WP/18/37

The Status of GDP Compilation Practices in 189 Economies

and the Relevance for Policy Analysis

by Francien Berry, Massimiliano Iommi, Michael Stanger, and Louis Venter

IMF Working Papers describe research in progress by the authors and are published
to elicit comments and to encourage debate. The views expressed in IMF Working
Papers are those of the authors and do not necessarily represent the views of the IMF, its
Executive Board, or IMF management.
 2

© 2018 International Monetary Fund WP/18/37

IMF Working Paper

Statistics Department

The Status of GDP Compilation Practices in 189 Economies and the Relevance
for Policy Analysis

Prepared by Francien Berry, Massimiliano Iommi, Michael Stanger, and Louis Venter

Authorized for distribution by Claudia Dziobek

March 2018

IMF Working Papers describe research in progress by the authors and are published
to elicit comments and to encourage debate. The views expressed in IMF Working
Papers are those of the authors and do not necessarily represent the views of the IMF, its
Executive Board, or IMF Management.

Abstract

This paper examines the status of GDP compilation in 189 economies against six key criteria
that describe national accounts compilation practices: whether the benchmark year is up to
date, the availability and timeliness of annual and quarterly GDP, whether GDP by
production and expenditure approaches are compiled independently to allow for
comparisons, whether estimates by the income approach are available, and the vintage of the
System of National Accounts (SNA) applied. We used publicly available information
including from the IMF’s Dissemination Standards Bulletin Board (DSBB), and, for 108
developing economies, information provided by the IMF’s real sector advisors stationed in
the Fund’s 10 Regional Technical Assistance Centers (RTACs). The data were compared
with the UNSD and World Bank databases. We find that 50 percent of economies have
acceptable benchmark years, 72 percent report timely annual GDP data, while 55 percent of
economies report timely data for quarterly GDP. The study presents some conclusions for
priorities of capacity development.

JEL Classification Numbers: E01

Keywords: National Accounts, GDP, Data Quality, Data Dissemination
Author’s E-Mail Address: [email protected]; [email protected]; [email protected];
[email protected]
 3

Contents Page

Abstract ......................................................................................................................................2

I. Introduction ............................................................................................................................4

II. Years Elapsed since the Benchmark Year.............................................................................5

III. Availability of Annual and Quarterly GDP .........................................................................8

IV. Timeliness of Annual and Quarterly GDP.........................................................................10

V. GDP by Production, Expenditure, and Income ...................................................................12

VI. Data on GDP by Production and Expenditure Derived Independently .............................14

VII. Vintage of the SNA Applied ............................................................................................15

VIII. Conclusions and Further Research ..................................................................................16

Tables
1. Status of GDP Compilation in 189 Economies: Six Criteria...........................................................5
2. Annual GDP - Age of Benchmark Year in 189 Economies ..................................................7
3. Annual and Quarterly GDP by Production, Expenditure, and Income and Vintage of SNA 9
4. Annual and Quarterly GDP - Timeliness of Data Release in 189 Economies.....................10

Figures
1. Annual GDP - Age of Benchmark Year in 189 Economies ..................................................6
2. Annual and Quarterly GDP Compilation by Region .............................................................8
3. Availability of Advance Release Calendars in 189 Economies...........................................11
4. Availability of Advance Release Calendars Across Regions ..............................................11
5. Quarterly GDP Compilation by Region and Method ..........................................................13
6. Are Production and Expenditure GDP Derived Independently? .........................................14
 4

I. INTRODUCTION1

Reliable GDP data are critical for macroeconomic and financial policy analysis
including IMF surveillance. Data users rely on national accounts compilers to provide
consistent, timely, and relevant GDP statistics in line with internationally acceptable
standards laid out in the System of National Accounts (SNA). The IMF Statistics Department
provides technical assistance to well over 100 economies to improve GDP compilation.2 It
also advises Fund economists when questions arise on the quality of GDP data for
surveillance purposes. Based on the experiences gained with these activities over the past
decades, this paper identifies six key criteria that are routinely applied to assess whether GDP
data are fit for purpose.3 In this paper, we highlight the results for the Fund’s 189-member
economies mainly in terms of the World Economic Outlook (WEO) regional groups.

