Recent productivity developments in the world economy: an overview from The Conference board total economy database.
Chen, Vivian ; Gupta, Abhay ; Therrien, Andre 等
AGAINST THE BACKDROP OF the recession, global productivity,
measured as output per person employed, fell 1.0 per cent in 2009
according to preliminary estimates published as part of The Conference
Board Total Economy Database in January 2010. This decline in world
productivity is the first significant contraction of that benchmark
since the early 1980s (Chart 1). The current productivity slowdown has
hit countries across the world. Still there remains a large diversity
between advanced economies, on the one hand, and emerging and developing
economies, on the other. When measured as a group, advanced economies
saw the growth rate of labour productivity fall from 1.3 per cent in
2007 to -1.2 per cent in 2009. In emerging and developing economies it
dropped from 6.3 per cent in 2007 to 1.8 per cent in 2009. (2) Between
the advanced economies themselves there has also been high diversity,
with the United States performing at considerably higher labour
productivity growth rates than most European countries during 2009. Also
Asian economies have generally performed better than other emerging
economies.
[GRAPHIC 1 OMITTED]
While short-term productivity movements are highly volatile during
peaks and troughs, the long-term trend worldwide over the past three
decades has been toward faster productivity growth (Chart 2). This is
mainly due to emerging and developing economies that have rapidly taken
over leadership in productivity growth since the early 2000s. In
contrast, the average long-term growth of labour productivity in
advanced economies has stalled since 2000, even though the trend in the
United States started slowing after 2004 only. How ever, levels of
productivity in emerging and developing countries are still much lower
than in the advanced economies, leaving substantial scope for catching
up and a strengthening of competitiveness relative to advanced
economies.
Total factor productivity (TFP) growth, which measures the change
in GDP growth over the compensation-share weighted growth of combined
factor inputs (labour and capital inputs, adjusted for change in their
quality), has weakened in advanced countries, dropping from 0.4 per cent
per year in the period between 1995 and 2005 to -0.1 per cent from 2005
to 2008. TFP growth in emerging and developing economies, on the other
hand, has strongly improved from 1.0 per cent in 1995-2005 to 2.2 per
cent in 2005-2008 (Chart 3). During the most recent years (2007-2009),
total factor productivity has remained at much higher growth rates in
emerging and developing economies than in advanced economies.
The above summary of measures of productivity, output and input is
based on The Conference Board Total Economy Database. It is a detailed
dataset that provides output, input and productivity for 123 countries
around the world since 1950. The purpose of this database is to
facilitate international comparisons of productivity performance at the
macroeconomic level by providing consistent and reliable data; to
support empirical and theoretical research in the area of productivity
and growth accounting; to examine long term growth trends; and to
provide a basis for growth forecasts and projections.
In this overview article, we first discuss the key characteristics
of the database. We then introduce the growth accounting methodology and
the construction of the variables. In the final section we present a
brief analysis of some of the major results observed from the January
2010 release of the database.
Organization, Data Sources and Methodology of the Total Economy
Database
Organization of the database and public access
The Conference Board Total Economy Database was originally
developed by the Groningen Growth and Development Centre at the
University of Groningen in the Netherlands in the early 1990s. Since the
late 1990s, it has been produced in partnership with The Conference
Board, and as of 2007 the database was transferred from the University
of Groningen to The Conference Board in New York, which has maintained
and extended the database since then.
Two distinguishing features of the database are its wide country
coverage and its timeliness. The scant and inconsistent data in emerging
and developing economies is the bottleneck for most international
comparisons. The Total Economy Database makes use of information from
the latest national accounts, labour force surveys, and other employment
statistics available for individual countries. In order to maximize
international consistency, the figures are largely derived from the most
reliable international sources, such as the Organization for Economic
Cooperation and Development (OECD); the Statistical Office of the
European Union (Eurostat); the International Monetary Fund (IMF); the
International Labour Organization (ILO); and the World Bank. However,
for many counties data from international sources have been supplemented
with those from national statistical offices to increase timeliness when
possible. Hence in most cases the estimates are updated to the year t-1,
where t is the current year (in this case 2010), using a combination of
forecasts and estimates up to the third quarter of the year.
The database provides annual estimates of the levels and growth
rates of GDP, total population, employment, hours, labour quality,
capital services, labour productivity, and total factor productivity
starting from 1950 for 123 economies in the world, representing 97 per
cent of the world's population and 99 per cent of global output.
The level estimates are expressed in 2009 U.S. dollars, and converted at
purchasing power parity to adjust for differences in relative price
levels between countries. The database is publicly released every year
in January, with series covered up to year t-2 (2008 in the current
version), preliminary estimates for the year t-1 (i.e. 2009) and
projections for the current year t (i.e. 2010).
[GRAPHIC 2 OMITTED]
With the latest release (January 2010), the database has been
extended with a module on sources of growth, including labour quantity
and quality, capital services (non-ICT and ICT), and total factor
productivity. The extended module aims to integrate two previous data
sets: the world economy productivity data set created by Dale Jorgenson
and Khuong Vu of Harvard University (Jorgenson and Vu, 2009) (3) and the
Total Economy Growth Accounting Database of the Groningen Growth and
Development Centre (Ypma, Timmer and van Ark, 2003; Timmer and van Ark,
2005). (4)
[GRAPHIC 3 OMITTED]
The data series are publicly available without charge from The
Conference Board Total Economy Database website (http://www.conference
board.org/economics/database.cfm). The main results have been discussed
in a short publication for members of The Conference Board in January,
The Productivity Brief (The Conference Board, 2010), which is a prelude
to The Conference Board's annual Performance Report to be released
in the Fall with an updated version of the database.
Below we provide a brief description of data sources and
methodology including the latest changes, in particular to the use of
PPPs, the aggregation to regional and world averages and the
introduction of TFP estimates. More detail on the methodology can be
found in the document Methodological Notes and detailed source
descriptions are given in Detailed Sources. These files are all
downloadable from the database website.
Output: GDP and Purchasing Power Parities
The most frequently used measures of efficiency of an economy is
labour productivity, which is the average output produced per unit of
labour. Labour productivity estimates are obtained by dividing the real
output measure (Gross Domestic Product) by the total labour input used
to produce that output. Two measures of labour productivity are
included: output per person employed and output per hour for countries
with total hours data available.
