based on available surveyed information, expert knowledge
and the proportionality assumption:

- Surveyed information of import shares (ratios). Census of manufacturing industries, if

asked, may provide import shares of products used as inputs by every industry. This
information must be analysed together with information provided by expert knowledge.
The share of a product being imported into a particular industry should not be assumed
to be the same across industries. This means that the share of a product which is imported
and used by a different industry or by final demand is different.

- Expert knowledge. The use of surveyed information is not enough to fill up the import
matrix in table 11.1, expert knowledge is also needed. The use of information from import
statistics at the very detailed classification level will help identify whether an imported
commodity is complementary and fully imported. For example, natural rubber which is
not produced in Europe or the United States must be fully imported. It is also possible to
know by expert knowledge whether the imported commodity is used for final or
intermediate consumption or for capital formation. For example, imported machinery
certainly will be allocated to gross capital formation and imported motorcycles could be
allocated to final household expenditures. Also at the detailed level it is possible to guess
in what industries the imported products are used.

- Proportionality assumption. After having fully utilized the available information as

discussed above, the residual of the import of a commodity can be proportionally
allocated. One may allocate the residual to all appropriate users (namely to the elements
of the row in the use table corresponding to that commodity) if the residual is large and
if one is uncertain about the consumption of the import by the users that have not received
any allocation. One may also decide that some flows need no additional allocation if one
230

is sure about their consumption of the imported commodity. The criterion for allocation
is the share of use of a particular user within the group to be distributed. For example, in
table 2.2 of chapter II, if the imports of goods are all distributed proportionally to all users,
then U11/U1, U12/U1, ..., HC1/U1 and K1/U1 will be used as shares respectively for the
goods industry, the market services industry,..., household final expenditures and gross
capital formation. Even without surveyed information and expert knowledge, the
proportionality assumption is more useful if it is applied at the detailed level of
commodity classification instead of at the more aggregate level. Applying the
proportionality assumption at the detailed level will provide a different import ratio for
each element in the use table when commodities in the use table are aggregated.

(b) Deflation by types of products

11.10. The deflation procedure that is applied to the SUT framework varies by type of products. For the
purpose of demonstration, products are classified into three groups: goods, market services and other non-
market services.

11.11. Goods and market services are normally treated similarly since information on their unit price is
normally obtainable, except of course for some market services for which it is not easy or even possible to
determine their quantities and thus their unit prices. Prices of other non-market services are not available
because other non-market services are by definition not for sale. In table 11.1, it is easily recognized that only
goods have trade margins and that other non-market services are assumed to be not subject to product taxes.

11.12. The basic deflation procedure is as follows:

First, deflate goods and market services in the supply table at basic prices (table 2.1 of chapter II) in
order to derive industry output at constant prices and implicit industry output price indices;

Second, calculate the price indices for other non-market services based on production cost using the
use table at basic prices (table 11.1);

Third, deflate intermediate consumption and final uses in the use table using basic price indices,
import price indices and the price indices for non-market services calculated in step two;

Fourth, deflate gross value added at basic prices residually by taking the difference between industry
output in constant prices and intermediate consumption in constant prices;

Fifth, deflate taxes on products by multiplying total product supply at basic prices (table 2.1 of
chapter II) with the base year tax rates, applying different rates to imports and domestic goods;

Sixth, obtain GDP at constant market prices as the sum of real gross value added at basic prices and
real taxes on products. (GDP at constant prices can also be obtained by summing components of final
demand in constant prices.)

Below is further discussion of how goods, market services, other non-market services and taxes on products,
value added at basic prices and other components of value added are deflated in the above procedure.
 231

Deflation of goods

11.13. Domestically produced goods should first be deflated by basic price indices in the supply table across
every row from column 1-3. The result of the deflation is the implicit price indices of industry outputs at
constant prices and the implicit indices of industry outputs.

11.14. The application of basic price indices is also carried out for all the uses of products in use table 11.1.
Here, obviously it is assumed that price affects all users similarly in terms of production costs and the
differential effects on users come from changes in trade margins and taxes on products.

Deflation of goods sold at different prices to different users

11.15. For some products such as energy products which include electricity, gas, gasolines which are
produced and/or supplied by the public sector but sold at different prices to different users even when there
are no trade margins involved, the application of the same price index in the use table for all users is not
logical. Different price indices to different users should be applied, but then it would not be possible to get
the same implicit price index for the supply table and the use table. This is the well-known shift-share problem
in price deflation. The problem created is that output quantity produced may decline while output uses at
constant prices may increase and vice versa. The example in table 11.2 will make this problem clearer. Say
electricity rates for the household sector are different from those set for to other sectors while their respective
shares of quantity used changed.

11.16. In the example in table 11.2, prices set for households increased 10% from the base year to the current
year while prices set for other sectors remained unchanged. Because of the price increase, households
consumed less and with electricity cheaper and an increasing scale of production other sectors consumed more.
One can see in the indices that there is no change in quantity index but there is a reduction in volume index.
In fact, with a shift in shares it is quite possible to have an increase in quantity index and a reduction in volume
index and vice versa. The above example reflects the situation in many countries where different users of
energy are charged different prices.

Table 11.2. Shift share problem in price and quantity indices

Household Other sectors Total

Base year

Quantity 20 10 30

Unit price 6 5

Value 120 50 170

Current year

Quantity 15 15 30

Unit price 6.6 5

Value 99.00 75.00 174.00

Value at constant prices, quantity and volume indices

Value 90 75 165

Quantity index 75.00 150.00 100.00

base year = 100

Volume index 75.00 150.00 97.06

base year = 100
232

11.17. In this situation, the price index applied in the supply table is the one reflecting the cost of production
and the price indices applied to the use table are users' price indices. In some countries, the implicit price
indices derived from the use table are applied to the supply table to "harmonize" the two tables, but in this case
volume indices in the use or supply tables would not reflect real quantity indices.

Deflation of imported commodities

11.18. Imported commodities used for intermediate and final consumption shown in table 11.1 are deflated,
commodity by commodity, by import price indices.

Deflation of market services

11.19 Market services are produced for sale, therefore it seems natural that there exist quantities of services
sold and their unit prices. In practice, it is not easy to define quantities of services and to collect statistics on
their unit prices. There are normally three alternative methods that statisticians use in dealing with the problem.
The first alternative is to use some proxies for unit price indices. For example, many service outputs are
simply deflated by general consumer price index or by special price indices among consumer price indices that
seem appropriate to the service outputs to be deflated. The second alternative is to construct price indices for
the services on the basis of their costs of production. The approach is similar to that used to deflate other non-
market services that will be discussed later. When price indices are constructed from the costs of production,
the operating surplus is ignored, which means that it will be deflated by the constructed price index itself when
output is so deflated. The third alternative is to derive quantity indices first and extrapolate base year values
by these quantity indices. For example, the number of patients may be used as proxy for the quantitative index
of hospital output. By the last method, price indices are derived implicitly given output at current and base
year prices. It is not possible to describe in detail, product by product, how volume of a service is
approximated or how current values are deflated in this chapter, therefore only some general common practices
are described below.

Deflation of trade margins

11.20. Trade margins, which normally are separated into wholesale and retail trade margins, are the output
of the trade industries. There should be quantity and price indices developed. But this is almost impossible.
Changes in the ratios of trade margins over purchases are assumed to measure changes in price indices of trade
margins even though changes in the ratios may also include changes in technology. In order to derive trade
margins in constant prices, the ratios of trade margins of the base year are applied to the uses of goods at basic
prices shown in table 11.1. It is important to note that the ratios used here are the ratios at basic prices which
are calculated from table 11.1 of the base year, i.e they are equal to TM1/UB1 (UB1 is the total commodity
uses of goods at basic prices) . These ratios are different from the those that are used to calculate trade margins
at current prices, TM1/SP1 shown in the supply table 2.1 of chapter II.

Deflation of other market services

11.21. Other market services may be deflated by one of the methods described in paragraph 11.19. The
practice among countries can be as diverse as one could imagine since no international standards have been
developed in this area.

11.22. For financial intermediary services, some countries deflate imputed charges by obtaining details on
various kinds of loans and deposits and their respective interest rates and deflate interest received and paid by
changing their respective interest rates. Volume of explicit interest charges is normally extrapolated from the
 233

base year by a volume index of transactions. Other countries compute volume index by the sum of credits and
debits of financial institutions deflated by the price indices of their final uses.

11.23. For passenger transport, price indices of transport fees are normally used to deflate outputs at current
prices. For goods transport, volume indices based on ton-kilometres are normally used to extrapolate real
outputs from the base year and implicit price indices are obtained as a result.

11.24. Hotel and lodging room rates can be used to deflate the output of hotel and lodging services by type
of hotel and lodging.

11.25. The output of restaurants and the like may be deflated by input price indices or consumer price indices.

11.26. For life and casualty insurance, it is not easy to deflate their outputs by deflating the components that
are used to calculate them. They are either deflated by general consumer price index or special indices
according to kind of benefits.

11.27. Business services such as accounting, legal services, data-processing, engineering and architectural
services, advertising services, machinery and equipment rental and leasing services, etc. are commonly deflated
by input prices and wage rates. Volume indices are created in some countries to measure real outputs of legal
services. Volume indices are based on the number of court cases and proceedings, property transfers and
mortgage advances. Some legal services may be deflated by hourly rate index for lawyers.

11.28. Rents are deflated by rent indices. Owner-occupied housing services may be deflated either by rent
index or by costs depending on how outputs of the services are estimated.

