ICT, Corporate Restructuring and Productivity
Laura AbramovskyRachel Griffith
IFS and UCL
ZEW – November 2007Workshop on Innovative Capabilities
and the Role of Consultants in the Information Economy
US productivity growth has far outpaced the EU
Technology diffusion
• A lot of attention to knowledge production– but technology and knowledge diffusion probably more important
for productivity
• Information and Communication Technology (ICT) has been one of largest technological advances
Investment in ICT in the US has continued to accelerate
Investment in ICT/GDP
0%
1%
2%
3%
4%
5%
6%
7%
19
80
19
85
19
90
19
95
20
00
US
UK
EU-15
Austria
Gross fixed capital formation in software divided by GDP, both in 2000 pricesSource: Timmer, Ypma and van Ark (2003), updated June 2005
US productivity growth driven by ICT using sectors
US EU-15
ICT-producing sectors
+ 1.9% + 1.6%
ICT-using sectors
+ 3.5% - 0.1%
other sectors - 0.5% - 1.1%
Change in annual growth in output per hour from 1990–95 to 1995–2001
ICT and productivity - focus of literature
• Allows firms to restructuring internally– more optimal structure of the workplace – change decision authority– change workers’ tasks
• Improves organisational efficiency and productivity
• Adjustment process, requires complementary investments – in skills – in organisational capital
ICT adoption also facilitates corporate external restructuring
• Transactions previously conducted face-to-face can now be conducted at arms-length
• Feasible to outsource and off-shore activities that it was previously not possible to– optimal boundaries of the firm change– transactions can more cheaply be carried out over large distances
How can outsourcing affect productivity?
• Specialised providers of intermediate goods and services can exploit economies of scale/scope
• Improving suppliers incentives to innovate
• Purchasers of services can become more productive in core activities by outsourcing non-core activities if previously overloaded
Rapid growth in business service providers
• Consistent with one of the biggest recent changes to the US and UK economies - business services have been very fast growing– in 1980 business services very small sector– employment in UK business services grew by 92% or 1.9m jobs
from 1984 to 2001– accounted for over half of the total growth– from 1995-2001 accounted for 1/3 of growth– in 2002, 4m people employed in this sector– accounting for around 1 in 7 jobs in the whole economy
Business services in the UK
0%
5%
10%
15%
20%
25%
Rentin
g of
mac
hinery
Comput
er se
rvice
s
Lega
l acti
vities
Accou
nting
Mark
et re
sear
ch
Mana
gem
ent c
onsu
ltanc
y
Archit
ectura
l & e
ngine
ering
Technic
al te
sting
Adver
tising
Human
reso
urce
s ser
vices
Inve
stiga
tion &
sec
urity
Indus
trial
clean
ing
Other
serv
ices
% o
f to
tal e
mpl
oym
ent
in B
usin
ess
Ser
vice
s
Source: ABI 2001
Is there evidence for this?
• Does ICT facilitate corporate external restructuring in the form of outsourcing of intermediate services
• Is this correlated with productivity growth?
• Econometric study– use large nationally representative data for the UK
• firms that use more ICT are more productive• firms that outsource more services and use ICT are
even more productivity
Empirical approach
• Estimate production function
• allow the way that ICT affects productivity to depend on amount of outsourcing of services
• measures the extent of complementarities between purchased services and ICT
Empirical approach
• Establishments face a production function of the form
i: establishments j: industry
Y: outputL: employment K: total capitalC: information and communication technologiesS: purchased servicesG: purchased goodsA: total factor productivity
iiiiijii GSCKLFAY ,,,,
Empirical approach
• Estimate in deviations from the industry mean, all variables take the form:
• allows us to control for unobserved industry specific factors (including price deflators)
• without imposing too many restrictions on the production technology
jii YYy lnln~
Empirical approach
itjripartgimneiusiage
iics
igisicikili
partgmneusage
cs
gsckly
)~*~(
~~~~~~
Correlations, not causal relationship
• We estimate by ordinary least squares (OLS)– problem if firms anticipate future shocks
• Decline in price of ICT exogenous to the firm– leads to increased investment in ICT
• Increased levels of ICT mean that it is now feasible to outsource many more business processes– leads to further ICT investment
• Heterogeneity across firms arises due to adjustment costs; we use this to identify complementarity between ICT and outsourcing in production from cross-section data
Data
• UK Annual Census of Production (ARD-ABI)
– Collected by the Office of National Statistics (ONS)– Establishment level (firm line of business)– Random stratified sample– Legal obligation to reply– Cross-section 2000-2003 (few repeated observations)– Manufacturing and services industries– 70,044 observations; median size 48 employees
Data
• Outsourcing of services (s)– intermediate expenditure on services
(legal and accountancy, marketing, renting of machinery, telecommunication and transport, etc.)
– on average 15% as a share of gross output
• ICT (c) – software capital stock (purchased and developed by own staff) – only partial measure of ICT investment– on average 1.25% as a share of gross output
How does ICT affect productivity?