The output measures in the database represent Gross Domestic
Product at market prices, which are obtained from national accounts
sources from international organizations and national statistical
institutes. (5) The post-1990 measures are obtained from a variety of
sources, including the National Accounts and Economic Outlook of the
OECD, national accounts data from Eurostat, and the IMF World Economic
Outlook Database. Pre-1990 measures are mostly obtained from historical
series, collected by Angus Maddison (2007). (6)
To allow international comparisons of the levels of labour
productivity, output levels are adjusted by purchasing power parities
(PPPs) to take into account the differences in price levels. The
measures of GDP and productivity levels are expressed in constant
U.S.-dollar market prices for 2009, and are adjusted for cross-country
differences in the relative prices of goods and services using PPPs. Two
measures of GDP in dollars are available from the database, one which is
converted at EKS PPPs and the other at Geary-Khamis (GK) PPPs. (7) The
original EKS series, which are measured in constant 2005 U.S. dollars
and are extrapolated to 2009 with GDP deflator changes, are unpublished
estimates from Penn World Tables (PWT). These estimates, which were
kindly provided by Alan Heston and will be used in the upcoming version
of PWT 7, are benchmarked on the 2005 PPPs from the International
Comparisons Project (ICP) at the World Bank (World Bank, 2008), and are
available for 111 of the 123 countries in the database. (8) The
adjustments made by PWT reflect: (9)
1) an adjustment for global weighting for individual countries
using EKS weights over domestic absorption (DA) for all countries
rather than over five main regions as was done in the ICP by the
World Bank;
2) an adjustment for the net foreign balance using the PPP for
domestic absorption (DA) rather than the exchange rate as in the ICP;
and
3) a downward adjustment in the PPP for China, which originally was
based on relatively high prices for 11 cities, in order to better
reflect the impact of relatively lower prices in rural areas in China,
which were not adequately reflected in the original World Bank estimate.
The effect of the first two adjustments increased GDP (in U.S.
dollars) for the global economy (all countries excluding the United
States) by 7.6 per cent relative to the U.S. in 2005. The China
correction adds another 2 percentage points to global GDP (excluding the
United States, which is the benchmark country). For China specifically
the first two effects lead to an upward adjustment in GDP of 13 per cent
relative to the World Bank measure, and together with the adjustment for
prices (the third effect) even to an upward adjustment of 28.5 per cent
of the World Bank GDP level for China. (10)
Geary-Khamis series of GDP are expressed in 1990 U.S. dollars and
are available for all of the 123 countries in the database. The
benchmark year estimates were in almost all cases derived from Maddison
(2007). Maddison used a PPP for China which was constructed for 1986,
and which is much lower than the recent PPP obtained by the ICP/World
Bank. As a result Maddison's GDP level for China in U.S. dollars is
roughly 40 per cent higher than that of the World Bank. We adjusted
Maddi son's GDP level for China downwards by 22.6 per cent, which
brings it closer to the adjustments for China in the PWT PPP index, as
described above.
Labour quantity: Employment and hours worked
From the perspective of productivity measurement, it is very
important that the measures of employment used are consistent with the
measures of output. In this regard, the key concern is that employment
figures need to cover all persons engaged in productive activity that
fall
within the production boundary of the system. (11) In terms of
production boundary, the domestic concept is adopted which includes all
workers employed domestically, but excludes any nationals working
abroad. Employment therefore should include employees, self-employed as
well as unpaid family members that are economically engaged, apprentices
and the military.
The employment data for most advanced countries since the 1990s are
from the National Accounts (domestic concept) from the OECD and
Eurostat, supplemented by the growth rates mostly from labour force
surveys to extrapolate backward the employment levels for earlier years.
For many developing countries, the employment figures do not strictly
follow the international standard defined above due to the lack of
qualified data sources. (12)
Still, output per person employed is a crude measure of labour
productivity. Total hours worked measures the quantity of labour input
more accurately, and is defined as the aggregate number of hours
actually worked during the year in employee and self-employment jobs.
Series of hours worked are currently available for 51 countries in the
database with OECD and Eurostat National Accounts being the major data
sources for recent years. These data sources aim to ensure that the
total hours worked is within the production boundary and that it is
consistent with the employment data used in the database.
Growth accounting
Output and labour quantity allow for the calculation of labour
productivity. Another type of productivity measure is total factor
productivity (TFP), which is average output produced by a combination of
multiple inputs, including labour and capital input, and with
adjustments for changes in the quality of labour and changes in the
composition of capital assets. To obtain total factor productivity
estimates, a standard growth accounting framework is used to compute the
contribution of these inputs to aggregate output (GDP) growth. The
growth accounting methodology has been pioneered by Solow (1957) and
further developed by Jorgenson and associates (Jorgenson and Griliches,
1967; Jorgenson, Gollop and Fraumeni, 1987).
In the general production function below, output (Y) is produced by
an input bundle X, consisting of capital services (K) and labour
services ([L.sub.Q]). Capital services are decomposed into six asset
types: computer hardware; software; telecommunications equipment;
dwellings, buildings and structures; transport equipment; and machinery.
Labour services ([L.sub.Q]) are the product of labour quantity (L) and
labour quality (Q). Input (X) is augmented by a Hicks-neutral total
factor productivity (A).
1) Y = AX([L.sub.Q], K)
Under the assumption of perfect competitive factor markets (where
the marginal product of each input equals its price) and constant
returns to scale, the above production function can be transformed into
the following growth accounting framework:
2) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
where [DELTA]lnX denotes the growth rate of variable X over two
time periods, v's stand for the average input shares in total
factor income and because of constant returns to scale, [MATHEMATICAL
EXPRESSION NOT REPRODUCIBLE IN ASCII]. Equation (3) can be arranged to
per hour/ worker terms:
3) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
where y is labour productivity, defined as y = Y/L, the ratio of
total output to labour quantity, and k is capital deepening, defined as
k = K/L, the ratio of capital services to labour quantity. Total hours
worked is a preferred measure of labour quantity. When this variable is
not available in most developing and emerging economies, total
employment is used instead under the assumption that the average hours
worked per person do not change and the change in total hours worked
equals the change in total employment. Equation (2) and (3) illustrate
that the output growth is driven by a share weighted input growth and
TFP growth, a residual that captures all sources of growth which are
left unexplained by labour and capital services in the production
function.