11.29. Deflators of education services and health services that serve specific individuals may be derived
implicitly by first extrapolating real outputs from the base year by volume indices respectively by the number
of students or patients. General education and general health services that serve the community at large may
be deflated by general consumer price index. But generally, most countries calculate volume indices using
employment based on persons engaged or hours worked in these areas.

Deflation of other non-market services

11.30. Non-market services are deflated by price indices that are constructed from production costs with
operating surplus assumed to be zero. The cost structures of the use table in basic prices should be used for
this purpose. As can be seen from column 3 of table 11.1, the price index for this column is the weighted price
index of all the elements in the column including components of gross value added at basic prices.

11.31. In constructing a price index, compensation of employees can be deflated by changes in wage rates,
other taxes on production can be similarly deflated as taxes on product and consumption of fixed capital is
obtained automatically when the perpetual inventory method is applied. If the perpetual inventory method is
not used, the same ratios used in the base year should be used to calculate consumption of fixed capital in real
terms.

Deflation of taxes on products

11.32. Tax deflators can be approximated by multiplying the change in tax rate and the price change of the
product taxed if taxes on products like VAT are levied as percentages of the value. A better way to achieve
234

the same thing is to equalize the volume index of the tax and of the product taxed. The method is shown in
table 11.3 below.

Table 11.3. Calculation of tax at constant prices

t t+1 Index (previous year = 100)

Quantity 100 200 200

Price 3 4.5 150

Value 300 900

Tax rate 3% 3.6% 120

Tax collected 9 32.4

Tax deflator = Change in tax rate x change in price 180

Tax collected in constant price = 32.4/1.8 18

or = volume index of product x tax collected at base year = 2 x 9 18

11.33. This approach however is elaborate and very time consuming. A short-cut method can be used instead,
i.e. taxes on products in real terms are calculated by applying the average tax rates TX/SB of the base year
(shown in the supply table 2.1 of chapter II) to both flows of domestic and imported goods and services at basic
prices (shown in the use table 11.1). Taxes on products should be separated into: import duties which are
applied only to imports, and other taxes on products which are applied to both domestic products and imports.
In the revised SNA, taxes on products either imposed directly on producers before sales or on sales are
collected and reorganized by commodities as shown in table 2.1 of chapter II. Subsidies are similarly treated.

Deflation of other components of value added

11.34. Compensation of employees can be deflated by wage rates. Consumption of fixed capital as said above
can be deflated in the process of calculating gross and net capital stocks by the perpetual inventory method.
Operating surplus similar to value added can only be deflated residually, that is:

- Value added at constant price = industry output at constant price minus intermediate
consumption at constant prices;

- Operating surplus at constant price = value added at constant prices minus other components
of value added at constant prices.

(c) Choice of index number formulas

11.35. The two commonly used types of indices are the Laspeyres and the Paasche indices. They are fixed
weighted. The Laspeyres indices are based on the weights of the base year while the Paasche indices are based
on the weights of the current year. The formulas are shown below:
 235

Volume indices
Laspeyres volume index

j p0q0 q
qt
0 j p0q t
j p0q0 j p0q0
'

Fisher volume index ' Laspeyres(Paasche

Paasche volume index

j p tq t j p tq t
j p tq0
'
j p tq t q
q0
t

Price indices

Laspeyres price index

j p0q0
pt
p0 j p1q0
j p0q0 j p0q0
'

Paasche price index

j p tq t j p tq t
j p0q t
'
j p tq t
p0
pt

Fisher price index ' Laspeyres(Paasche

11.36. “In general, a Laspeyres index tends to register a larger increase over time than a Paasche index.”3
“The Laspeyres index provides an upper bound to the theoretic index [and] the Paasche index can be shown
to provide a lower bound to the theoretic index”4 (see columns 2 and 5 in table 11.4). This means that, over

3
System of National Accounts 1993., para. 16.20.

4
Ibid., para. 16.22.
236

time, a Laspeyres tends to produce a higher rate of growth in the GDP. The Fisher index is preferable since
it has been proved that “if the utility function can be represented by a homogeneous quadratic function (which
is homothetic) Fisher's Ideal Index is equal to the underlying theoretic index.5

Table 11.4. Volume growth rate (t/t-1) of GDP of the Netherlands6

Growth rates Laspeyres Laspeyres Fisher (chain Paasche (chain Paasche (weights
(weights of 1986) (chain index) index) index) of 1986)
1987 1.4 1.4 1.4 1.4 1.4
1988 3.4 2.6 2.6 2.6 2.7
1989 4.8 4.7 4.6 4.4 4.6
1990 4.2 4.1 4.0 3.9 3.5
1991 2.3 2.3 2.2 2.2 2.0
1992 2.0 2.0 2.0 2.0 2.0
1993 1.3 0.8 0.8 0.8 0.7

11.37. However, the Fisher index has some drawbacks:

- “[It is] demanding in its data requirements as both Laspeyres and the Paasches indices
have to be calculated, thereby not only increasing costs but also possibly leading to delays
in calculation and publication;

- “[It] is not so easy to understand as Laspeyres or Paasche indices which can be interpreted
simply as measuring the change in the value of a specified basket of goods and services;
-
“The particular preference function for which Fisher provides the exact measure of the
underlying theoretic index is only a special case;

- “The Fisher index is not additively consistent.”7

11.38. The weighted Laspeyres volume index is based on the prices of a fixed base period and thus tends to
become progressively less relevant to the economic situations of the later periods to the point that it becomes
unacceptable, i.e. the rate of growth is overestimated. It may be necessary to update the base period and then
to link the old series to the series of the new base period. When weights are changed every year, the index is
called chained index. Time series results connecting year t to year 0 can be obtained by multiplying separately
estimated year-to-year volume indices:

5
Ibid., para. 16.23.

6
Sake de Boer, Jan van Dalen and Piet Verbiest, “The Use of Chain Indexes in the Netherlands”, Department of
National Accounts, Statistics Netherlands, 1997.

7
System of National Accounts, para. 16.25.
 237

j p0q1 j p1q2 j pn&1qn

j p0q0 j p1q1 j pn&1qn&1
Lq ' x x ... x

Hence the name “chain indices.” The overestimation of growth rates is averted. The chain volume Laspeyres
index can serve as an alternative to the Fisher index. The SNA says, “The chain Laspeyres index should
provide a very close approximation to the chain Fisher in situations in which it is too difficult or time
consuming to calculate the Fisher.”8 A chain index suffers from the lack of additive consistency, i.e. the
deflated total is not the sum of the deflated components or the sum of the share is not equal to unity.

11.39. The Paasche price index is recommended for use together with the Laspeyres volume index to satisfy
the factor reversal test under which the product of the volume index and the price index must be equal to the
ratio of values in current prices.

3. Some important remarks

11.40. The comprehensive double deflation method described above requires full information on industry
and commodity outputs, intermediate consumption by industries, imports, exports, taxes on products and price
indices by detailed commodities. Uses of commodities have to be broken down into values at basic prices,
trade margins and taxes on products with values at basic prices broken down further into domestic and
imported components. If the list of commodities is 500, then all fields of statistics from censuses of production
to foreign trade, surveys of consumer expenditures, government expenditures, and gross capital formation
must be broken down into 500 commodities. The number of activities or industries, however, need not be that
detailed. Many countries classify activities in no more than 100 industries. Normally, the number of
commodities is much higher than the number of industries. The information required normally is available
only for benchmark years and therefore the method cannot be easily applied annually. In countries where it
is it normally takes three years past the year with changes in real terms. Thus there is a need for a quicker
method, to provide provisional annual estimates. This will be discussed in part C below. Final annual estimates
will be prepared when more complete information is available.

11.41. A study of the Austrian table showed that9:

(i) A highly aggregated table would significantly affect bias in constant price values. Deflating
54 products instead of 1,000 domestically produced products and 3,500 imported products
would let estimates of value added at constant prices deviate from the reference solutions
(based on detailed information) by over 10% in 15 cases out of 54 and over 20% in 8 cases;

(ii) A separate deflation of domestically produced and imported products would somewhat reduce
the overestimation or underestimation of value added at constant prices made by deflation
without separation of domestically produced and imported products.

8
Ibid., para. 16.67.

9
Josef Richter, “The Role of Import-Matrices in Compiling Constant Price Input-Output Tables” in Problems of
Compilation of Input-Output Tables, edited by Afred Franz and Norbert Rainer, Orac, Vienna, 1986.
238

This confirms the advantages of deflating at the detailed commodity level and of separation of domestically
produced products from imported products.

11.42. In addition, the method shows that the implicit price index for household final expenditures may not
be the same as the consumer price indices (CPI) that are compiled independently. The differences are due to
three factors:

- Basic price indices, not the CPI, are used to calculate the implicit price index for household
final expenditures;

- Some components of expenditures are covered by the SNA household final expenditures but
not by the CPI, namely owner-occupied housing, production for own final consumption,
financial intermediary service charges;

- Normal statistical errors.

But the CPI provides information to check the implicit price indices of household final consumption
expenditures at purchasers' prices at both the aggregate and detailed commodity level. For the commodity level,
implicit indices that combine price indices of goods and services, trade and transport margins, and product
taxes consumed must be used.

11.43. The SNA implicit price index should reflect more accurately cost increases in the household sector.
The information provided by the SNA and CPI is useful for policy analysis.

11.44. Similarly to the CPI, there might also be deviations of price indices calculated within the SNA
framework from those used elsewhere. For instance, price indices for imports and exports in the SNA are also
more comprehensive since they cover both goods and services, including both expenditures by residents abroad
and expenditures of non-residents at home. Checking and adjusting implicit price indices for exports f.o.b.,
and prices against unit price based on data of external trade statistics should also be done.