We assess the status of GDP by examining six key features of the national accounts.
These features are (1) years elapsed since the most recent benchmark year to indicate
whether the current structure of the economy is appropriately represented, (2) the availability
of annual and quarterly data, for forward looking policy analyses and forecasting, (3) the
timely dissemination of annual and quarterly data, (4) the availability of GDP by production,
expenditure, and income approaches, (5) the availability of independently estimated GDP
approaches, and (6) the vintage of the SNA applied. The paper is based on publicly available
information including the IMF’s Dissemination Standards Bulletin Board (DSBB), and for
108 developing economies, information provided by the IMF’s real sector advisors located in
the Fund’s 10 Regional Technical Assistance Centers (RTACs). The data were also
compared with the UNSD’s National Accounts Official Country Data database and the
World Bank’s database on Statistical Capacity Indicators. The dataset used in this paper is
available in the technical appendix published separately.

1
The authors would like to thank Thomas Alexander, Serkan Arslanalp, Rob Dippelsman, Claudia Dziobek,
Laurent Kemoe, Jimmy McHugh, Iana Paliova, Gabriel Quiros, Roberto Rosales, Patrizia Tumbarello, and
Yingbin Xiao for comments on an earlier draft. The authors are also grateful to Zia Abbasi, Pamela Audi,
Hubert Gbossa, Donna Grcman, Gregory Legoff, Achille Pegoue, Brooks Robinson, Todor Todorov, Martha
Tovar, and Richard Wild for their contributions to put the dataset together.
2
The term “economy,” as used in this paper, does not in all cases refer to a territorial entity that is a state as
understood by international law and practice. The term also covers some non-sovereign territorial economies,
for which statistical data are maintained and provided internationally on a separate and independent basis.
3
To help users identify quality-related features of their macroeconomic statistical systems, the IMF developed
the Data Quality Assessment Framework (DQAF). The DQAF is rooted in the UN Fundamental Principles of
Official Statistics and is organized around a set of prerequisites and five dimensions of data quality—assurances
of integrity, methodological soundness, accuracy and reliability, serviceability, and accessibility. This paper is
focused on a subset of the DQAF.
 5

Table 1. Status of GDP Compilation in 189 Economies: Six Criteria

1. Most recent benchmark year 4. Availability of GDP by different

approaches

2. The availability of annual 5. Independent compilation of GDP by the

and quarterly GDP production and expenditure approach

3. Timeliness of dissemination 6. The vintage of the SNA applied

II. YEARS ELAPSED SINCE THE BENCHMARK YEAR

The years elapsed since the update of the benchmark year is a key indicator of data
quality of national accounts. The benchmark year refers to the year in which an economy`s
transactions such as production—including input costs—, consumption, capital formation,
international trade, and taxes are comprehensively measured and accurately reflect the
structure of the economy (in current price GDP). The base year refers to the year of the
weights that are used to measure constant price GDP. Benchmark and base years do not
necessarily coincide. When a country adopts the chain-linking approach, the base year is
updated annually.4 As the structure of an economy evolves and as more or better data become
available, revisions are needed to realign the national accounts with economic reality. Likely
the most significant element of national accounts revisions is the updating of the benchmark
year although other reasons for major revisions exist, for example changes in statistical
methods and changes in concepts, definitions, and classifications.5

A good practice is to benchmark GDP estimates every 10 years or less. Updating the
benchmark year generally improves the quality of national accounts estimates because of the
incorporation of newly available and revised data sources, enhanced coverage as well as
improved estimation methods. GDP is compiled based on the information and data sources
available at a given period and thus benchmark revisions would require, for example,
comprehensive household budget surveys, recent population censuses, business structural
surveys or censuses, and up to date business registers. Compiling these statistics (including
data collection and processing) is costly and may take several months. Often, the compiling
agency will require additional funding for those activities.

About 50 percent of the economies updated their benchmark year within the last
10 years. For 65 economies (34 percent) the benchmark year is between 11 to 15 years old,

4
Dippelsman, Robert, Venkat Josyula, and Eric Métreau. 2016. “Fixed Base Year vs. Chain Linking in National
Accounts: Experience of Sub-Saharan African Countries.” IMF Working Paper No. 16/133.

5
See System of National Accounts, 2008, the Guidelines on Revisions Policy and Analysis (Organization for
Economic Co-operation and Development and Eurostat, 2008) and Carson, Carol S, Sarmad Khawaja, and
Thomas K Morrison. 2003. "Revisions Policy for Official Statistics: A Matter of Governance." IMF Working
Paper No. 04/87.
 6

and older than 15 years in 30 economies (16 percent), as shown in Figure 1. From a regional
perspective, Table 2 shows that almost all Advanced Economies and over 80 percent of
Emerging and Developing Europe have updated base years. In the Commonwealth of
Independent States, Sub-Saharan Africa, Emerging and Developing Asia, Middle East, North
Africa, Afghanistan and Pakistan, updated base years exist in about 36 to 64 percent of the
economies. In Latin America and the Caribbean, the percent of economies with updated
benchmark years falls to 16 percent, mainly reflecting the status quo in the Caribbean and
Central American countries.

