World Development Vol. 33, No. 5, pp. 681–700, 2005
Ó 2005 Elsevier Ltd. All rights reserved
Printed in Great Britain
www.elsevier.com/locate/worlddev
0305-750X/$ - see front matter
doi:10.1016/j.worlddev.2005.01.004
The Impact of Investment in IT
on Economic Performance:
Implications for Developing Countries
ROUBEN INDJIKIAN
UNCTAD, Geneva, Switzerland
and
DONALD S. SIEGEL *
Rensselaer Polytechnic Institute, Troy, NY, USA
Summary. — We review quantitative and qualitative research on the impact of IT on economic per-
formance in developed and developing countries. In general, studies from the developed world have
yielded evidence of a strong positive correlation between IT and economic performance, as well as
IT-induced changes in workforce composition in favor of highly skilled or educated workers and
organizational changes that allow firms to implement IT more effectively. To maximize social re-
turns to IT investment, policymakers in developing countries must address two key deficiencies:
(1) a lack of knowledge of ‘‘best practices’’ in IT usage and (2) IT-related skill deficiencies in the
workforce.
Ó 2005 Elsevier Ltd. All rights reserved.
Key words — information technology, developing countries, skill-biased technological change,
productivity
1. INTRODUCTION
more comprehensive understanding of the im-
pact of IT and the Internet on productivity in
During the 1990s, there was a rapid diffusion
various parts of the world economy requires a
of information technology (henceforth, IT) and
synthesis and integration of various studies on
the use of the Internet throughout the globe,
the use of IT in e-business processes at various
especially in developed countries. The average
levels of aggregation (e.g., establishments or
household in the USAand other developed
plants, firms, industries, and countries) and in
countries now contains at least one personal
many countries. This type of assessment might
computer with access to the Internet. More
also be useful for policymakers in countries
importantly, the use of the Internet is pervasive
that are producers of IT hardware and
in the private sector in most developed coun-
tries. The use of IT and the Internet is also
beginning to increase in many developing and
* We thank Robin Mansell, Matti Pohjola, Oliver
transition economies.
Coomes, two anonymous referees, and participants at
Aprecipitous decline in stock prices starting
the High-Level Regional Conference for Transitional
in 2000, along with a concomitant slowdown
Economies on ICT and E-business Strategies, organized
in economic growth, has dampened the enthusi-
by UNCTAD and UNECE, in Geneva, Switzerland, in
asm for IT. Although many policymakers and
October 2003. The views expressed in this working paper
producers and users of IT continue to be enthu-
are those of the authors and do not necessarily reflect
siastic about its long term economic impact, a
those of UNCTAD. Final revision accepted: January 10,
sober analysis of this question is warranted. A
2005.
681