Labour quality
The labour input, whether in terms of total employment or hours
worked, represents a series of labour quantity. In order to measure
labour's contribution to output growth, an adjustment for changes
in the quality of labour is needed. The labour quality index, which is
constructed from a weighted summation of the percentage of labour force
in low, medium and high skill levels using relative wages as weights for
three skill levels respectively, ranges between 1 and 2.8 for developing
economies and between 1 and 2.25 for advanced economies. (13) Thus a
labour quality index of 1 indicates that all working force population is
of low skill and an index of 2.8/2.25 shows that all is of high skill.
These measures are calculated on an annual basis to determine trends in
the labour quality index.
The labour force skill level distribution is compiled from three
databases: (1) Barro and Lee (2000), (2) EU KLEMS Growth and
Productivity Accounts (Timmer et al., 2007), and (3) projections by the
International Institute for Applied System Analysis (IIASA, 2010). Both
Barro and Lee and the IIASA projection paper classify the population of
15+ into no schooling, primary, secondary and tertiary schooling for
5-year intervals, whereas EUKLEMS categorizes the percentage of total
hours worked into low, medium and high skill level on an annual basis.
There are discrepancies between the three datasets in terms of both
coverage (i.e. population vs. hours worked) and definitions, and we lack
information to consolidate the three datasets into a unified one due to
these data limitations. Instead we have used a statistical relationship
among these three datasets to construct an annual labour quality index
for 104 countries from 1960 onwards divided into three skill levels for
the labour force. The weights for three skill levels are calculated
based on the EUKLEMS data, and reflect average relative wages for the
aggregate of advanced and for that of emerging and developing countries,
and subsequently allocated to each country belonging in one of the two
country groups. (14)
Capital services: ICT and non-ICT
We obtained measures of capital services for two major asset
groups, each including three asset types: non-ICT capital, including
non-residential construction, transport equipment and machinery; and ICT
capital, including IT hardware, telecommunication equipment and
software. For each type of asset, a capital stock series, [K.sub.i,t] is
constructed from the investment data, [1.sub.i,t]. The perpetual
inventory method with a geometric depreciation rate is used as follows:
4) [K.sub.i,t] = (1 - [[delta].sub.i])x [K.sub.i,t] 1 + [l.sub.i,t]
All values in the above equation are in real terms (quantities).
The initial capital stocks [K.sub.0,i] are obtained by assuming initial
values equal to [[l.sub.i],0]/[g + [[delta.sub.i] , where g is the
average GDP growth rate and [l.sub.i,0] is the investment in asset type
i in the initial period. The same set of depreciation rates,
[[delta].sub.i] , is used for all countries:
* for non-ICT capital: construction - 0.03, Transport - 0.2,
Machinery - 0.13
* for ICT capital: IT Hardware - 0.3, Telecom Equipments - 0.12,
Software - 0.46
Growth in capital services flow is measured by the weighted sum of
growth in different types of capital stock.
[DELTA]lnK = ln[K.sub.t] - ln[K.sub.t-1] = [summation over
i][[bar.V].sub.i,t][DELTA]ln[K.sub.i,t]
The weights in the above equation are two-period average shares
(that is, the average of the shares in period t and period t-1) of each
asset type in the value of total capital compensation.
6) [[bar.V].sub.i,t] = [[[V.sub.i,t] + [V.sub.i,t-1]/2]
7) [[bar.V ].sub.i,t] =
[p.sub.i,t][K.sub.i,t]/[summation.sub.t][[P.sub.i,t][K.sub.i,t] The
rental price Pi,t of capital services from asset type i in period t is
defined as
8) pi = [r.sub.t] + [[delta].sub.i] - [[pi].sub.i,t]
In the above rental price equation, [r.sub.t] is the nominal rate
of return and [[pi].sub.i,t] is the asset price inflation (or capital
gains). The asset price inflation is calculated using current price and
constant price investment series. The ex-post or internal rate of return
is calculated by estimating the capital revenue for each time period
from the labour compensation share data:
9) (1 - [LabShare.sub.t]) x [GDP.sub.t.sup.Current]=
[summation.sub.[iota]][p.sub.i] x [K.sub.i,t]
If labour share data are not available or the estimated internal
rates of return are negative or very large, ex-ante rates of return are
taken from the IMF International Financial Statistics series on central
bank discount rate, government bond yield and lending rate.
The aggregated growth rates of capital services (ICT and non-ICT)
are calculated as the weighted sum of growth rates of individual capital
stocks, using the shares in capital compensation obtained from equation
6, 7 and 8 as weights. The total compensation share of capital input in
output is derived as the residual, i.e. one minus the share of labour
compensation in total factor income.
The data on non-ICT investment by asset type is based on the Penn
World Tables investment dataset, as described and reworked by Erumban
(2008). For OECD countries, the Penn World Tables dataset is extended by
linking it with OECD investment series from the year 2004 onwards. For
non-OECD countries, data are available only for aggregate gross fixed
capital formation from United Nations National Accounts, and we used the
2004 distribution from Penn World Tables for later years. The
asset-specific investment price deflators are obtained by using the
rates of asset-price inflation in 2004 for individual countries obtained
from PWT for the later years as well.
The Jorgenson & Vu (2009) dataset is used to integrate the data
for ICT investment with non-ICT investment. These datasets are extended
to more recent years using WITSA Digital Planet Report 2008, published
by the World Information Technology and Services Alliance (WITSA). Since
WITSA only reports total expenditure on ICT, the purchases made by
consumers need to be removed to obtain the estimates for the investments
in ICT assets. The estimation of business investment (out of the total
reported in WITSA) is based on the latest values in the Jorgenson &
Vu dataset for the year 2005. These series also need to be converted to
constant prices or volume series. The price deflators for the years
2006-2008 are estimated by assuming the same rate of ICT asset-price
inflation as in 2005 for the later years as well.
Labour share
The labour share, defined as the ratio of total labour compensation
to gross value added at basic prices, is used to assign weights to
labour and capital inputs in the growth accounting equation. The labour
shares from EUKLEMS are used whenever the data are available. OECD and
Eurostat also report data on labour compensation for employees, which
are used to fill the gaps. In those cases we assume that the
compensation for self-employed can be imputed from the average
compensation for employees by adjusting the employee labour compensation
share for the employee share among total employment to obtain the total
labour compensation share among GDP.
For a number of large non-OECD, non-EU countries, we estimated the
labour share using alternative sources. In the case of China, the labour
share is estimated from input-output tables. For Brazil, India and
Russia the labour share is calculated using compensation data from the
ILO. For the other emerging and developing economies, we use 0.5 as the
labour share. In much of the growth accounting literature, a labour
share of 0.7 is widely used across time and countries. (15) However, we
decided to use 0.5 as the labour share for emerging and developing
economies, because capital is relatively scarce in most of those
remaining economies, and thus its return is high, while labour is cheap
compared to advanced countries, leading to a lower labour share. Also,
the adjustment for the labour share that is allocated to self-employed
remains relatively large in many developing economies.
