Download OECD Compendium of Productivity Indicators 2006

OECD Compendium of
Productivity Indicators
2006
OECD COMPENDIUM OF PRODUCTIVITY
INDICATORS
2006
ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT
ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT
th
Pursuant to Article 1 of the Convention signed in Paris on 14 December 1960, and which came
th
into force on 30 September 1961, the Organisation for Economic Co-operation and Development
(OECD) shall promote policies designed:
•
To achieve the highest sustainable economic growth and employment and a rising standard of
living in member countries, while maintaining financial stability, and thus to contribute to the
development of the world economy.
•
To contribute to sound economic expansion in member as well as non-member countries in
the process of economic development; and
•
To contribute to the expansion of world trade on a multilateral, non-discriminatory basis in
accordance with international obligations.
The original member countries of the OECD are Austria, Belgium, Canada, Denmark, France,
Germany, Greece, Iceland, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain,
Sweden, Switzerland, Turkey, the United Kingdom and the United States. The following countries
th
became members subsequently through accession at the dates indicated hereafter: Japan (28 April
th
th
th
1964), Finland (28 January 1969), Australia (7 June 1971), New Zealand (29 May 1973), Mexico
th
st
th
(18 May 1994), the Czech Republic (21 December 1995), Hungary (7 May 1996), Poland
nd
th
th
(22 November 1996), Korea (12 December 1996) and the Slovak Republic (14 December 2000).
The Commission of the European Communities takes part in the work of the OECD (Article 13 of the
OECD Convention).
© OECD 2006
FOREWORD
Over the past few years, productivity and economic growth have been an important focus of
OECD work. This work has included both efforts to improve the measurement of productivity growth,
as shown in the development of the OECD Productivity Manual, published in 2001, as well as work to
enhance the understanding of the drivers of productivity performance. In the course of this work,
questions about data choices and the measurement of productivity were examined at several
occasions. At the same time, OECD was confronted with a growing interest in internationally
comparable data on productivity growth.
The continued interest of many OECD member countries in productivity led to a decision to
develop an OECD Productivity Database, based on data that were considered to be as comparable
and consistent across countries as possible. This database and related information on methods and
sources is available through the OECD Internet site and free of charge at:
www.oecd.org/statistics/productivity
In 2005, a large number of indicators on productivity were being combined in one document for
the first time to coincide with an OECD workshop on productivity measurement, held in Madrid. The
present document constitutes the 2006 update of the productivity compendium. Its primary source is
the OECD Productivity Database, although indicators are drawn from other sources, such as the
OECD STAN database, which enables productivity calculations for individual industries.
The compendium includes indicators as well as methodological notes and describes the
measurement challenges and data choices that were made as well as the remaining measurement
problems. Further detail is available in a number of specific annexes.
The compendium was prepared for an OECD workshop on productivity measurement and
analysis, held in Bern from 16-18 October 2006. Agnes Cimper, Julien Dupont, Paul Schreyer and
Colin Webb prepared the text and tables in the compendium. Additional contributions to text and
tables from Nadim Ahmad and Pascal Marianna are gratefully acknowledged.
© OECD 2006
3
Compendium of Productivity Indicators
TABLE OF CONTENTS
Highlights
7
Background: why productivity matter?
11
A) Economy-wide indicators of productivity growth
A.1. Growth in GDP per capita
13
14
A.2. Growth in GDP per hour worked
16
A.3. Alternative measures of labour productivity growth
18
A.4. Capital productivity
20
A.5. Growth accounts for OECD countries
22
A.6. The contribution of multi-factor productivity and ICT capital to GDP growth
24
A.7. Labour productivity growth in the business sector
26
B) Productivity levels
B.1. Income and productivity levels
29
30
B.2. Levels of GDP per capita and GDP per hour worked, 1950-2005
32
B.3. Alternative measures of output
34
B.4. Differences in labour productivity by economic activity
36
B.5. Labour productivity and heterogeneity
38
C) Productivity growth by industry
C.1. Contribution of key activities to aggregate productivity growth
41
42
C.2. Productivity growth in manufacturing
44
C.3. Productivity growth in services
46
C.4. Labour productivity of foreign affiliates
48
D) Official multi-factor productivity statistics for OECD member countries
D.1. Australia
51
52
D.2. Canada
54
D.3. New Zealand
56
D.4. Switzerland
58
D.5. United States
60
Annex 1 – OECD productivity database
62
Annex 2 – OECD estimates of labour productivity levels
68
Annex 3 – OECD databases relevant to productivity analysis
References
73
© OECD 2006
76
5
Compendium of Productivity Indicators
HIGHLIGHTS
This compendium presents productivity indicators in four broad areas. The first section presents
economy-wide indicators of productivity growth. It shows that:
•
Over 1995-2005, growth in GDP per hour worked was highest in Ireland, Korea and the Slovak
Republic. In the Czech Republic, Greece, Hungary and Iceland, labour productivity grew much
faster during 2000-2005 compared with the period 1995-2000. Over the same period, labour
productivity growth slowed down in many other OECD countries. Alternative measures of labour
productivity growth, based on net domestic product or gross domestic income, show a very similar
picture, with some exceptions.
•
Capital productivity has declined in most OECD countries over the last fifteen years. The fall in
capital productivity since 1990 has been very pronounced in Japan but also in Spain, Canada,
Portugal and Denmark. Notable exceptions to the decline in output per unit of capital input are
Ireland and Finland where capital productivity grew over most of the last decade.
•
Stronger growth in the major seven countries in the second half of the 1990s was due to several
factors, including higher labour utilisation, capital deepening, notably due to investment in
information and communications technology (ICT), and more rapid multi-factor productivity (MFP)
growth. In France, Germany and in the United States, the contribution of labour input to growth was
positive for the period 1995-2000 but negative for the period 2000-2005. The fall in the contribution
of labour input to growth has also been most pronounced in the United Kingdom. The decline in the
contribution of labour input to growth slowed markedly in Japan. In Canada, France, Germany, Italy
and the United Kingdom, MFP growth fell between the period 1995-2000 and 2000-2005, but it
rose in Japan and in the United States.
•
Investment in ICT accounted for between 0.3 and 0.6 percentage points of growth in GDP over the
period 1995-2005. Australia, Denmark, Sweden, the United Kingdom and the United States
received the largest boost from ICT capital; Japan and Canada a more modest one; and Austria,
France and Germany a much smaller one. In several countries, ICT accounts for the bulk of
capital’s contribution to GDP growth.
The second section presents indicators of productivity levels. It shows that:
•
Iceland, Ireland, Luxembourg, Norway, Switzerland and the United States had the highest levels of
per capita income in 2005, while Belgium, France, Ireland, the Netherlands, Norway and the United
States had the highest levels of GDP per hour worked. The different ranking of certain countries on
these two measures is due to labour utilisation; many European countries have lower levels of
labour utilisation than the United States. This implies that they have fewer people contributing to
GDP; levels of GDP per capita are therefore lower than levels of GDP per hour worked.
•
These rankings change little when alternative measures of output, such as net domestic product or
gross national income, are used. Exceptions are Ireland and Luxembourg that have a lower ranking
on gross national income per hour worked than on GDP per hour worked.
•
Since the 1950s, cross-country differences in income and productivity levels have eroded
considerably. Japan and Korea have experienced the highest rates of catch-up since 1950, while
many European countries experienced strong catch-up with the United States until 1980, but have
fallen back since, Ireland and Korea being among the most notable exceptions.
© OECD 2006
7
•
Australia, Canada, New Zealand and the United Kingdom already had relatively high income levels
in 1950 and have done little catching up since. Eastern European countries, Mexico and Turkey
started with low income levels in the 1950s and have only caught up a little.
•
Industries predominantly involved in the extraction, processing and supply of fuel and energy goods
produced the highest value added per labour unit. These industries were more than twice as
productive as the average industry. Besides the energy-producing industries, those that yield the
most value added per labour unit are industries considered more technology and/or knowledge
intensive. In manufacturing, the chemical industry has the highest relative labour productivity level,
while in services, finance, insurance and telecommunications lead the way.
•
The focus on labour productivity growth generally assumes that homogeneity exists within the predefined industry groups, in other words that the productivity growth observed at the aggregate level
is representative of the growth at the business level. However, comprehensive estimates of
productivity are dependent on business level data. The comparison of the coefficient of variation at
the 2 digit industry level for 6 large OECD countries show that there is considerable business size
class heterogeneity in labour productivity across countries and industries.
The third section presents indicators of productivity growth by industry, and also includes indicators on
the contribution of multinationals to productivity performance. It shows that:
•
In many OECD countries, notably in Spain, Greece, Czech Republic, Sweden, Hungary, the United
Kingdom, the United States, and Australia, business sector services have accounted for the bulk of
labour productivity growth over 2000-2005. However, the manufacturing sector remains important
in the Korea, Sweden, Hungary, Finland and Czech Republic.
•
Within manufacturing, large differences in the rate of productivity growth can be observed.
Electrical and optical equipment is often the industry with the highest rate of productivity growth,
over 15% annually in 1995-2000 for some OECD countries such as Finland, Hungary, Korea,
Sweden and the United States.
•
The variation in labour productivity growth across services sectors is also considerable. Industries
such as wholesale and retail trade, post and telecommunications are typically the services sectors
with the highest rate of productivity growth, while business services, hotels and restaurants, and
transport and storage often have lower rate of labour productivity growth.
•
Labour productivity of foreign affiliates in manufacturing in the United Kingdom is more than two
times higher than the average labour productivity of domestic firms in the manufacturing sector.
•
Foreign affiliates made an important contribution to labour productivity growth in the United States,
accounting for almost a quarter of manufacturing productivity growth over 1995-2001. In the Czech
Republic, France, Sweden and the United Kingdom, the bulk of productivity growth in
manufacturing was due to foreign affiliates. In Japan, foreign affiliates made only a minor
contribution to productivity growth.
The fourth and last section presents the methodology used for the calculation of official multi-factor
productivity statistics published by some OECD countries and how MFP measures differ from those
computed by the OECD. It shows that:
•
8
The Australian Bureau of Statistics has computed and published time series of multi-factor
productivity indices for several years. The ABS’ MFP measures differ in several aspects from the
MFP measures computed by the OECD. First, national data is based on more detailed source data
than the international data. Second, ABS adjusts labour input measures to reflect the composition
of the labour force e.g., by age, education and experience whereas the OECD labour input data is
a simple aggregate of hours worked. Thirdly, capital input as computed by ABS is based on a
© OECD 2005
Compendium of Productivity Indicators
broader scope of capital assets than used by the OECD. In particular, the national data includes
agricultural land and inventories, two assets that are absent from the OECD capital computations.
•
Statistics Canada has computed and published time series of multi-factor productivity indices for a
number of years. The MFP measures published by the Statistics Canada differ in several aspects
from the MFP measures computed by the OECD. First, the national data is significantly more
detailed and also more timely than the international data. Second, labour input measures have
been adjusted by Statistics Canada to reflect the composition of the labour force e.g., by age,
education and experience whereas the OECD labour input data is a simple aggregate of hours.
Thirdly, capital input as computed by Statistics Canada is based on a broader scope of capital
assets than used by the OECD. In particular, the national data includes land and inventories, two
assets that are absent from the OECD capital computations
•
In 2006, Statistics New Zealand released for the first time an official time series of multi-factor
productivity growth. This first dataset relates to the ‘measured sector’, consisting of industries for
which estimates of inputs and outputs are independently derived in constant prices. Excluded are
those industries —mainly government non-market industries whose services such as
administration, health and education, are provided for free or at nominal charge— whose real
value-added is measured in the national accounts largely using input methods, such as number of
employees. Labour input is measured as the total number of hours paid, the number of ordinary
and overtime hours for which an employee is paid. The elements of capital input are compiled at a
detailed and broader level than the OECD’s own estimates of capital services (which exclude for
example, residential buildings). The basic methodology behind the New Zealand MFP estimates
closely follows the methods presented in international documents such as the OECD Productivity
Manual. In some specific aspects, the national measures differ from MFP measures computed by
the OECD (scope of capital assets, sector coverage).
•
In 2006, The Swiss Federal Statistical Office published a first set of MFP estimates for Switzerland.
The basic methodology behind the Switzerland MFP estimates closely follows the methods
presented in international documents such as the OECD Productivity Manual. In some specific
aspects, however, the national measures differ from MFP measures computed by the OECD. The
national data is based on more detailed source than the six-way asset classification used by the
OECD. There is also important difference in the scope of capital measures; in particular the
estimates by the Swiss Federal Office include residential assets which are excluded from the
OECD data.
•
The United States Bureau of Labour Statistics has computed and published time series of multifactor productivity for a number of years. BLS MFP measures differ in several aspects from the
MFP measures computed by the OECD. First, the national data is significantly more detailed and
also more timely than the international data. Second, labour input measures have been adjusted by
BLS to reflect the composition of the labour force e.g., by age, education and experience whereas
the OECD labour input data is a simple aggregate of hours worked.
As a general rule, the national source is to be preferred over the international source for analyses that
relate to the country only whereas the international source is often better suited for comparisons between
countries.
© OECD 2006
9
Compendium of Productivity Indicators
WHY PRODUCTIVITY MATTERS
Productivity isn’t everything, but in the long run it is almost everything. A country’s ability to improve its
standard of living over time depends almost entirely on its ability to raise its output per worker.
Paul Krugman, The Age of Diminishing Expectations (1994)
Productivity is commonly defined as a ratio of a volume measure of output to a volume measure of
inputs and measures how efficiently production inputs are being used in the economy to produce outputs.
While there is no disagreement on this general notion, a look at the productivity literature and its various
applications reveals very quickly that there is neither a unique purpose for, nor a single measure of,
productivity. There is also a general understanding that productivity matters for the standard of living and
economic growth but to answer more specific analytical questions, different measures of productivity are
required. This Compendium has as one of its objectives to show various productivity measures that are
available at the OECD, along with brief information on their interpretation and methodology. In one way or
another, the various measures relate to the broader objectives of productivity measurement, tracing
technology, technical change and efficiency in the economy, in an industry or in a sector. More specific
analytical reasons why the OECD is interested in the measurement of productivity include:
•
Productivity growth is considered a key source of economic growth and competitiveness and
as such forms a basic statistic for many international comparisons and country assessments;
•
Productivity data are also used in the analysis of labour and product markets of OECD
countries. For example, Conway et al. (2006) investigate the link between productivity and
product market regulation across OECD countries;
•
Productivity change constitutes an important element in modelling the productive capacity of
OECD economies. This permits computation of capacity utilisation measures, themselves
important to gauge the position of economies in the business cycle and to forecast economic
growth. In addition, the degree to which an economy’s capacity is used informs analysts about
the pressures from economic demand and thereby about the risk of inflationary developments.
The international perspective typically embraced by the OECD gives rise to some additional
possibilities for analysis but poses also additional difficulties for measurement. Some of these analytical
possibilities as well as the associated measurement issues are indicated in the text accompanying the
productivity indicators.
© OECD 2006
11
Compendium of Productivity Indicators
A. ECONOMY-WIDE INDICATORS OF PRODUCTIVITY GROWTH
© OECD 2006
13
GROWTH IN GDP PER CAPITA
• Growth in GDP per capita remains one of the core
indicators of economic performance. Estimates of
the increase in GDP per capita for OECD countries
for 1995-2000 and 2000-2005 show that rates of
growth in GDP per capita were high for both periods
in the Slovak Republic, Hungary, Korea and Ireland.
In most OECD countries, growth of GDP per capita
was higher in 1995-2000 than in 2000-2005. The
Czech Republic, Greece and the Slovak Republic
were the countries with the largest acceleration
between the two periods; while in most of countries,
notably Ireland, Portugal and Mexico, GDP per
capita grew much slower in 2000-2005 than in 19952000. The estimates shown here are not adjusted for
differences in the business cycle; cyclically adjusted
estimates might show a somewhat different pattern.
• The growth in GDP per capita can be broken down
in a part that is due to labour productivity growth
(see indicator A.2.) and a part that is due to
increased labour utilisation, measured as hours
worked per capita. Growing labour utilisation can
have considerable impacts on the growth of GDP per
capita. A slowing or declining rate of labour
utilisation combined with high labour productivity
growth can be indicative of a greater use of capital or
A.1.
of a dismissal (or failure to employ) low-productivity
workers.
• Compared with the second half of the 1990s, in the
early 2000s, many European countries have
experienced a decrease in their rate of labour
utilisation. This slowdown in labour utilisation growth
was also accompanied by a sharp decline in labour
productivity growth in many European countries.
Only Greece and Japan experienced a pick-up in
both labour utilisation and labour productivity growth
from 1995-2000 to 2000-2005, showing that there
need not be a trade-off between labour productivity
growth and increased labour use.
Sources
 OECD Productivity Database, September
www.oecd.org/statistics/productivity
2006:
 OECD, Annual National Accounts Database, September
2006.
For further reading
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
Growth in GDP per capita
Total economy, percentage change at annual rate
%
9
8
2000-2005
1995-2000
7
6
5
4
3
2
1
Sl
ov
a
k
R
ep
H ubl
un ic
g
G ary
C
re
ze
e
ch K ce
R o re
ep a
ub
Ire lic
l
Po and
la
Tu nd
r
Lu Ic key
x e
N em lan
ew b d
Z e our
al g
Fi and
U
ni A nla
te u n
d st d
K i ra
ng lia
d
Sw o m
ed
e
S n
C pai
an n
N ada
or
w
U
ni O ay
te E
d C
St D
at
e
Ja s
pa
E n
D Uen 19
m
Be a r
lg k
iu
Au m
st
N F ri a
et ra
he n
rl c e
G and
er s
m
a
Sw M ny
it z exic
er o
la
nd
Po It a
rtu ly
ga
l
0
-1
EU19 includes all EU members that are also OECD member countries.
14
© OECD 2005
Compendium of Productivity Indicators
A.1.
GROWTH IN GDP PER CAPITA
Growth in GDP per capita – the contribution of productivity and labour utilisation
Total economy, percentage change at annual rate
Average 1995-2000
=
Growth of GDP per capita
Average 2000-2005
Growth of GDP per hour
worked
+
Growth in labour
utilisation (1)
Slovak Republic
Hungary
Greece
Korea
Czech Republic
8.5
Ireland
Iceland
Luxembourg
New Zealand
Finland
Australia
United Kingdom
Sweden
Spain
Canada
Norway
United States
Japan
OECD19 (2)
Denmark
EU15
Belgium
Austria
France
Netherlands
Germany
Mexico
Switzerland
Italy
Portugal
-2
0
2
4
6
-2
0
2
4
6
-2
0
2
4
6
1. Labour utilisation is measured as hours worked per capita.
2. OECD19 includes Japan, EU15 and NAFTA.
© OECD 2006
15
GROWTH IN GDP PER HOUR WORKED
• Productivity growth can be measured by relating
changes in output to changes in one or more inputs
to production. The most common productivity
measure is labour productivity which links changes
in output to changes in labour input. This key
economic indicator is closely associated with
standards of living.
• In 1995-2000, estimates of GDP per hour worked for
OECD countries show that rates of labour
productivity growth were the highest in Ireland,
Korea and the Slovak Republic; and, while Ireland
marked a slowdown over those last five years,
labour productivity growth for Korea and the Slovak
Republic remained very high in 2000-2005. On the
other hand, labour productivity growth was the
lowest in 1995-2000 for Italy, the Netherlands and
Spain and it remained quite low for Italy over the last
five years.
• Over the past 10 years, labour productivity growth
has varied considerably; in the Czech Republic,
Greece, Hungary and Iceland it grew much faster
during 2000-2005 compared with the period 19952000. Over the same period, labour productivity
growth slowed down in other OECD countries, such
as Austria, Canada, Ireland, Luxembourg, Mexico
and Portugal.
• Since 2000, most OECD countries have experienced
a marked slowdown in labour productivity growth,
exception made of some small countries, such as
A.2.
the Czech Republic, Hungary and Iceland which had
among the highest labour productivity growth over
this last decade, together with Greece, Ireland and
the Slovak Republic.
• The estimates shown here are not adjusted for
differences in the business cycle; cyclically adjusted
estimates might show a somewhat different pattern.
Sources
 OECD Productivity Database, September 2006:
www.oecd.org/statistics/productivity
 OECD, Annual National Accounts Database, September
2006.
For further reading
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 Ahmad, N., F. Lequiller, P. Marianna, D. Pilat, P.
Schreyer and A. Wölfl (2003), “Comparing Labour
Productivity Growth in the OECD Area: The Role of
Measurement”, STI Working Papers 2003/14, OECD,
Paris.
 Beffy, P-O., Patrice Ollivaud, Pete Richardson and
F.Sédillot (2006), “New OECD Methods for Supply-Side
and Medium-Term Assessments: A Capital Services
Approach”, Economics Department Working Paper
2006/482, OECD, Paris.
 Pilat, D. and P. Schreyer (2004), “The OECD Productivity
Database – An Overview”, International Productivity
Monitor, Number 8, Spring, pp. 59-65.
 OECD (2004), “Clocking In (and Out): Several Facets of
Working Time”, OECD Employment Outlook 2004,
Chapter 1, OECD, Paris.
OECD measures of labour productivity growth
The OECD Productivity Manual. There are many different approaches to the measurement of productivity. The calculation and
interpretation of the different measures are not straightforward, particularly for international comparisons. To give guidance to
statisticians, researchers and analysts who work with productivity measures, the OECD released the OECD Productivity Manual in
2001. It is the first comprehensive guide to various productivity measures and focuses on the industry level. It presents the theoretical
foundations of productivity measurement, discusses implementation and measurement issues and is accompanied by examples from
OECD member countries to enhance its usefulness and readability. It also offers a brief discussion of the interpretation and use of
indicators of productivity. See: www.oecd.org/sti/measuring-ind-performance
The OECD Productivity Database. Productivity measures rely heavily on the integration of measures of output and input. Some of
the most important differences among studies of labour productivity growth are linked to choice of data, notably the combination of
employment, hours worked and GDP. To address this problem, OECD has developed a reference database on productivity at the
aggregate level, with a view to resolving the problem of data consistency. In deriving estimates of labour productivity growth for the
economy as a whole, the database combines information on GDP, employment and hours worked. For employment and hours
worked, a special effort is made to use the best available information for each country, based on a consistent matching of data on
employment and annual hours worked per person employed (see Annex 1).