---

682
WORLD DEVELOPMENT
software. Hence the presentation of evidence
Table 1. Average annual percentage of GDP devoted to
on the antecedents and consequences of IT
expenditure on ICT (1993–2001)
investment and of diffusion of Internet technol-
Country
Percentage
ogies in both developed and developing coun-
New Zealand
10.3
tries is important.
Sweden
8.8
We believe that a comprehensive review of re-
Australia
8.7
cent research on this topic is of special interest
Switzerland
8.4
for developing countries for two reasons: First,
Singapore
8.3
these countries have not yet fully reaped the
UK
8.0
benefits of IT. Second, they have fewer re-
USA7.8
sources to devote to IT and thus have a smaller
Canada
7.7
‘‘margin for error’’ than developed countries.
Netherlands
7.5
An objective assessment and resolution of the
Denmark
7.3
debate on this subject has important policy
Hong Kong
7.2
implications, since an adjudication of this dis-
South Africa
7.1
pute could allow these countries to formulate
Japan
7.1
an optimal set of IT and e-business strategies
Colombia
7.0
that are adapted to local institutions.
France
6.9
Our article begins with a discussion of IT dif-
Czech Republic
6.8
fusion in various regions of the world. This is
Israel
6.6
followed by an extensive review of the literature
Belgium
6.5
on the relationship between IT, the Internet,
Finland
6.4
and productivity growth at the firm, industry,
Hungary
6.2
and national levels. Next, we examine interna-
Germany
6.2
tional evidence on the widespread phenomenon
Norway
6.1
of skill-biased technological change (hence-
Ireland
5.8
forth, SBTC) and the organizational dynamics
Malaysia
5.8
of e-business diffusion in traditional sectors
Korea
5.8
of economy. These include IT-induced orga-
Austria
5.6
nizational changes, such as changes in human
Slovakia
5.5
resource management practices and organi-
Chile
5.5
zational structures.
The following section
Brazil
5.4
provides examples of IT and e-business applica-
Portugal
5.5
tions in developing countries. We conclude with
Vietnam
4.7
a summary of the key findings and policy rec-
Italy
4.6
ommendations. These recommendations may
Taiwan
4.6
be useful to policymakers who wish to use IT
Greece
4.4
and e-business as a means to bridge the techno-
Spain
4.2
logical gap between the developed and develop-
Venezuela
3.9
ing world, which might ultimately lead to
Slovenia
3.7
convergence in economic growth and develop-
Poland
3.7
ment across countries.
China
3.7
Argentina
3.7
Mexico
3.5
Turkey
3.3
2. COMPARING IT IN VARIOUS
Bulgaria
3.1
REGIONS
Philippines
3.1
Thailand
3.1
There have been many attempts to measure
Russia
2.9
the e-readiness of countries, in terms of IT
India
2.7
use and production and its impact on their
Indonesia
2.1
economy. Table 1 presents statistics of the
Egypt
2.2
intensity of investment in IT, computed as the
Saudi Arabia/
1.8
ratio of IT expenditure to GDP for 51 coun-
Gulf States
tries. The relatively low rate of investment
Romania
1.5
in IT in developing countries appears to
refute the hypothesis that diffusion patterns
Source: Pohjola (forthcoming).