Aggregation of levels and growth rates
Growth rates for individual countries are calculated using the log
difference in levels instead of the percentage change in the actual
level. We chose this method in order to facilitate aggregation as well
as decomposition of the growth for individual countries and components.
With regard to the aggregation to regional country groupings, the
following methods are used for GDP, labour input and labour productivity
growth respectively:
10) [DELTA]ln[Y.sub.region] =
[[summation].sub.i][[bar.W].sub.i]ln[Y.sub.i]
11) [DELTA]ln[L.sub.region] =
[DELTA]ln[[summation].sub.i][L.sub.i]]
12) [DELTA]ln[y.sub.region] =
[[summation].sub.i][[bar.W].sub.i][DELTA]ln[y.sub.i] +
[([[summation].sub.i][[bar.W].sub.i][DELTA]ln[L.sub.i] - [DELTA]ln
[[summation].sub.i][L.sub.i])] =
[[summation].sub.i][[bar.W].sub.i][DELTA]ln[Y.sub.i]] + R]
with [w.sub.i] as the country share in PPP adjusted nominal GDP of
the region for each year and a bar denoting the two-period average.
Hence aggregate GDP growth is the weighted sum of the country GDP
growth. Growth in labour quantity (employment or hours) is simply the
log difference of the summed total labour quantity values for all the
countries in one region. The aggregate labour productivity growth is the
weighted sum of country productivity growth rates plus a reallocation
term R. This reallocation term is positive if employment shifts from low
productivity countries towards high productivity countries.
The levels of regional GDP and labour productivity are calculated
by applying the above PPP-adjusted current GDP-weighted growth rates to
the benchmark year, 2009. (16) The labour productivity in 2009 is simply
calculated as
[[summation].sub.i][Y.sub.i]/[[summation].sub.i][L.sub.i.]
Aggregate total factor productivity growth rates for different
regions are calculated by using the aggregate output, input and labour
shares from the growth accounting equation. Aggregate output and input
(quality adjusted labour input (17) and capital services) growth rates
are calculated by taking the weighted average of individual country
growth rates. The weights used are two period averages of the country
shares in the PPP-adjusted nominal GDP of the group for each year. The
aggregate labour compensation share for each year, i.e.,
[sup.[upsilon]L,region], is obtained by summing up the labour
compensation (PPP adjusted) of individual countries and then dividing
this sum by total nominal GDP (PPP adjusted) of the group. A bar on the
regional labour compensation share indicates the two-period average.
Thus the regional TFP growth rates are calculated using the formula
below:
13) [DELTA]ln[A.sub.region] = [DELTA]ln[Y.sub.region] -
[[bar.V].sub.L,region] [DELTA]ln[L.sub.Q,region] - (1 -
[[bar.V].sub.L,region])[DELTA]ln[K.sub.region]
A Bird's Eye View of Recent Productivity Measures
Table 1 summarizes the productivity, output and total hours growth
rates for a selected group of advanced economies. Despite the deep
recession, labour productivity growth in the United States strengthened
in "per hour" terms in 2009 to 2.5 per cent, up from 1.4 per
cent in 2008. Productivity growth in other advanced economies was much
weaker in 2009: 0.3 per cent in Japan, -1.0 per cent in the Euro Area,
and as much as -1.9 per cent in the United Kingdom. Hence the
productivity growth differential between the United States and the Euro
Area in 2009 is 3.5 percentage points, and between the United States and
the United Kingdom, it stands at 4.4 percentage points. For comparison,
the productivity growth differential between the United States and the
Euro Area was only 1 percentage point between 1995 and 2005 and 0.2
percentage points between the United States and the United Kingdom.
While estimates for 2009 are still preliminary, and adjustments are
made to both the output and hours numbers once more definitive data from
the national accounts and employment statistics are published, the
productivity growth differential is so large that some important
observations already emerge. First, the differences among advanced
countries (or regions) partly reflect differences in output declines
during 2009, which were much higher in Japan (-5.6 per cent), the Euro
Area (-4.1 per cent), and the United Kingdom (-4.8 per cent) than in the
United States (-2.5 per cent). The second factor, however, is the much
larger number of hours lost in the United States as a result of a very
sharp reaction of firms to the crisis. Hours worked declined 5.1 per
cent in the United States, with 3.6 percentage points caused by job
shedding and another 1.5 percentage points attributable to a reduction
in the working hours of workers who still had jobs. Labour hoarding in
Europe was a much larger factor, as the Euro Area saw total working
hours decline 3.1 per cent, of which 1.9 percentage points were related
to job losses.
Greater flexibility in labour markets may be seen as one cause for
the divergent patterns between the United States and the Euro Area, but
it probably does not tell the whole story as Japan, which does not have
a particularly flexible labour market either, also saw a large decline
in total hours worked (-5.9 per cent), which reflected a loss of 3 per
cent in workers and almost 3 per cent in hours worked per worker. In
contrast, the United Kingdom which has among the most flexible labour
markets in the European Union, showed a pattern that was not all that
different from the Euro Area. In the United Kingdom hours fell 2.8 per
cent, of which 2.0 percentage points were due to a decline in employed
persons.
While productivity growth in 2009 was negative in most European
countries, there were significant variations between major countries.
Germany performed far worse than France (-2.2 per cent in Germany and
0.3 per cent in France) due to a deeper contraction in output in Germany
(-5.0 per cent) than France (-2.3 per cent). Interestingly, Germany lost
no jobs in net terms, as all the loss in working hours (-2.8 per cent)
was due to shortening of working hours of employees who stayed in
employment. France's more moderate decline in output and slight
increase in productivity relates to the smaller impact of exports on the
economy, which was severely hit in Germany, and the sustained growth of
the public sector. While both Italy and Spain suffered strong output
declines (-4.9 per cent in Italy and -3.7 per cent in Spain), Spain shed
hours five times as rapidly (-7.5 per cent) as Italy (-1.7 per cent).
Much of the employment losses were in construction and tourism, which is
characterized by a large share of temporary, less well-protected jobs.
Consequently, labour productivity increased by 3.8 per cent in Spain
while it declined by 3.2 per cent in Italy.