4. Alternative procedures

11.45. Some countries apply the double deflation method with a different procedure. The deflation process
is applied directly to intermediate consumption and final demand in purchasers' prices in table 2.2 of chapter
II, without separating imports from domestically produced products and without separating trade and transport
margins as well as taxes from basic values. Appropriate indices of purchasers' prices are then used for
deflation, which may include indices of import prices, export prices, consumer prices, prices of investment
goods, etc. This method is simpler since it does not require assumptions for separation. However, it may
generate biased estimates especially when imports are not separated from domestically produced goods as
noted in paragraph 11.40. The alternative procedures may be applied to calculate industry value added and
GDP at constant prices, but to calculate the I/O table at constant prices, it is necessary that assumptions be used
to create margins and to separate imports from domestically produced products. Nevertheless, more research
is still needed to study appropriate procedures to apply the double deflation method.

C. Double deflation method for annual national accounts

 239

11.46. For annual derivation of GDP, gross value added by industries and components of final demand in
constant prices, the above elaborate procedure cannot be applied because, normally, the following information
is not available:

- Intermediate consumption by kind of commodities (at best the information available is total
intermediate consumption over output for the corporate sector collected by annual surveys in
many countries);

- Household final expenditures (at best only information on retail trade is available by survey
which can be used to estimate household final consumption).

11.47. The information that is normally available in current prices includes:

- Gross output by industries;

- Imports by commodities;
- Exports by commodities;
- Gross capital formation by institutional sectors;
- Government final consumption expenditures.

It is necessary to estimate the supply and use tables in current prices on the basis of the above information and
the supply and use tables of the benchmark year before one can proceed with the deflation process as discussed
above. The benchmark year can also be the preceding year, when annual SUT tables are established on a firm
basis.

1. Estimation of main aggregates in current prices

11.48. The value added coefficients of the base year are needed to estimate current value added by industries.
In addition, gross value added at current prices from the corporate sector, the general government sector and
non-profit sectors may be estimated independently by taking the sum of compensation of employees, taxes on
production, consumption of fixed capital and profits plus dividends and interests paid out minus dividends and
interests received, which is operating surplus. The latter estimates may be used to adjust the estimates
previously obtained by using the base year value-added coefficients for the corporate sector. With all the total
aggregates available, the next step is to break down final demand aggregates into commodity components.
Here, it is necessary to use both available information and the coefficients available in the base year. The RAS
method which tries to estimate the components of an aggregate given the value of the aggregate, the shares of
the components in the base year and possibly values of some components can also be used here. The RAS
method was fully described in chapter IX.

2. Estimation of the supply table in current prices

11.49. The estimation of the supply table in current prices is in fact that of the total value of each commodity
in current prices produced in the year. The application of the product-mix assumption to industry outputs will
produce these estimates. This means that for the goods industry for instance the ratios X11/I1 are multiplied
to goods industry output to estimate every commodity produced by the goods industry.

3. Estimation of the intermediate consumption in current prices

240

11.50. The estimation of the matrix of intermediate consumption at current prices for table 11.1 is
implemented by the RAS method given value added by industries and final demand and commodity outputs
obtained from the supply table in current prices in step 2.

4. Estimation of the use table in constant prices

11.51. The estimation of the use table in constant prices proceeds as described in section B.

5. Deflation process

11.52. With the supply and use table available in current prices, one is able to proceed with the deflation
process as described in section B.

6. Special problem in the annual method

11.53. The comprehensive benchmark method and the annual method focus on estimating value added by
industries and components of final demand in current prices before deriving these values in constant prices.
Normally, annual surveys on various areas of statistics which national accountants have to rely on to get good
estimates come late, always after the end of the year. But there is always a need to have estimates by the end
of the year. To respond to this need, national accountants have to use monthly and quarterly statistics and rely
on the relationships of the last quarter to the previous three of the previous year to blow up the values of the
three quarters to the full year.

11.54. In many cases, they have to rely on quantity indices of production for some industries. The procedure
of estimating gross outputs is then applied in the reverse direction as previously discussed. In such cases, some
quantity index has to be used to estimate an industry output in constant prices first. The output in current
prices is obtained later. For agricultural production, quantities of products are always estimated first, then
multiplied by current prices to get outputs in current prices. In other cases when industry output is a composite
of many commodities, quantity indices are applied to previous year values and then inflated by constructed
price indices for the industry. The constructed price indices of the industry outputs are the weighted price
indices of the commodities produced by the industries. Product-mix shares of the base year have to be used
as weights.

11.55. In any case, the appropriate method is to obtain values in current prices so that they can be balanced
in the SUT framework with other values which are readily available in current prices.

7. Concluding remarks

11.56. Estimation of annual quantity and price indices should be as detailed as possible. However, the
process is normally time-consuming, particularly with regard to the availability of many annual basic statistics
long after the end of the year. It is therefore advisable to proceed by using the use table at a more aggregated
level for preliminary estimates, possibly at a 30-commodity level or less. Revised estimates will be based on
the detailed supply and use tables in current prices updated from the benchmark year.

11.57. Though the method provides real value added by detailed industries and all components of final
demand by detailed commodities, it may also produce a negative value added at the detailed level, especially
when the base year is too far away from the current year during which significant technical changes occurred.
242

XII. IMPACT ANALYSIS

12.1. Normally, policy makers want to see how a particular industry will evolve in the future so that they
can plan accordingly. They also want to see the importance of that industry in the economy in terms of how
much employment, income and taxes it generates and also what capital and imports it needs to grow. Thus,
impact analyses are usually focused in both directions: the impact of other activities on the industry under study
or the impact of that industry on other industries. This chapter will present various ways of modelling impact
analysis using input-output tables, including multipliers, backward and forward linkages as well as the use of
the Social Accounting Matrix (SAM).

A. Full-fledged impact analysis

1. Basic impact equation

12.2. An input-output model in its simplest form is a full articulation of inter-activity analysis; the
development of any single industry is interrelated to the development of all other industries. Thus, the right
approach to study an industry in both directions is first to predict the path of growth (or if prediction is not
possible, to look at different probable paths of growth) of the full vector of final demand, i.e. the growth of
final demand of every product in the economy and then use the following basic impact equation to calculate
total impacts on gross outputs:

(12.1) )X ' (I&A d)&1 )Y d

or

(12.2) )X ' C d)Y d

where Cd = (I-Ad)-1

12.3. It is important to further clarify the concepts used in equation 12.1 which have been discussed in
chapter X. The basic input-output relations are as follows:

(12.3) X = (I-A)-1 Y

where

Y=S+E-M

Y: The column vector of final demand

S: The column vector of final expenditures (i.e. final consumption expenditures of
households, NPISHs and general government, gross capital formation)
E: The column vector of exports
M: The column vector of total imports.
 243

As AX, the intermediate inputs, can be split into two components: one supplied domestically (AdX) and the
other supplied from abroad (Mid), equation 12.3 can be rewritten as

(12.4) X = (Ad X + Mid )+ Y

or

(12.5) X = Ad X + (S + E - M + Mid)

When the following are given

Y =S+E-M

Yd = S + E - (M - Mid)

equation 12.1 is obtained. This equation can be applied to the level of Yd or a change in Yd. (M-Mid) is
imports for final expenditures. ªYd is change in final demand. Ad stands for the domestic input coefficients.
Its calculation is explained in chapter VI. Yd stands for final demand of domestic goods and services.

2. Preparation of the initial impact vector

12.4. ªYd is the initial impact vector that is used to calculate total impacts on other industries in the
economy. It stands for a change in final demand of domestic goods and services. Final demand of domestic
goods and services is understood to be final consumption expenditures plus exports and minus imports for final
expenditures, i.e.:

Final expenditures (S)

.Household final consumption expenditures

.Final consumption expenditures of NPISHs
.Government final consumption expenditures
.Gross capital formation

(+) Exports (E)

(-) Imports for final expenditures (M - Mid)

A full ªYd has to be prepared because change in the output of a given industry is caused by change in final
demand not only for its product but also for the outputs of all other industries.

12.5. The result obtained by equation 12.2 can then be used to compute total impacts on the total economy.

3. Calculation of total impacts

12.6. The total impacts on various economic indicators such as value added, employment, capital stocks and
imports are further results of the calculation of total impacts on outputs.

Total impact on outputs

244

12.7. ªX is the total impact on outputs caused by the change in final demand of domestic goods and
services. It is calculated by equation 12.1.

Total impacts on value added

(12.6) ªV = v ªX

Total impacts on employment

(12.7) ªL = l ªX

Total impacts on capital formation

(12.8) ªK = k ªX

where:

ªV is the change in value added generated by the change in Y as defined in para. 12.3, v is the row vector of
value added and coefficients (i.e. value added per unit of output of each industry);

ªL is the change in employment generated by the change in Y, l is the row vector of employment coefficients
(i.e. employment -- hours worked -- per unit of output of each industry);

ªK is the change in fixed capital stocks required to satisfy the new level of final demand and k is the row vector
of total stock of produced and even non-produced assets (such as land) per unit of output of each industry if all
assets are summed together.

Vectors v, l and k may be replaced by matrices if value added, labour and capital can be split into types. For
example value added includes compensation of employees, operating surplus, consumption of fixed capital
and taxes on products and production (indirect taxes); labour can be divided into managers, skilled technicians
and the unskilled. For studying capital formation, some adjustment to the SNA concept may be needed. In the
SNA, capital stock is linked to the economic activity or industry that owns the capital. For this reason,
investment in roads, ports, airports and irrigation networks constructed and owned by the Government may
be linked to the activities producing government services instead of to the economic activities that benefit
directly from them such as the transport or agricultural industry. Analysts may need to reclassify capital stocks
to the appropriate industries which benefit.