---

IMPACT OF INVESTMENT IN IT
683
(and perhaps economic growth) are somehow
its larger IT-producing sector and faster growth
converging, and the digital divide is being re-
in IT-using service sectors such as wholesale
duced. In fact, it is conceivable that such statis-
and retail trade and financial brokerage. On
tics may actually understate the extent of the
the other hand, McKinsey (2004) asserts that
gulf between developed and developing econo-
a higher productivity in the USAis due to a
mies, since developed countries also have more
more favorable regulatory environment, stron-
favorable environmental conditions and more
ger competition, and superior corporate orga-
robust institutions to support technical ad-
nization in ‘‘traditional’’ sectors. 1
vance.
These findings are consistent with evidence
van Ark and Piatcovski (2004) analyzed IT
presented in van Ark et al. (2004), who stressed
investment patterns and their impact on eco-
the importance of accumulation of intangible
nomic performance in two sets of countries re-
capital, knowledge, and skilled labor in IT in
garded as being at different levels of economic
explaining growth in productivity in services.
development: the 15 countries of the European
As noted in Bresnahan, Brynjolfsson, and Hitt
Union (‘‘old’’ Europe) and 10 Central European
(2002), a superior corporate organization is
economies under accession (‘‘new’’ Europe).
likely to be determined by improvements in
They conclude that there is a trend toward con-
communication, networking, and coordination
vergence in investment in IT between ‘‘old’’ and
made possible by modern IT.
‘‘new’’ Europe. Investment in IT capital was
International organizations such as UNC-
also found to be an important source of pro-
TAD recognize the lack of hard, systematic data
ductivity growth in both sets of countries.
on the information economy in developing and
Given that the next phase of investment in
transition economies and are trying to address
IT for ‘‘new’’ Europe is likely to be concen-
this problem. Until such information becomes
trated in the service sector, the authors argue
available, UNCTAD is using data generated
for the establishment of institutions and the
by other sources, including sister organizations,
implementation of market-oriented reforms
such as ITU and the World Bank, to derive
that would enhance the effectiveness of IT in
aggregate IT development data and indices
services.
(UNCTAD, E-commerce and Development
Several
research
organizations
have
at-
Reports 2001–04). The UNCTAD Index of IT
tempted to quantify the extent of investment
diffusion consists of three dimensions: connec-
in IT at the national level, by deriving general-
tivity, access, and policy environment, with each
ized indexes of IT or ‘‘e-readiness.’’ It is impor-
factor assumed to have an equal weight. The
tant to note that such indexes are not precise
indicators of connectivity are per capita mea-
measures of this construct, since there are lim-
sures of the number of Internet hosts, PCs, tele-
ited data on this phenomenon and some rather
phone mainlines, and cellular subscribers.
heroic assumptions (e.g., perfect competition in
Proxies for access consist of per capita GDP,
input or output markets) must be invoked to
the number of Internet users per capita, the per-
derive them.
centage of the population that is literate, and
Some researchers address these difficulties by
the cost of a local call. Indicators for the policy
developing their own taxonomies. For example,
environment are the presence of an Internet ex-
van Ark, Frankema, and Duteweerd (2004) di-
change and measures of competition in three
vide the economy into three distinct sectors: IT
markets: local loop telecoms, domestic long dis-
producing industries, IT using industries, and
tance, and the Internet Service Provider market.
non-IT industries. The second and third catego-
Each of these indicators is a composite of subin-
ries are defined on the basis of their ‘‘IT inten-
dicators that are aggregated by applying the fol-
sity,’’ or IT capital per worker or per unit of
lowing formula as weights: value achieved/
output. Given that such industry-level statistics
maximum reference value.
exist only for the USA, they extrapolate the US
Other aggregate indexes measuring the ability
data to European countries. Interestingly, they
of national economies to exploit the economic
find that non-IT industries constitute two thirds
potential of the Internet and e-commerce yield
of the US and European economies, and an
a similar picture. The two most comprehensive
even higher fraction in emerging economies.
indexes in that respect gauging the e-readiness
Based on similar data, van Ark, Inclaar, and
of approximately half of the countries of the
McGuckin (2003) conclude that superior rela-
globe are the ‘‘E-Readiness Index’’ (ERI) of
tive economic performance in the USA(i.e., rel-
the Economist Intelligence Unit of the UK
ative to Europe) can be partially attributed to
(EIT, 2003) and the ‘‘Networked Readiness