Figure 1. Annual GDP - Age of Benchmark Year in 189 Economies

Source: Fund staff estimates as of October 2017.

 7

Table 2. Annual GDP - Age of Benchmark Year in 189 Economies

Older than 15
1-10 years 11 -15 years Total
Region years
Economies
Number Percent Number Percent Number Percent
Advanced Economies 35 95 2 5 0 0 37
Commonwealth of
7 64 1 9 3 27 11
Independent States
Emerging and Developing
12 40 11 37 7 23 30
Asia
Emerging and Developing
10 83 1 8 1 8 12
Europe
Latin America and the
5 16 19 59 8 25 32
Caribbean
MENA, Afghanistan, and
8 36 9 41 5 23 22
Pakistan
Sub-Saharan Africa 17 38 22 49 6 13 45

Memo: G-20 Economies 18 90 2 10 20

Total Economies 94 50* 65 34 30 16 189

Source: Fund staff estimates as of October 2017; Economies are grouped according to IMF WEO regional classification.
* This represents 90.5 percent of the world GDP (current values of 2016)

The effects of updating outdated benchmark years vary widely across economies and
can present significant difficulties for policy makers. Updating the benchmark year can
have a large impact on both the level and growth rate of GDP, particularly if prices and
volumes have changed significantly. The update may show an increase or decrease
depending on the developments in the economy since the previous benchmarking exercise,
the robustness of the compilation method, and the quality of the source data.

Some examples illustrate this point: In Africa, after benchmarking, the level of GDP in
current prices was 89 percent higher in Nigeria, 60 percent higher in Ghana, 25 percent in
Zambia, but 10 percent lower in Botswana. After benchmarking, national accounts typically
provide more accurate estimates of the size of the economies (as measured by GDP) but also
of their production and consumption structures (because, for instance, the benchmark update
incorporates new or existing activities or products, which were previously not captured or ill-
measured). In Latin America, it has generally resulted in increases in nominal GDP levels
(for the base year assessed under the old and new methodology) with a median increase of
8.8 percent.6

6
Olinto Ramos Roberto, Gonzalo Pastor, and Lisbeth Rivas. 2008. “Latin America: Highlights from the
Implementation of the System of National Accounts 1993 (1993 SNA).” IMF Working Paper No. 08/239.
 8

III. AVAILABILITY OF ANNUAL AND QUARTERLY GDP

Annual GDP data are needed for a longer, more in-depth view of trends and cycles in
the economy while quarterly GDP data are key for short-term surveillance, forecasting,
and policy analysis. Annual GDP is produced by 188 of the 189 economies in this study
(Figure 2). During the past 15 years, many economies launched efforts to compile quarterly
national accounts. Capacity development activities by the IMF, World Bank, and others have
focused on higher frequency data, particularly quarterly national accounts.

Figure 2. Annual and Quarterly GDP Compilation by Region

100
90
80
70
60
Percent of region

50
40
30
20
10
0
Advanced Commonwealth Emerging and Latin America Sub-Saharan MENA, Emerging and
Economies of Independent Developing and the Africa Afghanistan, and Developing Asia
States Europe Caribbean Pakistan

Regions
Annual GDP - Expenditure or Production Approach Quarterly GDP - Expenditure or Production Approach

Source: Fund staff estimates as of October 2017; Economies are grouped according to IMF WEO regional classification.

While annual and quarterly national accounts use the same principles and definitions,
data sources and statistical techniques differ. Quarterly national accounts require source
data that are more frequent but possibly less complete and less comprehensive and are
typically supplemented with extrapolation and other statistical techniques. An additional
process is required to transform, and integrate quarterly source data within the SNA
framework and to make quarterly estimates consistent with available annual estimates.7

Currently, 133 economies (70 percent of the total) disseminate quarterly GDP (Table 3).
The share of economies that compile quarterly GDP is about 53 percent for Sub-Saharan
Africa, 50 percent in the Middle East, North Africa, Afghanistan and Pakistan, and 40
percent in Emerging and Developing Asia. It is between 80 and 100 percent in the rest of the
world. Overall, the economies that compile quarterly GDP represent approximately 97
percent of world GDP (current values of 2016).

