Maurizio Martinelli, Irma Serrecchia, Michela Serrecchia
Institute for Informatics Institute for Informatics and Telematics (IIT-CNR)
Keywords: Internet diffusion, digital divide, domain names.
Abstract: The digital divide can occur either as a “local” (within a given country) or “global” (between developing
and industrialized countries) phenomenon. Our study intends to offer an important contribution by
analyzing the digital divide in Italy and the factors contributing to this situation at the territorial level (i.e.,
macroareas: North, Center, South and at the regional level). To do this, we used the registration of Internet
domains under the “.it” ccTLD as proxy. In particular, we analyzed domain names registered by firms. The
analysis produced interesting results: the distribution of domains registered by firms in Italian regions is
more concentrated than the distribution according to income and population, suggesting a diffusive effect.
Furthermore, when analyzing the factors that contribute to the presence of a digital divide at the regional
level, regression analysis was performed using social, economic and infrastructure indicators. Results show
that Italian provinces that have good productive efficiency, a high cultural level, and greater spending for
investment in telephony and telematics are the best candidates for utilization of the Internet.
The analysis of the Internet presence in various
social activities and economic and political areas
indicatess a serious problem: the existence of a
“digital divide” between those who possess the
material and cultural conditions to exploit the new
technologies, and those who do not, or who lack the
crucial ability to adapt to the rapid continual change
that characterizes the Internet today (Warschauer,
2001; OECD, 2001).
The term “digital divide” first appeared in the
1990s during the privatization phase of the Internet,
to indicate a condition of relative disadvantage
experienced by certain user categories in terms of
accesssibility and cost - for example, by residents of
inland areas compared to those on the East and West
coasts (NTIA, 1995, 1998, 1999, 2000, 2001). Many
studies have been conducted at the national and
international levels concerning Internet diffusion,
and reveal the presence of a digital divide between
developing and industrialized countries (“global
digital divide”) as well as within a given country
(“local digital divide”).
Furthermore, the presence of the digital divide
has stimulated researchers and governments
worldwide to identify contributing factors and
propose methods for reducing this gap.
There are two schools of thought regarding this
subject. Some authors, for example Dasgupta et al.
(2001), maintain that the digital divide is a problem
which exists and will continue to persist because
“development researchers” do not see the potential
benefits of Internet use in poor countries. Duncombe
(2000) insists that Internet access is not justified
among the African poor because lack of information
will prevent them from using technology. Others
believe that in the future, the benefits of the web will
also spread among those who are disadvantaged in
terms of accessibility and cost; it is only a question
of time, whereas those who are currently more apt to
use the new technology will reach a saturation level
(Norris, 2001; Roger, 1995). Other studies (Lal,
1996; Anand, 2000; IDC, 2000), in agreement with
this theory, confirm that where Internet services are
available, poor families do not appear to hesitate or
be incapable of using the new technology. Bayes
(1999) noted the rapid adoption of e-mail by
families in Bangladesh as a cost-effective alternative
to telephone calls. In a recent study involving more
than 100 companies in developing countries,
Rajkumar (2000) discovered that Internet use was
accompanied by increased exports and sales, and
encouraged greater customer loyalty. Kenny et al.
(2005) showed that since the mid-1990s the increase
in Internet users in the developing world has been
Martinelli M., Serrecchia I. and Serrecchia M. (2005).
In Proceedings of the First International Conference on Web Information Systems and Technologies, pages 431-435
DOI: 10.5220/0001232504310435
more rapid than the growth rate in wealthy countries.