---

684
WORLD DEVELOPMENT
Index’’ (NRI) representing a joint effort by
to have an idea of network or e-readiness of
INSEAD, World Economic Forum and the
the countries.
Infodev program of the World Bank. The latest
Akey implication of the data presented in
NRI was presented in ‘‘The Global Informa-
this section is that developing nations are likely
tion Technology Report 2002–03’’ (Dutta, Lan-
to fall even further behind the developed world
vin, & Paua, 2003).
unless appropriate policies are formulated and
The NRI consists of three dimensions related
implemented to reverse the palpable divergence
to IT: environment, readiness, and usage. The
in rates of diffusion of IT and the Internet.
first dimension measures the extent to which a
country’s markets, political and legal system,
and infrastructure (e.g., telecommunication
infrastructure) are beneficial to the develop-
3. IT AND PRODUCTIVITY: A REVIEW
ment and usage of IT. Readiness relates to
OF RECENT EMPIRICAL EVIDENCE
the ability of three key economic agents: indi-
viduals, firms, and government to capitalize
In recent years, numerous scholars have ana-
on the use of IT. The third dimension measures
lyzed the relationship between IT and economic
the incidence of IT usage by these same agents.
performance. Many of these studies examine the
The authors of the above joint report also re-
impact of IT on productivity growth, although
late IT expenditure to the NRI, in an effort to
some researchers also examine its effects on firm
determine how much ‘‘bang for the buck’’ na-
profitability and stock prices. Asummary of
tions are generating from such expense. They
several of these papers is presented in Table 3,
conclude that the USA, Finland, and Spain
where each study is presented by its methodol-
‘‘overperform,’’
while
Vietnam,
Colombia,
ogy, country of origin of the data, level of aggre-
and New Zealand underperform. Another les-
gation, and by a parsimonious description of
son they derive from this exercise is that expen-
the key findings.
diture on IT is not sufficient to guarantee a high
These empirical studies have been conducted
NRI score, since New Zealand had the highest
at all levels of aggregation, that is, at the estab-
rate of investment in IT, while it has a rank of
lishment, firm, industry, and national levels.
23 for the NRI.
Many papers present econometric estimates of
Acomparison of the NRI and ERI indexes
a simple (Cobb–Douglas) production function,
and rankings is presented in Table 2. Although
with an additional input representing invest-
they are generally showing similar patterns, the
ment in IT capital, as opposed to conventional
comparison draws attention to some differences
physical capital (structures and equipment).
and variations. The ERI index is the more ex-
Other authors (e.g., Brynjolfsson & Hitt,
tended one and is biased toward higher denom-
1996) have derived estimates of IT labor input
inations for the best performers. Thus in the
(typically the number of employees classified
case of the ERI, the variation among 25 best
as information systems workers). Siegel (1997)
performers is between 8.67 (for the champion,
estimated the following reduced-form equation
i.e., Sweden) and 6.96 for the 25th country (Is-
to assess the relationship between computers
rael in this case). For the NRI, the highest rank
and productivity:
country is 5.92 (Finland) and the 25th country
(Spain) achieved a score of 4.67. Note that the
y ¼ a þ b X
X
1
1 þ b2
2 þ u;
ð1Þ
lowest indices for them are, respectively, 2.37
(for the ERI, the 60th country, i.e., Vietnam)
where y is the growth in total factor productiv-
and 2.07 (for the NRI, the 82nd country, i.e.,
ity (TFP), X1 is the rate of R&D investment, X2
Haiti). There are also certain gaps in both sys-
is the rate of investment in computers, u is the a
tems. Thus, in the ERI, some important coun-
classical disturbance term, a is the rate of dis-
tries (from the perspective of ‘‘best practices’’)
embodied external technical change, b1 is the
such as Estonia and Tunisia are missing. The
rate of return on R&D, and b2 is the rate of re-
world’s second leading IT producer, Japan, is
turn on computers. The author finds that both
ranked rather low in both systems. While Ban-
IT and R&D enhance productivity growth, also
gladesh is not included in the ERI, Iran is ex-
reporting evidence of significant complementa-
cluded in the NRI. However, in spite of some
rities between ICT and R&D capital.
gaps, both systems are undertaking an impor-
Much of the recent firm-level evidence sug-
tant task of measuring and weighting a large
gests that there are ‘‘excess’’ returns to IT,
variety of important indicators permitting us
that is, the marginal product of IT capital is