Table 2 provides a summary of the growth rates for seven leading
emerging economies: Brazil, China, India, Indonesia, Mexico, Russia and
Turkey. On average, labour productivity growth of these major emerging
economies group was 3.6 per cent in 2009, which was down by 1.7
percentage points from the 5.3 per cent rate in 2008. But there were
large differences among the emerging economies. China showed the
strongest output and productivity performance in 2009 at 7.7 per cent
and 8.2 per cent respectively. (18) This was largely the result of a
boom in bank loans and rapid investment in infrastructure, which
stimulated output growth and at least temporarily boosted the
investment-intensive activities of state-owned enterprises (SOEs).
Although employment estimates are extremely difficult to obtain, the
evidence from a variety of partial sources suggests that employment
growth in China stalled as a result of layoffs by private companies,
especially export-oriented firms. Hence, while overall productivity
increased, there may have been important underlying structural changes
that may impact China's productivity trend ahead. For example, SOEs
and infrastructure construction probably occupied a larger footprint in
the economy in 2009 than before.
While Brazil and Mexico have been among the weakest productivity
performers historically, their performance diverged as Brazil's
output, employment, and productivity performance strengthened in recent
years. Productivity growth in Brazil stood at 1.5 per cent in 2009, down
from 4.0 per cent in 2008, whereas productivity growth in Mexico
continued to be negative in 2009 (-0.3 per cent in 2009 following a 0.9
per cent decline in 2008). In contrast, in addition to China, Indonesia
and India strengthened their performance in recent years; as they were
relatively shielded from the global crisis because of moderate exposure
to exports and the global financial world, their performance remained
reasonably strong during the recession. Labour productivity in India
grew at 3.9 per cent in 2009 as a result of 5.8 per cent GDP growth and
2.0 per cent employment growth. The negative productivity growth in
Indonesia (-0.3 per cent) is a result of faster growth in employment
than GDP (4.7 per cent for GDP growth and 5.0 per cent for employment
growth). In contrast, Russia and Turkey, which were both strongly
exposed to the global crisis, suffered most in terms of output (-3.8 per
cent and -4.5 per cent respectively), employment (0 per cent and -1.3
per cent respectively), and productivity (-3.8 per cent and -3.2 per
cent respectively).
While the short-term improvements in productivity help countries
position themselves to exploit their growth potential, the actual
trigger for sustainable growth is the long-term productivity trend (see
also Chart 2), which is also the main source for improvements in living
standards. To accelerate the long-term productivity trend, growth needs
to come from not only investment in inputs, which equip workers with
higher skills and better tools to produce, but also from an increase in
the efficiency with which these inputs (such as labour, workforce
skills, machinery, and technology inputs) are used, i.e., the total
factor productivity growth.
Table 3 and Chart 4 show the sources of output growth for major
regions and countries. For the world economy, the output growth of 4.4
per cent from 2005 to 2008 was partly due to an increase in labour
input, which contributed 0.7 percentage points to output growth. Another
0.2 percentage point was due to an improvement in the quality of the
labour force, measured as the skill level of the labour force according
to their level of educational attainment. Most of the output growth in
the world, however, has been due to a rise in non-ICT capital: it
accounted for half (2.2 percentage points) of total output growth (4.4
per cent). ICT capital contributed 0.4 percentage points to output
growth from 2005-2008, leaving a residual growth (TFP) of 0.9 per cent.
In contrast, during the earlier period 1995-2005, ICT capital
contributed 0.5 percentage points to output growth, leaving a TFP
residual of only 0.6 per cent. The acceleration of TFP growth after 2005
might represent a more efficient use of capital, which may relate to
either ICT or non-ICT capital.
[GRAPHIC 4 OMITTED]
The panels for the aggregate advanced and emerging economies show
the diverging developments in TFP growth especially since 2005. While
Germany and the United Kingdom still generated some TFP growth, largely
related to the peak of the business cycle, TFP growth stalled in Japan
and even somewhat declined in France--reflecting the greater
inefficiency of the growth process.
The faster output growth rates of emerging and developing economies
(at 7.3 per cent, relative to 2.0 per cent in advanced economies from
2005 to 2008) is largely due to the faster increase in non-ICT capital,
especially in China and India, and a much higher efficiency by which the
inputs are being used, especially in China. In 2008, the overall TFP
growth rates for emerging and developing economies was still at 0.85 per
cent, although lower than the average of 2.2 per cent from 2005-2008.
This was mainly due to the start of the recession and the cooling of the
Chinese and Indian economies. However, even among emerging economies
there are substantial differences: China showed an increase of 1.3 per
cent in TFP growth in 2008, which was modest compared to 2.2 per cent in
Brazil and 5.1 per cent in Russia, but much better than the -0.9 per
cent decline in India. The Brazilian and Russian economies probably
received a TFP bonus from the price boom in natural resource production
in 2008, while in India, the contribution of capital to growth remained
at the same level despite a significant deceleration in output growth in
2008. The rise in the long-term TFP trend puts the emphasis for future
growth even more strongly on the emerging economies. This raises their
competitive strength, as it helps these countries to match higher costs,
such as rising wages, by their ability to lower costs and prices through
efficiency gains.
Conclusion
It should be stressed again that there are substantial
uncertainties concerning the productivity estimates for recent periods.
National statistical offices often make significant adjustments to their
output and employment estimates as the measures from a range of surveys
and administrative sources come in with a delay of several months, and
are sometimes adjusted significantly during the process of
reconciliation of the various sources. Annual GDP growth estimates can
be adjusted by as much as one per cent upward or downward for advanced
countries and sometimes more for emerging economies while the
adjustments for employment are usually much smaller.
Nevertheless, we believe that "real time" productivity
figures provide useful insights as they provide signals on how the
direction of the productivity trend may be affected by the latest
estimates, and how the differences between countries can play out. For
example, while the post-1995 productivity growth differential between
the United States and the EU-15 (i.e. the EU member state constellation
before 2004) has been adjusted following several statistical revisions,
the productivity growth differential remained at roughly 1 percentage
point with the United States showing an approximately 2.5 per cent
increase in GDP per hour in 1995-2005 visa-vis one and a half per cent
in the European Union. The semi-annual updates in The Conference Board
Total Economy Database, however, keep track of ongoing revisions in the
data.
Projections of productivity growth are surrounded by even more
uncertainties, particularly in times of structural shifts such as in
current times. However, on the basis of GDP forecasts and assumptions on
the degree of procyclicality of productivity growth, the following
projections may be seen as plausible given the current economic
situation in various countries.