12.8. Table 12.3 gives an example of an increase in final expenditures in the economy of Thailand which
is listed in column 1. On the assumption of domestic content listed in column 2, final expenditure of domestic
goods and services (ªYd) is calculated and listed in column 3. Column 4 shows the increase in outputs
calculated by equation 12.2. Column 5 shows the increase in intermediate inputs supplied by imports (ªMid)
calculated by (A - Ad)ªX. ªMy which is imports for final expenditures is the difference between column 1 and
column 3. Total impacts on imports or the sum of ªMid and ªMy are shown in column 6. Column 7 shows the
percentage of imports over outputs, industry by industry. The input-output coefficient matrix and the domestic
coefficient matrix shown in tables 12.1 and 12.2 provide information used in the calculation. In table 12.4,
total impacts on value added and its components are shown. No data on capital stocks by industry are available
so that increases in capital stocks can only be calculated in a manner similar to that of value added. Total
impacts on employment can also be similarly calculated.
 245

Table 12.1. Table of input-output coefficients in Thailand, 1985

1 2 3 4 5 6

1.. Agr. and mining .0984 .2233 .0276 .0000 .0001 .0213

2. Manufacturing .1676 .3295 .4342 .0536 .4302 .1781

3. Construction .0015 .0013 .0006 .0018 .0009 .0083

4. Trade .0260 .0450 .0545 .0054 .0213 .0339

5. Transport .0132 .0238 .1118 .0240 .0483 .0181

6. Others .0227 .0210 .0344 .1104 .0395 .0398

Compensation .1150 .0943 .1195 .2144 .1759 .3688

Operating surplus .4939 .1747 .1717 .5299 .1921 .2071

Consumption of fixed capital .0387 .0309 .0252 .0464 .0703 .0957

Taxes on production .0230 .0562 .0206 .0141 .0214 .0289

Input multipliers 1.646 2.263 2.402 1.363 2.173 1.608

Table 12.2. Table of domestic coefficients in Thailand, 1985

1 2 3 4 5 6

1. Agr. and mining .0980 .1691 .0265 .0000 .0001 .0207

2. Manufacturing .1250 .2243 .3238 .0487 .2805 .1610

3. Construction .0015 .0013 .0006 .0018 .0009 .0083

4. Trade .0260 .0450 .0545 .0054 .0213 .0339

5. Transport .0132 .0238 .1118 .0240 .0242 .0181

6. Others .0153 .0204 .0338 .1087 .0333 .0363

Compensation of employees .1150 .0943 .1195 .2144 .1759 .3688

Operating surplus .4939 .1747 .1717 .5299 .1921 .2071

Consumption of fixed capital .0387 .0309 .0252 .0464 .0703 .0957

Taxes on production .0230 .0562 .0206 .0141 .0214 .0289

Domestic output multipliers 1.443 1.770 1.912 1.294 1.613 1.457

246

Table 12.3. Initial impacts and total impacts

1 2 3 4 5 6 7

ªY Domestic ªYd ªX ªMid=(A-Ad)ªX ªM=ªMid+ªMy ªM/ªX

ratio

[Link]. and mining 3,350 95.9% 3,197 13,920 2,797 2,935 21.1%

[Link] 29,485 82.3% 24,278 51,109 8,501 13,708 26.8%

[Link] 6,743 100.0% 6,743 7,020 0 0 0.0%

[Link] 6,361 100.0% 6,361 10,300 0 0 0.0%

[Link] 6,227 96.7% 6,021 9,230 222 428 4.6%

[Link] 15,388 95.5% 14,739 18,995 281 929 4.9%

Total 67,554 61,336 110,574 11,796 17,994 16.3%

Table 12.4. Increases in value added

due to an initial increase in final expenditures shown in table 12.3.

1 2 3 4 5 6 Total

Compensation 1,600 4,822 839 2,208 1,624 7,005 18,098

Operating surplus 6,876 8,928 1,205 5,458 1,773 3,934 28,174

Consumption of fixed capital 538 1,577 177 478 648 1,817 5,235

Taxes on production 320 2,871 145 145 198 550 4,229

Total value added 9,334 18,198 2,366 8,289 4,243 13,306 55,736

4. Endogenization of household income effects

12.9. The total impacts shown in sections 1, 2 and 3 require that every component of final demand,
including household final consumption expenditure be estimated exogenously. Personal consumption
expenditure and its corresponding counterpart, value added, can be endogenized in the basic equation 12.1.
Table 12.5 will show how the household sector can be endogenized in a simple manner, using value added as
a proxy for household incomes. A more elaborate treatment of the household sector is the focus of the social
accounting matrix (SAM) technique that will be introduced in section C.

12.10. In table 12.5, the normal intermediate matrix has been extended into three rows and three columns,
that is, it includes an additional row for value added and an additional column of household or personal final
consumption expenditures (PCE). Value added becomes the source of revenue to finance final expenditures
of the household sector. Since not only is household disposable income normally smaller than value added
but also its total expenditures are normally smaller than its disposable income, the PCE coefficients must be
derived by dividing values in the PCE column by 220 (total value added). This derivation implicitly assumes
that (i) household income is a fixed percentage of the part of value added that is subject to distribution and
redistribution to the household sector (wages, dividends, net property income, net transfers), (ii) a fixed
percentage of household income is saved, (iii) the income left for consumption is spent on various goods and
services in fixed shares.
Table 12.5. Endogenization of the household sector
 247

Intermediate consumption PCE Other final Total

demand (FD)

1 2 3

Product (1) 10 20 35.1 34.9 100

Product (2) 20 30 58.5 91.5 200

Value added (3) 70 150 0 0 220

Total 100 200 93.6

12.11. Table 12.5 shows that the intermediate demand matrix includes three rows and three columns, ªYd
used in the basic equation 12.1 is the column “other final demand (FD)” with the last element equal to zero.
Thus, the extended A matrix is equal to

.100 .100 .159

A ' .200 .150 .266
.700 .750 .000

12.12. The endogenization of the household sector produces both advantages and disadvantages. On the
advantage side, it is not necessary to estimate directly household income and consumption for a given impact
study. They are the results of the calculation. On the disadvantage side, in addition to the assumption of
constant technical coefficients, it is necessary to make additional assumptions: the assumption of constant
consumption behaviour, constant income distribution shares and constant saving behaviour which are not
normally true even in the short term at a detailed level of input-output analysis.
12.13. When the household sector is endogenized, the initial impact vector ªYd would not include estimates
of household final consumption expenditures. These values will be automatically generated through the use
of value added that is subject to distribution and redistribution to the household sector. The economy is now
driven mainly by government final consumption expenditures, exports, changes in inventories, and gross
capital formation.

12.14. The above example shows the endogenization of only the household sector in a very simple way. In
order to endogenize additional sectors, for example the government sector, it is necessary to separate value
added into household incomes and government incomes. This will be discussed in section C.

5. Conclusion

12.15. To summarize the discussion of section A, it is important to re-emphasize that a study of any industry
in the economy must be implemented comprehensively, i.e. together with every other industry because of the
inter-industrial linkages.

12.16. In some studies, some economists enter the output of the industry under study into the vector of final
demand Y and calculate total impacts. The results obtained by using equation 12.2 are obviously much greater
than the total contribution of the industry. For example, if the flow matrix in table 1.2, the coefficient matrix
248

in table 1.3 and the inverse in table 1.6 of chapter I are used, and the vector of final demand as (100 0 0) is
entered, total impacts on outputs will be (107.7 35.1 14.1). This is nonsensical because the total output of
sector 1 in the economy is only 100 while the estimated output is 107.7. This type of analysis is similar to the
use of output multipliers that will be discussed in section B.

12.17. In order to clarify further the point made above, assume that a new level of output of a certain product,
for instance X2, would be obtained in the future and that one would want to see the impact of this output on
the economy. Let us call the vector of other products X1; then the basic I/O equation can be written as follows:

X = AX + Y
or
X1 A11 A12 X1 Y1
(12.9) ' %
X2 A21 A22 X2 Y2

Equation 12.9 shows that Y1 and X2 are known. Thus unknown variables are X1 and Y2.

(12.10) X1 ' A11X1 % [A12X2 % Y1]

(12.11) Y2 ' X2 & [A21X1 % A22X2]

In equation 12.10, X1 is computed given X2 and Y1. In equation 12.11, given that X1 and X2 are known, Y2
is calculated as a residual. This means that if X2 is oversupplied, the excess must be exported or kept in
inventory. These two equations can also be used to calculate the impacts of changes in X2 and Y1 on X1 and
Y2.

12.18. Thus it is not possible to isolate the total impact of a particular industry on the whole economy when
calculating its total impacts. Impact analysis of an economic problem would make sense only if it can be
formulated consistently as changes in the vector of final demand, i.e. changes in final consumption, exports,
fixed capital formation (i.e. investment on capital goods) and inventories.

12.19. The impact analysis described above requires an articulation of the initial impact vector which is
described in equation 12.5. It looks as though the full initial impact vector of the entire economy always has
to be formulated. It is not so, however. For instance, the total impact of a project on the economy can be
consistently formulated since change in final demand is only a vector of changes in capital formation (i.e.
investments). Similarly, effects of changes in exports on the economy can also be calculated.