Until 2005, estimates of productivity growth were published in the Annex Tables of the OECD Economic Outlook. These measures
were estimates of labour productivity growth for the business sector and were designed for different purposes, though considered of
equal value to those published in the Productivity Database. However, the two sets of series differed in the following ways: 1) The
measures of labour and multi-factor productivity in the OECD Productivity Database refer to the total economy. They are based on a
detailed assessment of labour and capital input, which incorporates adjustments for average hours worked per person employed and
for capital services. These economy-wide productivity measures provide a close link to changes in GDP per capita. 2) The measures
of labour productivity in the OECD Economic Outlook covered the business sector only and did not adjust for average hours worked
nor for capital services. The main advantage of these measures was that a large part of the economy was excluded, notably the
public sector in which productivity is typically poorly measured. However, reflecting the absence of consistent source information, the
quality and comparability of those series varied over time and between countries, therefore in 2006, OECD Economic Outlook shifted
to a total economy basis and business sector estimates are no longer published. In the longer term however, the intention is for
Economic Outlook to move back to a business sector basis as and when consistent and appropriate data are available from member
country sources. Further information regarding the background and use of such data is given by Beffy et al. (2006). More information
is available on the special website for the database: www.oecd.org/statistics/productivity
16
© OECD 2005
ov
ak
R
ep
ub
lic
C
K
ze
or
ch Ir ea
R e la
ep nd
H ubl
un ic
g
G ary
re
Ic ece
e
Sw lan
e d
U
ni Fi den
te nl
d an
St d
a
N tes
or
w
a
U
ni A Jap y
te u a
d st n
K
O ing ralia
EC d
L u D om
xe 19
m (1
bo )
u
Fr rg
a
Po n c
rt e
C uga
an l
G a
er da
m
Au any
st
EU r i a
SwBelg 15
it z ium
e
D
N e rl a n
ew n d
m
Z e ar
al k
N M and
et e
he x
rla ico
nd
s
It a
Sp l y
ai
n
Sl
Sl
ov
C ak
ze R
ch ep
R ub
ep lic
Hu u bl
ng i c
a
Ko ry
r
Ic ea
el
G and
re
e
Po ce
la
U
ni Ire nd
te la
d n
S d
Sw tate
ed s
e
Ja n
N pa
or n
U
ni F wa
te in y
d la
O Ki n nd
EC gd
D om
Au 19 (
st 1)
ra
Fr lia
G a nc
er e
m
Lu Den any
xe m
m ar
bo k
ur
EU g
C 1
an 5
N B ad
ew el a
g
Ze iu m
al
a
N Au nd
et s
h t
Sw erl ria
i tz and
er s
la
n
Sp d
a
M in
ex
ic
o
I
t
Po al
rtu y
ga
l
Compendium of Productivity Indicators
A.2.
GROWTH IN GDP PER HOUR WORKED
1995-2000 compared with 2000-2005
Total economy, percentage change at annual rate
%
6
© OECD 2006
2000-2005
1995-2000
5
4
3
2
1
0
Total economy, percentage change at annual rate
1995-2005
%
6
5
4
3
2
1
0
1. OECD19 includes Japan, EU15 and NAFTA.
17
ALTERNATIVE MEASURES OF LABOUR PRODUCTIVITY GROWTH
• Real depreciation has grown somewhat faster than real
GDP in the past decade in many OECD countries,
reflecting investment in new technologies and a shift in
the structure of investment and capital stocks towards
shorter-lived assets. As a consequence, NDP per hour
worked has risen somewhat slower than GDP per hour
worked. The gap turned out to be relatively large in the
United States, Denmark and Switzerland whereas in
Finland, in the Slovak Republic and in the United
Kingdom, NDP growth exceeded GDP growth.
• Looking at trend growth of real GDP and GDI per hour
worked over the past ten years shows that differences
between the two measures are relatively small, except
for a few countries (e.g. Finland, Sweden, Korea and
Switzerland). By definition, the difference between the
two measures is most important in those countries that
experienced the largest shifts in their terms of trade
and/or where foreign trade accounts for a large share of
GDP.
A.3.
Sources
 OECD Productivity Database, September 2006:
www.oecd.org/statistics/productivity
 OECD, Annual National Accounts Database, September
2006.
For further reading
 Commission of the European Communities, OECD, IMF,
United Nations, World Bank (1993), System of National
Accounts 1993, Brussels/Luxembourg, New York, Paris,
Washington DC.
 Boarini, Romina, Åsa Johansson and Marco Mira d’Ercole
(2006), Alternative Measures of Well-Being, OECD Statistics
Brief, n°11, May.
 Kohli, Ulrich (2004); “Real GDP, real domestic income and
terms of trade changes”; Journal of International Economics
62.
Alternative measures of output
GDP is a gross measure that does not account for capital used in production. The associated loss in value, depreciation, reduces the
net value of production that is available as net income in any given year. The observation has often been made that a growing part of
capital goods is short-lived (for example computers), and that this structural shift in the composition of assets brings with it a higher
overall depreciation. A case could thus be made to measure productivity on the basis of net output as well as GDP. Countries with a
structure of fixed assets that is biased towards short-lived assets would exhibit a relatively lower NDP per hour worked than GDP per
hour worked, reflecting relatively higher depreciation.
However, net measures require reliable estimates of depreciation and the empirical basis for depreciation estimates is generally not
well established. For similar types of assets, significant differences exist in the service lives and depreciation rates that are used by
different countries. These rates are sometimes based on assumptions more than in-depth empirical studies or are based on evidence
dating back a number of years. This reduces the quality of depreciation estimates as well as their international comparability. The
international GDP-NDP comparisons should thus be interpreted with caution.
It is also useful to compare GDP per hour worked with Gross Domestic Income (GDI) per hour worked over time. Real gross domestic
income (GDI) measures the purchasing power of the total incomes generated by domestic production (including the impact on those
incomes of changes in the terms of trade); it is equal to gross domestic product at constant prices plus the trading gain (or less the
trading loss) resulting from changes in the terms of trade. The terms-of-trade effect arises because real GDI is obtained by deflation
with the price index for domestic final demand rather than the price index for GDP. It measures the rate at which exports can be traded
against imports from the rest of the world. The difference between movements in GDP at constant prices and real GDI are not always
small. If imports and exports are large relative to GDP, and if the commodity composition of the goods and services which make up
imports and exports are very different, the scope for potential trading gains or losses may be large. If the prices of a country’s export
rise faster (or fall more slowly) then the prices of its imports — that is if its terms of trade improve — less exports are needed to pay for
a given volume of imports so that at a given level of domestic production of goods and services can be reallocated from exports to
consumption or capital formation. Thus, an improvement in terms of trade makes it possible for an increased volume of goods and
services to be purchased by residents out of the incomes generated by a given level of domestic production.
Despite its analytical usefulness, it should be borne in mind that GDI is a measure of income and not a measure of production.
However, as has been pointed out by some authors (Kohli 2004), terms of trade effects resemble technical change. Changes in
constant price GDP would only reflect technical change whereas GDI is a measure that reflects both technical change and terms of
trade. This makes GDI a meaningful measure even in the context of production analysis.
18
© OECD 2005
Compendium of Productivity Indicators
A.3.
ALTERNATIVE MEASURES OF LABOUR PRODUCTIVITY GROWTH
Growth in National Domestic Product per hour worked compared with growth in GDP per hour worked
1995-2005
Total economy, percentage change at annual rate
%
NDP per hour w orked
GDP per hour w orked
6
5
4
3
2
1
Ita
ly
D
Sw
en
i tz
m
ar
er
k
la
nd
(2
00
4)
N
et
he
rl a
nd
s
(2
00
4)
Be
lg
iu
m
M
ex
ic
o
Au
st
ria
(2
00
4)
an
y
a
er
m
G
C
an
ad
Fr
an
ce
Fi
nl
an
d
ni
te
d
Sw
ed
en
Ic
el
an
d
St
Au
at
es
st
ra
li a
(2
00
U
ni
4)
te
d
Ki
ng
do
m
Sl
ov
ak
U
R
ep
ub
lic
(2
00
4)
0
Growth in Gross Domestic Income per hour worked compared with growth in GDP per hour worked
1995-2005
Total economy, percentage change at annual rate
%
GDI per hour w orked
GDP per hour w orked
6
5
4
3
2
1
Sl
ov
ak
R
ep
ub
Ita
ly
lic
(2
00
4)
Ko
Ire
re
la
a
nd
(2
00
4)
Ic
el
an
d
Sw
ed
en
Fi
nl
U
an
ni
d
te
d
Au
St
st
at
ra
es
li a
(2
00
Ja
4)
pa
n
(2
U
ni
00
te
4)
d
Ki
ng
do
m
Fr
an
ce
G
er
C
m
an
an
ad
y
a
(2
00
N
4)
ew
Au
Ze
s
al
tri
an
a
d
(2
00
4)
Be
M
lg
ex
iu
m
ic
o
(2
00
4)
Sw
D
i tz
en
er
m
la
ar
nd
k
(2
00
N
4)
et
he
rl a
nd
s
0
© OECD 2006
19
CAPITAL PRODUCTIVITY GROWTH
• While labour productivity is the most common partial
productivity measure, capital productivity provides
another, supplementary piece of information about
productivity growth. Capital productivity is measured
as the ratio between output and capital input.
• Capital productivity has to be distinguished from the
rate of return to capital. The former is a physical,
partial productivity measure; the latter is an income
measure that relates capital income to the value of
the capital stock.
• Capital productivity has declined in most OECD
countries over the last fifteen years. The fall in
capital productivity since 1990 has been very
pronounced in Japan but also in Spain, Canada,
Portugal and Denmark. Notable exceptions to the
decline in output per unit of capital input are Ireland
and Finland where capital productivity grew over
most of the last decade.
• Two important drivers shape capital productivity:
overall efficiency or multi-factor productivity growth
and the amount of labour input per unit of capital in
production. The fewer hours worked are available
per unit of capital, the lower capital productivity.
Generally, the cost of using capital has declined
relative to labour, so that the amount of labour input
per capital input has declined as well, leading to the
observed fall in capital productivity.
A.4.
• Although most countries experienced a decline in
capital productivity, the rates at which capital
productivity fell has varied significantly over time. In
Japan, the United States, Sweden, Spain, Australia
and New Zealand, the decline in capital productivity
slowed noticeably between the period 1990-95 and
1995-2000, before to decline markedly between the
period 1995-2000 and 2000-2005. The most
remarkable shift could be observed in Finland where
capital productivity declined in the first half of the
1990s and then rose in the period afterwards. Note,
however, that like other productivity measures,
capital productivity varies considerably with the
business cycle as no adjustments have been made
for variations in the rate of capacity utilization.
Sources
 OECD Productivity Database, September 2006:
www.oecd.org/statistics/productivity
For further reading
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 Schreyer P. (2003), Capital Stocks, Capital Services and
Multi-factor Productivity, OECD Economic Studies No 37.
 Schreyer, P., P-E. Bignon and J. Dupont (2003), “OECD
Capital Services Estimates: Methodology and A First Set
of Results”, OECD Statistics Working Papers 2003/6,
OECD, Paris.
Box 1. Measuring capital input
From the viewpoint of the economic theory, the objective is to measure the flow of productive services that capital delivers in
production. This measure relies on the assumption that capital services are a fixed proportion of the productive capital stock. In which
case, the rate of change of capital services coincides with the rate of change of the capital stock. The latter is estimated by
cumulating investment flows and correcting them for retirement, wear and tear and obsolescence. The aggregate flow of capital
services is obtained by weighting the flow of capital services of each type of asset by its share in total capital income. More
information on capital measurement can be found on the documentation OECD Productivity Database.
Box 2. International comparability of price indices
Price indices are vital for in measuring volume investment, capital services and user costs. Accurate price indices should be constant
quality deflators that reflect price changes for a given performance of ICT investment goods. Thus, observed price changes of
‘computer boxes’ have to be quality-adjusted for comparison of different vintages. There are differences how countries deal with
quality adjustment with possible consequences for the international comparability of price and volume measures of ICT investment.
In particular, those countries that employ hedonic methods to construct ICT deflators tend to register a larger drop in ICT prices than
countries that do not. The OECD uses a set of ‘harmonised’ deflators to control for some of the differences in methodology and
assumes that the ratios between ICT and non-ICT asset prices evolve in a similar manner across countries, using the United States
as the benchmark. Although no claim is made that the ‘harmonised’ deflator is necessarily the correct price index for a given country,
we feel that the possible error due to using a harmonised price index is smaller than the bias arising from comparing capital services
based on national deflators. However, from an accounting perspective, adjusting the price index for investment goods for any country
implies an adjustment of the volume index of output. In most cases, such an adjustment would increase the measured rate of volume
output change. At the same time, effects on the economy-wide rate of GDP growth appear to be contained (see Schreyer (2001) for a
discussion). More information is available on the special website for the database: www.oecd.org/statistics/productivity
20
© OECD 2005
Compendium of Productivity Indicators
A.4.
CAPITAL PRODUCTIVITY GROWTH
Growth in capital productivity, 1995-2000 compared with 2000-2005
Total economy, percentage change at annual rate
2000-2005
%
1995-2000
5
4
3
2
1
0
-1
-2
-3
Fr
an
Ire
ce
la
nd
(
2
Sw
00
ed
3)
en
(2
00
Be
3)
lg
iu
m
(2
00
3)
G
er
Au
m
an
st
U
ra
y
ni
lia
te
d
(2
Ki
00
ng
4)
do
m
(2
00
U
ni
3)
te
d
S
ta
G
te
re
ec
s
N
e
et
(2
he
00
rl a
3)
nd
s
(2
00
3)
C
an
ad
Ja
a
pa
n
(2
00
Au
4)
st
r ia
(2
00
3)
Ita
ly
(
20
Po
03
rtu
)
ga
l
(2
D
00
en
3)
m
ar
k
(2
00
3)
Fi
nl
an
d
(2
00
Sp
3)
N
ai
ew
n
(
2
Ze
00
al
4)
an
d
(2
00
2)
-4
1
Growth in capital productivity, 1990-1995 compared with 1995-2000
Total economy, percentage change at annual rate
%
1995-2000
1990-1995
5
4
3
2
1
0
-1
-2
-3
Ja
pa
n
Po
r tu
ga
ni
l
te
d
St
at
es
U
Sp
ai
n
Ita
ly
Be
lg
iu
m
G
er
m
an
y
C
an
ad
a
G
re
ec
N
ew
e
Ze
al
U
an
ni
d
te
d
Ki
ng
do
m
Sw
ed
en
D
en
m
ar
k
Fr
an
ce
N
et
he
rla
nd
s
Au
st
ra
lia
Au
st
r ia
Ir e
la
nd
Fi
nl
an
d
-4
1. 1991-1995 for Germany.
© OECD 2006
21
GROWTH ACCOUNTS FOR OECD COUNTRIES
• Stronger growth in the major seven countries in the
second half of the 1990s was due to several factors,
including higher labour utilisation, capital deepening,
notably due to investment in information and
communications technology (ICT), and more rapid
multi-factor productivity (MFP) growth. In France,
Germany and in the United States, the contribution
of labour input to growth was positive for the period
1995-2000 but negative for the period 2000-2005.
The fall in the contribution of labour input to growth
has also been most pronounced in the United
Kingdom. The decline in the contribution of labour
input to growth slowed markedly in Japan. In
Canada, France, Germany, Italy and the United
Kingdom, MFP growth fell between the period 19952000 and 2000-2005, but it rose in Japan and in the
United States.
• Investment in ICT accounted for between 0.3 and
0.6 percentage points of growth in GDP over the
period 1995-2005. Australia, Denmark, Sweden, the
United Kingdom and the United States received the
largest boost from ICT capital; Japan and Canada a
more modest one; and Austria, France and Germany
a much smaller one. In several countries, ICT
accounts for the bulk of capital’s contribution to GDP
growth.
A.5.
• In Canada, Finland, Ireland, the Netherlands, New
Zealand and Spain, increased labour utilisation
made a large contribution to growth of GDP over
1995-2005.
• In Ireland, Finland, Greece and the United States,
MFP growth was also an important source of GDP
growth. In Austria, Denmark, Italy, the Netherlands
and Spain, MFP growth was very low or negative
between 1995 and 2005.
Source
 OECD Productivity Database, September 2006
www.oecd.org/statistics/productivity
For further reading
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 Schreyer, P., P.E. Bignon and J. Dupont (2003), “OECD
Capital Services Estimates: Methodology and a First Set
of Results”, OECD Statistics Working Paper 2003/6,
OECD, Paris.
 Schreyer, P. (2004), “Capital Stocks, Capital Services
and Multi-factor Productivity Measures”, OECD
Economic Studies No. 37, 2003/2, OECD, Paris, pp. 163184.
Growth accounting
Economic growth can be increased in several ways; by increasing the amount and types of labour and capital used in production, and
by attaining greater overall efficiency in how these factors of production are used together, i.e. higher multi-factor productivity. Growth
accounting involves breaking down growth of GDP into these contributions; i.e. labour input, capital input and MFP. The growth
accounting model is based on the microeconomic theory of production and rests on a number of assumptions, among which the
following are important: i) production technology can be represented by a production function relating total GDP to the primary inputs
labour L and capital services K; ii) this production function exhibits constant returns to scale; and iii) product and factor markets are
characterised by perfect competition.
For any desired level of output, the firm minimises costs of inputs, subject to the production technology discussed above. Factor input
markets are competitive, so that the firm takes factor prices as given and adjusts quantities of factor inputs to minimise costs. The
rate of growth of output is a weighted average of the rates of growth of the various inputs and of the multi-factor productivity term.
The weights attached to each input are the output elasticities for each factor of production. Output elasticities cannot be directly
observed, however, and the factor shares of labour and capital are often used as weights.
22
© OECD 2005
Compendium of Productivity Indicators
A.5.
GROWTH ACCOUNTS FOR OECD COUNTRIES
Contributions to growth of GDP, G7 countries, 1995-2000 and 2000-2005
Percentage points
Labour input
%
ICT capital
Non-ICT capital
Germany
Italy
1
Multi-factor productivity
1995-2000
2000-2005
5
4
3
2
1
0
-1
Canada
France
Japan
United
Kingdom
Contributions to GDP growth 1995-2005
Percentage points
Labour input
ICT capital
Non-ICT capital
United
States
2
Multi-factor productivity
%
9
8
7
6
5
4
3
2
1
0
Ja
pa
n
ly
an
y
Ge
rm
Ita
De
nm
ark
Be
lgi
um
nc
e
Fra
Au
str
ia
Ca
na
da
Un
ited
St
ate
s
Ne
w
Ze
ala
Un
nd
ite
dK
ing
do
m
Ne
th e
rla
nd
s
Po
rtu
ga
l
Sw
ed
en
lia
Au
str
a
Sp
ain
Fin
lan
d
Ire
lan
d
Gr
ee
ce
-1
1. 1995-2003 for Italy and the United Kingdom; 1995-2004 for Japan.
2. 2000-2002 for New Zealand; 2000-2003 for Austria, Belgium, Denmark, Finland, Greece, Ireland, Italy, Netherlands,
Portugal, Sweden, and the United Kingdom; 2000-2004 for Australia, Japan and Spain.
© OECD 2006
23
GROWTH ACCOUNTS – THE CONTRIBUTION OF MULTI-FACTOR
PRODUCTIVITY AND ICT CAPITAL TO GDP GROWTH
• Multi-factor productivity growth was one of the
factors that helped strengthen growth in Greece, the
United States, Spain, Japan, Sweden and New
Zealand between the periods 1995-2000 and 20002005. In other countries, including Ireland, Portugal,
Australia, Finland, Italy, Canada Germany, Italy, the
United Kingdom, Austria, Belgium, Denmark, the
Netherlands and Spain, MFP growth slowed down
from 1995-2000 to 2000-2005. Multi-factor
productivity rose during the period 1995-2000 and
declined in the period 2000-2005 in the Netherlands,
Austria, Denmark, Portugal and Italy.
• The contribution of ICT capital to GDP growth
decreased in most of OECD countries between
1995-2000 and 2000-2005. The decrease over this
period was particularly large for Sweden and the
United States and Japan, and smallest for the United
Kingdom, Canada, the Netherlands and Denmark.
The contribution of ICT capital to GDP growth has
A.6.
increased from 1995-2000 to 2000-2005 in Greece,
New Zealand, Finland, Italy and Austria.
Source
 OECD Productivity Database, September 2006
www.oecd.org/statistics/productivity
 Groningen Growth and Development Centre, June 2005.
For further reading
 Lequiller, F., N. Ahmad, S. Varjonen, W. Cave and
K.H. Ahn (2003), “Report of the OECD Task Force on
Software Measurement in the National Accounts”, OECD
Statistics Working Paper 2003/1, OECD, Paris.
 Ahmad, N. (2003), “Measuring Investment in Software”,
STI Working Paper, 2003/6, OECD, Paris.
 Schreyer, P., P.E. Bignon and J. Dupont (2003), “OECD
Capital Services Estimates: Methodology and a First Set
of Results”, OECD Statistics Working Paper 2003/6,
OECD, Paris.
The contribution of ICT capital to GDP growth
Correct measurement of investment in ICT in both nominal and volume terms is crucial for estimating its contribution to economic
growth and performance. Data availability and measurement of investment in ICT based on national accounts (SNA93) vary
considerably across OECD countries, especially as regards measurement of investment in software, deflators applied, breakdown by
institutional sector and temporal coverage. In the national accounts, expenditure on ICT products is considered as investment only if
the products can be physically isolated (i.e. ICT embodied in equipment is considered not as investment but as intermediate
consumption). This means that investment in ICT may be underestimated and the order of magnitude of the underestimation may differ
depending on how intermediate consumption and investment are treated in each country’s accounts.