Following its dismal performance in 2009, global productivity is
expected to improve sharply to 2 per cent or even more in 2010. This
increase will be the result of the combination of a projected recovery
in world GDP of more than 3 per cent and a modest increase in world
employment. Advanced economies will mostly see jobless productivity
growth as labour markets recover slowly, although limited technology and
innovation gains could lead to higher than expected job growth in less
productive (in several non-market) service industries. Most emerging and
developing economies will experience a combination of productivity and
employment growth in 2010. This not only reflects their growing
contribution to world output growth, but also a strengthening of their
global competitiveness based not only on their low relative cost, but
also on increasingly higher productivity.
A long-term improvement in the productivity trend will depend on a
revival of global demand, stimulated by technological change and
innovation. The growth accounting approach, which is now integrated in
The Conference Board Total Economy Database, provides the framework for
the next step which is to develop medium term projections of the sources
of growth to strengthen the forecasts of GDP growth in international
comparative perspective. This sets the agenda for the next version of
the database, which will also include an increase in the number of
countries included.
References
Azeez Erumban, Abdul (2008) "Measurement and analysis of
capital, productivity and economic growth," University of Groningen
(http://dissertations.ub.rug.nl/faculties/feb/2008/a.e.erumban/)
Bonthuis, Boele (2009) "Measuring Labour Quality," The
Conference Board and University of Groningen, 2010,
(http://www.conferenceboard.org/economics/downloads/Paper_b_bonthuis.pdf)
Deaton, Angus, and Alan Heston (2009) "Understanding PPPs and
PPP-based national accounts," Princeton University and University
of Pennsylvania, NBER Working Papers No. 14499,
(www.nber.org/papers/w14499)
Gollin, Douglas (2002) "Getting Income Shares Right,"
Journal of Political Economy, Vol. 110, No. 2, pp. 458 - 474.
IIASA (2010) "IIASA Education Forward Projections for
2000-2050," available at http://
www.iiasa.ac.at/Research/POP/Edu07FP/index.html?sb=12.
Jorgenson, Dale W., Frank M. Gollop, and Barbara M. Fraumeni
(1987), Productivity and U.S. Economic Growth. Cambridge, MA: Harvard
University Press.
Jorgenson, Dale W., and Zvi Griliches (1967), "The Explanation
of Productivity Change," Review of Economic Studies, Vol. 34, No.
3, pp. 249-83.
Jorgenson, Dale W. and Khuong Vu (2009),
"Growth Accounting within the International Comparison
Program," ICP Bulletin, Vol. 6, No. 1, March, pp. 3-19.
Maddison, Angus (2007) Statistics on World Population, GDP, and Per
Capita GDP, 1-2006 AD, OECD, updated in 2009 (see www.ggdc.net/
Maddison/Historical_Statistics/horizontalfile_03-2009.xls).
Ravallion, Martin (2010) "Price levels and economic growth :
making sense of the PPP changes between ICP rounds," World Bank,
Policy Research Working Paper 5229, Washington D.C.
Solow, Robert M. (1957) "Technical Change and the Aggregate
Production Function." Review of Economics and Statistics, Vol. 39,
No. 3, pp. 212-20.
The Conference Board (2010) "The 2010 Productivity Brief:
Productivity, Employment, and Growth in the World's
Economies," Executive Action Series No. 319.
Timmer, Marcel P. and Bart van Ark (2005) "IT in the European
Union: A driver of productivity divergence?" Oxford Economic
Papers, Vol. 57 No. 4, pp. 693-716.
Timmer, Marcel P., Gerard Ypma and Bart van Ark (2003), "IT in
the European Union: Driving Productivity Divergence?" Groningen
Growth and Development Centre Research Memorandum GD-67, Groningen:
University of Groningen, June (update).
Timmer, Marcel P., Mary O'Mahony and Bart van Ark (2007)
"EU KLEMS Growth and Productivity Accounts: An Overview,"
International Productivity Monitor, Number 14, Spring, pp. 71-85.
World Bank (2008), International Comparison Program, Tables of
final results, Washington D.C.
Vivian Chen, Abhay Gupta, Andre Therrien, Gad Levanon and Bart van
Ark (1)
The Conference Board
(1) Vivian Chen and Abhay Gupta are economists at The Conference
Board, Andre Therrien is an economic analyst, Gad Levanon is associate
director macroeconomic research, and Bart van Ark is The Conference
Board's chief economist as well as a professor of economics at the
University of Groningen. Emails: vivian.chen@conference-board.org;
abhay.gupta@conference-board.org; andre.therrien@conference-board.org;
gad.levanon@conference-board.org; bart.vanark@conference-board.org.
(2) Advanced economies include United States, EU15, Japan, Canada,
Switzerland, Norway, Israel, Iceland, Cyprus, Korea, Australia, Taiwan,
Hong Kong, Singapore and New Zealand. Emerging and developing economies
include China, India and countries in developing Asia, Latin America,
Middle East, Africa, Central & Eastern Europe, Russia and other
Commonwealth of Independent States countries.
(3) See also Dale Jorgenson's website:
http://www.economics.harvard.edu/faculty/jorgenson.
(4) See also The Groningen Growth and Development Centre website:
http://www.ggdc.net/databases/ted_growth.htm
(5) Our focus here is on the aggregate economy, which includes
non-market and semi-public services such as health and education
services, as well as public administration and defense. Measurement
problems in these industries are substantial, and in several cases (in
particular for government), output growth is measured using input growth
which is affecting the aggregate productivity numbers. Another problem
case is real estate where output mostly reflects imputed housing rents
rather than sales of firms. Productivity measures of the market economy,
excluding these industries, are not available for as many countries as
in this study.
(6) For the latest Maddison estimates, see:
http://www.ggdc.net/Maddison/Historical_Statistics/horizontal-file_03-2009.xls
(7) "EKS" stands for the originators of this PPP formula,
Eltoto, Kovacs and Szulc, which essentially is a multilateral Fisher
index. Geary and Khamis are the originators of a PPP formula, which is a
multilateral index similar to binary Paasche index, giving relatively
large weights to large countries.
(8) The following 12 countries in the Total Economy Database are
not covered by the PWT PPPs thus do not have GDPEKS series: Algeria,
Barbados, Costa Rica, Dominican Republic, Guatemala, Jamaica, Myanmar,
St. Lucia, Trinidad & Tobago, Turkmenistan, United Arab Emirates and
Uzbekistan.