---

IMPACT OF INVESTMENT IN IT
685
Table 2. Comparison between Network Readiness Index (GITR, 2003) and E-Readiness rankings (EIU, 2003)
Country
NRI score
NRI rank
E-R score
E-R rank
Finland
5.92
1
8.38
6
USA5.79
2
8.43
3
Singapore
5.74
3
8.18
12
Sweden
5.58
4
8.67
1
Iceland
5.51
5
–
–
Canada
5.44
6
8.20
10
UK
5.35
7
8.43
5
Denmark
5.33
8
8.45
2
Taiwan
5.31
9
7.43
20
Germany
5.29
10
8.15
13
Netherlands
5.26
11
8.40
3
Israel
5.22
12
6.96
25
Switzerland
5.18
13
8.26
8
Korea
5.10
14
7.80
16
Australia
5.04
15
8.20
9
Austria
5.01
16
8.09
14
Norway
5.00
17
8.20
7
Hong Kong
4.99
18
8.20
11
France
4.97
19
7.76
19
Japan
4.95
20
7.07
24
Ireland
4.89
21
7.81
15
Belgium
4.83
22
7.78
17
New Zealand
4.70
23
7.78
18
Estonia
4.69
24
–
–
Spain
4.67
25
7.12
23
Italy
4.60
26
7.37
21
Luxembourg
4.55
27
–
–
Czech Republic
4.43
28
6.52
27
Brazil
4.40
29
5.25
36
Hungary
4.30
30
6.23
29
Portugal
4.28
31
7.18
22
Malaysia
4.28
32
5.55
33
Slovenia
4.23
33
–
–
Tunisia
4.16
34
–
–
Chile
4.14
35
6.33
28
South Africa
3.94
36
5.50
32
India
3.89
37
3.95
46
Latvia
3.87
38
–
–
Poland
3.85
39
5.57
30
Slovak Republic
3.85
40
5.47
34
Thailand
3.80
41
4.22
42
Greece
3.77
42
6.83
26
China
3.70
43
3.75
50
Botswana
3.68
44
–
–
Argentina
3.67
45
5.41
35
Lithuania
3.65
46
–
–
Me´xico
3.63
47
5.56
31
Croatia
3.62
48
–
–
Costa Rica
3.57
49
–
–
Turkey
3.57
50
4.63
39
Jordan
3.51
51
–
–
Morocco
3.50
52
–
–
Namibia
3.47
53
–
–
(continued next page)

---

686
WORLD DEVELOPMENT
Table 2—continued
Country
NRI score
NRI rank
E-R score
E-R rank
Sri Lanka
3.45
54
4.13
44
Uruguay
3.45
55
–
–
Mauritius
3.44
56
–
–
Dominican Republic
3.40
57
–
–
Trinidad and Tobago
3.36
58
–
–
Colombia
3.33
59
4.86
37
Jamaica
3.31
60
–
–
Panama
3.30
61
–
–
Philippines
3.25
62
3.93
47
El Salvador
3.17
63
–
–
Indonesia
3.16
64
3.31
53
Egypt
3.13
65
3.72
51
Venezuela
3.11
66
4.75
38
Peru
3.10
67
4.47
41
Bulgaria
3.03
68
4.55
40
Russian Federation
2.99
69
3.88
48
Ukraine
2.98
70
3.28
54
Vietnam
2.96
71
2.91
56
Romania
2.66
72
4.15
43
Guatemala
2.63
73
–
–
Nigeria
2.62
74
3.19
55
Ecuador
2.60
75
3.79
49
Paraguay
2.54
76
–
–
Bangladesh
2.53
77
–
–
Bolivia
2.47
78
–
–
Nicaragua
2.44
79
–
–
Zimbabwe
2.42
80
–
–
Honduras
2.37
81
–
–
Haiti
2.07
82
–
–
Sources: GITR (2003) and EIU (2003).
substantially higher than the marginal product
(1993) also found very low returns on IT invest-
of non-IT capital. There is also some evidence
ment by Canadian banks. On the other hand,
that these private or firm-level returns have be-
Siegel and Griliches (1992) reported a positive
come higher in recent years. This is important
and significant correlation between a manufac-
because there was a lack of a consensus regard-
turing industry’s rate of investment in IT and
ing empirical results, at least in some of the
its productivity growth for all time periods.
early studies. Using country-level data, Oliner
At the firm level, Lichtenberg (1995) estimated
and Sichel (1994) concluded that IT did not
production functions and (as mentioned ear-
make a significant contribution to output
lier) found strong evidence of excess returns
growth. Catherine Morrison (1997) reached a
to information systems equipment and labor.
similar conclusion using industry-level data.
Most of the recent papers seem to find a
Based on estimation of a more elaborate set
strong relationship between IT and improve-
of cost function equations, including the price
ments in economic performance. Stiroh (2001)
and the quantity of IT equipment as separate
and Jorgenson and Stiroh (2000) report some
arguments in the cost function, she reported
good news regarding the aggregate impact of
that IT capital had only a very small impact
investment in information technology in the
on technical progress. Berndt, Morrison, and
USA. In their early studies (Jorgenson &
Rosenblum (1992) estimated production func-
Stiroh, 1995), the authors reported that com-
tions at the industry level and found that IT
puters did not make a large contribution to
was uncorrelated with productivity growth in
economic growth, reporting low estimates of
most industries. Parsons, Gottlieb, and Denny
the returns to IT that were quite similar to