In particular, expenditure on software has only very recently been treated as capital expenditure in the national accounts, and
methodologies still vary considerably across countries. Difficulties for measuring software investment are also linked to the ways in
which software can be acquired, e.g. via rental and licences or embedded in hardware. Moreover, software is often developed on own
account. To tackle the specific problems relating to software in the context of the SNA93 revision of the national accounts, a joint
OECD-EU Task Force on the Measurement of Software in the National Accounts has developed recommendations concerning the
capitalisation of software (Lequiller, et al., 2003; Ahmad, 2003). These are now being implemented by OECD member countries.
24
© OECD 2005
Compendium of Productivity Indicators
A.6.
GROWTH ACCOUNTS – THE CONTRIBUTION OF MULTI-FACTOR
PRODUCTIVITY AND ICT CAPITAL TO GDP GROWTH
Multi-factor productivity growth, 1995-2000 and 2000-2005
In percentage points
2000-2005
1
1995-2000
5%
4%
3%
2%
1%
0%
U
Ire
l
an
G d
ni ree
te
ce
d
St
at
e
Sw s
ed
e
Fi n
nl
an
U
d
ni
Ja
te
d
p
Ki a n
ng
do
Au m
N
ew stra
Z e lia
al
an
d
Fr
an
c
G
er e
m
a
C ny
an
a
Be da
lg
iu
m
S
N
p
et
he ain
rla
nd
s
Au
st
r
D
en ia
m
a
Po rk
rtu
ga
l
It a
ly
-1%
The contributions of ICT capital, 1995-2000 and 2000-2005
1
In percentage points
2000-2005
1995-2000
0.9%
0.8%
0.7%
0.6%
0.5%
0.4%
0.3%
0.2%
0.1%
It a
l
Ja y
pa
n
C
an
ad
Sw a
ed
e
Fr n
an
c
Au e
N
s
et
he tria
rl a
nd
Ire s
la
nd
Sp
a
Po i n
rtu
G ga
er
l
m
an
y
N
Au
s
tra
D
en lia
ew ma
Z e rk
al
a
Be n d
lg
iu
U
m
ni
te Gre
d
Ki ece
ng
do
m
U Fin
ni
l
a
te
d nd
St
at
es
0.0%
1. 2000-2002 for New-Zealand; 2000-2003 for Austria, Belgium, Denmark, Finland, Greece, Ireland, Italy, the Netherlands,
Portugal, Sweden and the United Kingdom, 2000-2004 for Australia, Japan and Spain.
© OECD 2006
25
LABOUR PRODUCTIVITY GROWTH IN THE BUSINESS SECTOR
•
Productivity growth can be measured by relating
changes in output to changes in one or more inputs to
production. The most common productivity measure is
labour productivity which links output to labour input.
The OECD publishes several labour productivity
measures: those presented in the OECD Productivity
Database refer to the total economy and relate
changes in output to changes in hours worked, while
those published in the OECD Economic Outlook N. 78,
relate output per employed person in the business
sector. In 2006, due to difficulties in obtaining reliable,
consistent and comparable data of labour input for the
business sector, OECD Economic Outlook shifted to a
total economy basis and business sector estimates are
no longer published, see text box below for details.
•
As with other productivity indicators, the growth in
business sector labour productivity is quite variable
from year to year, because employment generally
moves more slowly than value added. Businesses do
not immediately employ more staff when there is an
upturn and they do not immediately lay off staff when
there is a down turn.
•
The following charts show considerable differences in
business sector labour productivity growth between
countries. In 1995-2005, Greece, Iceland, Korea,
Poland and Turkey had the highest growth in business
sector labour productivity, while growth for Italy,
Germany, Mexico, the Netherlands, Spain and
Switzerland was slower over the same period. In the
Czech Republic, Norway, Turkey, Greece and the
A.7.
United States, business sector labour productivity grew
much faster from 2000-2005 than from 1995-2000,
while it slowed down over the period in other OECD
countries, notably Australia, Italy, Luxembourg,
Mexico, Poland and Portugal. From 2000 onwards,
business sector productivity slowed down in most
OECD countries and only picked up in a few countries.
Sources
 OECD Economic Outlook, No. 78.
For further reading
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 Ahmad, N., F. Lequiller, P. Marianna, D. Pilat, P.
Schreyer and A. Wölfl (2003), “Comparing Growth in
GDP and Labour Productivity: Measurement Issues”,
OECD Statistics Brief, No. 7 December, OECD, Paris.
 Ahmad, N., F. Lequiller, P. Marianna, D. Pilat, P.
Schreyer and A. Wölfl (2003), “Comparing Labour
Productivity Growth in the OECD Area: The Role of
Measurement”, STI Working Papers 2003/14, OECD,
Paris.
 Beffy, P-O., Patrice Ollivaud, Pete Richardson and
F.Sédillot (2006), “New OECD Methods for Supply-Side
and Medium-Term Assessments: A Capital Services
Approach”, Economics Department Working Paper
2006/482, OECD, Paris.
 Pilat, D. and P. Schreyer (2004), “The OECD Productivity
Database – An Overview”, International Productivity
Monitor, Number 8, Spring, pp. 59-65.
Atkinson, Tony (2005) Atkinson Review: Final report;
available on http://www.statistics.gov.uk.
Measuring business sector productivity
Business sector output is defined as economy-wide GDP less the government wage bill, less net indirect taxes and government
consumption of fixed capital. Business sector employment is defined as total economy employment less public sector employment.
The business sector thus excludes outputs and inputs that relate to government. The main advantage of business sector measures is
that they exclude a large part of the economy, i.e. the public sector, in which productivity is typically poorly measured. In the absence
of market production and market prices (as is the case for example for general administration), government output is normally
measured by the inputs that are used to produce this output, implying zero productivity growth.
Measures of business sector productivity are also important because this sector ultimately determines the development of an
economy’s potential output and of the economy's tax base. At the same time, the precise definition of the business sector is not always
straightforward. Also, it can be difficult to allocate certain enterprises to the market or to the non-market sector. In their national
statistics, countries use different approaches towards measuring the output and inputs of the business sector. Differences arise in
particular in the treatment of health or education services that may be considered part of non-market production in some countries and
part of market production in others.
Over the past few years, several countries have started to develop output-based measures of production for non-market activities. For
example, the United Kingdom has put in place measures of output of education and health services in their national accounts. A report
to the U.K. Office of National Statistics (Atkinson 2005) provides a comprehensive discussion of the issues involved in measuring nonmarket output.
Despite their conceptual advantages, business sector measures of labour productivity are subject to greater constraints of data
availability than labour productivity measures for the total economy. In particular, it is difficult to obtain reliable and internationally
comparable data for the total number of hours worked in the business sector, making it necessary to revert to the total number of
persons employed as a measure of labour input. As long as average hours worked per person do not change significantly over time,
this makes little difference to the resulting productivity measures. However, when there are marked shifts in working hours, the two
measures of labour input may give rise to quite different rates of labour productivity growth. In 2006, OECD Economic Outlook shifted
to a total economy basis and business sector estimates are no longer published. In the longer term however, the intention is for
Economic Outlook to move back to a business sector basis as and when consistent and appropriate data are available from member
country sources. Further information regarding the background and use of such data is given by Beffy et al. (2006).
26
© OECD 2005
la
Ic nd
el
an
Ko d
re
Tu a
r
C
ze G key
ch re
e
N Rep ce
ew u
b
U Z e lic
ni al
te an
d
St d
a
Sw tes
ed
Fi en
nl
an
Au d
U
ni De stria
te
d nm
Ki a r
ng k
d
Au om
st
ra
C lia
an
Po ada
rtu
g
Fr al
an
B
c
Lu elg e
xe iu
m m
bo
u
N rg
or
w
a
M y
e
G xic
N e rm o
et
he any
Sw rlan
it z ds
er
la
nd
Sp
ai
n
Ita
ly
Po
Po
la
n
Tu d
rk
G ey
re
C
e
ze Ic ce
ch el
R and
ep
ub
li
Ko c
re
U N a
ni or
te w
d
St a y
at
U
e
ni
te Sw s
d
e
Ki de
ng n
do
Fi m
n
D lan
en d
m
a
A u rk
st
r
Fr ia
an
N Be ce
ew lg
i
Z e um
al
Au and
st
ra
C lia
an
G ad
N e rm a
et
he a ny
rl a
nd
s
Sw Sp
a
itz in
er
l
Po and
rtu
g
M al
ex
ico
Lu
xe It
m aly
bo
ur
g
Compendium of Productivity Indicators
A.7.
LABOUR PRODUCTIVITY GROWTH IN THE BUSINESS SECTOR
Growth in GDP per employee in business sector, 1995-2000 compared with 2000-2005
Percentage change at annual rate
%
7
2000-2005
%
7
© OECD 2005
1995-2000
6
5
4
3
2
1
-1
0
-2
-3
Growth in GDP per employee in business sector, 1995-2005
Percentage change at annual rate
1995-2005
6
5
4
3
2
1
0
-1
-2
-3
27
Compendium of Productivity Indicators
B. PRODUCTIVITY LEVELS
© OECD 2005
29
INCOME AND PRODUCTIVITY LEVELS
• In 2005, GDP per capita in the OECD area ranged
from over USD 35 000 in Iceland, Ireland,
Luxembourg, Norway, Switzerland and the United
States to less than USD 15 000 in Mexico, Poland
and Turkey. For most OECD countries, income
levels are 70-85% of US income levels.
• The differences in income reflect a combination of
labour productivity, measured as GDP per hour
worked, and labour utilisation, measured as hours
worked per capita. A country’s labour productivity
level is typically the most significant factor in
determining differences in income, particularly in
countries with low levels of GDP per capita.
• Most OECD countries have higher levels of GDP per
hour worked than GDP per capita because they
have lower levels of labour utilisation. The difference
between income and productivity levels is largest in
European countries; in 2005, GDP per hour worked
surpasses the US productivity level in Belgium,
France, Ireland, the Netherlands and Norway,
whereas income levels in most of these countries
are substantially lower than in the United States.
• In many OECD countries, labour use, as measured
by hours worked per capita, is substantially lower
B.1.
than in the United States. This is because of
disparities in working hours but also in several
countries because of high unemployment and low
participation of the working-age population in the
labour market. In Iceland and Korea, however,
labour input per capita in 2005 is considerably higher
than in the United States, owing to relatively long
working hours and high rates of labour force
participation. Labour input per capita is also
relatively high in Canada, the Czech Republic,
Japan, New Zealand and Switzerland.
Sources
 OECD, Productivity Database, September 2006:
www.oecd.org/statistics/productivity.
 OECD, Annual National Accounts Database, September
2006.
For further reading
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 Pilat, D. and P. Schreyer (2004), “The OECD Productivity
Database – An Overview”, International Productivity
Monitor, No. 8, Spring, pp. 59-65.
 OECD (2004), “Clocking In (and Out): Several Facets of
Working Time”, OECD Employment Outlook 2004,
Chapter 1, OECD, Paris.
Comparisons of income and productivity levels
Comparisons of income and productivity levels face several measurement problems (see also Annex 2 for further detail). First, they
require comparable data on output. In the 1993 System of National Accounts (SNA), the measurement and definition of GDP are
treated systematically across countries. Most countries have implemented this system; in the OECD area, Turkey is the only
exception, and its output is likely to be understated relative to other OECD countries. Other differences, such as the measurement of
software investment, also affect the comparability of GDP across countries, although the differences are typically quite small.
The second problem is the measurement of labour input. Some countries integrate the measurement of labour input in the national
accounts; this may ensure that estimates of labour input are consistent with those of output. In most countries, however, employment
data are derived from labour force surveys which are not entirely consistent with the national accounts. Labour input also requires
measures of hours worked, which are typically derived either from labour force surveys or from business surveys. Several OECD
countries estimate hours worked from a combination of these sources or integrate these sources in a system of labour accounts,
which are comparable to the national accounts. The OECD Productivity Database includes estimates of total hours worked which aim
at consistency between estimates of employment and hours worked. The cross-country comparability of hours worked remains
somewhat limited, however, with a margin of uncertainty in estimates of productivity levels.
Third, international comparisons require price ratios to convert output expressed in a national currency into a common unit. Exchange
rates are of limited use for this purpose because they are volatile and reflect many influences, including capital movements and trade
flows. The alternative is to use purchasing power parities (PPP), which measure the relative prices of the same basket of
consumption goods in different countries. The estimates shown here are based on official OECD Purchasing Power Parities for 2005.
30
© OECD 2005
Compendium of Productivity Indicators
B.1.
INCOME AND PRODUCTIVITY LEVELS
Income and productivity levels, 2005
Percentage point differences with respect to the United States
Percentage gap in GDP per capita
Effect of labour utilisation1
Gap in GDP per hour worked
Turkey (2)
Mexico
Poland
Slovak Republic
Hungary
Portugal
Czech Republic
Korea
Greece
New Zealand
Spain
EU-19 (3)
Italy
OECD
Germany
Euro-zone (4)
France (5)
Japan
Finland
United Kingdom
Belgium
Sweden
Austria
Australia
Canada
Denmark
Netherlands
Iceland
Switzerland
Ireland
Norway
-80
-60
-40
-20
0
-80
-60
-40
-20
0
-80
-60
-40
-20
0
1. Based on total hours worked per capita.
2. GDP for Turkey is based on the 1968 System of National Accounts.
3. EU member countries that are also member countries of the OECD.
4. Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain.
5. Includes overseas departments.
© OECD 2005
31
LEVELS OF GDP PER CAPITA AND GDP PER HOUR WORKED, 1950-2005
• Cross-country differences in GDP per capita and
labour productivity in the OECD area have eroded
considerably since the 1950s. Over the 1950s and
1960s, income levels in OECD countries were
catching up with those of the United States except in
Australia, New Zealand and the United Kingdom. In
the 1970s, this phenomenon was less widespread
and the rate of catch-up fell except in Korea. In the
1980s, there was even less catch-up, as GDP per
capita grew more slowly than in the United States in
19 OECD countries. The same was true for
16 OECD countries from 1990-2005, with Ireland
and Korea being the most notable exceptions.
• Japan and Korea had the highest rates of catch-up
over the period 1950-2005, with GDP per capita
growing more rapidly than in the United States, by
2.5% and 3.3%, respectively. Rates of catch-up were
much lower, typically below 1% a year, in most of
western Europe. Australia, New Zealand, the United
Kingdom and Canada already had relatively high
income levels in 1950 and have done little catching
up with the United States. Switzerland has seen a
marked decline in its relative income level. Eastern
European countries, Mexico and Turkey started with
B.2.
low income levels in the 1950s and have only caught
up a little.
• Changes in levels of GDP per hour worked show a
slightly different pattern. Out of 27 OECD countries,
for which data are available for a long period, only
Mexico, Canada, Switzerland and Australia have not
been catching up almost continuously with
US productivity levels over the post-war period.
Several European countries now stand even with the
United States in terms of average labour productivity
and some have surpassed US productivity levels.
Sources
 OECD, Productivity database, September 2006.
www.oecd.org/statistics/productivity.
 OECD, Annual National Accounts Database, September
2006.
For further reading
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 Angus Maddison (2001), The World Economy: A
Millennial Perspective, Development Centre Studies,
OECD, Paris.
Comparisons of income and productivity levels over time
Comparisons of income and productivity levels for a particular year (see B.1) can be updated over time by using time series of GDP,
population, employment and hours worked. Time series for GDP, population, employment and hours worked are all derived from the
OECD Productivity Database. However, this database only dates back to the early 1970s, therefore for earlier years, estimates were
derived by using data for GDP and population from Angus Maddison (2001), The World Economy: A Millennial Perspective, OECD
Development Centre, OECD, Paris.
GDP per hour worked in the OECD area, 1950, 1973 and 2005
United States = 100
1950
1973
2005
120
United States = 100
100
80
60
40
20
32
Ita
ly
Sp
ai
Ic n
el
an
Ja d
pa
Ne Gr n
w eec
Ze e
al
a
Sl
ov Po nd
a k rt u
R ga
ep l
ub
C
ze Hu lic
ch ng
R a ry
ep
ub
li
Ko c
re
Po a
la
n
M d
ex
ic
o
No
rw
Be a y
lg
iu
m
N Irel
et
a
he nd
rla
nd
s
U
F
ni ra
te n c
d
e
St
G ate
er s
m
D any
en
m
Sw ark
ed
U
e
ni
te Au n
d
s
Ki tria
ng
d
Au om
st
ra
F lia
Sw inla
itz nd
er
la
C nd
an
ad
a
0
© OECD 2005
Compendium of Productivity Indicators
B.2.
LEVELS OF GDP PER CAPITA AND GDP PER HOUR WORKED, 1950-2005
Catch-up and convergence in OECD income levels, 1950-2005, United States = 100
Medium rate of catch-up (<=1.1% annually)
140
140
120
High-income, no catch-up (<0.1% annually)
Belgium
Germany
Finland
Netherlands
France
Norw ay
120
Italy
100
100
80
80
60
60
40
40
20
20
Australia
New Zealand
Sw itzerland
Denmark
Canada
Sw eden
United Kingdom
0
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
0
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Rapid catch-up (>1.1% annually)
Low starting point, low rates of catch-up (< 0.3%
annually)
140
140
Greece
Ireland
Japan
Korea
120
120
Portugal
Spain
Austria
100
100
80
80
60
60
40
40
20
20
0
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Czech Republic
Mexico
Hungary
Poland
Turkey
Slovak Republic
0
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Source: 2005 income levels from indicator B.1; previous years based on OECD productivity database and Angus Maddison (2001),
The World Economy: A Millennial Perspective, Development Centre Studies, OECD, Paris.
© OECD 2005
33
ALTERNATIVE MEASURES OF OUTPUT
• When data in the national accounts are taken at face
value, a comparison between GDP and NDP per hour
worked hardly changes the ranking of countries with
respect to relative levels of labour productivity. On
average, NDP accounts for about 85% of GDP, although
there is some variation across countries. Nonetheless,
in Japan, the Czech Republic and the Slovak Republic,
depreciation is large in relation to GDP. By contrast,
Greece, Mexico, or the United States are countries
where NDP is relatively high compared to GDP. Similar
to the level comparisons of NDP and GDP per hour
worked, the growth rates of NDP and GDP per hour
worked can be compared (see A.3).
• In most OECD countries, the difference between GDP,
NDP and GNI per hour worked is small since gross
income inflows from abroad tends to be offset by gross
outflows. This indicates that GDP per hour worked is a
useful proxy for other measures of output in productivity
calculations, especially when data availability restricts
international comparability. However, the difference is
significant for a few countries, notably Ireland and
Luxembourg. GDP per hour worked, GNI per hour
worked and GDI per hour worked increased in all OECD
countries between 1995 and 2005 (or latest year
B.3.
available), but the largest increase is registered for
Ireland, which is also an example for a country where
GDP was significantly higher than GNI, whereas
Switzerland is a case where the opposite holds.
Sources
 OECD Productivity Database, September 2006.
www.oecd.org/statistics/productivity
 OECD, Annual National Accounts Database, September
2006.
For further reading
 Commission of the European Communities, OECD, IMF,
United Nations, World Bank (1993), System of National
Accounts 1993, Brussels/Luxembourg, New York, Paris,
Washington DC.
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 Ahmad, N., F. Lequiller, P. Marianna, D. Pilat, P. Schreyer
and A. Wölfl (2003), “Comparing Labour Productivity Growth
in the OECD Area: The Role of Measurement”, STI Working
Papers 2003/14, OECD, Paris.
 Boarini, Romina, Åsa Johansson and Marco Mira d’Ercole
(2006), Alternative Measures of Well-Being, OECD Statistics
Brief, n°11, May.
Alternative measures of output – level comparisons
GDP is a gross measure that does not account for capital used in production. The associated loss in value, depreciation, reduces the
net value of production that is available as net income in any given year. The observation has often been made that a growing part of
capital goods is short-lived (for example computers), and that this structural shift in the composition of assets leads to higher overall
depreciation. A case could thus be made to measure productivity on the basis of net domestic product as well as GDP. Countries with
a structure of fixed assets that is biased towards short-lived assets would exhibit a relatively lower Net Domestic Product (NDP) per
hour worked than GDP per hour worked, reflecting relatively higher depreciation.
However, net measures require reliable estimates of depreciation and the empirical basis for depreciation estimates is generally not
well established. For similar types of assets, significant differences exist in the service lives and depreciation rates that are used by
different countries. These rates are sometimes based on assumptions more than in-depth empirical studies or are based on evidence
dating back a number of years. This reduces the quality of depreciation estimates as well as their international comparability. The
international GDP-NDP comparisons should thus be interpreted with caution.
It is also useful to compare GDP with Gross National Income (GNI). This comparison permits to take into account the payments
received from or sent to the rest of world. GNI is equal to GDP less net taxes on production and imports, less compensation of
employees and property income payable to the rest of the world plus the corresponding items receiving from the rest of the world. For
example, when company profits are transferred abroad, this leads to GNI being lower than GDP. Conversely, when foreign affiliates or
domestic firms or residents abroad transfer payment to the domestic economy, this will raise GNI relative to GDP.
At the same time it should be noted that income-based measures, while useful from a welfare perspective, are not necessarily the
preferred indicator when it comes to measuring productivity. The purpose of productivity measures is to capture inputs and outputs in a
production process and GNI may not be the best measure of output for this purpose.
