Transcript on US

University of Groningen
What do We Know about Services
Productivity in Europe?
Bart van Ark
University of Groningen
and The Conference Board
CPB Workshop on "Productivity in services:
Determinants, international comparison, bottlenecks, policy"
10 June 2004, The Hague
1
Set Up of Presentation
 Evidence on services productivity as source of
productivity slowdown in EU-15:
 56-industry level
 taxonomies (ICT using, skills, innovation type)
 The suspects explaining the European productivity
slowdown:
 measurement
 ICT and innovation in services
 the role of markets
 Roads forward to support services productivity growth:
 innovation policies and improved framework
conditions
 business strategies focused on intangible capital
formation
2
GGDC Studies





Bart van Ark, Robert Inklaar and Robert H. McGuckin (2003), "Changing
Gear: Productivity, ICT and Service Industries in Europe and the United
States", in J.F. Christensen and P. Maskell, eds., The Industrial Dynamics
of the New Digital Economy, Edward Elgar, pp. 56-99 (with TCB, updated)
Mary O’Mahony and Bart van Ark, eds. (2003), EU Productivity and
Competitiveness: An Industry Perspective. Can Europe Resume the
Catching-up Process?, DG Enterprise, European Union, Luxembourg
(downloadable from http://www.ggdc.net/) (with NIESR; updated)
Bart van Ark, Lourens Broersma and Pim den Hertog (2003), "Services
Innovation, Performance and Policy: A Review", Research Series No. 6,
Directorate-General for Innovation, Ministry of Economic Affairs, The
Hague (with Dialogic).
Robert Inklaar, Mary O'Mahony and Marcel Timmer (2004), “ICT and
Europe's Productivity Performance; Industry-level Growth Account
Comparisons with the United States,” Research Memorandum GD-68,
Groningen Growth and Development Centre
Robert McGuckin, Matthew Spiegelman and Bart van Ark (2004), “Retail
Productivity in Europe and U.S.”, The Conference Board (forthcoming)
3
Two GGDC data bases
 Industry Labour Productivity Database: series on
nominal and real value added, employment and hours,
56 industries for 15 EU countries and U.S., 1979-2002
(updated)
 applies U.S. hedonic deflators for ICT to ICTproducing industries
 industry aggregation on the basis of Tornqvist
weighting
 Industry Growth Accounting Database: above + series
for six asset types (of which three ICT), three skill
levels, 26 industries for 4 EU countries (France,
Germany, UK and Netherlands) and U.S., 1979-2001
4
Can the European Union Resume the
Catching Up-Process?
Labour Productivity (GDP per hour worked) in 1999 US$
40
38
-8%
-5%
36
34
-4%
32
-10%
30
28
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
European Union
United States
5
Much variation across across industries
Labour Productivity Growth for 12 Main Sectors, EU and U.S., 1979-2002
Total Economy*
Agriculture, Forestry and Fishing
Mining and quarrying
Manufacturing
Electricity, gas and water supply
Construction
Distributive trades
Transport
Communications
Financial Services
Real estate
Business Services
Other community, Social and
Personal Services
Public Administration, Education
and Health
1979-90
2.16
EU-15
1990-95
2.41
1979-90
1.20
US
1990-95
1.14
1995-02
1.64
1995-02
2.46
5.3
2.8
3.4
2.2
1.7
1.3
2.8
5.2
1.9
-0.7
0.5
5.5
12.8
4.0
4.6
0.6
1.7
3.5
6.1
1.1
0.1
0.7
3.6
2.1
2.9
6.0
0.8
1.0
1.8
8.7
2.3
-0.6
0.1
5.2
4.2
3.2
1.2
-0.8
1.6
1.4
2.5
1.2
0.6
-1.0
-0.2
4.8
3.7
1.7
0.3
1.6
1.1
3.6
1.9
1.7
-0.1
4.2
1.7
4.4
1.6
0.2
5.4
2.2
6.7
5.6
1.2
0.5
-1.7
0.2
0.0
0.5
1.4
0.3
0.3
1.2
0.8
-0.4
-0.8
-0.4
6
Most important results on labour
productivity at level of 56 industries
 U.S. productivity growth advantage over Europe is not
ubiquitous:


In just over 50% of industries, U.S. labour productivity growth
is faster than in EU (market services and high tech
manufacturing) from 1995-2002
Only a limited number of service industries account for U.S.
advantage in productivity growth
 At the same time there is a lot more dynamism in U.S.:



Industries with above 2% productivity growth are much more
present in U.S. than in EU
In two-thirds of industries, U.S. labour productivity growth
accelerates in 1995-2002 over 1990-1995
Whereas in almost three quarters of industries, EU labour
productivity growth slows down
7
Contribution of largest contributors in U.S. is
substantial strongly dominated by services
Contribution to aggregate labour productivity of 5 largest and 5 least
contributing industries in U.S., U.S. and EU, 1995-2002
US 1995-2002
%-point
%-contribution
contribution
EU-15 1995-2002
%-point
%-contribution
contribution
5 Largest contributors in US
Retail trade, except of motor vehicles
Wholesale trade and commission trade
Electronic valves and tubes
Financial intermediation,
Activities auxiliary to financial intermediation
0.36
0.35
0.31
0.23
0.22
15%
14%
13%
9%
9%
0.06
0.08
0.11
0.09
0.02
4%
5%
6%
6%
1%
5 Least contributors in US
Food, drink & tobacco
Mechanical engineering
Printing & publishing
Textiles
Scientific instruments
-0.09
-0.05
-0.03
-0.03
-0.02
-4%
-2%
-1%
-1%
-1%
-0.01
0.00
0.00
-0.01
0.00
0%
0%
0%
-1%
0%
Aggregate Labour productivity growth
2.46
100%
1.64
100%
8
Contribution of largest contributors in Europe is
smaller and mainly in high tech manufacturing
Contribution to aggregate labour productivity of 5 largest and 5 least
contributing industries in EU., U.S. and EU, 1995-2002
US 1995-2002
EU-15 1995-2002
%-point
%-contribution %-point
%-contribution
contribution
contribution
5 Largest contributors in EU
Communications
Computer and related activities
Legal, technical and advertising
Electronic valves and tubes
Health and social work
0.18
0.09
0.07
0.31
0.08
7%
4%
3%
13%
3%
0.22
0.14
0.13
0.11
0.10
13%
9%
8%
6%
6%
5 Least contributors in EU
Mining and quarrying
Insurance and pension funding
Textiles
Clothing
Food, drink & tobacco
-0.01
0.02
