International Comparisons of
Industry Output, Inputs and
Productivity Levels:
Methodology and New Results
Presentation prepared for Discussion session
on Productivity Levels and PPPs, Bern, 18 October 2006
Marcel Timmer,
Groningen Growth of Development Centre,
University of Groningen
This project is funded by the European Commission, Research Directorate
General as part of the 6th Framework Programme, Priority 8, "Policy
Support and Anticipating Scientific and Technological Needs".
Motivation
•
•
Recent changes in global structures of production and
trade
To what extent do countries differ in their production
structure in terms of:
•
• ICT capital intensity
• Skilled labour usage
• Materials and Energy usage
• Services inputs
• Productivity levels?
This paper provides new evidence on relative levels
of output, inputs and productivity at a detailed
industry level
For 1997 for Australia, Canada, France, Germany, the
Netherlands, the UK and the US.
•
Methodology
•
•
•
•
Basically: deflation of Input-Output tables using relative prices
of inputs and output across countries
Multilateral index number approach advocated by Caves,
Christensen and Diewert (CCD, 1982), following Conrad and
Jorgenson (1985),
PPPs following Nishimizu and Jorgenson (1978); Jorgenson,
Kuroda and Nishimizu (1987)
Two new characteristics in this paper
• Sectoral measures of output, input and productivity
(Gollop, 1979) to eliminate effect of differences in industry
structure
• Relative prices based on producer, rather than expenditure
prices
Example of Aggregation with and without integration
Labour input:
5 hours
5 hours
10 hours
Production of car parts
Intermediate input:
10 car parts
Production of cars
Car assembly
Sectoral output:
Gross output:
5 cars
5 cars
5 cars plus
10 car parts
5 cars
Table 1 Comparison of input shares in output,
with and without integration,
transport equipment manufacturing, 1997
Netherlands
as % of sectoral output
Sectoral Energy input
Sectoral Material input
Sectoral Services input
ICT capital input
Non-ICT capital input
Labour input
Sectoral output
as % of gross output
Energy input
Material input
Services input
ICT capital input
Non-ICT capital input
Labour input
Gross output
Gross output as % of sectoral output
U.S.
0.8%
52.6%
21.1%
0.5%
7.2%
17.9%
100.0%
0.8%
40.7%
24.8%
1.8%
6.7%
25.1%
100.0%
0.7%
60.1%
16.1%
0.4%
6.5%
16.2%
100.0%
110.5%
0.6%
55.2%
18.8%
1.4%
5.1%
19.0%
100.0%
132.4%
Variables for level accounting
•
•
•
•
•
•
Supply-Use tables for all seven countries in 1997
•
•
SUTs turned into symmetric Input-output table (industry by industry)
at basic prices plus net taxes, using fixed product-sales structure
assumption,
Split of SIOT into domestic table and import
26 industries covering total economy
45 intermediate products (aggregated to materials, energy
and service inputs)
Six capital assets (including ICT and non-ICT capital)
Two Labour categories into university & non-university
labour
PPPs for sectoral output, sectoral intermediates, capital
and labour for 1997
Output and input PPPs
•
•
•
•
Output PPP: A mix of production PPPs and adjusted
expenditure PPPs for industry (gross) output, basic
prices (Timmer, van Ark & Ypma, 2006)
Intermediate input PPPs: assume same basic price
for all deliveries from industry j (domestic table) and
exchange rates for import table
Labour PPPs: relative wages of university & nonuniversity labour
Capital PPPs: investment PPPs for 6 assets
combined with annualization factor (relative user
cost of capital)
PPPt K  PPPt I
u tK,C
u
K
t ,US
Table 2 Relative price levels of sectoral output and
inputs, transport equipment manufacturing (PPPs /
exchange rate, US$=1), 1997
France
Sectoral output
Energy input
Material input
Services input
ICT capital input
Non-ICT capital input
Labour input
1.52
1.67
1.09
1.33
1.35
1.40
1.56
Germany
1.49
1.86
1.09
1.17
1.06
1.05
2.03
Netherlands
1.64
2.05
1.07
1.04
1.16
1.24
1.01
U.K.
1.64
2.64
1.11
1.15
1.39
1.56
1.25
U.S.
1.00
1.00
1.00
1.00
1.00
1.00
1.00
Australia
1.46
1.40
1.15
1.10
1.11
0.87
0.70
Canada
0.86
0.56
0.92
0.92
0.89
0.84
0.92
Methodology for level accounting
Using the CCD approach translog multilateral indices of output
relative to the U.S. can be derived as follows:
ln Yc  ln YUS   vˆi ,c ln Yi ,c  ln Yi    vˆi ,US ln Yi ,US  ln Yi 
i
with vˆi ,c
1
1

 v i , c   v i , c  , v i , c 
2
n C

i
piY,c Yi ,c
p
I
Y
i ,c i ,c
Y
1
and ln Yi   ln Yi ,c
n C
(9)
Methodology for level accounting
•
E.g. the relative energy intensity (E-INT) of industry j is given by:
ln E  INT  ln
Ec
E
 ln US
Yc
YUS
Decompose relative sectoral output (Y) between countries c and
US into relative input (X) and productivity (A):
•
ln
Ac
Y
X
 ln c  ln c
AUS
YUS
X US
Figure 1, Relative TFP levels by industry group in the
Netherlands, France, Germany, UK, Canada and
Australia in 1997, U.S.=1
Unweighted industry average TFP levels, 1997, US=1.0
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
AUS
CAN
FRA
GER
NLD
market services
UK
AUS
CAN
FRA
GER
goods production
NLD
UK
U.S. most ICT intensive; non-ICT capital intensity
in Europe still larger
2.5
Unweighted industry average capital input per unit of output, 1997,
US=1.0
2
1.5
1
0.5
ICT capital
Non-ICT capital
market services
ICT capital
UK
NLD
GER
FRA
CAN
AUS
UK
NLD
GER
FRA
CAN
AUS
UK
NLD
GER
FRA
CAN
AUS
UK
NLD
GER
FRA
CAN
AUS
0
Non-ICT capital
goods production
Source: pre-EU KLEMS data, GGDC
European countries least energy and servicesintensive but more materials-intensive
Unweighted industry average intermediate input use per unit of output,
1997, US=1.0
2.5
2
1.5
1
0.5
Energy input
Material input
Market services
Services input
AUS
CAN
FRA
GER
NLD
UK
AUS
CAN
FRA
GER
NLD
UK
AUS
CAN
FRA
GER
NLD
UK
AUS
CAN
FRA
GER
NLD
UK
AUS
CAN
FRA
GER
NLD
UK
AUS
CAN
FRA
GER
NLD
UK
0
Energy input
Material input
Services input
goods production
Source: pre-EU KLEMS data, GGDC
Main findings
•
•
•
European TFP levels on par with U.S., but lower in
Canada and Australia
U.S. production is more intensive per unit of output in
use of
• ICT capital,
• Skilled labour,
• Energy,
• Purchased services,
But less intensive in
• Materials
• non-ICT capital
Next steps
•
•
•
•
•
Full KLEMS data set for EU (see
www.euklems.com)
Are differences in productivity growth and levels
in services real or figments of measurement error?
Adequate treatment of net taxes in IO-framework
Aggregation to total economy
Evaluate index number alternatives for output and
productivity level comparisons