34
© OECD 2005
Lu
xe
m No
bo
r
ur way
g
(2
00
Be 4 )
l
N
et giu m
he
rl a
nd
s
F
Ire
ra
n
la
n d ce
(
U
ni 2 00
te
4)
d
St
at
e
G
er s
m
an
D
en y
m
a
Sw rk
ed
en
U
ni Aus
te
t
r
d
Ki ia
ng
do
m
Au
s t Fin
ra
l
li a and
(2
00
Sw
4)
i tz
er
la
Ita
nd
ly
(2
00
4)
Sp
ai
C
Ic n
a
N
ew na elan
d
Ze a ( d
2
al
an 0 04
d
)
J a (20
04
pa
)
n
(2
00
4
)
Sl
G
ov
re
ec
ak
e
P
R
C
ze ep ort
ug
ch ub
a
li
R
ep c (2 l
u b 00
4)
li c
(2
00
4)
M
ex Kor
ea
ic
o
(2
00
4)
Lu
xe
N
m
bo orw
ur
ay
g
(2
00
Be 4 )
l
N
et giu m
he
rl a
nd
s
F
Ire
ra
la
n d nce
(
U
ni 2 00
te
4)
d
St
at
e
G
er s
m
an
D
en y
m
a
Sw rk
ed
en
U
ni Aus
te
tr
d
Ki ia
ng
do
Au
m
s t Fin
ra
l
li a and
(2
00
Sw
4)
i tz
er
la
I
nd tal
(2 y
00
4)
Sp
ai
n
C
Ic
a
N
ew na ela
n
d
d
Ze a
al (20
an
04
d
)
J a (20
04
pa
)
n
(2
00
4
)
Sl
G
ov
re
ec
ak
e
P
R
C
ze ep ort
ug
ch ub
a
li
R
ep c (2 l
0
ub
li c 04)
(2
00
4)
M
K
ex
or
ea
ic
o
(2
00
4)
Compendium of Productivity Indicators
B.3.
USD mln
USD mln
© OECD 2006
ALTERNATIVE MEASURES OF OUTPUT
Gross Domestic Product and National Domestic Product per hour worked, 2005
Total economy, USD, current prices, current PPPs
70
NDP per hour w orked
GNI per hour w orked
GDP per hour w orked
60
50
40
30
20
10
0
Gross Domestic Product and Gross National Income per hour worked, 2005
Total economy, USD, current prices, current PPPs
70
GDP per hour w orked
60
50
40
30
20
10
0
35
DIFFERENCES IN LABOUR PRODUCTIVITY BY ECONOMIC ACTIVITY
•
The ratio of value added to employment provides
an indication of which industries yield relatively
high value added per unit of labour input. Although
total employment is not the best measure of labour
input for this purpose (see box), a reasonably
clear pattern emerges.
•
In 2002, industries predominantly involved in the
extraction, processing and supply of fuel and
energy goods continued to produce the highest
value added per labour unit. These industries
were more than twice as productive as the
average industry. They account for about 4% of
total OECD value added and are typically highly
capital-intensive.
•
•
Besides the energy-producing industries, those
that yield the most value added per labour unit are
those considered more technology and/or
knowledge intensive (see box). In manufacturing,
the chemical industry has the highest relative
labour productivity level, while in services, finance,
insurance and telecommunications lead the way.
•
B.4.
The relatively low ratios of value added to
employment for business services are partly due to
the fact that this category covers a diverse range of
activities from renting of machinery and equipment,
computer services and R&D to legal, accounting,
architectural and engineering services. Use of a more
detailed activity breakdown would allow more high
value added activities to be revealed.
Data Resources:
 OECD, STAN Indicators Database, August 2005.
 OECD, STAN Database, August 2005.
For further reading
 Ahmad, Nadim, Francois Lequiller, Pascal Marianna, Dirk
Pilat, Paul Schreyer and Anita Wölfl (2003), “Comparing
Labour Productivity Growth in the OECD Area: The Role
of Measurement”, STI Working Paper 2003/14, OECD,
Paris.
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 OECD (2001), OECD Science, Technology and Industry
Scoreboard, D.4. OECD, Paris.
 OECD (2003), OECD Science, Technology and Industry
Scoreboard section D and Annex 1. OECD, Paris.
Construction, wholesale and retail trade, hotels
and restaurants and textiles show relatively low
levels of labour productivity in all three major
OECD regions. These industries are typically
highly labour-intensive, have a high proportion of
low-skilled jobs and are not considered hightechnology sectors.
Measuring labour productivity by industry
The indicators shown here are determined by data availability and simply measure value added per person employed. Further
adjustments to labour input, including adjustment for part-time work and hours worked per worker, can be made for certain OECD
countries but international comparisons by industry are not yet feasible. A few other notes apply to the indicators:
•
For the labour productivity levels, 2002 value added at current prices was used. For the European Union, member countries'
value added data were aggregated after applying 2002 US dollar GDP purchasing power parities (PPPs). Industry-specific PPPs
are preferable, but are not available for all sectors and countries.
•
The labour productivity levels by industry are relative to the total non-agriculture business sector. This consists of all industries
except i) Agriculture, hunting, forestry and fishing (ISIC 01-05); ii) Real estate activities (ISIC 70 - for value added this includes
imputed rent of owner occupied dwellings); and iii) Community, social and personal services (ISIC 75-99 - this includes mainly
non-market activities such as public administration, education and health for which, in some countries, there are still some
measurement problems). This activity-based definition is not to be confused with an institutional definition of the business sector
(see A.7).
•
Sectors that are considered technology- and/or knowledge-intensive are highlighted in the graphs. For manufacturing, R&D
intensity is the main determinant broadly identifying Chemicals (ISIC 24), Machinery and equipment (ISIC 29-33) and Transport
equipment (ISIC 34-45) as high or medium-high technology sectors (see Annex 1, of STI Scoreboard 2003 for more details). For
services, preliminary analyses of embodied technology (based on input-output tables) and composition of workforce skills have
complemented limited R&D intensity information to describe the following broad service sectors as knowledge intensive: Post and
telecommunications (ISIC 64); Finance and Insurance (ISIC 65-67) and Business services (ISIC 71-74).
•
The European Union in this section covers the 15 member countries up to 2004 plus two new members, the Czech Republic and
Hungary.
•
Data on value added for the EU countries is expressed in basic prices; for Japan and the United States, it is at factor costs.
36
© OECD 2006
Compendium of Productivity Indicators
B.4.
DIFFERENCES IN LABOUR PRODUCTIVITY BY ECONOMIC ACTIVITY
Labour productivity levels relative to the total non-agriculture business sector, 2002
United States
E l e c t r ic it y , g a s a n d w a t e r s u p p ly
P e t r o l e u m r e f in in g
M i n in g a n d q u a r r y in g
C h e m ic a l s
F in a n c e a n d I n s u r a n c e
P o s t a n d t e l e c o m m u n ic a t io n s
T r a n s p o r t e q u ip m e n t
P a p e r , p r i n t in g , p u b li s h in g
M a c h i n e r y a n d e q u ip m e n t
F o o d , d r in k , to b a c c o
N o n - m e t a lli c m in e r a l s
R u b b e r a n d p la s t ic s
B a s ic m e t a l s a n d p r o d u c t s
B u s in e s s s e r v ic e s
T ra n s p o rt a n d s to r a g e
W o o d a n d m a n u f a c t u r in g n e c
C o n s t r u c t io n
T e x t il e s , c lo t h in g
T r a d e , h o te ls a n d r e s t a u r a n ts
0
1
2
3
4
5
Japan
P e tr o le u m r e fin in g
E l e c t r i c i t y , g a s a n d w a t e r s u p p ly
C h e m i c a ls
F in a n c e a n d In s u ra n c e
P o s t a n d te le c o m m u n ic a tio n s
T ra n s p o rt e q u ip m e n t
B u s i n e s s s e r v ic e s
P a p e r , p r in t in g , p u b l i s h i n g
M i n in g a n d q u a r r y i n g
M a c h in e ry a n d e q u ip m e n t
N o n - m e t a l li c m in e r a ls
B a s ic m e ta ls a n d p r o d u c ts
R u b b e r a n d p la s tic s
T ra n s p o rt a n d s to ra g e
F o o d , d rin k , to b a c c o
T r a d e , h o t e ls a n d r e s t a u r a n t s
C o n s tr u c tio n
W o o d a n d m a n u fa c tu rin g n e c
T e x t i le s , c lo t h i n g
0
1
2
3
4
5
European Union
P e t r o l e u m r e fin in g
M in in g a n d q u a r r y in g
E le c t r ic it y , g a s a n d w a t e r s u p p ly
C h e m ic a ls
P o s t a n d te le c o m m u n ic a ti o n s
F in a n c e a n d I n s u r a n c e
T r a n s p o r t e q u ip m e n t
P a p e r, p r in tin g , p u b lis h in g
T ra n s p o rt a n d s to ra g e
N o n - m e t a ll ic m in e ra ls
B u s in e s s s e r v ic e s
M a c h in e r y a n d e q u ip m e n t
F o o d , d r in k , to b a c c o
R u b b e r a n d p la s t ic s
B a s ic m e ta ls a n d p r o d u c ts
C o n s tr u c tio n
W o o d a n d m a n u f a c t u rin g n e c
T r a d e , h o te ls a n d r e s t a u r a n ts
T e x tile s , c lo th in g
0
© OECD 2006
1
2
3
4
5
37
LABOUR PRODUCTIVITY AND HETEROGENEITY
•
Many examples of productivity analyses focus on
relatively aggregated industries, typically at the 2,
or sometimes, 3-digit industry level. The focus is
typically on productivity growth and, so, these
analyses generally assume that homogeneity
exists within these pre-defined industry groups. In
other words: that the productivity growth observed
at the aggregate level is representative of the
growth at the business level. Where this
assumption does not hold it is possible that
changes in 2-digit labour productivity, say, could
be attributed to labour merely because of changes
in the market share of heterogeneous businesses
(as well as entries and exits into the population).
•
Comprehensive estimates of productivity are
therefore dependent on business level data.
Unfortunately
internationally
comparable
databases of businesses covering whole
economies are not currently available; usually
because of restrictions that preserve the
anonymity and confidentiality of businesses and
their data. The OECD’s Structural and
Demographic Business Statistics database
however provides a useful compromise between
the need for very detailed data on businesses and
the need for international comparability. The
database provides information on value-added
and employment in most OECD countries at the 4
digit industry level and by size class, and, so,
allows researchers to conduct their analysis of
productivity levels at a more detailed level;
although information on prices and price change
are also needed when the focus is on productivity
growth. That said the database is able to provide a
diagnostic assessment of the robustness of the
homogeneity assumption.
•
The accompanying table to this section tries to do
this by comparing the coefficient of variation at the
2 digit industry level for 6 large OECD countries.
The coefficient is calculated as the standard
deviation of labour productivity estimates across 5
size classes normalised by the simple un-weighted
average of the labour productivity in these 5 size
classes.
•
The table shows considerable heterogeneity in
labour productivity figures across industries; in
other words that labour productivity figures can
differ significantly by business. Of course such a
comparison can overstate the true scale of
heterogeneity across a sector since businesses
within a given size class may indeed be
homogeneous and indeed those in the largest size
class will contribute disproportionately to overall
value-added and employment. But the point of the
table is merely to illustrate that heterogeneity can
be considerable. Moreover coefficients of variation
within a 2 digit industry are similarly sized when
one compares labour productivity estimates
38
B.5.
across its 4 digit industries (see OECD Structural
and Demographic Business Statistics, 1996-2003).
•
The table also shows that in many industries,
particularly in manufacturing, and in most
countries, the larger the business the higher the
labour productivity. This will possibly, in part,
reflect higher degrees of capital investment in
larger businesses but it may also indicate
economies of scale.
•
In Japan virtually all sectors show increasing
productivity the larger the business (90%). But in
the
United
States
the
percentage
for
manufacturing is close to zero. Interestingly this
statistic increases dramatically (81%) for the
United States if the smallest businesses are
excluded. This may reflect the fact that small
businesses tend to be newer and more innovative
than older, medium size, businesses, reflecting in
part the role of creative destruction, but it may also
reflect data comparability. Ideally the estimates
would have been produced using full-timeequivalent measures or number of hours worked
but this has not been possible and instead, for
France, Germany, Italy, Japan and the United
Kingdom the number of persons engaged were
used and for the United States number of
employees. Because small businesses will
typically have a higher percentage of persons
engaged, and not on the pay-roll, it is possible that
the estimates of relative labour productivity for
small businesses in the United States are biased
upwards. Additionally the numerator for the United
States uses turnover, whereas in Japan (census
based) and EU countries (factor costs) valueadded is used.
•
What is also clear from the table and perhaps not
surprising is that the coefficients in service sectors
are typically higher than in the manufacturing
sector,
reflecting
possibly
the
greater
heterogeneity in services. The table also shows a
lower correlation between business size and
labour productivity in the service sector than in
manufacturing; perhaps again reflecting the larger
heterogeneity.
Data Resources:
 OECD, SDBS Database, September 2006.
For further reading
 Ahmad et al (2003), “Comparing Labour Productivity
Growth in the OECD Area: The Role of Measurement”,
STI Working Paper 2003/14, OECD, Paris.
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 OECD (2003), OECD Science, Technology and Industry
Scoreboard section D and Annex 1. OECD, Paris
 OECD (2006), Structural and Demographic Business
Statistics, 1993-2004. OECD, Paris.
© OECD 2006
Compendium of Productivity Indicators
B.5.
LABOUR PRODUCTIVITY AND HETEROGENEITY
Size Class Heterogeneity in Labour Productivity
ISIC Revision 3 and code description
15 Food products and beverages
16 Tobacco products
17 Textiles
18 Wearing apparel and fur
19 Leather and articles of leather; footwear
20 Wood & products of wood & cork, exc. furniture
21 Paper and paper products
22 Printing, publishing and recorded media
23 Coke and petroleum products; nuclear fuel
24 Chemicals and chemical products
25 Rubber and plastic products
26 Other non-metallic mineral products
27 Basic metals
28 Fabricated metal products, exc. machinery
29 Machinery and equipment, nec
30 Office, accounting & computing machinery
31 Electrical machinery & apparatus, nec
32 Radio, TV, communication equipment
33 Medical, precision & optical instrum.; watch & clocks
34 Motor vehicles, trailers and semi-trailers
35 Other transport equipment
36 Furniture; manufacturing, nec
37 Recycling
40-41 Electricity, gas and water supply
40 Electricity, gas, steam and hot water supply
41 Collection, purification and distribution of water
45 Construction
50 Sale, maint.,repair vehic/cycle, person., h/hold gd.
51 W/sale. trade & commission exc motor veh. & cycle
52 Retail trade exc. motor veh. & cyc.; repair h/hold gds.
55 Hotels and restaurants
60 Land transport, transport via pipelines
61 Water transport
62 Air transport
63 Support. & aux. transport act., travel agencies activ.
64 Post and telecommunications
65 Financial intermediation, except insurance and pension funding
66 Insurance and pension funding, except compulsory social security
67 Activities auxiliary to financial intermediation
70 Real estate activities
71 Renting of mach. without operator; rent. of pers. gds.
72 Computer and related activities
73 Research and development
74 Other business activities
Average
Average
Germany
France
Italy
Japan
United Kingdom United States
C of Var Mono C of Var Mono C of Var Mono C of Var Mono C of Var Mono C of Var Mono
0.29
Y
0.42
N
0.36
N
0.84
N
0.34
Y
0.12
N
0.24
Y
0.35
Y
0.07
N
0.23
N
0.36
N
0.26
N
0.57
Y
0.30
Y
0.11
N
0.37
N
0.31
N
0.36
Y
0.28
Y
0.29
N
0.22
N
0.31
Y
0.04
N
0.32
Y
0.44
Y
0.15
N
0.20
N
0.31
Y
0.19
N
0.33
Y
0.74
Y
0.34
N
0.30
N
0.21
N
0.22
Y
0.55
Y
0.45
Y
0.20
Y
0.32
N
2.55
N
0.38
N
0.37
N
0.85
N
0.29
Y
0.19
Y
0.27
Y
0.58
Y
0.33
N
0.43
N
0.28
N
0.08
N
0.23
Y
0.51
Y
0.17
Y
0.18
N
0.18
Y
0.55
Y
0.41
Y
0.41
Y
0.23
N
0.24
N
0.36
Y
0.26
N
0.21
Y
0.69
Y
0.16
N
0.37
N
0.26
Y
0.06
N
0.23
Y
0.43
Y
0.24
Y
0.23
Y
0.13
Y
0.20
Y
0.43
Y
0.09
N
0.40
N
0.37
N
0.43
N
0.42
N
0.43
N
0.31
Y
0.15
Y
0.30
Y
0.44
Y
0.07
Y
0.17
N
0.35
Y
0.76
Y
0.22
N
0.35
N
0.21
N
0.32
N
0.47
Y
0.32
N
0.34
N
0.89
N
0.19
N
0.64
Y
0.12
N
0.62
N
0.28
N
2.10
N
0.30
Y
0.53
Y
0.27
N
0.39
N
0.25
Y
0.15
Y
0.39
Y
0.06
N
0.31
N
0.06
N
0.32
N
0.88
N
0.13
N
0.76
N
0.29
N
0.93
N
0.16
N
0.36
N
0.12
N
0.37
Y
0.28
Y
0.11
Y
0.36
Y
0.12
Y
0.22
N
0.09
Y
0.28
N
0.45
Y
0.65
N
0.04
N
0.23
N
0.09
N
0.38
N
0.06
N
0.21
N
0.10
N
0.10
N
0.06
N
0.23
N
0.15
N
0.27
N
0.17
Y
0.16
N
0.22
N
0.16
N
0.20
N
0.50
N
0.22
N
0.56
N
0.29
N
0.38
N
0.80
N
0.17
N
0.40
N
0.83
N
0.29
N
0.14
N
0.24
N
0.22
N
0.12
N
0.09
N
0.46
N
0.39
N
0.55
Y
0.29
N
0.19
N
0.27
N
0.37
N
0.38
N
0.13
N
0.27
N
0.67
N
0.35
N
0.22
N
0.29
N
0.66
N
0.85
N
0.07
N
0.17
N
0.36
Y
0.13
Y
0.37
Y
0.11
N
0.22
N
0.13
N
0.35
N
0.32
Y
0.37
N
0.07
N
0.21
N
0.14
N
0.16
N
0.19
N
Total
0.46
0.25
0.53
0.90
0.16
Manufacturing
0.59
0.32
0.79
0.90
0.19
0.03
0.00
Services
0.20
0.14
0.21
0.07
0.00
Total: Excluding smallest size class
0.61
0.41
0.56
0.95
0.30
0.61
Manufacturing
0.76
0.43
0.84
0.95
0.43
0.81
Services
0.30
0.29
0.21
0.07
0.44
Notes: "C of Var"reflects the coefficient of variation. It measures the standard deviation of labour productivity normalised by the average of labour productivity by size class. It is important to note that the denominator is a simple
average of the 5 size classes used and, so, is not weighted by the relative contributions of a particular size class to value-added or employment. "Mono" returns a "Y" if labour productivity increases monotonically by size class and
"N" otherwise.Figures in bold ("N") refer to cases where labour productivity is monotonically increasing when the smallest size class is excluded. Averages are based on "Y"=1 and "N"=0. For averages excluding the smallest size
class bolded "N"s are also equal to 1.
© OECD 2006
39
Compendium of Productivity Indicators
C. PRODUCTIVITY GROWTH BY INDUSTRY
© OECD 2006
41
CONTRIBUTION OF KEY ACTIVITIES TO AGGREGATE PRODUCTIVITY
GROWTH
• A breakdown of productivity growth by activity can
show which industries are particularly important for
overall productivity performance. In many OECD
countries, notably in Spain, Greece, Czech Republic,
Sweden, Hungary, the United Kingdom, the United
States, and Australia, business sector services have
accounted for the bulk of labour productivity growth
over 2000-2005.
• However, the manufacturing sector remains important
in the Korea, Sweden, Hungary, Finland and Czech
Republic.
• For several countries, as Spain, Hungary, Czech
Republic, the contribution of business services has
increased between the periods 1995-2000 and 20002005. This is linked to their growing share in total value
added. However, it also reflects stronger labour
productivity growth in some OECD countries, such as
Greece, Hungary, and Spain. In Portugal, Luxembourg,
C.1.
Italy, Canada and Australia, on the other hand, labour
productivity growth in business services slowed over
the past decade, a trend that can also be observed at
the aggregate level (see A.7).
Data Resources:
 OECD, Annual National Accounts Database, September
2006.
For further reading
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 Wölfl, A. (2003), “Productivity Growth in Service Industries:
An Assessment of Recent Patterns and the Role of
Measurement”, STI Working Paper 2003/7, OECD, Paris.
 Wölfl, A. (2005), “The Service Economy in OECD
Countries”, STI Working Paper 2005/3, OECD, Paris.
Measuring labour productivity growth by industry
Labour productivity growth can be calculated as the difference between the rate of growth of output or value added and the rate of
growth of labour input. Calculating a sector’s contribution to aggregate productivity growth requires a number of simple steps, as
explained in the OECD Productivity Manual. First, the aggregate rate of change in value added is a share-weighted average of the
industry-specific rate of change in value added, with weights reflecting the current price share of each industry in value added. On the
input side, aggregation of industry-level labour input is achieved by weighting the growth rates in total employment (since hours worked
by industry are not available across OECD countries) with each industry’s share in total labour compensation. Aggregate labour
productivity growth can then be calculated as the difference between aggregate growth in value added and aggregate growth in labour
input. An industry’s contribution to aggregate labour productivity growth is therefore the difference between its contribution to total
value added and total labour input. If value added and labour shares are the same, total labour productivity growth is a simple weighted
average of industry-specific labour productivity growth. Similar approaches can be followed when production, instead of value added, is
used as the output measure. Difficulties in measuring output and productivity in services sectors should also be taken into
consideration when interpreting the results (see Wölfl, 2003).
42
© OECD 2006
Compendium of Productivity Indicators
C.1.