(9) We thank Alan Heston for providing the PWT rework of the ICP
PPP data. For a detailed description on the PWT PPPs, see Angus Deaton
and Alan Heston (2009).
(10) See also Ravaillon (2010) who defends the World Bank measures
of the China PPP, but admits to an upward bias due to the undercoverage
of rural prices in China.
(11) Employment has been defined by the International Labour
Organization (ILO) in the "Resolution concerning statistics of the
economically active population, employment, unemployment and
underemployment," adopted by the thirteenth International
Conference of Labour Statisticians. It is defined consistently in the
System of National Accounts 1993 (1993 SNA) and European System of
Accounts 1995 (1995 ESA). ILO: http://
www.ilo.org/public/english/bureau/stat/res/index.htm; 1993 SNA XVII:
Population and Labour Inputs: http:/
/unstats.un.org/unsd/sna1993/toctop.asp; 1995 ESA Chapter11 Population
and labour Inputs: http://
circa.europa.eu/irc/dsis/nfaccount/info/data/esa95/en/esa95en.htm.
(12) For example, the employment figures for most African and
Mid-East countries are actually labour force data, which are unadjusted
for unemployment or underemployment.
(13) The following weights are used: 1 for low skill, 1.42 (1.36)
for medium skill, 2.8 (2.25) for high skill for developing (advanced)
economies.
(14) See Bonthuis, 2009, for a more detailed explanation.
(15) For example, Gollin (2002) identified and compared several
adjustments for calculating labour shares and concluded that factor
shares are approximately constant across time and countries within a
range of 0.65 to 0.80.
(16) 2009 is set as the benchmark year for the aggregate levels in
order to be consistent with other tables in the Total Economy Database
(January 2010). Although the choice of benchmark year affects the
levels, it does not affect the growth rates of GDP and labour
productivity.
(17) For countries with missing labour quality data, labour input
reflects the change in labour quantity only, i.e., the change in
employment or total hours worked. Consequently, total factor
productivity growth in those countries is somewhat overstated due to
this missing input component. However, because of the generally small
contribution of the labour quality in the output growth, the TFP
overestimation is relatively low in magnitude.
(18) Note that growth rates are based on the difference in the log
of the levels of each variable. For example, China's GDP growth
rate in 2009 changed from 8.0 per cent, when calculated in percentage
terms, to 7.7 per cent when using log differences.
Table 1
Growth of Labour Productivity, Real GDP and Total Hours Worked
for Advanced Countries, 1995-2009
United Euro United
States Japan Area Kingdom
Labour Productivity Growth
(GDP per hour, annual average,
per cent)
1995-2005 2.4 2.1 1.4 2.2
2005-2009 1.5 0.8 0.5 0.9
2007 1.4 1.8 1.1 2.3
2008 1.4 0.1 0.1 1.0
2009 2.5 0.3 -1.0 -1.9
(estimate)
Real GDP Growth
(annual average, per cent)
1995-2005 3.3 1.1 2.2 2.9
2005-2009 0.7 -0.5 0.6 0.4
2007 2.1 2.4 2.7 3.0
2008 0.4 -0.7 0.6 0.5
2009 -2.5 -5.6 -4.1 -4.8
(estimate)
Growth in Total Hours Worked
(annual average, per cent)
1995-2005 0.9 -1.0 0.8 0.7
2005-2009 -0.9 -1.3 0.1 -0.5
2007 0.7 0.5 1.6 0.7
2008 -0.9 -0.8 0.4 -0.5
2009 -5.1 -5.9 -3.1 -2.8
(estimate)
France Germany Italy Spain Canada
Labour Productivity Growth
(GDP per hour, annual average,
per cent)
1995-2005 1.8 1.6 0.5 0.5 1.5
2005-2009 0.8 0.2 -0.8 2.0 0.1
2007 -0.2 0.7 0.1 1.6 0.4
2008 0.4 -0.3 -0.5 1.5 -0.7
2009 0.3 -2.2 -3.2 3.8 -0.2
(estimate)
Real GDP Growth
(annual average, per cent)
1995-2005 2.2 1.3 1.4 3.6 3.6
2005-2009 0.6 0.4 -0.6 1.2 0.6
2007 2.2 2.4 1.4 3.5 2.5
2008 0.3 1.0 -1.1 0.9 0.4
2009 -2.3 -5.0 -4.9 -3.7 -3.2
(estimate)
Growth in Total Hours Worked
(annual average, per cent)
1995-2005 0.4 -0.3 0.9 3.1 1.7
2005-2009 -0.2 0.1 0.2 -0.8 0.5
2007 2.4 1.8 1.3 1.9 2.1
2008 -0.1 1.3 -0.5 -0.6 1.1
2009 -2.6 -2.8 -1.7 -7.5 -3.0
(estimate)
Note: Growth rates are based on the difference in the log of the
levels of each variable
Source: The Conference Board Total Economy Database, January 2010
Table 2
Growth of Labour Productivity, Real GDP and Persons Employed
for Major Emerging Economies, 1995-2009
Major Brazil Russia India
Emerging
Economies
Labour Productivity Growth
(GDP per persons, annual average,
per cent)
1995-2005 4.1 0.3 3.7 4.2
2005-2009 5.9 2.6 3.7 5.2
2007 7.6 3.5 7.0 6.2