7
IMF, 2017, “Quarterly National Accounts Manual - Concepts, Data Sources, and Compilation.”
 Table 3. Annual and Quarterly GDP by Production, Expenditure, and Income and Vintage of SNA
(Number of Economies and Percent of Region)

Memo1:
Emerging Common- Emerging Emerging Latin MENA,
Sub-
Advanced G20 and wealth of and and America Afghanistan
Saharan Total
Economies Emerging Independent Developing Developing and the and
Africa
EU States Asia Europe Caribbean Pakistan
Economies

Economies
Percent of
Number

Number

Number

Number

Number

Number

Number

Number

Number
Percent

Percent

Percent

Percent

Percent

Percent

Percent

Percent

Total
Annual GDP
Production - current prices 37 100 15 100 11 100 30 100 12 100 32 100 22 100 43 96 187 99
Production - constant prices 37 100 15 100 11 100 29 97 12 100 32 100 22 100 43 96 186 98
Expenditure - current prices 37 100 15 100 10 91 20 67 12 100 31 97 21 95 43 96 174 92
Expenditure - constant prices 37 100 13 87 10 91 18 60 12 100 22 69 15 68 43 96 157 83
Memo: annual GDP by
37 100 15 100 11 100 30 100 12 100 32 100 22 100 44 98 188 99
production or expenditure
Memo: annual GDP by
37 100 15 100 10 91 20 67 12 100 31 97 21 95 42 93 173 92
production and expenditure
Memo: GDP - Income

9
37 100 12 80 10 91 12 40 9 75 16 50 11 50 24 53 119 63
Approach2
Quarterly GDP
Production - current prices 34 92 15 100 11 100 12 40 10 83 25 78 11 50 15 33 118 62
Production - constant prices 35 95 15 100 10 91 12 40 10 83 27 84 10 45 23 51 127 67
Expenditure - current prices 37 100 14 93 8 73 9 30 11 92 17 53 5 23 11 24 98 52
Expenditure - constant prices 37 100 13 87 7 64 9 30 11 92 17 53 3 14 10 22 94 50
Memo: quarterly accounts:
production or expenditure 37 100 15 100 11 100 12 40 11 92 27 84 11 50 24 53 133 70
Vintage of SNA
SNA 1968 1 3 1 5 2 1
SNA 1993 1 3 1 7 8 73 18 60 2 17 15 47 16 73 29 64 89 47
SNA 2008 / ESA 2010 36 97 14 93 3 27 11 37 10 83 17 53 5 23 16 36 98 52
Total number of Economies
37 61 15 29 11 2 30 21 12 3 32 7 22 4 45 2 189 100
and percent of World GDP
Sources: Fund staff estimates as of October 2017, UN National Accounts Statistics: Main Aggregates and Detailed Tables, 2016 and World Bank Statistical Capacity Indicator
Database; Economies are grouped according to IMF WEO regional classification.

1/ The economies covered are Argentina, Brazil, Bulgaria, People's Republic of China, Republic of Croatia, Hungary, India, Indonesia, Mexico, Republic of Poland, Romania,
Russian Federation, Saudi Arabia, South Africa, and Turkey.
2/ GDP-Income Approach also includes economies that compile Gross Operating Surplus.
 10

IV. TIMELINESS OF ANNUAL AND QUARTERLY GDP

Timeliness of annual and quarterly GDP data is a key requirement for policy decisions.
Timeliness is defined in the IMF Special Data Dissemination Standard Guide as the lapse of
time between the end of a reference period (or a reference date) and the date on which the
data are disseminated. The IMF’s Data Standards Initiatives (SDDS and SDDS Plus) require
economies to disseminate annual data within nine months of the reference period and
quarterly GDP within one quarter. Using these criteria, Table 4 shows that timely annual data
are disseminated by 73 percent of economies while timely quarterly data are disseminated by
55 percent (this is 77 percent of economies compiling quarterly data). In 32 economies
(17 percent of the total), annual data are disseminated with a lag of more than 15 months and
25 economies (13 percent) disseminate quarterly estimates with a lag of more than six
months.

To promote timeliness and transparency, the IMF’s Data Standards Initiatives

recommend that economies disseminate Advanced Release Calendars (ARCs). An ARC
should show the release dates of data for the current month and the following three months.8
A publicly available ARC provides users with information on when data will be released, it
signals good management and transparency of statistical compilation, and helps compilers
take an active and organized approach to acquiring the inputs for their work. It also promotes
communication with users including the media. Currently 57 percent of economies have
ARCs (Figures 3 and 4), of which 7 percent correspond to SDDS Plus economies, 32 percent
to SDDS economies and 17 percent to e-GDDS economies.