CONTRIBUTION OF KEY ACTIVITIES TO AGGREGATE PRODUCTIVITY
GROWTH
Contribution to growth of value added per person employed
Annual average growth rates and contributions in percentage points
1995-2000
Business sector services
Other services
Manufacturing
Other industries
5
3.0
4
total economy labour productivity grow th
3
2.5
2.1
2.4
2.9
2.2
1.4
2
2.6
1.3
1.8
1.5
2.8
3.1
2.5
1.9
1.9
1.3
1.3
1.1
0.8
1
0.9
0
Sp
ai
n
Ita
ly
U
G
re
ni
ec
te
e
d
Ki
ng
do
m
Au
st
ra
li a
N
or
w
N
a
et
he y
rla
nd
s
C
an
ad
a
Po
C
r tu
ze
ch
ga
l
R
ep
ub
li c
Ko
re
a
Ja
pa
n
Fi
n
la
Lu
nd
xe
m
bo
ur
g
D
en
m
a
Sw rk
ed
en
Au
st
ria
G
er
m
an
y
Fr
an
ce
Be
lg
iu
m
H
un
ga
ry
-1
2000-2005
Business sector services
5
1
Other services
Manufacturing
Other industries
4.3
4
3.6
3.8
3.4
2.9
total economy labour productivity grow th
0.9
3
1.8
2
1.6
1.6
1.4
1
1.4
1.6
1.0
1.3
1.0
1.1
0.7
1.1
0.8
0.1
0
0.0
-0.5
Ja
pa
n
Ko
re
a
Be
lg
N
i
u
et
he m
rla
nd
s
C
an
ad
a
D
en
m
ar
k
Fi
nl
an
d
Au
st
ria
G
er
m
an
y
Fr
an
Lu
ce
xe
m
bo
ur
Po g
rtu
ga
l
Ita
ly
Sp
ai
n
C
G
ze
re
ec
ch
e
R
ep
ub
l
Sw ic
ed
en
H
un
U
ni
ga
te
ry
d
Ki
ng
U
do
ni
te
m
d
St
at
es
Au
st
ra
lia
N
or
w
ay
-1
1. Or latest year available.
© OECD 2006
43
PRODUCTIVITY GROWTH IN MANUFACTURING
• The manufacturing sector has traditionally been the
main driver of aggregate productivity growth in
OECD countries. While its contribution to aggregate
productivity growth has become less important in
recent years (see C.1), notably in some OECD
countries, it still shows strong performance in many
industries.
• For
most
OECD
countries,
manufacturing
productivity growth was slower over 2000-2003 than
over 1995-2000, with the exception of a few
countries, notably Australia and Switzerland. The
difference is particularly marked in Canada, Korea
and Poland, where the productivity growth in
manufacturing marked a slowdown in the early 2000
thus revealing a strong structural shift in the
manufacturing sector.
C.2.
1995-2000, the productivity growth for most
countries was quite high and over 15% annually for a
few countries, such as Finland, Hungary, Korea,
Sweden and the United States. Nevertheless, most
OECD countries experienced slower productivity
growth in electrical and optical equipment in 20002003 than in 1995-2000. Due to rapid technological
progress, measuring productivity growth in this
industry is particularly difficult. Some countries have
introduced so-called hedonic deflators to address the
rapid changes in quality and characteristics that
characterise this industry. Other countries have not,
which implies that the international comparability of
productivity growth rates in this industry is likely to
be somewhat limited.
Data Resources:
 OECD, STAN Indicators Database, November 2005.
• Within manufacturing, large differences can be
observed. High- and medium-high technology
industries (see Box), such as machinery, electrical
and optical equipment, chemicals and motor vehicles
have typically experienced relatively high rates of
productivity growth. Low-technology manufacturing
industries, such as wood and textiles, have often
observed slightly lower rates of productivity growth.
However, growth rates remained quite high in 20002003 for some countries, in particular in the textiles
sector where the Czech Republic, France, the United
Kingdom and the United States had a yearly
productivity growth above 7%.
 OECD, STAN Database, November 2005.
For further reading:
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 Pilat, D., A. Cimper, K. Olsen and C. Webb (2006), “The
Changing Nature of Manufacturing in OECD Economies”,
STI Working Paper, OECD, Paris.
 Triplett, Jack (2006), Handbook on Hedonic Indexes and
Quality Adjustment in Price Indexes: Special Application
to Information Technology Products, OECD, Paris.
 Wölfl, A. (2003), “Productivity Growth in Service
Industries: An Assessment of Recent Patterns and the
Role of Measurement”, STI Working Paper 2003/7,
OECD, Paris, www.oecd.org/sti/working-papers
• Developments in certain industries are also
noteworthy. Electrical and optical equipment is
among the manufacturing industries with the highest
rate of productivity growth. Over the time-period
Hedonic methods
Examining the role of ICT-producing sectors in economic growth is heavily influenced by measurement problems, both regarding
outputs and inputs. The key measurement problem for the manufacturing of ICT goods on both the output and input side concerns
prices, in particular how to statistically capture significant quality improvements associated with technological advances in goods such
as computers and semi-conductors. The use of hedonic deflators is generally considered as the best way to address these problems.
Several countries currently use hedonic methods to deflate output in the computer industry (e.g. Canada, Denmark, France, Sweden
and the United States). However, these countries do not use exactly the same method. Some countries, such as the United States,
apply their own hedonic deflator, others apply the US hedonic deflator adjusted for exchange rates, and yet other OECD countries
apply conventional deflators.
Adjusting for these methodological differences in computer deflators for the purpose of a cross-country comparison is difficult, since
there are considerable cross-country differences in industrial specialisation. Only few OECD countries produce computers, where price
falls have been very rapid; many only produce peripheral equipment, such as computer terminals. Similar differences in industry
composition exist in Radio, Television and Communication Equipment (ISIC 32), which includes the semi-conductor industry. The
differences in the composition of output are typically larger than in computer investment, where standardised approaches have been
applied (e.g. Schreyer, et al. 2003).
44
© OECD 2006
ze
ch
%
© OECD 2006
U
e
15
10
5
-5
0
20
Wood
15
10
5
-5
0
-10
U Sw
ni
t e
ze ed den
ch Sta
R te
ep s
ub
l
Ja ic
p
Fi an
nl
an
Ko d
H rea
un
D ga
en ry
m
a
Fr rk
an
c
A
Sw u e
it z stria
er
U
l
a
ni
te M nd
d
Ki exic
ng o
do
m
Sp
G ain
re
P e
L u o r ce
xe tug
m al
bo
u
N rg
or
Be wa
y
N
l
et giu
he m
rl
G and
er s
m
an
y
Sl
ov C It a
ak an ly
R ad
ep a
ub
lic
Machinery2
C
m
bo
ur
g
ni Belg
te
iu
d
m
Ki
ng
U
do
ni
te
m
d
St
at
es
D
en
m
ar
k
Fr
an
ce
C
an
ad
a
Ja
pa
n
Fi
nl
an
d
H
un
ga
ry
Sp
ai
n
Ko
re
a
ni
te
d
S
Au tate
Sl
st s
ov
ak K rali
R or a
ep ea
u
G blic
re
e
Ja ce
p
Ic a
e n
H lan
un d
g
Po ary
U
ni
te S lan
d w d
Ki ed
ng en
C
d
ze
ch M om
R ex
Sw ep ico
it z ubl
i
D erla c
en n
d
m
Be a
lg rk
i
Fi um
nl
a
Au nd
Lu F str
xe ra ia
m nc
bo e
N urg
or
w
a
N Ge Spa y
ew rm in
Z e an
al y
N Po and
et rt
he ug
rla al
C nd
an s
ad
a
It a
ly
U
1995-2000 compared with
High- and medium-high technology manufactures
%
8
%
Chemicals
10
5
-5
0
-10
-15
%
14
12
10
8
6
4
2
0
-2
-4
-6
-8
%
2000-2003
It a
G
ly
er
m
an
y
G
re
ec
e
Sl
M
ov
ex
ak
ico
R
ep
ub
lic
U
Lu
xe
14
12
10
8
6
4
2
0
-2
-4
-6
St
at
e
G s
re
e
N ce
or
G wa
er y
m
an
y
Ja
p
U
ni D e an
nm
te
d
Ki ark
ng
do
Po m
rtu
Be gal
lg
iu
m
Fr
an
Sw ce
N
e
et
he den
rla
nd
Fi s
nl
an
M d
ex
ico
It a
Au ly
st
ria
Ko
r
C ea
an
ad
a
Sw Sp
it z ain
er
la
nd
nc
s
20
ni
te
d
Fr
a
la
nd
om
pa
n
ng
d
Ja
et
he
r
Ki
an
y
ni
te
d
St
a
Sw tes
ed
e
Ja n
p
Fi an
nla
n
Fr d
an
L u K ce
xe o r
m ea
Sw b o
it z urg
er
l
H an
N un d
et g
he ar
rla y
D nd
en s
m
U
a
ni
te A rk
d us
Ki tri
ng a
d
G om
er
m
Be a ny
lg
i
Po um
rtu
ga
Sp l
ai
n
I
C t aly
an
a
G da
re
Sl
ec
ov
e
M
ak
e
R xic
ep o
ub
lic
U
%
12
U
N
d
nd
St
at
es
er
la
ar
y
n
%
15
G
er
m
ni
te
d
it z
un
g
e
H
ni un
te g
d ar
St y
a
Po tes
rtu
U
ni
te N gal
d or
Ki w
ng ay
D do m
en
m
a
M rk
ex
Au ico
st
Fr ria
an
Lu Gr ce
e
xe e
m ce
bo
Sw urg
ed
e
Ja n
pa
Sp n
C
ze Ge ain
ch rm
R an
ep y
ub
F l
Sw in ic
it z land
er
la
n
Ko d
r
N C a ea
et na
he d
rla a
nd
s
Sl
ov B It a
e
a k lg ly
R ium
ep
ub
lic
U
%
ni
te
U
Sw
H
ed
e
ec
bo
ur
g
G
re
m
%
Sw
Lu
xe
C.2.
U
U nite
ni d
t
C ed K Sta
ze i te
ch ng s
R do
ep m
ub
N
ew Fr lic
Z e anc
al e
a
Au nd
s
G tria
re
L u Hu e ce
xe n g
m ar
bo y
u
Ko rg
Fi rea
nl
a
N nd
o
Be rwa
lg y
Sw ium
ed
M en
D e
N en xico
et m
he a
rla rk
n
Sw J ds
it z apa
er n
la
n
S d
Po pai
n
G rtug
er a
m l
an
y
C It a
Sl
a l
ov A na y
d
ak us a
R tra
ep lia
ub
lic
K
R or
Lu e p ea
xe u b
m lic
bo
Be urg
lg
iu
Au m
s
C tria
an
a
G da
re
Sw ece
e
U Po de n
ni
te rtu
d ga
St l
at
Fi e s
n
H lan
un d
ga
Sw M e ry
it z xic
er o
la
N nd
or
U
w
ni
te F ay
d ra
Ki nc
ng e
G do m
er
m
D an
en y
m
ar
N
k
et
he It a
rla ly
nd
Ja s
Sl
p
ov
a k S an
R pa
e p in
ub
lic
C
Compendium of Productivity Indicators
PRODUCTIVITY GROWTH IN MANUFACTURING
Value added per person employed, percentage change at annual rate
1
Total manufacturing
10
8
6
4
2
0
-2
Low technology manufactures
6
4
2
-2
0
-4
30
25
20
15
10
5
0
-5
-10
-15
-20
Electrical & Optical equipment 2
Motor vehicles 3
10
8
6
4
2
0
-2
-4
-6
-8
-10
Textiles
1. Or latest year available.
2. Data for the United States refer to ISIC sector 29-33 “machinery and equipment”.
3. Data for Belgium, Japan, Korea, Netherlands, Portugal, Spain and UK refer to ISIC sector 34-35 “transport equipment”.
45
PRODUCTIVITY GROWTH IN SERVICES
C.3.
• In several OECD countries, productivity growth in
business services accelerated from 1995-2000 to
2000-2003. This was notably the case in Australia,
Hungary, Japan, Korea, New Zealand and the
United States. In most other countries, including
most large European countries, productivity growth
in business services was less rapid over 2000-2003
than over 1995-2000.
Data Resources:
• The variation across services sectors and across
countries is considerable. Industries such as
wholesale and retail trade, and post and
telecommunications are typically the services
sectors with the highest rate of productivity growth.
Productivity growth in business services, hotels and
restaurants as well as transport and storage are
typically lower, although these industries have been
relatively important in some OECD countries.
Productivity growth in social services is not shown in
the charts, as many OECD countries still measure
the output of these services through their inputs.
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
 OECD, STAN Indicators Database, November 2005.
 OECD, STAN Database, November 2005.
For further reading
 Ahmad, Nadim, Francois Lequiller, Pascal Marianna, Dirk
Pilat, Paul Schreyer and Anita Wölfl (2003), “Comparing
Labour Productivity Growth in the OECD Area: The Role
of Measurement”, STI Working Paper 2003/14, OECD,
Paris.
 OECD (2005), Enhancing Services Sector Performance,
OECD, Paris.
 Wölfl, A. (2003), “Productivity Growth in Service
Industries: An Assessment of Recent Patterns and the
Role of Measurement”, STI Working Paper 2003/7,
OECD, Paris.
 Wölfl, A. (2005), “The Service Economy in OECD
Countries”, STI Working Paper 2005/3, OECD, Paris.
Measuring productivity in services
The service sector now accounts for about 70 to 80% of aggregate production and employment of OECD economies and
continues to grow. But measuring output and productivity growth in many services is not straightforward. What exactly does a
lawyer or economist produce? How can the rapidly-changing pricing schemes of telecommunications providers be compared
over time? And how should one measure the ‘quantity’ of health services provided by public hospitals? These and similar
questions arise when statisticians attempt to measure the production of services and the output of service industries and the
difficulty of this task is hard to overstate.
Generally, measuring volumes in the national accounts requires that current-price values of flows of goods and services can be
divided into volume and price components. Typically, this is more difficult for services than for goods. Characteristics of goods
can normally be identified and changes in quantities and qualities are measurable in principle, whereas for services even
quantitative changes are often hard to measure, let alone quality change. Omitting qualitative changes does not necessarily
mean that volume of services will be underestimated. For example, technical progress may improve the quality of medical
services but may also lead to a decline in quality when, for example, self-service in post offices, longer queuing times or
increased distances to the nearest post office all of which reflect an increased burden on customers and a lower quality of
services.
At the level of individual industries, two types of measurement issues arise: comparability of current-price measures, and
comparability of the price indices used to deflate the latter or, in some cases, the comparability of volume indices. When it
comes to the impact of service industry measures on the comparability of aggregate GDP series, two other factors come into
play. The first is the extent to which a particular service industry’s output is bought as intermediate input by other domestic
producers as opposed to the share of output that is delivered to final demand (private consumption, investment, government
and export). The larger the share of the former, the smaller the impact of a measurement error at the industry level on
measures of aggregate GDP growth, and vice versa: if a service industry delivers exclusively for final demand purposes, any
possible measurement will carry over to measures of total GDP. The second factor is the size of the industry in question, or
more precisely, its share in economy-wide value-added. The larger the share, the bigger the potential impact of an incorrect
measure of industry-level output.
46
© OECD 2006
Au
st
ni Ic rali
te el a
d an
St d
at
Ko es
N rea
or
w
Ja ay
G pa
r n
H ee
un ce
D g
en ar
m y
N
U ew P ar
ni
ol k
te Z an
d ea d
Ki l a
ng nd
d
M om
C exic
G ana o
er d
a
Be ma
lg ny
Fi i um
n
Sw lan
d
N A ede
et u n
he st
rl ri a
Po and
Sl
rtu s
ov
ak
ga
S
R p l
ep ai
ub n
lic
Lu F It a
x e ra l y
Sw mb nce
it z ou
er rg
la
nd
U
%
5
4
3
2
1
0
-1
-2
-3
%
14
12
10
8
6
4
2
0
-2
-4
-6
-8
%
14
12
10
8
6
4
2
0
-2
-4
-6
-8
© OECD 2006
Wholesale & Retail trade
Transport & Storage
Ic
Au ela
str nd
Koalia
r
G
Cz
r ea
ec A eec
h
u
Ne R st e
w ep ria
Un Ze ubl
ite Fala ic
d in nd
Ki la
n n
Un Sgd d
ite weom
d de
St n
Frate
Ca ancs
n e
M ada
e
Spxico
No ai
De rw n
n a
Lu Po ma y
xe rtu rk
m ga
b l
Sw Hunour
Ne itz gag
e
th rla ry
er n
Belandd
lgi s
um
Sl
Po Ital
ov G la y
ak er n
Re ma d
pu ny
bl
ic
H
Un Gung
ite re ary
Un d S ec
t e
i
Czted Poate
ec Kin lans
h g d
Re do
p m
Noubl
i
Fi rwac
n
C la y
Geanand
rm da
Sw Bel an
itz giu y
er m
Auland
Ne
s
w Ko tria
Ze re
Auala a
s n
Swtra d
li
Lu De edea
xe nm n
m ar
b
Ne ou k
th Sprg
er ain
l
Iceand
lans
Po Itad
rtu ly
Sl
M ga
ov
ak Frexicl
Re an o
pu ce
bl
ic
C.3.
Sw
itz
er
la
n
Sp d
a
M in
e
Ge xic
rm o
a
Au ny
s
No tria
r
Au way
str
ali
a
De Ital
Ne nm y
th ar
er k
lan
Gr ds
ee
Un Fin ce
ite la
n
d
St d
at
e
Fr s
a
Sw nce
ed
Un
e
ite Can n
d
Ki ada
ng
do
Ja m
pa
n
Au
str
a
No lia
D rw
Un en ay
ite ma
r
d
St k
at
Gr es
ee
c
Sp e
Ca ain
na
Un
da
ite
d
Ki Italy
ng
d
Sw om
ed
Fi en
nla
n
M d
ex
i
Au co
str
ia
Ne Fra
th nc
er e
la
Ge nd
s
Sw rma
itz ny
er
la
n
Ja d
pa
n
Compendium of Productivity Indicators
PRODUCTIVITY GROWTH IN SERVICES
Value added per person employed, percentage change at annual rate
1995-2000 compared with
2000-2003
%
14
10
6
2
-2
-6
-10
-14
%
18
16
14
12
10
8
6
4
2
0
1
Business services 2
Hotels & Restaurants
Post & Telecommunications
1. Or latest year available.
2. In STAN, business services include ISIC Rev.3 sectors from 50 to 74.
47
LABOUR PRODUCTIVITY OF FOREIGN AFFILIATES
• Foreign affiliates are often considered to make an
important contribution to productivity performance.
OECD data on foreign affiliates in manufacturing and
services help point to their role in economic
performance.
• In 2002, labour productivity of manufacturing foreign
affiliates in the United Kingdom was more than two
times higher than the average labour productivity of
all manufacturing domestic firms.
• In the United States, Finland and France, labour
productivity of manufacturing foreign affiliates was
almost at the same level as the national average.
• With respect to the service sector, foreign affiliates in
Portugal were two times more productive than the
national average, while in Finland and the
United States, labour productivity of foreign affiliates
was lower than the national average.
• Between 1995 and 2001, in the manufacturing
sector, foreign affiliates in Sweden, Finland and the
United States recorded the highest growth of labour
C.4.
productivity, while in Spain and Portugal the growth
of labour productivity of these affiliates was negative.
• During the same period, the growth of labour
productivity of foreign affiliates in the service sector
was high in Japan, but negative in the Netherlands,
Portugal, Finland, the Czech Republic and France.
Sources
 OECD, Activity of Foreign Affiliates (AFA) database,
June 2005.
 OECD, Foreign Affiliates’ Trade in Services (FATS)
database, June 2005.
For further reading
 OECD (2005), Measuring Globalisation – OECD
Handbook on Economic Globalisation Indicators, OECD,
Paris.
Available
at:
www.oecd.org/sti/measuringglobalisation
 OECD (2005), Measuring Globalisation – Economic
Globalisation Indicators, OECD, Paris.
 C. Criscuolo (2005), “The Contribution of Foreign
Affiliates to Productivity Growth: Evidence from OECD
Countries”, STI working papers 2005-8, OECD, Paris.
Available at: www.oecd.org/sti/working-papers
Measuring labour productivity of foreign affiliates in host countries
In the calculations shown here, the choice of measurement of labour productivity was largely determined by the availability of data.
Even at the level of total manufacturing, it was not possible to calculate total factor productivity, since no data on capital stock were
available for foreign affiliates. Labour productivity is measured on the basis of gross value added divided by the number of employees.
Relative labour productivity of foreign affiliates is the ratio of labour productivity of foreign affiliates compared to the national total for
labour productivity in the manufacturing and services sectors.
To measure labour productivity in terms of growth for total manufacturing and services, value added data for foreign affiliates have
been deflated using sectoral deflators of the national industry and re-weighing them according to the sectoral structure of foreign
affiliates. These calculations have been made where possible. For some sectors, this approach was not possible due to missing data.
In these cases, deflators at a higher level of aggregation were used.
48
© OECD 2006
Compendium of Productivity Indicators
C.4.