Notes: from Table 2 in paper; each column is a separate regression with 70,044 observations;** significant at 1%
Dep variable: ln (Gross output)i
ln (L)i 0.344 0.342 0.341
Labour (0.005)** (0.005)** (0.005)**
ln (K)i 0.214 0.207 0.210
Capital (0.005)** (0.005)** (0.005)**
ln (S)i 0.165 0.165 0.182
Purchased Services (0.005)** (0.005)** (0.005)**
ln (G)i 0.248 0.249 0.250
Purchased Goods (0.005)** (0.005)** (0.005)**
ln (ICT)i 0.010 0.020
ICT (0.001)** (0.002)**
ln (ICT)i *ln (Services)i 0.005
(0.000)**
Interpretation of results
• Positive coefficient on interaction of ICT with outsourced services– Firm that do both more ICT and more outsourcing (compared to
their industry) are more productive
• Coefficient is elasticity of output with respect to ICT
– an establishment that has 1% more ICT than the average in its 4-digit industry (and outsources services at its 4-digit industry average) is 0.02% more productive and
• if outsources services by 1% more than average has 0.025% higher productivity
• if outsources services by 1% less than average has 0.015% higher productivity
Interpretation of results
level of outsourcing of services
elasticity of output with respect to ICT
25th percentile 0.0006
50th percentile 0.010
75th percentile 0.018
mean 0.020
Alternative explanations? Multinational establishments
Notes: from Table 2 in paper; each column is a separate regression with 70,044 observations; ** significant at 1%
Dep variable: ln (Gross output)i
ln (S)i 0.165 0.185
Purchased Services (0.005)** (0.005)**
ln (ICT)i 0.010 0.020
ICT (0.001)** (0.002)**
ln (ICT)i *ln (Services)i 0.005
(0.000)**
USi 0.187 0.140
(0.016)** (0.017)**
USi*ln(ICT) i 0.008 -0.013
(0.006) (0.006)*
MNEi 0.125 0.080
(0.009)** (0.009)**
MNEi*ln(ICT)I 0.001 -0.021
(0.003) (0.004)**
Alternative explanations? Skills
Dependant variable: ln (Gross output) i
ln (S)i 0.171 0.189
Purchased Services (0.006)** (0.006)**
ln (ICT)i 0.026 0.035
ICT (0.004)** (0.004)**
ln (ICT)i *ln (Services)i 0.006
(0.000)**
ln(SK)j 0.101 0.094
Skills (0.010)** (0.010)**
ln(ICT)i* Ln(SK)j 0.009 0.007
(0.002)** (0.002)**
Notes: from Table 4 in paper; each column is a separate regression with 61,135 observations;** significant at 1%
Alternative explanations? Allow ICT to interact with all other inputs
Dependant variable: ln (Gross output) i
ln (S)i 0.212
Purchased Services (0.007)**
ln (ICT)i 0.021
ICT (0.002)**
ln (ICT)i *ln (Services)i 0.016
(0.002)**
ln (ICT)i *ln (Labour)i -0.020
(0.002)**
ln (ICT)i *ln (Capital)i 0.008
(0.002)**
ln (ICT)i *ln (Goods)i -0.003
(0.002)Notes: from Table 2 in paper; each column is a separate regression with 70,044 observations;
** significant at 1%
Consistent story across manufacturing and servicesDependant variable: ln (Gross output)i
Manufacturing Services
ln (L)i0.321 0.341
Labour (0.008)** (0.006)**
ln (K)i0.185 0.217
Capital (0.008)** (0.007)**
ln (S)i0.164 0.178
Purchased Services (0.009)** (0.007)**
ln (G)i0.317 0.238
Purchased Goods (0.010)** (0.005)**
ln (ICT)i0.017 0.022
ICT (0.002)** (0.002)**
ln (ICT)i *ln (Services)i0.005 0.006
(0.001)** (0.001)**
Observations 20,785 44,539Notes: from Table 3 in paper; ** significant at 1%
Consistent story across 2-digit industries
Dependant variable: ln (Gross output) i
Mean2-digit industry
ln (L)i 0.324
Labour
ln (K)i 0.198
Capital
ln (S)i 0.176
Purchased Services
ln (G)i 0.272
Purchased Goods
ln (ICT)i 0.020
ICT
ln (ICT)i *ln (Services)i 0.004Notes: from Table 3 in paper; ** significant at 1%
Composition effect?
• Split industries by mean labour productivity in business services– coefficient on ICT*services is:
• high labour productivity: 0.005 (0.001)**
• low labour productivity: 0.006 (0.000)**
• Also split by average wage,– coefficient on ICT*services is:
• high average wage: 0.007 (0.001)**
• low average wage: 0.005 (0.000)**
Notes: from Table 3 in paper; ** significant at 1%
Conclusion
• Has ICT played an important role in productivity growth by facilitating corporate external restructuring, via outsourcing intermediate services?
• We find that ICT increases productivity (the elasticity of output with respect to ICT is higher) for firms that make greater use of outsourced services than other firms in their industry
• This is consistent with complementarities between ICT and outsourced services