2008 5.3 4.0 4.7 4.0
2009 (estimate) 3.6 1.5 -3.8 3.9
Real GDP Growth (annual average,
per cent)
1995-2005 5.5 2.4 3.8 6.3
2005-2009 7.1 3.8 4.2 7.6
2007 9.1 5.5 7.8 8.6
2008 6.8 5.7 5.4 6.5
2009 (estimate) 4.1 0.0 -3.8 5.8
Growth in Persons Employed
(annual average, per cent)
1995-2005 1.4 2.0 0.1 2.0
2005-2009 1.3 1.2 0.5 2.3
2007 1.5 2.0 0.8 2.4
2008 1.6 1.7 0.8 2.4
2009 (estimate) 0.5 -1.5 0.0 2.0
China Mexico Indonesia Turkey
Labour Productivity Growth
(GDP per persons, annual average,
per cent)
1995-2005 6.7 1.4 1.5 3.6
2005-2009 9.6 0.5 2.1 1.0
2007 11.5 1.6 3.6 3.1
2008 8.6 -0.9 1.0 -1.2
2009 (estimate) 8.2 -0.3 -0.3 -3.2
Real GDP Growth (annual average,
per cent)
1995-2005 7.8 3.6 3.1 4.2
2005-2009 10.0 2.0 5.5 1.9
2007 12.2 3.2 6.1 4.6
2008 9.2 1.4 5.9 0.9
2009 (estimate) 7.7 -1.5 4.7 -4.5
Growth in Persons Employed
(annual average, per cent)
1995-2005 1.1 2.2 1.6 0.7
2005-2009 0.4 1.5 3.4 0.9
2007 0.8 1.7 2.5 1.5
2008 0.6 2.2 5.0 2.1
2009 (estimate) -0.5 -1.2 5.0 -1.3
Note: Growth rates are based on the difference in the log of the
levels of each variable
Source: The Conference Board Total Economy Database, January 2010
Table 3
Contribution of the Change in Inputs and TFP to Average Annual Output
Growth by Country and Region, 1995-2005, 2005-2008, 2007, 2008
(percentage points per year)
Country Period Labour Labour Non-ICT ICT
Quantity Quality Capital Capital
World 1995-2005 0.55 0.29 1.64 0.47
2005-2008 0.68 0.18 2.23 0.35
2007 0.76 0.18 2.26 0.36
2008 0.40 0.18 2.32 0.32
Advanced 1995-2005 0.41 0.26 1.16 0.43
Economies 2005-2008 0.55 0.17 1.01 0.35
2007 0.77 0.17 1.04 0.35
2008 -0.09 0.17 0.98 0.32
Emerging 1995-2005 0.71 0.31 2.37 0.46
Economies 2005-2008 0.79 0.20 3.90 0.23
2007 0.77 0.20 3.96 0.24
2008 0.79 0.19 4.06 0.22
United 1995-2005 0.63 0.19 1.37 0.48
States 2005-2008 0.41 0.13 0.97 0.30
2007 0.50 0.13 1.01 0.30
2008 -0.59 0.13 0.87 0.26
Japan 1995-2005 -0.58 0.35 0.97 0.53
2005-2008 0.12 0.16 0.82 0.29
2007 0.28 0.16 0.78 0.31
2008 -0.44 0.15 0.85 0.23
France 1995-2005 0.29 0.31 0.92 0.30
2005-2008 0.41 0.24 0.94 0.28
2007 1.61 0.24 0.95 0.27
2008 -0.03 0.24 0.99 0.27
Germany 1995-2005 -0.23 0.02 0.44 0.31
2005-2008 0.71 0.04 0.59 0.38
2007 1.11 0.04 0.64 0.38
2008 0.82 0.04 0.70 0.39
united 1995-2005 0.49 0.36 1.12 0.41
Kingdom 2005-2008 0.19 0.14 1.03 0.33
2007 0.47 0.14 1.07 0.34
2008 -0.31 0.13 1.08 0.32
Italy 1995-2005 0.57 0.22 0.71 0.20
2005-2008 0.55 0.06 0.52 0.21
2007 0.87 0.06 0.55 0.21
2008 -0.34 0.05 0.47 0.21
Spain 1995-2005 1.98 0.53 1.49 0.19
2005-2008 0.85 0.36 1.84 0.22
2007 1.11 0.36 1.90 0.22
2008 -0.36 0.35 1.80 0.21
Canada 1995-2005 1.12 -0.19 1.34 0.34
2005-2008 1.06 0.16 1.51 0.35
2007 1.33 0.16 1.51 0.35
2008 0.67 0.16 1.32 0.30
china 1995-2005 0.52 0.32 5.63 0.13
2005-2008 0.30 0.19 7.37 0.18
2007 0.32 0.19 7.41 0.18
2008 0.25 0.19 7.23 0.18
Brazil 1995-2005 0.77 0.42 1.29 1.13
2005-2008 0.82 0.17 1.99 0.23
2007 0.82 0.17 1.96 0.23
2008 0.69 0.17 2.45 0.22
Russia 1995-2005 0.01 1.34 -2.78 1.26
2005-2008 0.34 0.26 -0.55 0.05
2007 0.37 0.26 -0.53 0.05
2008 0.38 0.26 -0.32 0.05
India 1995-2005 0.62 0.19 4.18 0.19
2005-2008 0.71 0.14 6.15 0.37
2007 0.71 0.14 6.29 0.36
2008 0.70 0.14 6.21 0.38
Country Period Total Factor Total
Productivity GDP
World 1995-2005 0.62 3.6
2005-2008 0.94 4.4
2007 1.53 5.1
2008 -0.18 3.0
Advanced 1995-2005 0.43 2.7
Economies 2005-2008 -0.07 2.0
2007 0.34 2.7
2008 -0.92 0.5
Emerging 1995-2005 1.04 4.9
Economies 2005-2008 2.19 7.3
2007 2.95 8.1
2008 0.85 6.1
United 1995-2005 0.63 3.3
States 2005-2008 -0.09 1.7
2007 0.18 2.1
2008 -0.24 0.4
Japan 1995-2005 -0.14 1.1
2005-2008 -0.15 1.2
2007 0.84 2.4
2008 -1.50 -0.7
France 1995-2005 0.39 2.2
2005-2008 -0.29 1.6
2007 -0.86 2.2
2008 -1.14 0.3
Germany 1995-2005 0.75 1.3
2005-2008 0.46 2.2
2007 0.27 2.4
2008 -0.98 1.0
united 1995-2005 0.52 2.9
Kingdom 2005-2008 0.42 2.1
2007 0.96 3.0
2008 -0.67 0.5
Italy 1995-2005 -0.32 1.4
2005-2008 -0.54 0.8
2007 -0.24 1.4
2008 -1.45 -1.1
Spain 1995-2005 -0.57 3.6
2005-2008 -0.50 2.8
2007 -0.09 3.5
2008 -1.15 0.9
Canada 1995-2005 0.67 3.3
2005-2008 -1.18 1.9
2007 -0.87 2.5
2008 -2.03 0.4
china 1995-2005 1.20 7.8
2005-2008 2.74 10.8
2007 4.12 12.2
2008 1.32 9.2
Brazil 1995-2005 -1.23 2.4
2005-2008 1.84 5.1
2007 2.34 5.5
2008 2.21 5.7
Russia 1995-2005 4.33 4.2
2005-2008 6.77 6.9
2007 7.60 7.8
2008 5.10 5.4
India 1995-2005 1.09 6.3
2005-2008 0.76 8.1
2007 1.12 8.6
2008 -0.94 6.5