LABOUR PRODUCTIVITY OF FOREIGN AFFILIATES
Relative labour productivity of foreign affiliates, 2002
Manufacturing sector
1
Service sector
2
2.3
2.2
2.1
2
1.9
1.8
1.7
1.6
1.5
1.4
1.3
1.2
1.1
Re
pu
bl
ic
Sw
ed
en
Fr
an
ce
Ne
th
er
la
nd
s
Fi
nl
an
Un
d
ite
d
St
at
es
Cz
ec
h
Ja
pa
n
Po
rtu
ga
l
H
un
ga
ry
0.9
Un
ite
d
Ki
ng
do
m
Ne Spa
in
th
er
la
nd
s
Ja
pa
n
P
Cz
or
ec
tu
ga
h
Re
l
pu
bl
i
No c
rw
Hu ay
ng
a
S ry
Un we
d
ite
en
d
St
at
e
Fi s
nl
an
d
Fr
an
ce
1
Average annual labour productivity growth, 1995-2001
Percentage points
Manufacturing sector
Foreign affiliates
1
Service sector
2
Foreign affiliates
All firms
All firms
10%
12%
8%
10%
6%
8%
4%
6%
2%
4%
0%
2%
-2%
-4%
0%
-6%
-2%
-8%
Po
rtu
ga
l
Sw
ed
en
Un
ite
d
St
at
es
Hu
ng
ar
y
Fr
an
ce
Ja
pa
n
Ne
th
er
lan
ds
Cz
ec
h
Re
pu
bli
c
Fi
nla
nd
-10%
Sp
ai
n
Fi
nl
an
d
Fr
Un
an
ite
c
d
e
Ki
ng
do
m
Hu
ng
ar
y
Ja
pa
Ne
n
th
er
la
nd
s
No
rw
ay
Po
rtu
ga
Sw l
ed
Un
en
ite
d
St
at
es
Cz
ec
h
Re
pu
bl
ic
-4%
1. Or nearest available year: Czech Republic 1997-2002; United Kingdom 1995-1999; Finland 1995-2002; Hungary 1996-2002; Spain
1999-2001 and Portugal 1996-2002.
2. Or nearest available year: Czech Republic 1995-2002; Sweden 1997-2000; Hungary 1998-2002; Netherlands 1997-2001; Japan
1997-2000 and Portugal 1996-2002.
Sources: OECD, AFA, FATS and STAN databases, June 2005.
© OECD 2006
49
Compendium of Productivity Indicators
D. OFFICIAL MULTI-FACTOR PRODUCTIVITY STATISTICS FOR OECD MEMBER
COUNTRIES
© OECD 2006
51
MULTIFACTOR PRODUCTIVITY MEASURES IN AUSTRALIA
• The Australian Bureau of Statistics has computed
and published time series of multi-factor productivity
indices for several years. The headline figure, also
the one that is most timely available is MFP growth
for the market sector. The market sector is an
industry grouping comprising: agriculture, forestry
and fishing; mining; manufacturing; electricity, gas
and water supply; construction; wholesale trade;
retail trade; accommodation, cafes and restaurants;
transport and storage; communication services;
finance and insurance; and the cultural and
recreational services industries. These are industries
with marketed activities for which there are
satisfactory estimates of the growth in the volume of
output. In addition, MFP productivity measures are
presently being developed for individual ANZSIC
industries.
• The ABS derives its estimates of MFP by forming a
combined chain volume measure of labour and
capital and dividing it into a chain volume measures
of the gross value added of the market sector. The
elements of capital input are compiled at a detailed
level. There are capital input measures for 14 asset
types for the corporate and unincorporated sectors
for each of the industries included in the market
sector. For each capital there is a volume indicator of
the flow of capital services and a rental value to
weight the service flow with the service flows of other
capital inputs. An aggregate chain volume measure
of capital services for the whole market sector is
then combined with a measure of hours worked
using estimates of capital and labour income
weights. For more details see ABS (2000).
• The ABS’ MFP measures differ in several aspects
from the MFP measures computed by the OECD.
First and importantly, national data is based on more
detailed source data than the international data.
Second, ABS adjusts, on an experimental basis,
labour input measures to reflect the composition of
the labour force e.g., by age, education and
experience. There are thus two MFP series, one
based on simple hours worked (akin to the OECD
labour input data) and one based on hours worked
52
D.1.
adjusted for compositional changes. Both series are
shown in the graph below. It is apparent that MFP
based on unadjusted hours rose more quickly than
MFP based on adjusted hours. This reflects the fact
that unadjusted hours rose less quickly than
adjusted hours which in turn means that the
composition of labour input has gradually shifted
towards more qualified, more experienced workers.
In other words, more of output growth can be
attributed to labour input, and therefore, a smaller
part remains as the ‘unexplained’ MFP residual.
• Thirdly, capital input as computed by ABS is based
on a broader scope of capital assets than used by
the OECD. In particular, the national data includes
agricultural land and inventories, two assets that are
absent from the OECD capital computations. These
and some other technical differences may lead to
differences in reported MFP growth between the
national and the international source. As a general
rule, the national source is to be preferred over the
international source for analyses that relate to
Australia only whereas the international source is
often better suited for comparisons between
countries.
• Productivity indexes fluctuate according to the
business cycle. One way to measure the trend rate
of growth of productivity is to calculate the average
annual growth rate between growth cycle peaks.
Growth cycle peaks for multifactor productivity are
identified as local maximum positive deviations of
the productivity index from its long-term trend.
Growth rates of MFP between cycle peaks are
shown in the figure below.
Sources
 Australian Bureau of Statistics, available at:
http://www.abs.gov.au
For further reading
 Australian Bureau of Statistics (2000); Australian
National Accounts: Concepts, Sources and Methods,
Catalogue No 5216.0, Chapter 27.
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
© OECD 2006
Compendium of Productivity Indicators
D.1.
MULTIFACTOR PRODUCTIVITY MEASURES IN AUSTRALIA
Multi-factor productivity (value-added based) in the market sector,
1985=100 (fiscal years)
MFP based on unadjusted hours w orked
MFP based on hours w orked adjusted for labour composition
130.0
125.0
120.0
115.0
110.0
105.0
100.0
95.0
Ju
n.
19
85
Ju
n.
19
87
Ju
n.
19
89
Ju
n.
19
91
Ju
n.
19
93
Ju
n.
19
95
Ju
n.
19
97
Ju
n.
19
99
Ju
n.
20
01
Ju
n.
20
03
Ju
n.
20
05
90.0
Source: Australian Bureau of Statistics, Catalogue No 5204.0 - Australian System of National Accounts, 2004-05.
Multi-factor productivity (value-added based) in the market sector during growth cycles
Percentage changes at average annual rate between MFP growth cycle peaks (fiscal years)
1.8%
1.6%
1.4%
1.2%
1.0%
0.8%
0.6%
0.4%
0.2%
0.0%
64
19
-65
6
19
to
9
8-6
68
19
-69
7
19
to
4
3-7
73
19
-74
8
19
to
2
1-8
81
19
-82
8
19
to
5
4 -8
84
19
-85
8
19
to
9
8 -8
Source: Australian Bureau of Statistics, Catalogue No 5204.0 - Australian System of National Accounts, 2004-05.
© OECD 2006
53
MULTIFACTOR PRODUCTIVITY MEASURES IN CANADA
• Statistics Canada has computed and published time
series of multi-factor productivity indices for a
number of years. The headline figure, also the one
that is most timely available is MFP growth for the
business sector. In addition, MFP productivity
measures are published for many 2-digit and 3-digit
NAICS industries.
• MFP indices are computed as the ratio between
value-added and combined labour and capital input,
i.e., they constitute value-added based MFP
measures. Statistics Canada’s MFP measures differ
in several aspects from the MFP measures
computed by the OECD. First and importantly, the
national data is significantly more detailed and also
more timely than the international data. Second,
labour input measures have been adjusted by
Statistics Canada to reflect the composition of the
labour force e.g., by age, education and experience
whereas the OECD labour input data is a simple
aggregate of hours worked. Thirdly, capital input as
computed by Statistics Canada is based on a
broader scope of capital assets than used by the
OECD. In particular, the national data includes land
and inventories, two assets that are absent from the
OECD capital computations. These and some other
technical differences may lead to differences in
reported MFP growth between the national and the
international source. As a general rule, the national
source is to be preferred over the international
source for analyses that relate to Canada only
whereas the international source is often better
suited for comparisons between countries.
D.2.
• Productivity measures at the industry level are
derived from a set of industry accounts. Under this
approach, a variety of productivity series at the
industry level are constructed using alternate
measures of output along with their corresponding
inputs. Industry data on outputs and inputs permit
the construction of bottom-up MFP measures for
major sectors as a weighted average of industry
productivity growth rates. More detailed information
on the methodology underlying Statistics Canada’s
productivity series can be found in the series
description Productivity Measures and Related
Variables (CANSIM Record No 1402) and in Baldwin
and Harchaoui (2005).
Sources
 Statistics Canada, CANSIM Database, Tables 383-0016
to 383-0019.available at: http://www.statcan.ca
For further reading
Baldwin, John and Tarek Harchaoui (2005);The
Integration of the Canadian Productivity Accounts within
the System of National Accounts: Current Status and
Challenges Ahead; Research Paper, Statistics Canada
Economic analysis methodology paper series: National
accounts; 11F0026 No 004.
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
Multi-factor productivity (value-added based) in the business sector, 1981=100
115.0
110.0
105.0
100.0
95.0
01
99
97
95
93
03
20
20
19
19
19
19
91
19
87
85
83
89
19
19
19
19
19
81
90.0
Source: Statistics Canada, CANSIM Table 383-0016.
54
© OECD 2006
Compendium of Productivity Indicators
D.2.
MULTIFACTOR PRODUCTIVITY MEASURES IN CANADA
Multi-factor productivity (value-added based) in 2-digit and selected 3-digit NAICS industries, 1997-2003
Percentage change at annual rate
Printing and related support activities
Wood product manufacturing
Non-metallic mineral product manufacturing
Paper manufacturing
Plastics and rubber products manufacturing
Fabricated metal product manufacturing
Educational services (except universities)
Furniture and related product manufacturing
Chemical manufacturing
Retail trade
Primary metal manufacturing
Accommodation and food services
Wholesale trade
Information and cultural industries
Clothing manufacturing
Manufacturing
Transportation equipment manufacturing
All industries
Food manufacturing
Grant-making, civic, and professional and similar organizations
Machinery manufacturing
Other services (except public administration)
Transportation and warehousing
Agriculture, forestry, fishing and hunting
Warehousing and storage
Arts, entertainment and recreation
Construction
Miscellaneous manufacturing
Electrical equipment, appliance and component manufacturing
Health care and social assistance (except hospitals)
Publishing industries, information services and data processing
Professional, scientific and technical services
Administrative & support, waste management
Personal and laundry services and private households
Petroleum and coal products manufacturing
Utilities
Beverage and tobacco product manufacturing
Textile and textile product mills
Support activities for mining and oil and gas extraction
Computer and electronic product manufacturing
Oil and gas extraction
-6.0%
-4.0%
-2.0%
0.0%
2.0%
4.0%
6.0%
Source: Statistics Canada, CANSIM Table 383-0013.
© OECD 2006
55
MULTIFACTOR PRODUCTIVITY MEASURES IN NEW ZEALAND
• In 2006, Statistics New Zealand released, for the first
time, an official time series of multi-factor productivity
growth. This first dataset relates to the ‘measured
sector’, consisting of industries for which estimates
of inputs and outputs are independently derived in
constant prices. Excluded are those industries –
mainly government non-market industries whose
services, such as administration, health and
education, are provided free or at nominal charges –
whose real value-added is measured in the national
accounts largely using input methods, such as
numbers of employees. Using an input series to
estimate the change in outputs implicitly assumes nil
productivity growth and to include these industries in
the 'measured sector' would lead to a bias in the
productivity series. Also excluded are a number of
private sector market industries that similarly use
some form of input measure to estimate real output,
for example the residential and commercial property
industries whose output is measured by the growth
in property assets.
• Labour input is measured as the total number of
hours paid, the number of ordinary and overtime
hours for which an employee is paid. It excludes
unpaid overtime but may include some hours that
are not actually worked, such as paid leave and
statutory holidays. While Statistics New Zealand
states a conceptual preference of a measure of
hours worked over hours paid, it has greater
confidence in the quality of its hours paid data which
are also available for longer time series. These data
considerations led to the choice of hours paid over
hours worked. A full description of data sources and
methodology can be found in Statistics New Zealand
(2006).
• Statistics
New
Zealand
follows
standard
methodology and derives its estimates of MFP by
subtracting a volume growth measure of labour and
56
D.3.
capital inputs from a volume growth measure of
output, constant price gross value added.
• The elements of capital input are compiled at a level
of 24 types of assets, and 22 industries. This is more
detailed and broader in scope than the OECD’s own
estimates of capital services (which exclude, for
example, residential buildings). In the national series
for New Zealand, for each type of asset in every
industry there is a volume indicator of the flow of
capital services and a rental value to weight the
service flow with the service flows of other capital
inputs. An aggregate chain volume measure of
capital services for the whole economy is then
combined with a measure of hours worked using
estimates of capital and labour income weights.
• The basic methodology behind the New Zealand
MFP estimates closely follows the methods
presented in international documents such as the
OECD Productivity Manual. In some specific
aspects, however, the national measures differ from
MFP measures computed by the OECD (scope of
capital assets, sector coverage). It follows that the
OECD’s estimates of New Zealand’s MFP growth
cannot be directly compared with the national
estimates published by Statistics New Zealand. As a
general rule, the national source is to be preferred
over the international source for analyses that relate
to New Zealand only whereas the international
source is often better suited for comparisons
between countries.
Sources
 Statistics New Zealand (2006); Productivity Statistics:
Sources and Methods; available at
http://www2.stats.govt.nz
For further reading
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
© OECD 2006
Compendium of Productivity Indicators
D.3.
MULTIFACTOR PRODUCTIVITY MEASURES IN NEW ZEALAND
Multi-factor productivity (value-added based) in total economy,
1988=100
140
135
130
125
120
115
110
105
100
19
88
19
89
19
90
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
95
90
MFP total economy
Source: Statistics New Zealand.
Growth accounts for New Zealand, 1990-1995, 1995-2000, 2000-2005 and 1995-2005
In percentage points
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
1990-1995
1995-2000
Contribution of labour input
2000-2005
Contribution of capital input
1995-2005
Multi-factor productivity grow th
Source: Statistics New Zealand.
© OECD 2006
57
MULTIFACTOR PRODUCTIVITY MEASURES IN SWITZERLAND
• In 2006, the Swiss Federal Statistical Office
published a first set of MFP estimates for
Switzerland. This first dataset relates to the total
economy and comprises thus the private and the
public sector. This differs from other national MFP
measures (e.g., Canada, Australia or the United
States) but is similar to the OECD MFP statistics.
• Labour input measures also correspond to the
simple unadjusted series of hours worked that are
used by the OECD. This reflects a constraint on data
availability that does not presently permit the Swiss
Federal Statistical Office to adjust hours worked for
compositional change although the desirability of
such an adjustment is clearly recognised by the
Statistical Office.
D.4.
however, the Swiss MFP measures differ from MFP
measures computed by the OECD. The national
data is based on more detailed source data than the
six-way asset classification used by the OECD.
There is also important difference in the scope of
capital measures. In particular, the estimates by the
Swiss Federal Statistical Office include residential
assets which are excluded from the OECD data. It
follows that international comparisons between the
national MFP results for Switzerland and OECD’s
MFP data for other countries have to be made with
the necessary caution.
Sources
 Swiss Federal Statistical Office
http://www.bfs.admin.ch
For further reading
• The Swiss Federal Statistical Office follows standard
methodology and derives its estimates of MFP by
deducting, from the volume measure of GDP a
weighted volume growth measure of labour and
capital inputs. The elements of capital input are
compiled at a level of 16 types of assets. For each
type of asset there is a volume indicator of the flow
of capital services and a rental value to weight the
service flow with the service flows of other capital
inputs. An aggregate chain volume measure of
capital services for the whole economy is then
combined with a measure of hours worked using
estimates of capital and labour income weights.
 Swiss Federal Statistical Office (2006), Analyse des
contributions à la croissance des facteurs de production,
de la productivité multifactorielle et du rôle joué par
l'intensité capitalistique de 1991 à 2004, Neuchâtel.
 Swiss Federal Statistical Office (2006), « Stock de
capital non financier », Rapport méthodologique,
Neuchâtel.
 Swiss Federal Statistical Office (2006), « Productivité
multifactorielle », Rapport méthodologique, Neuchâtel.
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
• The basic methodology behind the Swiss MFP
estimates closely follows the methods presented in
international documents such as the OECD
Productivity Manual. In some specific aspects,
58
© OECD 2006
Compendium of Productivity Indicators
D.4.
MULTIFACTOR PRODUCTIVITY MEASURES IN SWITZERLAND
Multi-factor productivity (value-added based) in total economy,
1991=100
107.0
106.0
105.0
104.0
103.0
102.0
101.0
100.0
99.0
98.0
97.0
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
Source: Swiss Federal Statistical Office.
Growth accounts for Switzerland, 1995-2000, 2000-2004 and 1995-2004
In percentage points
2.5%
2.0%
1.5%
1.0%
0.5%
0.0%
1995-2000
Contribution of labour input
2000-2004
Contribution of capital input
1995-2004
Multi-factor productivity grow th
Source: Swiss Federal Statistical Office.
© OECD 2006
59
MULTIFACTOR PRODUCTIVITY MEASURES IN THE UNITED STATES
• The United States Bureau of Labor Statistics has
computed and published time series of multi-factor
productivity indices for a number of years. The
headline figure, also the one most timely available
are MFP growth for the business and the non-farm
business sector. In addition, MFP productivity
measures are published for 20 2-digit SIC
manufacturing industries and 108 3-digit SIC
manufacturing industries, railroad transportation, and
air transportation.
• Business sector indices of MFP are computed as the
ratio between value-added and combined labour and
capital input, i.e., as value-added based MFP
measures.
• Industry-level multifactor productivity measures are
constructed by dividing an index of output by an
index of combined inputs. Combined inputs are a
weighted average of employee hours, capital
services
(land,
structures,
equipment
and
inventories), and intermediate purchases (materials,
energy, and purchased services).
• The indexes of output and of combined inputs are
Tornqvist indexes, developed for each industry by
computing a weighted average of the growth rates of
the various outputs or inputs between two periods,
with weights based on relative cost shares. The
weight for each item equals its average value share
in the two periods. Thus, industry-level MFP
measures follow a KLEMS-type approach which is
different from the value-added based approach for
the total business sector.
D.5.
• For a more complete discussion of the Tornqvist
methodology see "Industry Productivity Measures,"
Chapter 11 of the BLS Handbook of Methods.
• BLS MFP measures differ in several aspects from
the MFP measures computed by the OECD. First
and importantly, the national data is significantly
more detailed and also more timely than the
international data. Second, labour input measures
have been adjusted by BLS to reflect the
composition of the labour force e.g., by age,
education and experience whereas the OECD labour
input data is a simple aggregate of hours worked.
These and some other technical differences may
lead to differences in reported MFP growth between
the national and the international source. As a
general rule, the national source is to be preferred
over the international source for analyses that relate
to the United States only whereas the international
source is often better suited for comparisons
between countries.
Sources
 United States Bureau of Labor Statistics, Multifactor
Productivity statistics, available at:
http://www.bls.gov/mfp/home.htm.
For further reading
 Unites States Bureau of Labor Statistics
Handbook of Methods; BLS Bulletin 2490.
(1997),
 OECD (2001), Measuring Productivity – OECD Manual,
OECD, Paris.
Availability of U.S. BLS productivity measures for major sectors and sub-sectors of the economy
Productivity measure
Labour productivity1
Business
Non-farm business
Non-financial corporations
Manfacturing, total
Durable
Non-durable
Multifactor productivity1
Private business
Private non-farm business
KLEMS Multi-factor
productivity
Manufacturing and and 20 2digit SIC manufacturing
industries services
1
60
Input(s)
Index available
Labour
Labour
Quarterly
Quarterly
Labour, capital
Labour, capital
Annually
Annually
Labor, capital, energy, materials,
services
Annually
Includes government enterprises; multifactor productivity measures exclude such enterprises.
© OECD 2006
Compendium of Productivity Indicators
D.5.
MULTIFACTOR PRODUCTIVITY MEASURES IN THE UNITED STATES
Major sectors and manufacturing industries of the economy, 1987=100
Multi-factor productivity (value-added based) in the
business sector, 1987=100
Private non-farm business sector
Private business sector
KLEMS multi-factor productivity in manufacturing,
1987=100
125.0
125.0
120.0
120.0
115.0
115.0
110.0
105.0
110.0
100.0
105.0
95.0
100.0
00
01
20
99
20
19
98
97
19
19
95
96
19
94
19
93
19
19
92
91
19
19
90
89
19
19
88
87
19
19
05
20
03
20
01
20
99
19
97
19
95
19
91
89
93
19
19
19
19
87
90.0
KLEMS multi-factor productivity in 2-digit SIC manufacturing industries, 1987-2001
Percentage change at annual rate
Electric equipment
Ind mach. and computer equipment
Durable goods
Textile mills products
Instruments
Rubber and plastic products
Apparel products
Misc. Manufacturing
Furniture and fixtures
Stone, clay and glass
Lumber and wood products
Transportation equipment
Fabricated metal products
Primary metal industry
Nondurable goods
Chemicals
Petroleum refining
Food and kindred products
Paper products
Printing and publishing
-2.0%
-1.0%
0.0%
1.0%
2.0%
3.0%
4.0%
5.0%
Source: United States Bureau of Labor Statistics.
© OECD 2006
61
ANNEX 1: OECD PRODUCTIVITY DATABASE
Over several years, the analysis of productivity and economic growth has been an important
focus of OECD work. To carry out this work, OECD needs comparable data on productivity growth and
its underlying drivers. There has also been significant interest in productivity data by researchers and
analysts outside the OECD and this led to the development of the OECD Productivity Database,
designed to be as comparable and consistent across countries as is currently possible. The database
and related information on methods and sources were made publicly available through the OECD
Internet site on 15 March 2004.
Coverage
The OECD Productivity Database currently provides annual estimates of
•
Labour productivity growth, measured as GDP per hour worked, and all components data, for
29 OECD countries and the period 1970-2005;
•
Capital services and multi-factor productivity for 19 OECD countries for the period 1985-2005
or latest year available, depending on countries;
•
Labour productivity levels for 2005.
All of the above data relate to the economy as a whole. In addition, the Productivity Database
features industry-level measures of labour productivity growth, drawn from the OECD STAN
Database, see www.oecd.org/sti/stan .