12.20. In addition, for any type of impact analysis, it is important to differentiate short-term from long-term
effects. Increase in fixed capital investment is mostly short-term, the effects will disappear when production
of goods and services for the investment project stops. The long-term effects would be replacement investment,
maintenance and operation expenses to sustain the new production activities created by the project. How
would that be introduced in the model described in equation 12.2? The answer is that an increase in
replacement investment can be formulated as an increase in capital formation and an increase in maintenance
and operation expenses can be formulated in terms of an increase in final demand that directly and indirectly
requires the incremental level of maintenance and operation expenses.
 249

B. Other types of impact analysis

12.21. Other types of impact analysis are also used in input-output economics. These techniques are more
ad hoc as compared to the method described in section A. The more prominent ones are multipliers and
backward and forward linkages.

1. Multipliers

12.22. Multipliers measure the total effects on either output, employment or value added, given an increase
in one unit of output of a particular industry. The multiplier of an industry can be calculated by equation 12.2
to obtain ªX and then sum it up by using the final demand vector Yd in which only the value of final demand
of that industry product is set to 1 with the rest set to zero. But this produces the same result as calculating
the sum of the column corresponding to that industry of the domestic Leontief inverse Cd. This sum is called
the output multiplier. Column sums of the Leontief inverse C summarize total inputs required and thus are
called input multipliers. Input multipliers and output multipliers for each industry for the Thai economy in
1985 are shown at the bottom of tables 12.1 and 12.2. One can see that output multipliers should be smaller
than input multipliers because imported inputs would not have any effects on the economy. Multipliers are
mathematically defined as follows:

(12.12) Input multiplier of industry j ' j C

n

i
ij

(12.13) Output multiplier of industry j ' j C

n
d
i
ij

12.23. Multipliers in table 12.2 show that in Thailand in 1985, one unit of output of construction (industry
3) created more total outputs than any other sectors and that the output multiplier of agriculture and mining
(industry 1) is smaller than the output multiplier of "other" industry (industry 6) though potentially it may have
more impact (look at table 12.1 to see that the input multiplier of agriculture and mining is larger than that of
other industry) because of imports of inputs.

12.24 Other types of multipliers such as income and employment can also be calculated as follows:

Income multipliers

(12.14) Income multiplier of industry J ' j vi.C

n
d
ij i

Employment multipliers

(12.15) Employment multiplier of industry J ' j li.C

n
d
ij i

12.25. These multipliers calculated using the domestic Leontief inverse Cd are interpreted in the same way
as output multipliers. They show total increases in either income or employment given an increase in one unit
of output of each industry.
250

12.26. In an ad hoc sense, multipliers can be used as useful indicators to assess variation in effects for a
particular activity in microanalysis such as at the regional level. However, if multipliers are used as criteria for
development for the whole economy, there is a tendency to conclude (depending on the objective of the society
for either highest total output, highest total income or highest total employment) that the industry with the
highest multiplier, and that industry alone, will be selected for development. This conclusion tends to lead to
the misuse of multipliers. This type of analysis is similar to the calculation of total economic impacts of an
industry by using its output as final demand as analysed in section A. Any meaningful long-term analysis will
have to start with the formulation of the vector of final demand.

12.27. Finally, in relation to the use of income multipliers, it is noteworthy that all value-added multipliers
are the same, i.e. all equal to 1 if imports for intermediate goods and services are not eliminated from the I/O
coefficients. This is so because equation 12.14 is the same as the price equation discussed in chapter I (see
equation 1.16, para. 1.30). The "value-added" multipliers can be interpreted as follows: for a one-dollar
increase in value of final demand of any domestic product, the value added will increase by the value of 1 in
order to provide enough income for the final demand. In other words, a final demand of one dollar of any
domestic product, must be matched by a one-dollar increase in income. Given that only part of value added
is paid out as labour income with part of this labour income used for final consumption, and given again that
imports of goods and services are not eliminated from the I/O coefficients, one may calculate income
multipliers by endogenizing the household sector as in paragraphs 12.9-12.13.1 Again, all industries will have
the same income multiplier which is greater than 1 by a scalar. For example, the income multipliers of every
industry in matrix A in equation 12.14 are equal to each other and to 1.7405, which are elements of the third
row in the Leontief inverse. This means that for an increase in final consumption of one dollar, income must
increase by 1.7405 since only a part of income is spent on final consumer goods and services. Readers may
want to make up some examples of the A matrix and then proceed to calculate income multipliers in order to
check the above conclusion. Thus, the discussion here tries to convey the main idea that without introducing
imports of intermediate inputs into equation 12.14, income multipliers would not be meaningful for industry
analysis.

2. Backward and forward linkages

12.28. The terms “backward and forward linkages” of an industry are meant to measure the inter-industrial
linkages of a particular industry to other industries as suppliers of inputs (backward linkage) and as a provider
of input to other industries. These concepts are developed in the context of studying development strategy.
Many authors2 have proposed definitions as well as methods to measure these linkages.

12.29. The direct backward linkage of an industry is measured by the sum of the column corresponding to
that industry in the coefficient matrix A. The total backward linkage is measured by the sum of the industry
column in the Leontief inverse.

1
Readers can see the proof of this in Thijs ten Raa, Linear Analysis of Competitive Economies, Hemel Hempstead,
Harvester Wheatsheaf, 1995.

2
See P.N. Rasmussen, Studies in Sectoral Relations, Amsterdam, North Holland, 1956; Albert O. Hirshman, The
Strategy of Economic Development, New Haven, Yale University Press, 1958; Hollis B. Chenery and T. Watanabe,
“International Comparisons of the Structure of Production” in Econometrica, Vol. 5, No. 2, October 1958; Pan A.
Yotopoulos and J. B. Nugent, “A Balanced-Growth Version of the Linkage Hypothesis: A Test” in Quarterly Journal
of Economics, Vol. 90, No. 2, May 1973; Leroy P. Jones, “The Measurement of Hirshmanian Linkages”, ibid.
 251

12.30. The direct forward linkage is measured by the sum of the industry row in the coefficient matrix A.
The total forward linkage is measured by the sum of the industry row in the Leontief inverse.

12.31. The linkages also differ in whether A, Ad, C, Cd are used. The meaning of each concept derived by
using different types of data should be apparent to readers of part A of this chapter. Of course, all these
quantified concepts suffer from the same weakness of the multipliers since total backward linkages are nothing
but output multipliers.

C. The System of National Accounts (SNA) and Social Accounting Matrix (SAM) as an extension
of I/O model

12.32. In part B, the endogenization of the household sector does not explicitly show household incomes. The
extension of the input-output framework in the SNA would allow us to show how incomes of the household,
government and other sectors are formed.

12.33. As noted in chapter II, household incomes are not the same as compensation of employees. Household
incomes can be obtained only after the process of distribution and redistribution of incomes. For example, the
household sector receives, in addition to compensation of employees, dividends, interest on its deposits and
transfers from other sectors like social security benefits but it also has to pay out income taxes, interest on
loans, social security contributions, fines and penalties and other types of transfers. The net receipts form the
disposable income. Disposable incomes of other sectors are formed in a similar way. All these distributions
and redistributions can be modelled by extending the input-output framework to incorporate this type of
distribution and redistribution. The usefulness of this extension has already been discussed in part B. Table
12.6 will show an example of how the process of distribution and redistribution of incomes is modelled.
12.34. It is possible to observe from table 12.6 that the full input-output matrix is spelled out, including every
component of value added from column 1 to column 2 and from row 1 to row 5.

12.35. Columns 3-5 show the appropriation of value added shown in rows 3-5 by institutional sectors.
Column 3 shows how compensation of employees (130) generated in the production process in row 3 is
appropriated by the household sector. Column 4 shows taxes on products and production being all appropriated
by the Government. Column 5 shows how consumption of fixed capital and operating surplus are appropriated
by the household sector as mixed incomes and by other financial and non-financial sectors (i.e. other sectors)
before being distributed and redistributed.

12.36. Rows 6-8 show the receipts appropriated by various institutions before being distributed and
redistributed.

12.37. Columns 6-8 show the net distribution and redistribution of receipts appropriated in rows 6-8. Column
6 shows how the receipts received by the household sector (140) are paid to the Government as direct taxes
and contribution to social security (60), and to "other sectors" as interest payments, current transfers, etc.
(10), with the residual (70) as disposable income. Column 7 shows the government receipts paid out to the
household sector (30) as social security benefits, etc. and to "other sectors" (5) as subsidies, grants. Column
8 shows the net distribution of "other sectors" to the household sector (17) as interest, dividends, etc., to the
government sector (3) as business income taxes, and to "others" (30) as interest, dividends, or retained
earnings.

Table 12.6 . Endogenization of the household sector

252

1 2 3 4 5 6 7 8 9 10 11
PCE Other Total
FD

1. Industry 1 10 20 35.1 34.9 100

2. Industry 2 20 30 58.5 91.5 200

3. Compensation of employees 40 90 0 130

4. Taxes on products and 10 0 0 30

production 20

5. Other value added 20 40 0 60

6. Household 130 10 0 140

7. Government 30 60 3 0 93

8. Others 50 0 50

9. Household incomes 70 30 17 0 117

[Link] incomes 58

[Link]' incomes 10 5 30

Total 100 200 130 30 60 140 93 50 93.6

Note: PCE = personal final consumption expenditure

FD = final demand.

12.38. Row 9 shows net incomes received or retained by the household sector. The total disposable income
is equal to 117. Corresponding to this row is column 9 showing the consumption expenditures of the
household sector, which are equal to 93.6 only. This amount is smaller than the total disposable income, part
of which is saved.