Table 4. Annual and Quarterly GDP - Timeliness of Data Release in 189

Economies

(Number of Economies and Percent)

Release of Annual GDP
9 months or less 10 - 15 Months Later than 15 Months Total Economies
Number Percent Number Percent Number Percent
137 73 20 11 32 17 189
Release of Quarterly GDP
3 months or less 4 - 6 Months Later than 6 Months Total Economies
Number Percent Number Percent Number Percent
104 55 4 2 25 13 133 / 189
Source: Fund staff estimates as of October 2017.

8
The availability of an ARCs is a requirement for subscribing to the IMF’s SDDS and SDDS Plus. ARCs are
required for all SDDS and SDDS Plus data categories, except for the encouraged categories, and for those data
categories disseminated daily: interest rates, stock market (share price index), and exchange rates. e-GDDS
participants are encouraged to disseminate an ARC taking into considering each country’s own circumstances.
 11

Figure 3. Availability of Advance Release Calendars in 189 Economies

Advance Release
No Advance Release Calendar
Calendar 57%
43%

Sources: Fund staff estimates as of October 2017 and World Bank Statistical Capacity Indicator Database.

Figure 4. Availability of Advance Release Calendars Across Regions

100
90
80
70
Percent of region

60
50 ARC ARC
97 % 92% ARC
40 82 %

30
ARC ARC ARC
20 44% 44% 43 %
ARC
10 23%

0
Advanced Emerging and Commonwealth Sub-Saharan Latin America Emerging and MENA,
Economies Developing of Independent Africa and the Developing Asia Afghanistan,
Europe States Caribbean and Pakistan

Regions
Sources: Fund staff estimates as of October 2017 and World Bank Statistical Capacity Indicator Database.
 12

V. GDP BY PRODUCTION, EXPENDITURE, AND INCOME

Ideally, GDP is estimated according to three approaches; production, expenditure (both

in current and constant prices), and income (only current prices). Table 3 depicts
economies’ compilation scopes by production, expenditure, and income, and current versus
constant prices GDP. Virtually all economies compile annual GDP in current and constant
prices from the production approach. Current and constant price GDP from the expenditure
side are compiled by 174 and 157 economies respectively. Economies not reporting current
GDP by expenditure are mainly in Emerging and Developing Asia while several Latin
America and Caribbean economies and Middle East and North Africa economies only release
current expenditure GDP.

Often the first stage of developing a national accounts system is to compile GDP by type
of economic activity or industry (production approach). The availability of source data
drives the development of national accounts statistics. Industry and business statistics are
commonly more readily available on a continuous basis in most economies since they may
already be part of other statistical systems. On the other hand, key data sources for the
expenditure and income side require more advanced statistical systems covering household
and corporate income data on a continuous basis. For example, household final consumption
expenditure is generally derived from household surveys which are costly and thus not
always conducted on a regular, timely, and continuous basis.

Compiling constant price GDP requires a series of price statistics. Good practices on
deflation requires detailed price statistics for all GDP components by product groups.9
Detailed deflation is commonly carried out on annual estimates and the most relevant prices
required for deflation are Producer Prices (including agriculture, construction, and services),
Consumer Prices, and Export and Import Prices. These deflators are not available in some
economies. On the other hand, some countries rely on volume indicators to derive constant
prices GDP estimates directly (for instance, quarterly GDP).

About 64 percent of economies compile at least some components of annual GDP by the
income approach.10 The availability of GDP by income component supports the analysis of
profitability and distribution of income between capital and labor. However, with reliable
data on the compensation of employees, taxes less subsidies on products and production, the
other key component gross operating surplus/mixed income can be calculated as a residual.
GDP from the income approach is reported by 119 economies (Table 3) and the availability
is particularly limited in Sub-Saharan Africa (53 percent of economies), Middle East, North

9
Alexander, Thomas, Claudia Dziobek, Marco Marini, Eric Metreau, and Michael Stanger. 2017. “Measure up:
A Better Way to Calculate GDP.” IMF Staff Discussion Notes No. 17/02.
10
The availability of GDP by the income approach in this paper includes countries compiling some of the main
income components.
 13

Africa, Afghanistan and Pakistan (52 percent), Latin American and the Caribbean (50
percent), and Emerging and Developing Asia (41 percent).