The OECD Productivity Database is updated on a regular basis as new data become available. It
is accessible through the OECD Internet site, at: www.oecd.org/statistics/productivity .
Sources of data underlying the OECD Productivity Database
The OECD productivity database combines – to the extent possible – a consistent set of data on
GDP, labour input measured as total hours worked, and capital services. The following sources are
used in the productivity database.
Gross Domestic Product
Estimates for GDP are derived from OECD's Annual National Accounts (ANA). The data from
ANA are based on OECD’s annual national accounts questionnaire to OECD member countries. The
data resulting from this questionnaire differ somewhat from national sources and are more comparable
across countries than those derived from OECD’s quarterly national accounts, thanks to several small
methodological adjustments that are made. However, the differences with other OECD sources, such
as the Quarterly National Accounts and the Economic Outlook database, are minor for most countries.
Labour Input
The estimates of labour productivity included in the database refer to GDP per hour worked;
measures of GDP per person employed are not yet included. GDP per hour worked requires estimates
of trends in total hours worked that are consistent across countries. Generally, the default source for
total hours worked in the productivity database is the national accounts. However, for a number of
62
© OECD 2006
Compendium of Productivity Indicators
countries, the national accounts do not provide hours worked and other sources have to be invoked.
Consistency of data is achieved by matching the hours worked that are collected by the OECD for its
annual OECD Employment Outlook with the conceptually appropriate measure of employment for
each individual country, i.e. the measure of employment for that country that is consistent with the
measure of hours worked collected by the OECD.
Estimates of average hours actually worked per year per person in employment are currently
available on an annual basis for 29 OECD countries (See OECD Employment Outlook, Statistical
Annex Table F –www.oecd.org/els/employmentoutlook/statannex). The OECD Productivity Database
includes, in addition, hours of work per employee for Hungary and Korea. These estimates are
available from National Statistical Offices for 22 countries, 15 of which are consistent with National
Accounts concepts and coverage.
To develop these estimates, countries use the best available data sources for different categories
of workers, industries and components of variation from usual or normal working time (e.g. public
holidays, annual leave, overtime, absences from work due to illness and to maternity, etc.). For
example, in 2 countries (Japan and United States) actual hours are derived from establishment
surveys for regular or production/non-supervisory workers in employee jobs and from labour force
surveys (LFS) for non-regular or managers/non-supervisory employees, self-employed, farm workers
and employees in the public sector. In 3 other countries (France, Germany and Switzerland), the
measurement of annual working time relies on a component method based on standard working hours
minus hours not worked due to absences plus hours worked overtime. Standard working hours are
derived from an establishment survey (hours offered), an administrative source (contractual hours)
and the labour force survey (normal hours), respectively. The coverage of workers is extended using
standard hours reported in labour force surveys or other sources as hours worked overtime. Vacation
time is either derived from establishment-survey data on paid leave or the number of days of statutory
leave entitlements. Hours lost due to sickness are estimated from the number of days not worked from
social security registers and/or health surveys.
On the other hand, the national estimates for several other countries (i.e. Australia, Canada,
Czech Republic, Finland, Iceland, Mexico, New Zealand, Slovak Republic, Spain, Sweden and United
Kingdom) rely mainly on labour force survey results. Annual working hours are derived using a direct
method annualising actual weekly hours worked, which cover all weeks of the year in the case of
continuous surveys. But, for labour force surveys with fixed monthly reference weeks, this method
results in averaging hours worked during 12 weeks in the year and, therefore, necessitates
adjustments for special events, such as public holidays falling outside the reference week (i.e. Canada
and Finland). Finally, estimates of annual working time for 4 other EU member states are derived by
the OECD Secretariat by applying a variant of the component method to the results of the Spring
European Labour Force Survey (ELFS). A summary of the various measures available in September
2006 is shown in Table 1 below.
Two other considerations should be kept in mind. First, annual working-time measures are
reported either on a job or on a worker basis. To harmonise the presentation, annual hours worked
measures can be converted between the two measurement units by using the share of multiple job
holders in total employment, which is available in labour force surveys, albeit no further distinction is
1
possible between second and more jobs.
1.
© OECD 2006
For example, the BLS-Office of Productivity and Technology (OPT) estimates of annual hours of work for the United
States are reported on a (per) job basis and are later converted by the OECD Secretariat to a per worker basis by
multiplying the job-based annual hours of work by (1 + CPS based share of multiple jobholders in total
employment).
63
Table 1: Measures of annual working time and employment included in the OECD Productivity
Database as of September 2006
Source for annual actual working hours
Source for employment data
Australia
Australian Bureau of Statistics
Annual National Accounts
Austria
Annual National Accounts
Annual National Accounts
Belgium
OECD Employment Outlook
Annual National Accounts
Canada
Annual National Accounts
Annual National Accounts
Czech Republic
OECD Employment Outlook
Annual National Accounts
Denmark
Annual National Accounts
Annual National Accounts
Finland
Annual National Accounts
Annual National Accounts
France
Annual National Accounts
Annual National Accounts
Germany
Annual National Accounts
Annual National Accounts
Greece
Annual National Accounts
Annual National Accounts
Hungary
Annual National Accounts
Annual National Accounts
Iceland
OECD Employment Outlook
Annual National Accounts
Ireland
OECD Employment Outlook
Annual National Accounts
Italy
Annual National Accounts
Annual National Accounts
Japan
OECD Employment Outlook
Annual National Accounts
Korea
Annual National Accounts
Annual National Accounts
Luxembourg
OECD Employment Outlook
Annual National Accounts
Mexico
OECD Employment Outlook
Annual National Accounts
Netherlands
OECD Employment Outlook
Annual National Accounts
New Zealand
OECD Employment Outlook
Labour Force Statistics
Norway
Annual National Accounts
Annual National Accounts
Poland
OECD Employment Outlook
Annual National Accounts
Portugal
OECD Employment Outlook
Annual National Accounts
Slovak Republic
Annual National Accounts
Annual National Accounts
Spain
Annual National Accounts
Annual National Accounts
Sweden
Annual National Accounts
Annual National Accounts
Switzerland
Annual National Accounts
Annual National Accounts
Turkey
No data
Labour Force Statistics
United Kingdom
OECD Employment Outlook
UK Office for National Statistics
United States
US Bureau of Labor Statistics
US Bureau of Labor Statistics
Source: OECD (2006).
Second, given the variety of data sources, of hours worked concepts retained in data sources,
2
and of measurement methodologies (direct measures or component methods ) to produce estimates
of annual working time, the quality and comparability of annual hours worked estimates are constantly
questioned, and are subject to at least two probing issues:
•
Labour force survey-based estimates are suspected of over-reporting hours worked
compared to work hours reported in time-use surveys, in particular for those working
long hours, like managers and professionals.
•
Employer survey-based estimates do not account for unpaid overtime hours and are
sometimes suspected of under-reporting hours worked, with consequences on
productivity levels and growth.
The comparability of measures of hours worked across OECD countries thus remains an issue,
and work is currently underway, notably through the Paris Group, a UN city group on Labour and
Compensation, to further improve the available measures of hours worked.
2.
64
However, both methods can be summarised by the following identity: Annual hours per worker = Standard weekly
hours worked x Number of weeks actually worked over the year = Weekly hours actually worked x 52 weeks,
considering weekly reference period for reporting hours worked.
© OECD 2006
Compendium of Productivity Indicators
Capital input
The appropriate measure for capital input within the growth accounting framework is the flow of
productive services that can be drawn from the cumulative stock of past investments in capital assets
(see OECD, 2001a). These services are approximated by the rate of change of the ‘productive capital
stock’ – a measure that takes account of wear and tear, retirements and other sources of reduction of
the productive capacity of fixed assets. Flows of productive services of an office building, for instance,
are the protection against rain or the comfort and storage services that the building provides to
personnel during a given period (Schreyer et al., 2003). The price of capital services per asset is
measured as their rental price. If there are markets for capital services, as is the case for office
buildings, for instance, rental prices could be directly observed. For most assets, however, rental
prices have to be imputed. The implicit rent that capital good owners ‘pay’ themselves gives rise to the
terminology ‘user costs of capital’.
Capital input (S) is measured as the volume of capital services, assumed to be in a fixed
proportion to the productive capital stock (see Schreyer, et al., 2003 for a more extensive explanation
and for details of the computation of capital services). The productivity database publishes capital
services data with calculations based on the perpetual inventory method (PIM) The PIM calculations
are carried out by OECD, using service lives for different assets that are common across countries
and correcting for differences in deflators for information and communication technology assets.
Sources for the investment series by type of asset underlying the capital services series are national
3
statistical offices and the Groningen Growth and Development Centre Total Economy Growth
4
Accounting Database (http://www.ggdc.net).
Measures of multi-factor productivity in the OECD Productivity Database
The following methodology has been applied for the computation of multi-factor productivity
(MFP) measures:
Rates of change of output
Output (Q) is measured as GDP at constant prices for the entire economy. Year-to-year changes
are computed as logarithmic differences:
ln(
Qt
)
Q t −1
Rates of change of labour input
Labour input (L) is measured as total hours actually worked in the entire economy. Data on total
hours has been specifically developed for the present purpose, see above. Year-to-year changes are
computed as logarithmic differences: ln(
Lt
).
L t −1
3.
For Australia, Canada, France, Japan, Germany, New Zealand, Spain and the United States.
4.
For Austria, Belgium, Denmark, Finland, Greece, Ireland, Italy, the Netherlands, Portugal, Sweden, and the United
Kingdom.
© OECD 2006
65
Rates of change of capital input
i
Capital services are computed for six different types of assets ( S t
i = 1,2,...7 ) and aggregated
to an overall rate of change of capital services by means of a Törnqvist index:
 Si 
 S 
7
ln t  = ∑i =1 12 vit + vit −1 ln i t  with v it ≡
 St −1 
 St −1 
(
where
)
u it Sit
∑
7
i =1
u it Sit
v it is the share of each asset in the total value of capital services
expression, the value of capital services for each asset is measured by
cost price per unit of capital services and
∑
7
i =1
u itSit . In this
u itSit where u it is the user
Sit is the quantity of capital services in year t.
Cost shares of inputs
The total cost of inputs is the sum of the remuneration for labour input and the remuneration for
capital services. Remuneration for labour input has been computed as the average remuneration per
employee multiplied by the total number of persons employed. This adjustment was necessary to
correct for self-employed persons whose income is not part of the compensation of employees as
registered in the national accounts. The source for data on compensation of employees and for the
number of employees as well as the number of self employed is the OECD Annual National Accounts.
 COMPt
w t L t = 
 EE t

E t

where
w t Lt :
remuneration for labour input in period t
COMPt :
compensation of employees in period t
EE t :
number of employees in period t
Et :
total number employed (employees plus self-employed) in period t.
Total cost of inputs is then given by:
C t = w t L t + ∑i =1 u itSit and the corresponding cost shares are
7
s Lt ≡
s
66
S
t
w tLt
for labour input and
Ct
∑
≡
6
i =1
u itSit
Ct
for capital input.
© OECD 2006
Compendium of Productivity Indicators
Total inputs
The rate of change of total inputs is a weighted average of the rate of change of labour and
capital input with the respective cost shares as weights. Aggregation is by way of a Törnqvist index
number formula:
 X 
ln t  =
 X t −1 
1
2
(s
L
t
 L 
 S 
+ s Lt −1 ln t  + 12 s St + s St −1 ln t 
 L t −1 
 S t −1 
)
(
)
Multi-factor productivity
Multi-factor productivity is measured as the difference between output and input change, or as
‘apparent multi-factor productivity’:
 MFPt 
 Q 
 X 
 = ln t  − ln t  .
ln
 MFPt −1 
 Q t −1 
 X t −1 
© OECD 2006
67
ANNEX 2: OECD ESTIMATES OF LABOUR PRODUCTIVITY LEVELS
Introduction
International comparisons of productivity growth can give useful insights in the growth process,
but should ideally be complemented with international comparisons of income and productivity levels.
An examination of income and productivity levels may give insights into the possible scope for further
gains, and also places a country’s growth experience in the perspective of its current level of income
and productivity. OECD has published estimates of labour productivity levels in various studies (e.g.
Scarpetta, et al., 2000; OECD, 2003), and has released estimates of productivity on the OECD
Internet site, at: www.oecd.org/statistics/productivity.
Since the release of OECD estimates of productivity growth in the OECD Productivity Database
in March 2004 (see Annex 1), more attention has turned to the measurement of productivity levels,
since these serve as a yardstick of economic performance in many OECD countries. Several statistical
agencies and international organisations, including Eurostat, the UK Office for National Statistics, the
US Bureau of Labor Statistics, and the International Labour Organisation, now release estimates of
labour productivity levels, as do some academic institutions, such as the Groningen Growth and
Development Centre, and some private institutions, such as the Conference Board. In several
instances, notably in the case of Eurostat and the ONS, estimates of labour productivity levels serve
as official yardsticks of economic performance and are used to measure progress with regards to
explicit policy targets.
Given the importance attached to labour productivity levels, it is unfortunate that there is still
considerable variation in the currently available estimates. Primarily, this seems due to differences in
the choice of basic data. Indeed, much of the differences can be brought back to how different
organisations select and combine information on the three components of labour productivity levels at
the economy-wide level. These components are gross domestic product, labour input and a
conversion factor for total GDP (typically a purchasing power parity or PPP) that is needed to translate
output in national currency units to a common currency.
This annex briefly discusses some of the main measurement issues for these components, as
well as the different data choices that can be made. It focuses on the current OECD approach to
measuring labour productivity levels, but also refers to other possible approaches, where appropriate.
The discussion focuses on comparisons of labour productivity at the economy-wide level; the
estimation of productivity levels for individual industries raises additional measurement issues that go
5
beyond the scope of this annex.
Output: comparability and data choices
Most comparisons of labour productivity levels focus on GDP as the measure of output. Other
measures of aggregate output, such as GNP or national income, have also been used in a few
studies, but are not considered here. The measurement and definition of economic output is treated
systematically across countries in the 1993 System of National Accounts (SNA 93). Most countries in
the OECD area have now implemented the 1993 SNA, Turkey being the only exception, which implies
that its level of GDP is likely to be somewhat understated relative to other OECD countries. However,
despite the harmonisation of GDP estimates through the 1993 SNA, there are some differences in
estimation methods across countries (Ahmad, et al., 2003). These typically have only a small effect on
growth rates, but may be substantially more important for comparisons of output and productivity
5.
68
An OECD reader will discuss these issues in greater detail (OECD, forthcoming).
© OECD 2006
Compendium of Productivity Indicators
levels. Some of the main differences that are known to affect GDP levels are the following (Ahmad, et
al., 2003):
Expenditure on military equipment. The coverage of government investment in the US
National Income and Product Accounts (NIPA) is more extensive than that recommended by
the SNA, since it includes expenditures on military equipment (aircraft, ships, missiles) that
are not considered assets by the SNA. The national accounts in most other OECD countries
strictly follow the SNA in this matter. As the amount of public investment affects GDP, this
results in a statistical difference in the measurement of GDP. Convergence on this issue is
expected in the next edition of the SNA, in 2008. In the meantime, the OECD publishes data
in its Annual National Accounts Database for the United States which adjust for this
difference.
Financial Intermediation Services. Most banking services are not explicitly charged. Thus, in
the SNA, the implicit production of banks is estimated using the difference between interests
received and paid. All OECD member countries have estimated this part of bank production,
known as “Financial Intermediation Service Indirectly Measured” or “FISIM”. While it is
relatively straightforward to recognise and estimate FISIM, the key problem is breaking it
down between final consumers (households) and intermediate consumers (business and
government). Only the first part has an overall impact on GDP. In the United States, Canada
and Australia, such a breakdown has been estimated in the national accounts for some time,
in accordance with the SNA. A breakdown between final and intermediate consumers has
recently been implemented in most European countries, although the number of years of
historical data that has also been revised varies. Norway plans to make this change by the
end of 2006, Switzerland and the United Kingdom in 2007, and Japan in 2010.Turkey, which,
as mentioned above, has not yet moved towards SNA93 has no current plans in this area at
present, and New Zealand and Mexico have not yet made announcements.
Software investment. Another significant issue in the comparability of GDP concerns the
measurement of software. The 1993 SNA recommended that software expenditures be
treated as investment as long as the acquisition satisfied conventional asset requirements.
This change added nearly 2% to GDP for the United States, around 0.7% for Italy and
France, and about 0.5% for the United Kingdom. Doubts on the comparability of these data
were raised when comparing “investment ratios”, which are defined as the share of software
expenditures that are recorded as investment to total expenditures in software. These ratios
range from under 4% in the United Kingdom to over 70% in Spain (Lequiller, et al., 2003;
Ahmad, 2003). A priori, one would expect that these are roughly the same across OECD
countries. An OECD-Eurostat Task Force confirmed that differences in estimation
procedures contributed significantly to the differences in software capitalisation rates, and a
set of recommendations describing a harmonised method for estimating software were
formulated (Lequiller, et al, 2003; Ahmad, 2003). Many of these recommendations have
begun to be implemented by some countries but differences in software measurement will
nevertheless continue to have an impact on the international comparability of GDP levels for
some time to come.
The informal economy. Another factor that may influence the comparability of GDP across
countries is size of the non-observed economy. In principle, GDP estimates in the national
accounts take account of this part of the economy. In practice, questions can be raised about
the extent to which official estimates have full coverage of economic activities that are
included in GDP according to the SNA, or to which extent there some under-reporting is
involved. Large differences in coverage could substantially affect comparisons of productivity
levels.
It is not clear, a priori, how large the impact of these, and possible other, differences is on GDP
levels. What is clear, however, is that there is a margin of uncertainty associated with the
© OECD 2006
69
comparability of levels of GDP across countries. Consequently, there is also a range of uncertainty
associated with estimates of productivity levels; small differences between countries (of a few
percentage points) will obviously fall within this range of uncertainty. This is important in interpreting
estimates of productivity levels; countries within a small range of income and productivity levels may
not have income and productivity differences that are statistically or economically significant (Schreyer
and Koechlin, 2002).
The data choices for GDP are fairly uniform across different sources. In the OECD estimates of
productivity levels, data on GDP are derived from OECD's Annual National Accounts (ANA). The data
from ANA are based on OECD’s annual national accounts questionnaire to OECD member countries.
The data resulting from this questionnaire may differ somewhat from national sources and are more
comparable across countries than those derived from OECD’s quarterly national accounts (or the
OECD Economic Outlook database), thanks to some small methodological adjustments that are
made. For example, the US GDP estimates are adjusted for expenditure on military equipment, as
discussed above. However, the differences with other OECD sources, such as the Quarterly National
Accounts and the Economic Outlook database, are minor for most countries.
For two countries, Australia and New Zealand, the OECD’s Annual National Accounts provides
GDP estimates for fiscal years. This creates an inconsistency with other countries, since comparisons
of productivity levels ideally should correspond to the same (calendar) year. However, this problem is
considered relatively limited and no adjustment is made.
Labour input
Comparable measures of labour input are of great importance for international comparisons of
productivity levels. The OECD estimates of labour productivity levels are typically based on the same
6
data choices as the OECD estimates of labour productivity growth. Annex 1 therefore provides a
more detailed discussion of the measures of labour input used in the calculations of productivity levels.
Purchasing power parities for international comparisons
The comparison of income and productivity across countries also requires purchasing power
parity (PPP) data for GDP. Exchange rates are not suitable for the conversion of GDP to a common
currency, since they do not reflect international price differences, and since they are heavily influenced
by short-term fluctuations. The estimates used by the OECD are derived from its joint programme with
Eurostat and refer to current-price PPPs (Schreyer and Koechlin, 2002). For the current set of
comparisons, the most recent PPP benchmark comparison (for 2002) is used as the basis for the
estimates.
The OECD does not recommend the use of PPP-adjusted estimates of GDP in time series,
because of the difficulty to obtain PPPs that are consistent over time. This is why only one year of
productivity level comparisons is included in the OECD Productivity Database. Users interested in
adding a time dimension to this one year level comparison should use the corresponding database on
productivity growth, which gives appropriate indices of productivity growth for individual OECD
countries over a long time period.
OECD estimates of labour productivity levels for 2005, September 2006
Clearly, data for international comparisons of income and productivity are not perfect and some
choices between different sources have to be made. In the OECD approach, GDP is derived from the
6.
70
Some OECD countries have only very recently included estimates of hours worked in the national accounts. To the
extent possible, these estimates have been incorporated in the OECD estimates of productivity levels, as presented
in Annex Table 2. They have been fully incorporated in the OECD estimates of productivity growth and in the data
choices for those calculations, as shown in Annex Table 1.
© OECD 2006
Compendium of Productivity Indicators
OECD ANA database, which incorporates the latest comparative information on GDP from OECD
member countries. Data on employment for most countries are also from the OECD national accounts
as these should have a better correspondence to the estimates of GDP. For a limited number of
countries, no appropriate employment estimates are currently available from the national accounts, in
which case employment is derived from the OECD Labour Force Statistics. Estimates of hours worked
are either from the national accounts, or from the OECD Employment Outlook, as shown in Annex 1.
To convert GDP to a common currency, the OECD uses current PPPs, which are developed in the
OECD-Eurostat PPP programme.
Annex Table 2 presents the data choices and the resulting productivity level estimates for 2005.
These estimates still require further work in the following ways:
1. For several OECD countries, the estimates of annual hours worked per person are not
yet consistent with the national accounts. Data on hours worked currently reach the
OECD through two data collections. First, 22 OECD countries currently provide data on
hours worked to the annual data collection for the OECD Employment Outlook; 15 of
these countries provide the OECD with estimates of annual hours worked that are
consistent with national accounts concepts and coverage. Secondly, 18 countries
currently provide estimates of total hours worked in the framework of the national
accounts for inclusion in OECD’s Annual National Accounts. Further investigation of
these estimates of hours worked is needed.
2. The employment estimates that are currently incorporated in the national accounts are
not necessarily consistent across countries or with the corresponding estimate of GDP.