12.39 The endogenization of the household sector will turn the matrix between columns 1-9 and rows 1-9
into the intermediate flow matrix F. The corresponding output vector that is needed to convert the intermediate
matrix F into the coefficient matrix A is X = (100 200 130 30 60 140 93 50 117), the other final demand matrix
is Y = (34.9 91.5 0 0 0 0 0 0 0) and the extended A matrix is

.10 .10 .000 .00 .000 .000 .000 .000 .300

.20 .15 .000 .00 .000 .000 .000 .000 .500
.40 .45 .000 .00 .000 .000 .000 .000 .000
.10 .10 .000 .00 .000 .000 .000 .000 .000
A ' .20 .20 .000 .00 .000 .000 .000 .000 .000
.00 .00 1.000 .00 .167 .000 .000 .000 .000
.00 .00 .000 1.00 .000 .428 .000 .060 .000
.00 .00 .000 .00 .833 .000 .000 .000 .000
.00 .00 .000 .00 .000 .500 .322 .340 .000

It is possible to apply equation 12.2 to find X given Y but of course Ad is needed.

 253

12.40. In table 12.6, the government sector is not internalized, because if it was, it would be necessary to add
an additional corresponding column of final consumption expenditures for this sector.

12.41. The attempt to trace the formation of institutional incomes presented above is part of the SNA
framework which is actually further extended to trace also the distribution of savings and the uses of these
funds to finance investments in both financial and non-financial instruments, which form the capital and
finance accounts. The Social Accounting Matrix (SAM) is the extension of the SNA system where the
household sector is split into more sectors (high, low incomes, etc.). SAM may be extended further than the
example given in table 12.6 to include all capital accounts and financial accounts. Readers are advised to read
chapter XX of the SNA for a full discussion of SAM.

12.42. One may be tempted to think about the possibility of endogenizing every component of final demand,
but this would in fact lead us to the problem of optimal growth that is beyond the scope of this study.
However, it is already possible to perceive the limitation of this approach since it assumes that everything is
determined in a "fixed" manner, including economic behaviours. The linking of business incomes to capital
formation is more questionable than the linking of household disposable incomes to household final
consumption expenditures in a "fixed" manner because investment as discovered by Keynes is quite volatile.
Many general equilibrium models have relaxed this "fixed" relationship by the introduction of prices and other
effects. Other coefficients such as input-output coefficients and import coefficients that are assumed fixed can
also be relaxed. This is particularly true for import coefficients that were discussed in chapter VI.
254

XIII. AN ANALYTICAL APPROACH TO THE CALCULATION

OF THE "GREEN GDP"1

13.1. The input-output model is one widely used tool in analysing pollution problems. This chapter is aimed
at showing how I/O modelling may be utilized. In fact, there are many alternative ways in modelling pollution,
depending on the objectives of the model and data availability. The model presented in this chapter aims at
showing how I/O can be used to answer a “what if” question on the impact of pollution abatement on GDP.
The chapter does not in any way imply a recommendation since there currently are many models, from the very
simple to the highly sophisticated, that have been experimented with.2 The framework in this chapter
extends the simple I/O model to quantify explicitly pollution abatement activities, the information on which
at this time is lacking in almost every country. If these activities are not explicitly shown, the model would
be similar to many models commonly applied. In this chapter an "analytical concept" of Green GDP is
introduced. It is defined by deducting from GDP the cost of abatement activities projected in a "what if"
scenario, to avoid or eliminate degradation, as well the indirect feedback of these abatement activities including
the pollution generated by abatement activities themselves and also the additional value added created by the
latter. It differs from the "accounting concept" of Green GDP which is defined in the SNA Handbook on
Integrated Environmental and Economic Accounting (SEEA) (Sales No. [Link], 1993) and in other
environmental accounts systems, and which is calculated as the difference between GDP and the maintenance
cost value of degradation. For the above reasons, the analytical concept of GDP should be looked at as no
more than an exercise in using the I/O model.

13.2. The accounting concept of Green GDP is calculated for present accounting periods, the analytical
concept of Green GDP is computed for “what if” scenarios describing the future. Also, the analytical concept
cannot be standardized in the same manner as the accounting concept in SEEA. The value of the analytical
concept of Green GDP would depend on the changes that are modelled to take place in the future as a result
of shifts in resource uses to abatement activities; and those shifts are different, depending on the modelling
scenarios that are used.

13.3. The analytical concept of Green GDP is developed in this chapter not to replace but rather to
supplement the accounting concept of Green GDP. The differences between the two might be a measure of
the extent to which the economy would have to change in order to adapt to the shift in economic activities
needed to abate pollution in the future. Such adaptation might be considerable where pollution caused by
industries is severe and the resources available for abatement activities are limited. On the other hand, if
pollution is limited and many abatement activities have already taken place, the analytical and accounting
concepts of Green GDP might converge.

13.4. When in the remaining part of the chapter the term Green GDP is used, it refers to the analytical
concept of Green GDP. No further references will be made to the accounting concept as defined in the SEEA.

1
This chapter is a slight revision of the paper written by Viet Vu and Jan Van Tongeren for the second meeting of
the London Group on National Resources and Environment Accounting, Washington, D.C., 15-17 March 1995.

2
Readers can see references to some of the work done in this field at the end of the chapter.
 255

13.5. The analytical concept used in the chapter is mainly based on I/O relations and restrictions included
in the national accounts. With help of the I/O focus, the implications of the approach are illustrated. The I/O
approach should be considered as a first step in a more extended modelling exercise which will ultimately also
include income and financial restrictions on abatement activities that are reflected in the institutional sector
accounts of the SNA. The present chapter should therefore be considered as a report on work in progress.

13.6. The chapter is divided into four sections: The first section describes an input-output framework that
is adapted to the calculation of the Green GDP. The second section presents a simple I/O model in which the
Green GDP only allows for pollution generated by production activities, and all pollution is assumed to be
abated. The third section extends the I/O model to deal with pollution generated by household consumption,
and furthermore allows for partial abatement of pollutants generated by production and household consumption
beyond a socially tolerated level of pollution. Section four is exploratory, in the sense that it goes beyond the
scope of the I/O model and supply and use table (SUT) of the SNA and enters into elements that form part of
the Integrated Economic Accounts (IEA) of the SNA. These deal with payments for pollution abatement
through taxes and subsidies, and with saving and other elements financing investments in pollution abatement
equipment.

A. Adaptation of the I/O framework to the Green GDP

13.7. Below is presented an adapted version of an I/O model in which the Green GDP is defined as the
GDP that takes account of pollution abatement activities needed to reduce pollution generated within the
current accounting period below a socially accepted level. This definition excludes the effects on GDP of
abating pollution accumulated during previous periods. The model also assumes that there exists technology
that can abate all pollutants generated by production and consumption activities in the economy.

13.8. Two types of activities are distinguished in the model: all economic activities except pollution
abatement activities, and pollution abatement activities. The first group is referred to as economic activities
and the second group is called abatement activities. Inputs and outputs of economic activities are measured
in monetary terms. The output of the abatement activities is the services that result in elimination of the
pollutants generated by production and consumption.

13.9. The two I/O models below are based on n economic activities (excluding pollution abatement
activities), m pollutants and m pollution abatement activities. These define the following vectors and matrices:

A1 : the n x n direct input coefficient matrix of economic activities, presenting the inputs of
products from economic activities (monetary units) per unit (monetary unit) of output of
economic activities
X1 : the n x 1 vector of outputs of economic activities measured in monetary terms
Y1 : the n x 1 vector of final uses of products from economic activities, measured in monetary
terms
Xg : the m x 1 vector of outputs of pollution abatement activities (=pollutants abated) measured
in physical units
g1 : the m x n direct pollution coefficient matrix of economic activities, presenting the amount of
pollutants (physical units) generated per unit of output (monetary units) produced by
economic activities
g2 : the m x m direct pollution coefficient matrix of abatement activities, presenting the amount
of pollutants (physical units) generated per unit (physical unit) of pollutant abated (i.e. per
unit of output of pollution abatement activities)
256

g3 : the m x n direct pollution coefficient matrix of final uses, presenting the amount of pollutants
(physical units) generated per unit of product (monetary units) consumed by households
and/or otherwise used in final demand
H : the n x m direct input coefficient matrix of pollution abatement activities, presenting the
inputs of products from economic activities (monetary units) per unit (physical unit) of
pollutant abated (i.e. per unit of output of pollution abatement activities)
Yg : the 1 x m vector of pollutants unabated.

13.10. In defining above the vectors of pollution generated in economic activities, and the vectors and
matrices of pollution abatement activities, both the pollutants generated as well as the pollution abatement
services are expressed in physical terms, i.e grams, litres, etc. This type of mixed monetary-physical valuation,
which has become standard in environmental input-output analysis (see ref. 2-7), has been applied to the two
I/O models in sections B and CI. In section D, which deals with value added and incomes in monetary terms,
the physical units are replaced by the maintenance cost valuation of the SEEA.

13.11. Using the terminology defined above, two identities are defined that are closely related to each other:

(13.1) X1 = AX1 + HXg + Y1

(13.2) Xg = g1X1 + g2Xg + g3Y1 - Yg

13.12. Equation 13.1 is based on the conventional I/O system, showing the use of the outputs of economic
activities. AX1 is the input used in the production of outputs of economic activities. HXg is the input used
in pollution abatement activities: Xg is the pollutants abated, or --said differently-- the outputs of abatement
activities, measured in physical units; H being the direct input coefficient matrix of pollution abatement
activities, converts Xg to products of economic activities that are needed to abate pollutants generated. Y1 is
the product of economic activities that are used in final consumption, assuming there are no other final uses.

13.13. Equation 13.2 shows the total amount of pollutants abated Xg. It is equal to the pollutants generated
by the output of economic activities X1 (i.e. g1 X1 ), the pollutants generated by the output of pollution
abatement activities (i.e. the pollution generated by the abatement of pollutants) Xg (i.e. g2Xg), the pollutants
generated by final consumption (i.e. g3Y1), minus the pollutants that are not abated Yg with the latter
determined by the socially accepted level of pollution, which in many countries is set normatively by
government environmental protection agencies.