The development of annual GDP data in the ten G-20 Emerging Economies and the five
European Union Emerging Economies is broadly in line with that of Advanced
Economies. The most important exceptions are constant GDP from the expenditure approach
(available in 13 economies) and GDP from the income approach (available in 12 economies)
as shown in the second column of Table 3. These economies represent about 29 percent of
the world GDP while G-20 economies account for 86 percent of the world GDP.

Quarterly GDP from the production approach in constant prices (127 economies) is
more common than GDP from the expenditure approach (94 economies) as shown in
Table 3. The income approach is not as widely used as the two other approaches for
estimating quarterly GDP, partly because the required data might not be available at an intra-
annual frequency, and partly because the income approach may only be estimated at current
prices.11The availability of quarterly GDP from the expenditure approach at constant prices is
particularly limited in the Middle East, North Africa, Afghanistan and Pakistan
(14 percent), and Sub-Saharan Africa (22 percent) (Figure 5).

Figure 5. Quarterly GDP Compilation by Region and Method

100

90

80

70

60
Percent of region

50

40

30

20

10

0
Commonwealth Advanced Emerging and Latin America MENA, Emerging and Sub-Saharan
of Independent Economies Developing and the Afghanistan, and Developing Asia Africa
1 2
States Europe Caribbean Pakistan

Regions

Production - constant prices Production - current prices Expenditure - current prices Expenditure - constant prices

Source: Fund staff estimates as of October 2017; Economies are grouped according to IMF WEO regional classification.

11
See footnote 7.
 14

VI. DATA ON GDP BY PRODUCTION AND EXPENDITURE DERIVED INDEPENDENTLY

When the production and expenditure approaches are derived independently, GDP
estimates benefit from cross validation of each of these estimates. Different source data
and methods are used to estimate different components of GDP. Validation procedures are
carried out so that discrepancies between approaches are managed in favor of more reliable
data sources increasing the GDP robustness when consistency is achieved. Countries may not
be able to assign all discrepancies and thus report a statistical discrepancy in the GDP
estimates. Supply and use tables are the appropriate tool to perform a detailed reconciliation,
as recommended by the SNA.

In many economies, GDP from either the production or the expenditure approach is
estimated first and is considered the official measure. Thereafter, the sum of expenditure
aggregates or the sum of industry GDP are assumed to be equal to official GDP. For instance,
in most Sub-Saharan economies expenditure based GDP is not independently estimated and
frequently, private final consumption expenditure is calculated as a residual. Figure 6 shows
that production and expenditure GDP are derived independently by 48 percent of the
economies. The number of economies that derive independent estimates of GDP by
production and expenditure is proxied by those that produce supply and use tables or report
statistical discrepancies in GDP.

Figure 6. Are Production and Expenditure GDP Derived Independently?1

NO
52%
YES
48%

Sources: Fund staff estimates as of October 2017 and UNSD National Accounts Official Country Data.

1/ Of the 189 economies in this sample, 173 economies compile both Production and Expenditure GDP (See Table 3).
The number of economies that derive independent estimates of Production and Expenditure GDP is proxied by those that
produce supply and use tables or report statistical discrepancies in GDP.
 15

VII. VINTAGE OF THE SNA APPLIED

The most recent vintage of the SNA is the 2008 SNA. Of the 189 economies,
98 (52 percent of economies representing 94 percent of world GDP) have adopted the 2008
SNA methodology, 89 (47 percent) use the 1993 SNA, and two still apply the 1968 SNA.12
Implementation of the 2008 SNA varies across regions. Almost all Advanced Economies and
Emerging and Developing Europe economies, and 53 percent of Latin American and
Caribbean economies have adopted the 2008 SNA. In the other regions, most of the
economies produce their data according to the 1993 SNA, particularly in the Middle East,
North Africa, Afghanistan and Pakistan regions, where only five economies have adopted the
2008 SNA.

What does this mean for data users? How different is GDP when it is compiled
according to 2008 SNA versus 1993 SNA? The 2008 SNA retains the basic theoretical
framework of its predecessor the 1993 SNA. The 2008 SNA mainly introduces treatments for
new aspects of the economy, elaborates on aspects of analytical interest, and clarifies
guidance on a range of issues. The most important changes which impact GDP concern the
capitalization of expenditure on Research and Development (R&D) and Military Weapons
Systems.