---

Table 3. Recent empirical studies of the impact of ICT on economic performance
Author(s)
Methodology
Country/sector
Level of
Results
aggregation
Dunne et al. (2000)
Regressions of labor productivity
USA/manufacturing
Plant-level
Positive association
on computers
between computers and
labor productivity, which
appears to be growing
over time
McGuckin and Stiroh (1999)
Cobb–Douglas production
USA/manufacturing
Aggregate, major
Evidence of ‘‘excess’’
function with
and service
sector, and
returns to computer capital
IMPACT
computer capital
2-digit SIC
at each level of aggregation
industry-levels
Gera et al. (1999)
Cobb–Douglas production
USAand Canada/
Industry-level
Positive correlation
OF
function with
manufacturing
between investment in
computer capital
computers and labor
INVESTM
productivity growth
McGuckin et al. (1998)
Regressions of labor productivity
USA/manufacturing
Plant-level
Plants using advanced
on dummies
and service
computer-based
denoting whether
technologies have higher
ENT
the plant uses a computer-based
levels of productivity;
manufacturing
weaker evidence on the
IN
technology
relationship between
IT
technology usage and
productivity growth
Lehr and Lichtenberg (1998)
Cobb–Douglas production
USA/public sector
Organizational
‘‘Excess’’ returns to
function with computer
level
computer capital
capital and labor
(government
agencies)
Siegel (1997)
Latent variables model: regressions
USA/manufacturing
Industry-level
When controls are
of parametric
4-digit SIC
included in the model for
and nonparametric
measurement errors,
measures of TFP growth
computers have a positive
on the rate of investment
and statistically significant
in computers
impact on productivity
687

---

688
Table 3—continued
Author(s)
Methodology
Country/sector
Level of
Results
aggregation
Morrison and Siegel (1997)
Dynamic cost function estimation
USA/manufacturing
4-digit SIC
‘‘External’’ investments in
with ‘‘high tech’’ capital
industry-level
computers by related
industries (4-digit industries
within a 2-digit sector)
WORLD
enhance productivity
Greenan and Mairesse (1996)
Cobb–Douglas production function
France/manufacturing
Firm-level
Impact of computers is
with computer capital
and service
positive and at least as
large as for other types
DEVELO
of capital. Returns appear
to be higher in services
than in manufacturing
PMENT
Brynjolfsson and Hitt (1996)
Cobb–Douglas production function
USA/manufacturing
Firm-level
‘‘Excess’’ returns to
with computer capital and labor
and service
computer capital and labor
Jorgenson and Stiroh (2000)
Sectoral growth accounting methods
USAA
ggregate level
The growth contribution
of computers increased
substantially in the
mid-to-late 1990s
Siegel and Griliches (1992)
Correlation between nonparametric
USA/manufacturing
4-digit SIC
Positive correlation
measures of total factor productivity
industry-level
between rate of
and rate of investment in computers
investment in computers
and total factor
productivity growth
Source: Link and Siegel (2003, pp. 93–95).