13.14. The two identities are presented below in a matrix format, that is similar to a conventional I/O model
and may be handled in the same way:

X1 A H X1 Y1
(13.3) ' %
Xg g1 g2 Xg g3Y1&Y g

13.15. Without pollution abatement, the model would be reduced to:

(13.4) X = AX + Y
 257

in which X and Y are not necessarily the same as X1, Y1 in identity 13.3. The extent to which they deviate
depends on two factors that reinforce each other. The first one is the quantitative importance of the pollution
coefficient matrices g1, g2, g 3 which determines the pollution to be abated and thus the output of the pollution
abatement industries. The second factor is the input coefficient matrix H of pollution abatement activities,
which determines the amount of output that is diverted for use as inputs of the abatement industries, and is thus
no longer available for final use. The environmental accounts features are thus integrated with the
conventional input-output coefficient matrix A, without any modifications or adjustments of its coefficients.

B. Green GDP based on a simple I/O model

13.16. This section uses a very simple, but somewhat unrealistic, version of the I/O model presented in
section A, in order to explain the method adopted in this chapter to calculate the Green GDP. In this example
it is assumed that final use of goods and services does not generate pollution and that all pollutants generated
by industries are fully abated.

13.17. Thus, as g3 = 0 and Yg = 0, equations 13.1 and 13.2 are reduced as follows:

(13.5) X1 = AX1 + HXg + Y1

(13.6) Xg = g1X1 + g2Xg

13.18. By rewriting equation 13.6 and expressing Xg in terms of X1 and then substituting this value Xg into
equation 13.5, the following expressions can be derived:

(13.7) Xg = (I - g2)-1g1X1

(13.8) X1 = AX1 + H(I - g2)-1g1X1 + Y1

13.19. Further rewriting 13.8 gives:

(13.9) X1 = [ A + H(I-g2)-1g1] X1 + Y1
Direct inputs used Direct inputs used in Final use
in economic abatement activities
activities

In 13.9 Y1 is the Green GDP. Without pollution abatement, a larger part of output is available for final use,
i.e.,

(13.10) Y*= H(I-g2)-1g1X1 + Y1

Thus, Green GDP Y1 is smaller than the conventional GDP Y*, as some final use of Green GDP is diverted
as direct input in abatement activities.

13.20. A numerical example may further illustrate the above. In it, the first two rows and columns represent
economic activities, the third ones refer to the pollution abatement activity, and there is only one pollutant.
The direct input coefficient matrix is assumed to be as follows:
258

.15 .25 .002

(13.11)
.20 .05 .004
1 3 .020

13.21. In matrix 13.11, A1 is the 2 x 2 matrix in the top left quarter, g1 is the 1 x 2 vector on the bottom left,
H is the 2 x 1 vector in the top right, and g2 is represented by the bottom right cell. H in the top right describes
the abatement technology that is currently available; thus, in order to eliminate 1 unit of pollutant, .002 of
monetary value of product 1 and .004 of monetary value of product 2 are needed; this also generates .2 unit
of pollutant.

13.22. The point of departure for using coefficient matrix 13.11 is an output vector, i.e.,

1,000
(13.12) X1 '
2,000

The flow relations derived from this are shown in 13.13 below, using equation 13.9 to calculate the Green
GDP and equation 13.7 to calculate pollution abated Xg. The first two rows show the outputs of economic
products, intermediate consumption and final uses in monetary terms. The last row of the flow matrix shows
the pollutant in physical units generated by each industry, and fully abated.

Outputs Intermediate consumption Final use

1,000 150 500 14.3 335.7

(13.13) = +
2,000 200 100 28.6 1,671.4

7,142.8 1,000 6,000 142.8 0

13.23. Thus, the Green GDP is equal to the total value of final use, which is (335.7+1,671.4) = 2,007.1. The
GDP without pollution abatement is equal to the Green GDP plus the values of inputs used in pollution
abatement shown in the first two elements of the third column in the middle part of the equation, i.e. 2,007.1
+ (14.3 + 28.6) = 2,050. Full abatement of pollution will thus reduce GDP by 2.1% in this example. The last
row shows total output and output of pollution in each industry abated.

C. Green GDP based on a more realistic I/O model

13.24. In a more realistic I/O model, two of the assumptions of the simple model have been eliminated, and
all elements in equations 13.1 and 13.2 are utilized. Thus the realistic model includes pollution caused by
final uses (g3Y1 in equation 13.3) and in it not all pollution is abated, but rather a socially acceptable level of
pollution (Yg in equation 13.3) has been introduced.

By rewriting 13.1, Y1 is expressed in terms of X1, i.e.

(13.14) Y1 = (I-A)X1 - HXg

By substituting the above expression for Y1 in 13.2, Xg can be expressed in terms of X1. Thus:
 259

(13.15) Xg = [I - (g2 - g3H)]-1 [g1 + g3(I-A)]X1 - [I - (g2 - g3H)]-1 Yg

By substituting R = I - (g2 - g3H), 13.15 is simplified to:

(13.16) Xg = R-1 [g1 + g3(I-A)]X1 - R-1 Yg

13.25. Replacing Xg in equation 13.1 by that in equation 13.16, it is possible to write equation 13.1 in terms
of X1 and exogenous variables Yg and Y1 as follows:

(13.17) X1 = {A + HR-1[g1 + g3(I-A)]} X1 - HR-1 Yg + Y1

Direct inputs Direct inputs used in Minus: Direct inputs Final use
used to produce pollution abatement of pollution that is
output of activities (full not abated
economic abatement of pollution) (socially accepted
activities pollution)

13.26. The relationship presented in 13.17 above is similar to 13.9 in the case of the simple I/O model. In this
case, polluted GDP (Y*) is equal to polluted GDP (Y1), plus direct inputs used in pollution abatement activities
if all pollution were abated, minus direct inputs of pollution that is not abated, i.e.,

(13.18) Y* = Y1 + {HR-1[g1 + g3(I-A)]} X1 - HR-1Yg

13.27. The use of equation 13.18 in the calculation of the Green GDP is similar to what was described in
section B (equation 13.10).

13.28. Below is presented a numerical example similar to that in section B, which includes the following
additional values:

(13.19) g3 = (2 1.5)

(13.20) Yg = 1,000

13.29. Assuming again the same output vector as in the example of section B (equation 13.12), while using
the identity of equation 13.17, the flow matrix of outputs, intermediate consumption, final use and pollutants
to be abated, can be calculated as follows:

Outputs Intermediate consumption Final use

1,000 150 500 20.7 - 2.0 331.3

(13.21) +
2,000 = 200 100 41.4 - 4.0 1,662.6
9,343.4 1,000 6,000 186.9 3,156.5 - 1,000

13.30. The last element in the last column, 3,156.5, represents the total pollutants generated by final uses in
the case of full abatement, minus 1,000, which is the amount of pollutants tolerated in the economy. The
260

Green GDP (Y1) in the example is equal to (331.3 + 1,662.4) = 1,993.7, and the GDP without pollution
abatement (Y*) is equal to the sum of the elements in the vector below:

331.3 20.7 & 2 350

(13.22) Y( ' % '
1,662.6 41.4 & 4 1,700

D. Adaptation of the I/O model to study the economic impact of pollution abatement policies

13.31. The I/O models presented above reflect the scope of the discussion of Green GDP until now. In this
discussion only production and final use considerations play a role, while effects on income, redistribution of
income and the (financial) ability of industries to pay for pollution abatement are not dealt with. These effects
and the restrictions they pose are, of course, important in the further development of an analytical concept of
Green GDP as presented in this chapter. However, they can only be dealt with effectively if the data in the
institutional sectors of the SNA are included in the discussion and until now no clear conceptual framework
for doing this has been elaborated. In view of the importance of this aspect, however, some preliminary
thoughts have been elaborated below on how to adapt the I/O model to study other economic impacts of
pollution abatement policies. Further work on this continues in the Statistics Division.

13.32. The discussion below is in two parts. The first one discusses the incorporation into the model of output
of abatement activities in monetary terms, partial abatement, environmental taxes and subsidies, and payments
for abatement of pollution generated by final uses. The second part deals mainly with savings of institutional
sectors and forms of financing investments in pollution abatement equipment.

13.33. The discussion below does not enter into two other important limitations of the I/O model. One is that
neither prices of outputs nor intermediate consumption of industries are affected, even though the pollution
abatement activities affect the resources available for final use. Also, the I/O models do not enter into any
optimization of final use and Green GDP, which could only be achieved by applying linear programming
techniques in which input and pollution coefficient matrices of equation 13.3 would play a role, and which
would also take into account the capital stock needs of each industry.

1. Allocation of pollution abatement services to industries and final uses

13.34. The I/O models for the derivation of Green GDP discussed above only describe the pollutants
generated by industries and final uses, and do not identify the industries and final-use sectors paying for the
abatement services. Furthermore, Yg shows the pollutants that are tolerated but does not specify in which
industries or final use categories. Thus no information is available on the effects of abatement expenses on
the operating surplus of industries and how final uses are actually affected.