In most OECD economies, the methodological changes resulting from 2008 SNA
implementation resulted in a significant upward shift of the level of GDP while the
impact on growth rates was small.13 The overall impact on GDP-levels was 3.1 percent
(weighted average of all OECD economies in 2010), ranging from 1.2 to 5.1 percent. The
average increase of GDP due to capitalizing expenditure on R&D and Military Weapons
Systems was, respectively, 2.2 percent and 0.3 percent. Concerning GDP growth, over the
period 1992 to 2012 the difference in GDP growth for the OECD average was generally
within the boundaries of +/- 0.1 percentage points.

The impact for the member countries of the European Union was below the overall
OECD area. The methodological changes introduced by ESA 201014 increased the GDP of
the European Union by 2.3 percent.15The impact on the GDP level in the five biggest

12
The information on the vintage includes data as reported by economies.

Peter van de Ven, “New standards for compiling national accounts: what’s the impact on GDP and other
13

macro-economic indicators” OECD Statistics Brief, February 2015, No. 20.

14
The European System of National and Regional Accounts (ESA) sets out the harmonized methodology that
must be used for the compilation of national accounts data in the European Union (EU). The latest version is
ESA 2010 that was implemented in September 2014 (from that date onwards the data transmission from
Member States to Eurostat is following ESA 2010 rules). ESA 2010 is the counterpart of, and is consistent with,
the 2008 SNA.
15
Marianthi Dunn, Leonidas Akritidis and Luis Biedma, “The impact of ESA 2010 on key indicators of the
national accounts in Europe,” EURONA, 2/2014.
 16

European countries was less than 4 percent, while it was 4.4 percent in Sweden, 4.2 percent
in Finland. In the United States the level of the GDP increased by 4 percent.

How will the introduction of 2008 SNA impact developing economies? The impact on
GDP of the capitalization of R&D and Military Weapons Systems depends on the relative
size of these expenditures to GDP. In many emerging market economies and middle-income
economies, the impact could be of a magnitude similar to those of OECD/EU economies. In
many low-income economies, the component of R&D is likely to be small.

Beyond GDP, the implementation of the 2008 SNA affects other national accounts
aggregates such as gross fixed capital formation and saving. For instance, in the European
Union, the revision in gross fixed capital formation in 2010 was 12.9 percent for the
European Union as a whole, with the largest overall revisions in Ireland (35.4 percent) and
Sweden (30.2 percent).16 Although these refer to the combined impact of the implementation
of ESA 2010 and the associated statistical improvements, they are informative about the order
of magnitude of the revision due to the adoption of the 2008 SNA.

VIII. CONCLUSIONS AND FURTHER RESEARCH

A key finding of this study is that about 50 percent of the 189 economies estimate GDP
using benchmark data not older than 10 years. These countries represent 91 percent of
world GDP. The impact of updating the benchmark year depends on a variety of
circumstances such as the extent to which the structure of the economy has evolved since the
last benchmark year. GDP may be higher or lower, revisions may be large or small. Outdated
GDP benchmarks may result in misleading interpretations of GDP evolution. Updates in very
long intervals can produce large revisions and make the interpretation of the data more
difficult. Outdated benchmark years raise concerns about the usefulness of GDP estimates for
policy analysis.

Virtually, all economies compile annual GDP data while the compilation of quarterly
GDP data (71 percent of economies, representing about 97 percent of GDP), is less
developed. Timeliness of dissemination also varies across economies and many do not
publish Advance Release Calendars which makes it difficult to predict data release. Annual
data are disseminated with very long delays of 15 months or more in 17 percent of economies
and in 13 percent of economies, quarterly data are released with delays of six months or
more. Long delays in dissemination reduce the relevance of these data for policy analysis and
sometimes result in parallel data compilation efforts by other private or government agencies.
Such parallel systems typically use other data sources or nonstandard methodologies and
produce different results which may cause confusion. The adoption of an Advance Release
Calendar can be an effective first step in this respect.

16
See Dunn et al., “The impact of ESA 2010” (see footnote 2).
 17

The compilation of GDP based on the production approach is more developed than the
expenditure approach in many developing economies. The income approach is less
developed than the other two approaches except for Advanced Economies and the
Commonwealth of Independent States. Independent estimates of production and expenditure
approaches, which allow data comparisons are compiled by 48 percent of economies. The
lack of independently derived data in 52 percent of the economies reduces users’ ability to
judge the quality of the data.

On the vintage of the SNA, about 52 percent of the economies compile GDP data
according to the current vintage, the 2008 SNA, and most others apply the 1993 SNA. In
advanced economies, the adoption of the 2008 SNA has resulted in relatively small changes
to GDP growth, but larger changes, mostly upward, to the level of GDP with impacts on
ratios used for policy purposes such as debt-to-GDP. In low-income economies, the effect is
likely to be small especially as compared to changes that may result from updated benchmark
years or improved data sources. Improved data sources to capture the informal sector, for
example, are likely to lead to more significant revisions.