-0.03
-0.01
-0.09
0%
1%
-1%
0%
-4%
-0.02
-0.01
-0.01
-0.01
-0.01
-1%
-1%
-1%
-1%
0%
Aggregate Labour productivity growth
2.46
100%
1.64
100%
9
Much variation by industry across countries: Retail
LP growth in 1995-2002 and 1990-1995, Retail trade
8.00
6.00
1995-2002
1990-1995
2.00
-2.00
-4.00
nm
ar
k
in
De
Sp
a
Be
lg
iu
m
y
an
G
er
m
Ita
ly
l
Po
rtu
ga
la
nd
s
15
th
er
Ne
EU
-
an
ce
Fr
an
d
Fi
nl
m
bo
ur
g
ec
e
Lu
xe
G
re
Au
st
ria
la
nd
Ir e
de
n
Sw
e
UK
0.00
US
(in %)
4.00
Much variation by industry across countries: Banks
LP growth in 1995-2002 and 1990-1995, Financial Intermediation except insurance
12.00
10.00
8.00
(in %)
6.00
1995-2002
1990-1995
4.00
2.00
0.00
l
y
ga
an
tr u
m
er
Po
G
-2.00
-4.00
nd
a
nl
Fi
U
K
U
S
D
m
en
k
ar
e
ec
e
r
G
5
-1
EU
ria
st
u
A
ly
Ita
s
nd
lr a
he
et
in
a
Sp
N
F
e
nc
a
r
nd
la
e
Ir
Sw
en
ed
x
Lu
g
ur
bo
em
m
iu
lg
e
B
Much variation by industry across countries:
Telecommunication Services
LP growth in 1995-2002 and 1990-1995, Communications
16.00
14.00
12.00
(in %)
10.00
1995-2002
1990-1995
8.00
6.00
4.00
2.00
0.00
y
m
er
G
an
a
nl
Fi
nd
ly
Ita
U
s
en
nd
ed
lr a
Sw
he
et
5
K
-1
EU
N
F
e
nc
ra
g
ur
bo
m
xe
Lu
U
S
e
ec
re
G
ria
st
u
A
k
ar
D
m
en
l
in
a
Sp
Po
ga
rtu
um
gi
l
Be
nd
la
e
Ir
Variation across industries seems to be related to
aggregate productivity growth rate and somewhat
dependent on size of country: Total Economy
LP growth rates and standard deviation (all industries except ICT producing industries) by country,
8
PRT
7
6
Standard deviation by country
DNK
IRE
5
GRC
UK
4
LUX
3
BEL
NLD
SWE
GER
AUT
US
FIN
FRA
ESP
2
ITA
EU-15
1
-1.000
0
0.000
1.000
2.000
3.000
(% growth rate)
4.000
5.000
6.000
… although less so when looking at
Market Services only
LP growth rates and standard deviation of Market Services by Country
12.00
10.00
PR
T
Standard deviation
8.00
6.00
GRC
UK
GER
IRE
SWE
4.00
IT
LU
X
BEL
2.00
-1.00
-0.50
EU-15
0.50
1.00
FIN
NL
D
FRA
ESP
0.00
0.00
DNK
US
AUT
1.50
(% growth rate)
2.00
2.50
3.00
3.50
4.00
Suspect 1: Measurement problems hamper
adequate assessment of service productivity
 Few good studies on how big is the measurement
problem concerning services productivity:


on U.S.: Triplett, J.E. and B. Bosworth (2002) ““Baumol's
disease” has been cured: IT and multifactor productivity in U.S.
services industries,” Brookings Institution, Washington D.C.
International: Anita Wölfl (2003), “Productivity growth in
service industries – an assessment of recent patterns and the
role of measurement,” STI-Working Paper 2003-07, (Paris:
OECD)
 Some attempts to improve measurement of services
output:


Brookings workshops on “Measuring service sector output”
and methodological improvements by BEA and BLS, in
particular in area of financial services
Eurostat, Handbook on price and volume measures in national
15
accounts, Luxembourg, 2001
Source: Anita Woelfl (2003)
We cannot be sure of the bias in service
output measurement
 Over time:



Increased size of services has impact on aggregate (Griliches,
1992; 1994)
Increased complexity of services --> multidimensionality and
quality improvement
But methodological changes in e.g. financial services do not
show bias in only one direction
 Across countries:



Countries apply different methodologies (e.g., retail)
Part of service output measures is still based on input measures,
in particular in non-market services, but there are differences in
degree between countries
Measurement of PPPs in services is complicated, depending on
share of intermediate inputs in gross output
17
Measurement problems due to increased share of ICT
Industry
Services
Output
Primarily computers and
other ICT goods. Solvable
by using hedonic price
indices, which is possible
provided data availability
Primarily "customised" services and
public services (education, health,
etc.). Should be tackled by detailed
analysis of multiple dimensions of
output by industry. Difficult both in
methodological terms as well in
terms of data
Input
Primarily semiconductors.
Can be solved with
hedonic price indices,
provided data availability
and investment flow
matrices.
Primarily ICT capital input. Can be
solved by adjusting nominal input
series with hedonic price indices.
Feasible provided availability of
investment flow matrices.
B. van Ark, Measuring the New Economy, Review of Income and Wealth, March 2002
LP advantages in U.S. services are translated
in TFP advantages, as U.S. investment in
ICT is only slightly higher
Growth Accounting at Industry Level, Selected Service Industries, Netherlands and
United States, 1995-2001
Share of
Labour
Contribution of:
Employment Productivity
Labour
ICT
Non-ICT
in total
Growth
Quality
Capital
Capital
Employment
Growth Deepening Deepening
2001
1995-2001
NETHERLANDS
Market Services:
of which:
Wholesale trade
Retail trade
Financial intermediation
Business services
UNITED STATES
Market Services:
of which:
Wholesale trade
Retail trade
Financial intermediation
Business services
TFP
Growth
51.7
1.7
0.2
1.1
0.3
0.1
9.3
7.2
4.2
15.8
3.6
1.2
-0.6
1.0
0.2
0.0
0.4
0.3
0.9
0.5
2.8
1.0
0.1
0.3
0.3
0.6
2.4
0.4
-4.0
-0.9
49.9
3.7
0.3
1.6
0.5
1.2
6.1
10.3
4.5
12.3
6.5
6.3
5.4
0.0
0.1
0.1
0.1
0.2
1.8
0.7
2.9
1.7
0.5
0.3
0.9
0.2
4.1
5.2
1.5
19 -2.1
Source: Inklaar, O’Mahony and Timmer, 2003
Suspect 2: Service Industries do
Not Sufficiently Innovate
 ICT investment is an important enabler of innovation
and productivity growth
 … and U.S. has been more successful in obtaining
productivity effects from ICT investment than EU
 But productive use of ICT investment is strongly
dependent on various dimensions of non-technological
innovations
 Productivity effects are strongest in services with
supplier dominated innovations or strong organizational
innovations
20
ICT taxonomy points to main differentials
in ICT-using services
Labour productivity by Industry Group on the basis of ICT taxonomy,
EU and U.S., 1995-2002
Total Economy
ICT Producing Industries
ICT Producing Manufacturing
ICT Producing Services
1990-1995
EU-15
U.S.
2.4
1.1
1995-2002
EU-15
U.S.
1.6
2.5
7.0
13.2
4.6
8.3
17.4
3.1
8.5
16.0
5.9
9.3
22.6
3.1
ICT Using Industries
ICT Using Manufacturing
ICT Using Services
2.2
2.9
1.8
1.3
-0.2
1.7
1.7
2.1
1.6
4.8
1.7
5.3
Non-ICT Industries
Non-ICT Manufacturing
Non-ICT Services
Non-ICT Other
2.2
3.6
1.2
3.4
0.3
2.7
-0.3
0.8
1.0
2.0
0.3
2.2
0.5
1.4
0.4
0.6
Outside ICT-producing, EU manufacturing has
productivity advantage but this advantage is eroding
21
Complementarity of technological (ICT)
and non-technological innovations
 Case studies:



Company evidence (McKinsey)
SIID studies on service innovation combined with firm micro
firm level studies (van Ark et al., 2003)
Combined evidence from macro and sector studies (TCB Retail
study)
 Micro firm level studies


Special organizational and work practice surveys on U.S
(Brynjolffson and others, Black & Lynch)
Recent international work (OECD/Bartelsman, Hempell, van
Leeuwen/vd. Wiel)
 Macro approaches:



Analysis of TFP residuals with use of R&D, innovation
measures (OECD)
Cluster research with I/O and CIS tables (Broersma in van Ark
et al., 2003)
Intermediate input use of KIBS as proxy for organizational
22
innovation (Broersma and van Ark, 2004)
Innovators in services strongly combine
technological and non-technological innovations
Source: CBS, Kennis en Economie
characteristics of actual and
potential clients (market intelligence)
n
TECHNOLOGICAL
OPTIONS
(DIMENSION 4)
NEW CLIENT
INTERFACE
(DIMENSION 2)
r ib
uti
o
Marketing
Di
st
NEW SERVICE
CONCEPT
(DIMENSION 1)
on
ati t
nis men
ga
Or velop
de
Knowledge of the characteristics
of existing and competing
services (business intelligence)
A four dimensional typology of service innovation
used in SIID studies
NEW SERVICE
DELIVERY SYSTEM
(DIMENSION 3)
capabilities, skills & attitude of existing
and competing service workers
(Human Resource Management)
Source: den Hertog and Bilderbeek (1999)
© Dialogic
24
New measures of innovation according to
4-D innovation model look promising
Source: De Jong et al, EIM, 2004
25
Characteristics of service innovation
 Multidimensionality is the rule
 Dimensions are often renewed in other sectors
through new combinations
 The weights of dimensions change over time
 ICT facilitates in many cases, but is not sufficient nor
always necessary
 Next to industry characteristics, firm strategies matter
 Service innovations take place along the whole value
chain
 Co-operation (co-producing, co-innovating) takes
place a lot
 De- and re-regulation is important but impact is
diverse
26
The services sector is characterized by distinct innovation clusters
Towards a service innovation taxonomy (developed
from Pavitt, 1984)
US advantages are strongest in supplier
dominated services, organizational
innovative services and client led services
Labour productivity by Industry Group on the basis of combined Pavitt/
SIID taxonomy, EU and U.S., 1995-2002
1990-1995
1995-2002
EU-15
U.S.
EU-15
U.S.
Total Economy
2.4
1.1
1.6
2.5
Good producing industries
Supplier dominated manufacturing
Scale intensive industry
Specialised suppliers manufacturing
Science based manufactuirng
2.6
3.8
6.6
5.5
0.0
2.7
9.9
2.8
1.9
1.8
6.6
4.2
1.2
1.2
13.3
3.4
Service industries
Supplier dominated services
Specialised supplier services
Organizational innovative services
Client led services
Non-market services
2.9
0.5
2.4
1.2
1.2
2.3
0.0
1.1
1.3
-0.8
4.0
0.5
1.4
0.3
0.8
6.4
-0.3
2.6
4.2
-0.4
30
Non-technological innovations mainly
arise from investment in intangible inputs
31
Intangibles inputs are key in facilitating the
innovation process and creating more productivity
Intensive IT users have relatively high
intermediate purchases from knowledge
intensive business services (KIBS), which can be
used as proxy for organizational capital
Estimation results of model specification (5), 49 industries, The Netherlands
Dependent variable:
 Z KIBS , j , t 

 log
 Y

j ,t


1987-2001†
Intercept
 rI IT , j , t 


 rI

 tot , j , t 
Number of observations
1987-1994
1995-2001
0.016
(2.963)
0.020
(2.937)
0.012
(1.412)
0.094
(1.748)
685
0.014
(0.328)
342
0.072
(1.846)
343
Between parentheses are the t-values.
No data for services nec and personal services were included. Data No. 742 (maritime shipping 1993)
caused a severe outlier and was omitted here.
33
Productivity growth is enhanced by a combined
effect of ICT-use and KIBS purchases
Estimation results of labour productivity regressions (8) and (9) for different time periods,
Netherlands
1987-1994
(8)
(9)
Explanatory variables:*
Intercept
 K i , IT ,t 