---

IMPACT OF INVESTMENT IN IT
689
Table 4. The sources of US economic growth (1959–2001)
1959–73
1973–95
1995–2001
Output growth
4.18
2.78
4.07
Contribution of capital
1.77
1.40
2.03
Computers
0.07
0.20
0.49
Software
0.03
0.10
0.27
Communications capital
0.10
0.12
0.17
Other (noncomputer) capital
1.57
0.98
1.10
Contribution of labor
1.24
1.12
1.12
Aggregate total factor productivity
1.16
0.26
0.92
All values are average annual percentage growth rates. Input contributions are real growth rates, weight by average
nominal shares (following the convention in this literature).
Source: Jorgenson et al. (2002).
those presented by Oliner and Sichel (1994).
on the productivity of a given industry, that
However, in their recent papers (e.g., Jorgen-
is, IT investments undertaken by other indus-
son, Ho, & Stiroh, 2002), they conclude that
tries within the same broad sectoral category.
the impact of IT on the aggregate economic
The authors report that an increase in invest-
performance has increased over time, especially
ment in IT (and R&D) in a given industry has
in the last half of the 1990s. 2
a positive effect on the productivity perfor-
The key figures on sources of economic growth
mance of other industries (both their suppliers
in the USAare presented in Table 4, taken from
and customers). These results are broadly con-
Jorgenson et al. (2002). Based on a comprehen-
sistent with the notion that IT and the Internet
sive analysis of IT capital, the authors reported
constitute what leading economists refer to as
that computer hardware, software, and commu-
‘‘general purpose technology’’ (GPT) (Help-
nication equipment accounted for a much larger
man, 1998), or a technology that has wide
fraction of economic growth in the last six years
applications and productivity-enhancing effects
than in earlier periods. This may signify that
in numerous downstream sectors.
there are substantial adjustment costs in imple-
The OECD recently undertook a comprehen-
menting IT and that policymakers should not ex-
sive analysis of IT impact on productivity,
pect dramatic improvements in productivity
which was mainly based on data from devel-
growth in the short run. The economic payoff
oped countries. This analysis used indicators
comes only after a substantial increase in IT
such as the share of IT in nonresidential invest-
investment or activity.
ment, respective contributions of IT and other
In a similar vein, Morrison and Siegel (1997)
capital services into the output growth, as well
consider
the
possibility
that
conventional
as the impact of IT using and producing sectors
empirical studies of the connection between
as opposed to that of non-IT sectors, to the
IT and productivity actually underestimate
growth of economy and productivity of labor.
the returns to ICT, because they fail to take ac-
The result was fairly conclusive evidence sug-
count of externalities that arise from invest-
gesting that IT was a key driver of economic
ment in ICT. The authors extend the simple
growth in the USA, Canada, Nordic countries,
Cobb–Douglas production framework by esti-
and Netherlands, and also had a substantial
mating a dynamic, flexible cost function (i.e.,
positive impact on economic performance in
a Generalized Leontief functional form) for
other OECD countries. The USAand Finland
US manufacturing industries, which takes ac-
were the best performers in terms of labor pro-
count of adjustment costs that might arise from
ductivity growth in 1995–99 recording annual
IT (and other capital) investment. Their paper
growth of 2.5% and 2.75%, respectively. In
is a general critique and extension of various
both countries, 2% of labor productivity
new growth studies that use a simple produc-
growth can be attributed to the IT producing
tion function approach to assess the impact of
and IT using sectors, while only, respectively,
what the authors call external factors (invest-
0.5% and 0.75% were originated from sectors
ment in R&D, computers, and human capital)
with a low level of IT use (OECD, 2002, pp.
on growth. More importantly, they estimate
22–24). The evidence on the firm level in those
the effects of external investment in computers
countries also show the importance of time