From a capacity development point of view, priority should be given to updating the
benchmark year, developing quarterly GDP data, deriving production and expenditure
GDP data independently, and improving the timeliness of dissemination. In cases where
updating of the benchmark year and improved source data lead to significant revisions,
backcasting the data to address breaks in the series and effective communication with data
users are needed. The use of an Advance Release Calendar provides data users with
predictability and it signals sound management and transparency of statistical compilation. It
also helps compilers take an organized approach to data compilation.

Future studies may extend to other features of national accounts or more in-depth
analyses of the size of revisions due to improved methods and data sources. For example,
a further study could consider whether GDP captures the informal economy (where relevant)
or estimates of the size of household production or whether consistent time series are
available for an adequate period of time. The appropriate use of deflators to derive GDP in
constant prices could also be considered. Other research topics could include the coverage of
the other components of the national accounts for example the government sector (“G” in the
familiar formula Y= C+I+G+X-M). The System of National Accounts requires
comprehensive coverage of the general government sector but many economies compile data
only for the central government.17 The authors also recommend repeating this study in 3–5
years to monitor how the GDP data and compilation practices improve over time.

17
Dippelsman, Robert, Claudia Dziobek, and Carlos A. Gutiérrez Mangas. 2012. “What Lies Beneath: The
Statistical Definition of Public Sector Debt.” IMF Staff Discussion Note No. 12/09.
 18

REFERENCES

Alexander, Thomas, Claudia Dziobek, Marco Marini, Eric Metreau and Michael Stanger.
2017. “Measure up: A Better Way to Calculate GDP.” IMF Staff Discussion Notes
No. 17/02.

Carson, Carol S, Sarmad Khawaja, and Thomas K Morrison. 2003. "Revisions Policy for
Official Statistics: A Matter of Governance." IMF Working Paper No. 04/87.

Department of Economic and Social Affairs, Statistics Division. 2016. National Accounts
Statistics: Main Aggregates and Detailed Tables, 2016. 2017: United Nations.

Dippelsman, Robert, Claudia Dziobek, and Carlos A. Gutiérrez Mangas. 2012. “What Lies
Beneath: The Statistical Definition of Public Sector Debt.” IMF Staff Discussion Note
No. 12/09.

Dippelsman, Robert, Venkat Josyula, and Eric Métreau. 2016. “Fixed Base Year vs. Chain
Linking in National Accounts: Experience of Sub-Saharan African Countries.” IMF Working
Paper No. 16/133.

Dunn, Marianthi, and Leonidas Akritidis. 2014. "The impact of ESA 2010 on key indicators
of the national accounts in Europe." Eurostat Review on National Accounts and
Macroeconomic Indicators 7-27.

European Commission, IMF, OECD, UN, World Bank. 2009. System of National Accounts,
2008. New York: United Nations.

International Monetary Fund, Statistics Department. 2012. Data Quality Assessment

Framework for National Accounts Statistics. DC: IMF.

International Monetary Fund. Dissemination Standards Bulletin Board. Accessed September

2017. http://dsbb.imf.org/.

International Monetary Fund. 2016. Guide for Participants and Users: Enhanced General
Data Dissemination System.

International Monetary Fund. 2017. Quarterly National Accounts Manual - Concepts, Data
Sources, and Compilation. DC: IMF.

International Monetary Fund. 2013. The Special Data Dissemination Standard: Guide for
Subscribers and Users. DC: IMF.

Olinto Ramos Roberto, Gonzalo Pastor, and Lisbeth Rivas. 2008. “Latin America: Highlights
from the Implementation of the System of National Accounts 1993 (1993 SNA).” IMF
Working Paper No. 08/239.
 19

United Nations Statistics Division. 2017. National Accounts Official Country Data. August
31. Accessed October 2017. http://data.un.org/Explorer.aspx?d=SNA.

United Nations Statistics Division. 2017. National Accounts Statistics: Main Aggregates and
Detailed Tables, 2016. New York: United Nations.

Van de Ven, Peter. 2015. "New standards for compiling national accounts: what’s the impact
on GDP and other macro-economic indicators?" OECD Statistics Brief, February: No. 20.

World Bank. Data on Statistical Capacity. Accessed October 2017.

http://datatopics.worldbank.org/statisticalcapacity/.