log 

L
i
,
t


 K i ,non IT ,t 

log 

L
i ,t


 Z i ,t  Z KIBS ,i ,t
log 
Li ,t

 Z KIBS ,i ,t 

log 

L
i
,
t






1995-2001
(8)
(9)
1987-2001
(8)
(9)
2.379
(23.48)
2.376
(22.82)
2.540
(27.86)
2.540
(28.17)
2.466
(46.64)
2.474
(46.89)
0.051
(4.177)
0.052
(4.056)
0.061
(5.570)
0.050
(4.364)
0.032
(5.602)
0.030
(5.079)
0.157
(5.959)
0.157
(5.939)
0.197
(5.528)
0.207
(5.844)
0.148
(8.764)
0.157
(9.125)
0.435
(9.868)
0.437
(9.656)
0.315
(8.880)
0.311
(8.864)
0.391
(16.14)
0.380
(15.45)
0.073
(2.658)
0.072
(2.605)
0.152
(8.122)
0.109
(4.369)
0.186
(12.38)
0.170
(10.32)
-
-0.001
(-0.161)
-
0.020
(2.646)
-
0.011
(2.505)
0.995
368
0.994
368
0.996
321
0.995
368
0.991
689
0.992
689
Interaction term:
 K i , IT ,t
log 
 Li ,t

Z
  log  KIBS ,i ,t

 L
i ,t


Adjusted R2
Number of observations




IT-capital elasticity of output
5%
5%
6%
7%
3%
4%
The 48 industry fixed effect dummies are not reported for convenience.
The t-values are between parentheses.
Data on mining (No 31-60), No. 150 (oil processing 2001) and maritime shipping (No. 466-480) caused
severe outliers and were omitted
*
34
Suspect 3: Rigid Markets Hamper
Services Productivity Improvements
 Competition helps to increase entry and exit but the
effects on productivity greatly vary across industries
 Comprehensive reforms (in product and labour markets)
seem to be crucial
 Many (de-)regulations are very industry-specific
 Considerable time lags seem to be present before
productivity effects emerge
 Productivity may initially slow down after deregulations
(e.g. retail)
 A certain amount of experimentation with optimal level
of deregulation is necessary
35
Should policies focus on service innovation or
more broadly on framework conditions?
“Deepening”
1) Broaden R&D tax
credit schemes to
include
organisational
innovation.
2) Include service
firms explicitly in
policies aimed at
training and
mobility of
researchers and
personnel at large.
3) Include service
sectors in foresight
and roadmapping
activities.
4) Create and improve
science-industry
relationships in
services.
“Broadening”






Extend awareness
activities more explicitly
to non-technological
innovation and service
firms.
Support service
innovation management
in service industries.
Facilitate encounters
between manufacturing
and service innovators.
Assess comparative
advantages of services in
the Netherlands.
Invest in services
innovation research.
Develop policy
experiments on service
innovation.
“Horizontalization”







Spur innovation by sensible
deregulation and competition
Work on non-tariff trade
barriers in international trade
policies.
Use environmental
regulations to support
innovation.
Use general education and
science policies to create an
innovative, flexible and
service oriented labour force.
Invest in innovative
government services.
Consider the effects of policy
areas such as spatial and
transport planning on the
scope for innovation.
Improve basic statistics on
services.
For business, productivity is either not an
explicit target or at best part of its overall
value creation model
Cost
reductions
Rise in
technical change
Improvement in
operating efficiency
Productivity effect
Increase
economies of
scale
Fall
in
resource
prices
Value
Creation
Price
effect
Activity
effect
Improvement in
product
mix
Quality
Rise in
product prices
Innovation
Improvement in
resource mix
37
Conclusions & questions
 U.S. productivity advantages are not ubiquitous but is
strongly based in market services
 Measurement issues are important but biases should not
be automatically assumed
 There is a lot more dynamics in U.S. services (faster
growth and investment, more innovation, more changes
to markets) --> is more turbulence what Europe needs?
 Some of U.S. productivity advantages in services cannot
be easily adopted in Europe (e.g., scale effects)
 How can Europe develop productivity advantages in
services?
 Diversity and customization?
 Advance combination of manufacturing and service
functions?
 Set industry standards more easily?
38