---

690
WORLD DEVELOPMENT
to accumulate the critical mass of ITs and
employment or labor cost), constant returns
primarily Internet networks through their
to scale, homogeneity of degree one in prices,
extensive diffusion and make them an essential
and taking first differences yields
part of everyday practice for workforce thus
allowing for further improvements in produc-
dsN ¼ b þ b d lnðW
d lnðR=Y Þ
0
1
N =W jÞ þ b2
tivity of labor and corporate organization with
requisite efficiency gains (OECD, 2003). The
þ b d lnðC=IÞ þ u;
ð3Þ
3
OECD study confirms the findings and policy
conclusions of many researchers on the impor-
where the authors include two proxies for tech-
tance of IT investment in enhancing economic
nological change: R&D ‘‘intensity,’’ or the ratio
growth in most developed economies.
of R&D expenditures to sales (R/Y) and the
ratio of expenditures on computers to total cap-
ital investment (C/I), and u is a classical distur-
bance term. If b2 > 0 or b3 > 0, we have
4. EFFECTS OF ICT ON WAGES AND
evidence
of
‘‘skill-biased
technological
WORK ENVIRONMENT
change.’’ Paul and Siegel (2001) extend this
framework by estimating a dynamic flexible
The IT revolution has also heightened a phe-
cost function model. This approach allows for
nomenon known as ‘‘skill-biased technological
quasi-fixed inputs, a more general functional
change’’ (henceforth, SBTC) or the instance
form for the cost function (than a simple
when technological change results in a stronger
Cobb–Douglas or translog functional form),
demand for highly skilled and highly educated
and also includes measures of trade and out-
labor. This leads to an increase in the relative
sourcing as independent variables.
wages of these workers and shifts in workforce
Asummary of some recent studies of the im-
composition in favor of highly skilled and
pact of IT on wages and labor composition is
highly educated workers. In the long run, wage
presented in Table 5. Despite the fact that
competition from skilled labor in less developed
researchers have employed alternative method-
countries will diminish the magnitude of SBTC,
ologies and analyzed data from different
as companies in developed countries engage in
countries at different levels of aggregation (indi-
outsourcing to take advantage of lower wage
vidual, plant, firm, and industry levels), each
rates. Studies of SBTC are usually based on
study reports evidence that is consistent with
estimation of wage equations or cost functions,
the existence of SBTC. That is, these research-
typically including dummy variables that serve
ers generally find that some proxy for techno-
as proxies for technological change. The cost
logical change (R&D, computers, adoption of
function approach is desirable because it allows
advanced manufacturing technologies) is posi-
us to formally test whether technical change is
tively correlated with wages and shifts in labor
nonneutral, that is, it favors one factor of pro-
composition in favor of highly skilled or highly
duction relative to another. Note that under
educated workers.
SBTC, the assumption is that technological
Many economists who have studied SBTC
change favors one class of workers (e.g., highly
ignore the role of organizational change in
educated workers, at the expense of another
the implementation of new technologies. In re-
class of workers).
cent decades, many manufacturing firms have
Acommonly used cost function approach,
adopted new IT-based technologies, such as
employed by Berman, Bound, and Griliches
computer-aided design (CAD), computer-aided
(1994), is estimating the following restricted la-
manufacturing (CAM), computer numerical
bor cost function:
control (CNC), and just in time production
(JIT) systems. Implementation of these technol-
LCi ¼ f ðW i; TECH ; Y ; tÞ;
ð2Þ
ogies can have a dramatic impact on the work
environment since they may simultaneously re-
where LC is the labor cost, Wi is the wage
sult in downsizing (labor-saving innovations),
of the ith type of worker, TECH is a proxy
retraining of the remaining workforce (‘‘skill
for the technological change, Y is the output,
upgrading’’), and changes in job responsibilities
t is the time, and f is assumed to have a translog
resulting from integration across the functional
form. Invoking cost minimization and Shep-
areas of business (marketing, manufacturing,
hard’s lemma (SN = o ln LC/o ln PN, where SN
R&D, accounting/finance, logistics, purchas-
is the share of nonproduction labor in total
ing, and product design).

---

Document Outline

• The Impact of Investment in IT on Economic Performance: Implications for Developing Countries
  • Introduction
  • Comparing it in various regions
  • It and productivity: A review of recent empirical evidence
  • Effects of ICT on wages and work environment
  • ICT and Internet in developing countries: Some preliminary evidence
  • Tentative conclusions
  • References

---