13.35. To identify the effects of abatement expenses by industries and final uses, equation 13.2 of the I/O
model is transformed into a pollution abatement equation, presented as 13.2' in the table below. It assumes
that a specified percentage (*) of pollution by type is tolerated for each unit of output. In addition, the output
of abatement activities (Xg) must now be measured in monetary terms.
 261

(13.2') = Xg *1g1X1 + *2g2Xg *3g3Y1

Abatement costs assumed Abatement costs

to be paid by industries assumed to be paid
by final uses
(principally final
consumption by
households)

13.36. In the table, *i is a m x m diagonal matrix in which all off-diagonal elements are zeros and diagonal
elements are equal to the percentages of pollutants abated. Using the percentage of pollutant tolerated in each
industry3 to distribute Yg to intermediate inputs of each industry, the latter is reduced to zero. If X gis measured
in monetary terms, equation 13.3 which incorporates 13.2' becomes a conventional I/O model. The direct input
coefficient matrix of the I/O model measures both inputs to the production of economic activities and inputs
of pollution abatement activities.

13.37. The abatement cost assumed to be distributed to industries and final uses on the basis of a model
incorporating equation 13.2' is presented in table 13.1 below. In addition to the separate line for distributed
abatement cost, the table includes for each industry --i.e. economic and abatement activities-- and final use
sector, the output, inputs, value added and fixed capital requirements. The abatement cost corresponds to the
cost of pollution abated in each industry and in final uses, taking into account that pollution is only abated to
a specified degree. In the allocation, it is assumed that each industry and final use must pay for the pollution
it generates. The allocation is based on a maintenance cost valuation of abated pollution. Thus, pollutants in
physical units have been converted to costs of abatement, assuming linear I/O relationships and direct costs
of abating each unit of pollutant by type.

Table 13.1. Pollution abatement input-output structure

(based on monetary valuations abatement)

Economic ... Economic Abatement ... Abatement Final

activities activities activities activities uses
1 n 1 m

Output

Inputs

Abatement cost

Value added

Fixed capital
requirements

3
The assumption made in equation 13.2' is not entirely realistic. Pollution abatement is normally required when the
total level of a pollutant exceeds a specified tolerated level for a particular firm and not when a level of pollutant
generated by each unit of output exceeds a specified level of tolerance. However, this micro and more realistic scenario
cannot be well incorporated into an I/O model as the latter relies on a macro view of an industry in which establishments
involved in similar activities are aggregated into industries.
262

13.38. The model of 13.2' would require further refinement to the extent that payments for abatement services
are often not paid for by the industries causing the pollution. Thus in practice the following scenarios may be
encountered:

(a) Private firms, independent of industries produce abatement services and then collect fees from
industries and final users who generate pollution;

(b) Industries produce abatement services as ancillary activities within the same establishment;

(c) The Government produces abatement services (e.g. garbage collection) and then either
collects pollution taxes on products (i.e. indirect taxes) from industries and direct pollution
taxes from final users that generate pollution, or provides implicit subsidies by not charging
any fee.

13.39. To accommodate the three scenarios (a) - (c) mentioned above, the model needs to be modified. For
cases (a) and (b), in table 13.1, the value added and consequent operating surplus of an industry are reduced
by the abatement cost paid for by each industry. In case (b) abatement is carried out as an ancillary activity
of the establishments themselves. To deal with this case, ancillary activities mentioned in (b) are treated in the
model as separate establishments.

13.40. If in case (c) the Government carries out the abatement and charges for it through environmental
product taxes (for instance fees for garbage collection paid for by industries), the operating surplus is similarly
reduced by the abatement cost. On the other hand, if the Government pays for the cost of abatement, but does
not charge a fee, the value added of the industries concerned must be modified to include a subsidy equal to
the abatement costs so that their operating surplus will remain unchanged.

13.41. The costs to abate pollution generated by final users are part of final uses if paid for by final users,
otherwise a subsidy equal to abatement costs has to be entered into the column of final uses.

13.42. It is not possible to discuss all possible modifications but suffice it to say that table 13.1 has to be
analysed and modified to reflect government policies on the payment of abatement costs.
 263

2. Financing investments in abatement equipment by sectors

13.43. Institutional sectors are the entities that make investment decisions and therefore an impact study of
pollution abatement must pursue the additional capital costs that institutional sectors have to pay in order to
abate pollution. Because of the abatement costs incurred by each sector, operating surplus and therefore
savings needed to pay for fixed capital requirements of abatement equipment are affected. To determine these
effects on the financing capacity of each sector, a “Green Value Added” after deduction of abatement costs,
and fixed capital requirements by industries shown in table 13.1 must be allocated to the institutional sectors
that own the establishments.

13.44. To achieve this reallocation, the cross-classification matrix between industries and sectors (CCIS) --a
new analytical feature in the 1993 SNA-- might be used, as in table 13.2 below. In the CCIS matrix, common
elements between the SUT (supply and use table) and IEA (integrated economic accounts), such as value
added and capital formation, are cross-classified by industries and sectors.

13.45. To use the CCIS to reallocate industry data on output, inputs and value added to sectors, the CCIS
matrix of value added for the base year may be used and assumed to be a share coefficient matrix that remains
constant over time. Thus, in table 13.2 the shares of each sector are entered instead of levels. The sum of each
column then adds up to 1.0. Some further work needs to be done to determine which sectors are carrying out
each of the abatement activities and to estimate their shares.

Table 13.2. Cross-classification of valued added by industries and sectors (CCIS)

Economic .... Economic Abatement .... Abatement

activities activities activities activities
1 n 1 m

Non-financial sector

Financial sector

Government sector

Household sector

13.46. If the CCIS share matrix of value added is Svd and Vind is the diagonal matrix where the off-diagonal
elements are zero and the diagonal elements represent value added classified by industries, then value added
classified by institutional sectors (Vint) is:

(13.23) Vint = Svd Vind

13.47. The CCIS share matrix used to distribute abatement costs may not use the value added shares matrix
but rather the shares of operating surplus. Fixed capital requirements might be linked to output by
capital/output ratios. The allocation to institutional sectors of abatement costs and capital requirements is
otherwise similar to equation 13.23.

13.48. The capital requirement when classified by institutional sectors would show investment capital needed
to abate pollution by each institutional sector. These increased capital requirements are to be confronted for
each institutional sector with the lower value added, operating surplus and net saving, due to the effects of
pollution abatement.
264

13.49. The effects are illustrated in tables 13.3 and 13.4 below. Table 13.3 classifies the data by industries
and final uses and table 13.4 is a reclassification of the data by institutional sectors. In table 13.3 it is assumed
that an abatement cost of 10 is caused by economic activities of non-financial corporations and that the
Government carries out the abatement activity and assumes the cost. Thus an environmental subsidy (-10) is
imputed to economic activities, which eliminates the effect of the imputed abatement cost on operating surplus.

13.50. In table 13.4, the effects on the income and capital accounts of the non-financial sector and the
government sector are shown. To convert the data of table 13.3 to a sector classification in table 13.4, the CCIS
matrix is used. In this simple example it is easy to trace how the transactions of table 13.3 are reconstituted
in table 13.4. For the non-financial sector, all other things being equal, the operating surplus will remain as
before the introduction of pollution abatement and thus the net savings will remain 15 and the net lending
will too. As the government sector carries out the abatement activity, its net borrowing is equal to -40 and
the net savings due to its abatement policy is -10.

13.51. The above analysis could be made more interesting if the non-financial sector were broken down into
many more sub-sectors with significantly different levels of pollution.

Table 13.3. Pollution abatement input-output structure

(based on monetary valuations of abatement)

Economic Government Final

activity abatement uses
activity

Inputs 50 4
Abatement cost 10 1

Value added 40 6
Compensation of 35 6
employees
Environmental -10
subsidy
Operating 15 0
surplus
Output 100 11
Fixed capital requirement 30
 265

Table 13.4. Economic impact on institutional accounts

Non-financial Government
corp. sector sector

Output 100 11

Production Intermediate consumption 60 5

account (inputs + abatement cost)

Value added (including 40 6

environmental subsidy)

Operating surplus 15 0

Income accounts Environmental subsidies -10

Net savings 15 -10

Capital account Capital formation 30

Net lending(+)/net borrowing(-) 15 -40

REFERENCES

1. Bartelmus, Peter and Jan Van Tongeren. “Environmental Accounting: An Operational Perspective.” In National
Accounts and the Environment: Papers and Proceedings from a Conference in London. United Kingdom, 16-18 March,
1994. Ontario, Statistics Canada.

2. Baumol, William and Edward N. Wolff. “A Key Role for Input-Output Analysis in Policy Design.” Regional Science
and Urban Economics 24, No. 1 (February 1994), pp. 93-113.

3. Duchin, F.G.M., K. Thonstad Lange and Annemarth Idenburg. Strategies for Environmentally Sound Development.
Progress Report No.5. Institute for Economics Analysis, New York University, 1991.

4. Leontief, Wassily. “Environmental Repercussions and the Economic Structure: An Input-Output Approach.” Review
of Economics and Statistics 52, No. 3 (August 1970), pp.262-271.

5. Leontief, Wassily and Daniel Ford. “Air pollution and Economic Structure: Empirical Results of Input-Output
Computations.” In Input-Output Techniques, edited by Andrew Brody and Anne P. Carter. New York, American
Elsevier, 1972, pp. 9-30.

6. Kohn, R.E. “Input-Output Analysis and Air Pollution Control.” In Economic Analysis and Environmental Problems
edited by E.S. Mills. New York, National Economic Research, 1975, pp. 239-272 .

7. Miller, Ronald E. and Peter D. Blair. “Environmental Input-Output Analysis.” In Input-Output Analysis: Foundations
and Extensions. New Jersey, Prentice-Hall, 1985.

8. United Nations. SNA Handbook on Integrated Environmental and Economic Accounting. Sales No. [Link], 1993.

9. De Haan, Mark and Steven J. Keuning. “Taking the Environment into Account: The NAMEA Approach.” In The
Review of Income and Wealth. Series 42, No. 2, June 1996.