Addressing this problem will require further statistical work.
For analytical purposes, it is important that estimates of GDP per hour worked are combined with
estimates of GDP per capita and estimates of GDP per person in the labour force and GDP per
person of working age. The national accounts currently often do not include the necessary information
on working-age population and labour force, and such data have commonly been derived from labour
force statistics. The OECD’s change in method towards the national accounts as the main source of
employment information requires that the link between labour force statistics (i.e. national concepts)
and national accounts estimates of productivity (i.e. domestic concepts) is addressed.
© OECD 2006
71
Annex Table 2. OECD estimates of labour productivity for 2005 (September 2006)
GDP, million
national currency
units, based on
ANA
Australia 4
Austria
Belgium
Canada
Czech Republic
Denmark
Finland
France
Germany
Greece
Hungary
Iceland
Ireland
Italy
Japan
Korea
Luxembourg
Mexico
Netherlands
New Zealand 4
Norway
Poland
Portugal
Slovak Republic
Spain
Sweden
Switzerland
Turkey
United Kingdom
United States
OECD
G7
North America
OECD-Europe 5
EU-19 6
Euro-zone 7
PPP for total
GDP, 2004
GDP, million
USD
Employment
(1000
persons) 1
(1)
(2)
(3)
(4)
959,781
245,103
298,180
1,371,425
2,931,071
1,554,519
157,377
1,710,024
2,241,000
181,088
21,802,261
995,991
162,241
1,417,241
502,487,096
806,621,900
29,324
8,369,246
505,646
155,361
1,903,841
979,191
147,378
1,472,103
905,455
2,672,998
455,594
487,202
1,209,334
12,397,900
1.38
0.89
0.87
1.25
14.20
8.33
0.96
0.90
0.91
0.70
125.82
94.08
1.00
0.85
128.78
763.64
0.95
7.37
0.90
1.47
9.54
1.93
0.70
17.77
0.76
9.06
1.68
0.83
0.63
1.00
695,306
276,384
340,897
1,098,876
206,466
186,660
164,637
1,896,431
2,453,962
257,173
173,277
10,586
161,854
1,666,428
3,901,917
1,056,279
30,929
1,134,873
562,194
105,364
199,502
508,571
209,839
82,836
1,184,082
295,131
270,488
586,707
1,928,727
12,397,900
10,024
4,158
4,203
16,438
4,762
2,780
2,397
25,028
38,823
4,065
3,879
161
1,957
24,281
63,948
22,832
307
41,790
8,208
2,082
2,311
14,116
5,159
2,084
19,212
4,328
4,183
22,546
28,743
149,788
34,044,279
25,344,240
14,631,649
13,067,056
12,586,479
9,204,810
534,592
347,048
208,015
205,145
198,490
137,799
Annual average
hours worked, OECD source for Total hours
OECD source for
corresponding
to average hours worked (million
employment 2
employment
hours)
worked 2
3
estimates
(5)
ANA / LFS
ANA
ANA
ANA
ANA
ANA
ANA
ANA
ANA
ANA / LFS
ANA
ANA
ANA / EO
ANA
ANA / LFS
ANA
ANA
ANA / LFS
ANA
LFS
ANA
ANA
ANA / LFS
ANA
ANA
ANA
ANA
LFS
ONS
BLS
1,730
1,656
1,534
1,736
2,002
1,551
1,714
1,546
1,437
2,053
1,994
1,794
1,638
1,801
1,775
2,354
1,557
1,909
1,367
1,809
1,360
1,994
1,685
1,739
1,669
1,587
1,659
1,918
1,672
1,713
1749
1686
1754
1649
1652
1594
ABS
ANA
Empl. Outl.
ANA
Empl. Outl.
ANA
ANA
ANA / Empl. Outl.
ANA
ANA / EO
ANA
Empl. Outl.
Empl. Outl.
ANA / Empl. Outl.
Empl. Outl.
ANA
Empl. Outl.
Empl. Outl.
Empl. Outl.
Empl. Outl.
ANA
Empl. Outl.
Empl. Outl.
ANA
ANA
ANA
ANA / EO
GGDC
Empl. Outl.
BLS
GDP per
hour
worked,
USD
GDP per
hour
worked,
USA=100
(6)
(7)
(8)
17,345
6,885
6,447
28,543
9,534
4,313
4,107
38,702
55,804
8,347
7,733
289
3,206
43,733
113,532
53,742
478
79,776
11,221
3,765
3,143
28,147
8,693
3,624
32,068
6,869
6,937
43,244
48,071
256,587
40.1
40.1
52.9
38.5
21.7
43.3
40.1
49.0
44.0
30.8
22.4
36.6
50.5
38.1
34.4
19.7
64.7
14.2
50.1
28.0
63.5
18.1
24.1
22.9
36.9
43.0
39.0
13.6
40.1
48.3
83
83
109
80
45
90
83
101
91
64
46
76
104
79
71
41
134
29
104
58
131
37
50
47
76
89
81
28
83
100
934,887
584,972
364,906
338,353
327,983
219,692
36.4
43.3
40.1
38.6
38.4
41.9
75
90
83
80
79
87
1. The employment estimates for Austria, Canada, Greece, Japan, United Kingdom and United States refer to jobs.
2. ABS = Australian Bureau of Statistics; ANA = OECD Annual National Accounts; LFS = OECD Labour Force Statistics; Empl. Outl. = OECD Employment Outlook; ONS = UK Office for
National Statistics; GGDC = Groningen Growth and Development Centre; BLS = US Bureau of Labor Statistics.
3. The estimates of annual hours worked for Austria, Canada, Greece, Japan, United Kingdom and United States refer to hours worked per job. Data for France, Greece, Italy and
Switzerland are estimates.
4. GDP estimates refer to fiscal years.
5. Excluding Turkey.
6. All EU members that are also OECD member countries.
7. Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain.
Source : OECD Productivity Database, September 2006.
72
© OECD 2006
Compendium of Productivity Indicators
ANNEX 3: OECD DATABASES RELEVANT TO PRODUCTIVITY ANALYSIS
Annual National Accounts (ANA): The OECD’s Annual National Accounts presents a
consistent set of data mainly compiled on the basis of the 1993 System of National Accounts (SNA). It
contains data from 1970 whenever possible for OECD member countries. The database contains a
wide selection of accounts: main aggregates (GDP by expenditure, GDP by kind of activity, GDP by
income and disposable income, saving and net lending), detailed breakdown by kind of activity for
gross value added (at current and constant prices), components of value added, gross fixed capital
formation, employment and hours worked. It also includes final consumption expenditure of
households by purpose and simplified accounts for general government. Detailed accounts by
institutional sector are also available. The database also provides comparative tables based on
exchange rates and purchasing power parities.
Publications: OECD, National Accounts of OECD Countries, Volume I, 1993-2004, Main
Aggregates, 2006 Edition; OECD, National Accounts of OECD Countries: Detailed Tables 1993/2004,
Volume II, 2006 Edition. Also available annually on line on SourceOECD (www.sourceoecd.org).
Labour Force Statistics (LFS): The OECD’s Labour Force Statistics provides detailed statistics
on population, labour force, employment and unemployment, broken down by gender, as well as
unemployment duration, employment status, employment by sector of activity and part-time
employment. It also contains participation and unemployment rates by gender and detailed age
groups, as well as comparative tables for the main components of the labour force. Data are available
for each OECD member country and for OECD-Total, Euro zone and EU15. The time series cover 20
years for most countries. It also provides information on the sources and definitions used by Member
countries in the compilation of statistics.
Publication: OECD, Labour Force Statistics: 1985-2005, 2006 Edition. Available on SourceOECD
(www.sourceoecd.org).
Economic Outlook (EO): The OECD Economic Outlook Database is a comprehensive and
consistent macroeconomic database of the OECD economies, covering expenditures, foreign trade,
output, employment and unemployment, interest rates and exchange rates, the balance of payments,
outlays and revenues of government and of households, and government debt. For the non-OECD
regions, foreign trade and current account series are available. The database contains yearly and
quarterly information, both for the historical period and for the projection period. The historical data is
based on data from national statistical agencies and OECD sources such as the Quarterly National
Accounts, the Annual National Accounts, the Quarterly Labour Force Statistics, the Annual Labour
Force Statistics and the Main Economic Indicators.
Publication: OECD, Economic Outlook, twice yearly. The database is available on CD-ROM:
http://www.oecd.org/dataoecd/47/40/32108765.pdf A statistical annex to the OECD Economic Outlook
is available at: http://www.oecd.org/document/61/0,2340,en_2649_34109_2483901_1_1_1_1,00.html
Productivity Database: The OECD Productivity Database presents productivity at the aggregate
level and is designed to be as comparable and consistent across countries as is currently possible
(see Annex 1). The database and related information on methods and sources were first made
publicly available through the OECD Internet site on 15 March 2004. The OECD Productivity
Database currently provides annual estimates of:
•
Labour productivity growth, measured as GDP per hour worked, for 29 OECD countries and
the period 1970-2005 (PROD);
© OECD 2006
73
•
Capital services and multi-factor productivity for 19 OECD countries for the period 1985-2005
whenever possible (CAP);
•
Labour productivity levels for 2005 (see Annex 2; PRODL).
The OECD Productivity Database is updated on a regular basis as new data become available. It
is accessible through the OECD Internet site, at: www.oecd.org/statistics/productivity
STAN: The STAN database for Industrial Analysis includes annual measures of output, labour
input, investment and international trade by economic activity that allow users to construct a wide
range of indicators focused on areas such as productivity growth, competitiveness and general
structural change. The industry list based on the International Standard Industrial Classification (ISIC)
Rev. 3 provides sufficient detail to enable users to highlight high-technology sectors and is compatible
with those lists used in related OECD databases in the ‘STAN’ family (see below). STAN-Industry is
primarily based on member countries’ annual National Accounts by activity tables and uses data from
other sources, such as national industrial surveys/censuses, to estimate any missing detail. Since
many of the data points in STAN are estimated, they do not represent the official member country
submissions. See: www.oecd.org/sti/stan.
Publication: STAN-industry is available on line via SourceOECD (www.sourceoecd.org) with
tables showing data up to 2003. STAN-industry is also available on CDROM together with the latest
versions of STAN–R&D (ANBERD), STAN–Bilateral Trade Database (BTD) and a set of derived STAN
Indicators. See www.oecd.org/sti/stan/indicators.
AFA: The Activities of Foreign Affiliates database presents detailed data on the performance
of foreign affiliates in the manufacturing industry of OECD countries (inward and outward
investment). The data indicate the increasing importance of foreign affiliates in the economies of host
countries, particularly in production, employment, value added, research and development, exports,
wages and salaries. AFA contains 18 variables broken down by country of origin and by industrial
sector (based on ISIC Rev. 3) for 23 OECD countries.
Publication: OECD, Measuring Globalisation: Economic Globalisation Indicators, 2005 Edition.
Available on SourceOECD (www.sourceoecd.org).
FATS: This database gives detailed data on the activities of foreign affiliates in the services
sector of OECD countries (inward and outward investment). The data indicate the increasing
importance of foreign affiliates in the economies of host countries and of affiliates of national firms
implanted abroad. FATS contains five variables (production, employment, value added, imports and
exports) broken down by country of origin (inward investments) or implantation (outward investments)
and by industrial sector (based on ISIC Rev. 3) for 21 OECD countries.
Publication: OECD, Measuring Globalisation: Economic Globalisation Indicators, 2005 Edition.
Available on SourceOECD (www.sourceoecd.org).
74
© OECD 2006
Compendium of Productivity Indicators
Current country coverage of main databases used in this publication
Australia
Austria
Belgium
Canada
Czech Republic
Denmark
Finland
France
Germany
Greece
Hungary
Iceland
Ireland
Italy
Japan
Korea
Luxembourg
Mexico
Netherlands
New Zealand
Norway
Poland
Portugal
Slovak Republic
Spain
Sweden
Switzerland
Turkey
United Kingdom
United States
© OECD 2006
General databases
ANA
LFS
EO
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Productivity Database
PROD
CAP PRODL
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Industry-level
STAN
AFA
FATS
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
75
REFERENCES
Ahmad, Nadim (2003), “Measuring Investment in Software”, STI Working Paper 2003/6, OECD, Paris
Ahmad, Nadim (2003), “Towards More Harmonised Estimates of Investment in Software”, OECD
Economic Studies, No. 37, 2003/2, pp. 185-200.
Ahmad, Nadim, Francois Lequiller, Pascal Marianna, Dirk Pilat, Paul Schreyer and Anita Wölfl (2003),
“Comparing Labour Productivity Growth in the OECD Area: The Role of Measurement”, STI
Working Paper 2003/14 also published as OECD Statistics Working Papers 2003/5, OECD,
Paris.
Ahmad, Nadim, Francois Lequiller, Pascal Marianna, Dirk Pilat, Paul Schreyer and Anita Wölfl (2003),
“Comparing Growth in GDP and Labour Productivity: Measurement Issues”, OECD Statistics
Brief, No. 7 December, OECD, Paris.
Atkinson, Tony (2005) Atkinson Review of Measurement of Government Output: Final report; available
on http://www.statistics.gov.uk.
Australian Bureau of Statistics (2000); Australian National Accounts: Concepts, Sources and Methods,
Catalogue No 5216.0, Chapter 27.
Baldwin, John and Tarek Harchaoui (2005), “The Integration of the Canadian Productivity Accounts
within the System of National Accounts: Current Status and Challenges Ahead”; Research
Paper, Statistics Canada Economic analysis methodology paper series: National accounts;
11F0026 No 004.
Bassanini, Andrea and Stefano Scarpetta (2001), ”Does Human Capital Matter for Growth in OECD
Countries”, Economics Department Working Papers No. 282, OECD, Paris.
Bassanini, Andrea and Stefano Scarpetta (2001), “The Driving Forces of Economic Growth: Panel
Data Evidence for the OECD Countries”, OECD Economic Studies, No. 33, 2001/II, pp. 9-56,
OECD, Paris.
Beffy, P-O., Patrice Ollivaud, Pete Richardson and F.Sédillot (2006), “New OECD Methods for SupplySide and Medium-Term Assessments: A Capital Services Approach”, Economics Department
Working Paper 2006/482, OECD, Paris.
Boarini, Romina, Åsa Johansson and Marco Mira d’Ercole (2006), Alternative Measures of Well-Being,
OECD Statistics Brief, n°11, May.
Colecchia, Alessandra and Paul Schreyer (2001), “The Impact of Information Communications
Technology on Output Growth”, STI Working Paper 2001/7, OECD, Paris.
Colecchia, Alessandra and Paul Schreyer (2002), “The Contribution of Information and
Communications Technologies to Economic Growth in Nine Countries”, OECD Economic
Studies No. 34, pp. 153-171, OECD, Paris.
76
© OECD 2006
Compendium of Productivity Indicators
Commission of the European Communities, International Monetary Fund, Organisation for Economic
Co-operation and Development, United Nations, and World Bank (1993), System of National
Accounts 1993, Office for Official Publications of the European Communities Catalogue number
CA-81-93-002-EN-C, International Monetary Fund Publication Stock No. SNA-EA, Organisation
for Economic Co-operation and Development OECD Code 30 94 01 1, United Nations
Publication Sales No. E.94.XVII.4, World Bank Stock Number 31512.
Conway, Paul, Donato de Rosa, Giuseppe Nicoletti and Faye Steiner (2006), “Regulation,
Competition, and Productivity Convergence”, OECD Economics Department Working Papers,
No. 509, OECD, Paris.
Criscuolo, Chiara (2005), “The Contribution of Foreign Affiliates to Productivity Growth: Evidence from
OECD Countries”, STI Working Paper 2005/8, OECD, Paris.
De Serres, Alain (2003), ”Structural Policies and Growth – A Non-Technical Overview”, Economics
Department Working Papers No. 355, OECD, Paris.
Evans, John, Douglas Lippoldt and Pascal Marianna (2002), ”Trends in Working Hours in OECD
Countries”, Labour Market and Social Policy Occasional Paper 45, OECD, Paris.
Guellec, Dominique and Bruno van Pottelsberghe de la Potterie (2001), “R&D and Productivity
Growth: Panel Data Analysis of 16 OECD Countries”, OECD Economic Studies, No. 33, 2001/II,
pp. 103-126, OECD, Paris.
Kohli, Ulrich (2004); “Real GDP, real domestic income and terms of trade changes”; Journal of
International Economics 62.
Lequiller, Francois, Nadim Ahmad, Seppo Varjonen, William Cave, Kil-Hyo Ahn (2003), ”Report of the
OECD TaskForce on Software Measurement in the National Accounts”, OECD Statistics
Working Papers 2003/1, OECD, Paris.
Maddison, Angus (2001), The World Economy: A Millennial Perspective, Development Centre Studies,
OECD, Paris.
Nicoletti, Giuseppe and Stefano Scarpetta (2003), “Regulation, Productivity and Growth: OECD
Evidence”, OECD Economics Department Working Papers, No. 347, OECD, Paris.
OECD (1996), Industry Productivity: International Comparison and Measurement Issues, OECD, Paris
OECD (2001), Measuring Productivity - OECD Manual, OECD, Paris.
OECD (2001), OECD Science, Technology and Industry Scoreboard, D.4. OECD, Paris
OECD (2001), Measuring Capital – OECD Manual: Measurement of Capital Stock, Consumption of
Capital and Capital Services, OECD, Paris.
OECD (2003), The Sources of Economic Growth in OECD Countries, OECD, Paris.
OECD (2003), Science, Technology and Industry Scoreboard 2003, Section D and Annex 1, OECD,
Paris.
OECD (2003), ICT and Economic Growth – Evidence from OECD Countries, Industries and Firms,
OECD, Paris.
© OECD 2006
77
OECD (2004), The Economic Impacts of ICT – Measurement, Evidence and Implications, OECD,
Paris.
OECD (2004), “Recent Labour Market Developments and Prospects – Special Focus on: Clocking in
(and out): Several Facets of Working Time”, OECD Employment Outlook, Chapter 1, OECD,
Paris.
OECD (2005), Enhancing Services Sector Performance, OECD, Paris.
OECD (2005), Measuring Globalisation – OECD Handbook on Economic Globalisation Indicators,
OECD, Paris. Available at: www.oecd.org/sti/measuring-globalisation
OECD (2005), Measuring Globalisation – Economic Globalisation Indicators, OECD, Paris.
OECD (2005), OECD Economic Outlook No 78, December 2005, OECD, Paris.
OECD (2006), Structural and Demographic Business Statistics. OECD, Paris.
OECD (2006), Economic Policy Reforms: Going for Growth, 2006, OECD, Paris.
OECD (forthcoming), “Reader on the Measurement of Productivity Levels”, OECD, Paris.
Pilat, Dirk (2004), “The ICT Productivity Paradox: Insights from Micro Data”, OECD Economic Studies,
2004/1, pp. 37-65, OECD, Paris.
Pilat, D., A. Cimper, K. Olsen and C. Webb (2006), “The Changing Nature of Manufacturing in OECD
Economies”, STI Working Paper, OECD, Paris.
Pilat, Dirk, and Frank Lee (2001), “Productivity growth in ICT-producing and ICT-using industries: A
source of growth differentials in the OECD?”, STI Working Paper 2001/4, OECD, Paris.
Pilat, Dirk, Frank Lee and Bart van Ark (2002), “Production and Use of ICT: A Sectoral Perspective on
Productivity Growth in the OECD Area”, OECD Economic Studies No. 35, pp. 47-78, OECD,
Paris.
Pilat, Dirk and Paul Schreyer (2004), “The OECD Productivity Database – An Overview”, International
Productivity Monitor, Number 8, Spring, pp. 59-65.
Scarpetta, Stefano, Andrea Bassanini, Dirk Pilat and Paul Schreyer (2000), “Economic Growth in the
OECD Area: Recent Trends at the Aggregate and Sectoral Level”, Economics Department
Working Papers No. 248, OECD, Paris.
Schreyer, Paul (2000), “The Contribution of ICTs to Output Growth: A Study of the G7 Countries”, STI
Working Paper 2000/2, OECD, Paris.
Schreyer, Paul (2003), “Capital Stocks, Capital Services and Multi-Factor Productivity Measures”,
OECD Economic Studies, No. 37, 2003/2, pp. 164-184.
Schreyer, Paul and Francette Koechlin (2002), “Purchasing power parities – measurement and uses”,
Statistics Brief, March 2002, No. 3, OECD, Paris.
Schreyer, Paul and Dirk Pilat (2001), “Measuring Productivity”, OECD Economic Studies No. 33, pp.
127-169, OECD, Paris.
78
© OECD 2006
Compendium of Productivity Indicators
Schreyer, Paul, Pierre-Emmanuel Bignon and Julien Dupont (2003), “OECD Capital Services
Estimates: Methodology and A First Set of Results”, OECD Statistics Working Papers 2003/6,
OECD, Paris.
Swiss Federal Statistical Office (2006), Analyse des contributions à la croissance des facteurs de
production, de la productivité multifactorielle et du rôle joué par l'intensité capitalistique de 1991
à 2004, Neuchâtel.
Swiss Federal Statistical Office (2006), « Stock de capital non financier », Rapport méthodologique,
Neuchâtel.
Swiss Federal Statistical Office (2006), « Productivité multifactorielle », Rapport méthodologique,
Neuchâtel.
Triplett, Jack (2006), Handbook on Hedonic Indexes and Quality Adjustment in Price Indexes: Special
Application to Information Technology Products, OECD, Paris.
United States Bureau of Labor Statistics (1997), Handbook of Methods; BLS Bulletin 2490.
Wölfl, Anita (2003), “Productivity Growth in Service Industries: An Assessment of Recent Patterns and
the Role of Measurement”, STI Working Papers 2003-7, OECD, Paris.
Wölfl, Anita (2005), “The Service Economy in OECD Countries”, STI Working Paper 2005/3, OECD,
Paris.
© OECD 2006
79