South Asia, the Middle East, and Africa are far
behind in terms of hosting web sites. “The situation
of computers in education is similarly mixed
According to the economics literature, factors
determining the presence of the digital divide in the
world do not derive exclusively from the impact of
income level (Chinn and Fairlie, 2004; Norris,
Some studies show that low-income countries
that have undergone considerable reform towards
competition saw a growth of 80% in fixed and
mobile teledensity compared with non-reformed
countries (Kenny at al., 2003). In a study based on
the experiences of 86 countries from 1985 to 1999
Fink et al. (2001) show that sector reforms have a
higher percentage of “mainline provisions and a
higher percentage of labor productivity compared
with non-reformed countries. Studies show that
competitive provision reduces the cost of
infrastructures for Internet access (Rassotto et al.
2004; Qiang and Pitt, 2003; Reynolds et. al. 2004;
Wallsten, 1999). Other factors having some impact
on Internet use are schooling and illiteracy,
urbanization and electricity consumption (Kenny at
al. 2005; Dasgupta, Lall and Wheeler, 2001).
Regarding the extent of e-commerce, studies once
again suggest the critical role of the underlying
infrastructure but also point out the importance of
strong institutional structures in the form of the “rule
of law” and the availability of reliable methods of
payment such us credit cards (Kenny at al., 2005;
Oxley and Yeung, 2001).
Several metrics are available for measuring Internet
diffusion. The most convenient are the so-called
endogenous metrics which can be “obtained in an
automatic or semiautomatic way from the Internet
itself” (Diaz-Picazo, 1999). These metrics have the
undeniable advantage of accuracy; according to the
literature the most frequently used are Internet hosts
based on hostcount procedures (see studies
published by Internet Software Consortium or da
RIPE-NCC) and second-level domain names (Naldi,
1997; Zook, 1999; Bauer, Berneand and Maitland,
To measure the analysis of Internet diffusion in
Italy, we used the endogenous measure of second-
level domain names registered under the “.it
ccTLD, managed by the Institute of Informatics and
Telematics of CNR, Pisa.
Despite the advantages offered by endogenous
measures, there are also a few disadvantages, since
in some cases they tend to underestimate and in
others to overestimate the phenomenon being
studied (Zook, 1999, 2000, 2001). Overestimation
can occur when the number of hosts is used, often
associated with IP addresses, while if we consider
the number of domains registered, more than one
domain may be associated with the same registrant.
Underestimation can occur because not all internet
users register a domain name under their own
ccTLD, and in many countries the regulations allow
foreign citizens to register under their own ccTLD .
In the case of hosts, underestimation may be due to
the growing presence of firewalls and private
networks (Intranet) and the use of dynamic IP
addresses, increasingly accompanied by new tools
for access to the Net (for example, mobile phones).
In spite of these disadvantages, the numbers of hosts
and Internet domains are the principal means utilized
for analyzing Internet diffusion.
The Institute for Informatics and Telematics (IIT-
CNR), which manages the “.it” ccTLD Registry, is
conducting a study analyzing the diffusion of
Internet use in Italy. Data were extracted from the
databases of registrations managed by the IIT-CNR,
using automatic and semi-automatic procedures
[This means that we have created a new database
useful for analyzing Internet diffusion by initially
consulting the WHOIS
database using an automatic
procedure, for example in order to determine the
category of the applicant, the automatic procedure
verified whether a ORG field existed and if there
was, classified it as a firms. If the field was
erroneous, the LAR
database (semi-automatic
procedure) was consulted]. Approximately 550,000
domain names have been analyzed and grouped into
several categories (individuals, firms, universities,
associations, public groups, foundations,
The WHOIS database contains information regarding the
domain names registered under the “.it” ccTLD, applicants who
have signed a contract with IIT-CNR and technical and
administrative contacts.
The LAR is a letter requesting the registration of a domain
name, with which the applicant assumes full civil and penal
responsibility for the use of the domain name requested. To be
able to register a domain name under the “it.” ccTLD, firms must
send the IIT-CNR (Pisa) a LAR , containing the applicant’s
identifying data, declaration of the knowledge of basic principles
for use of resources and the Internet, and examination of the laws
established by the Rules Commission, concerning the technical
registration procedures.
committees, and other organizations) in order to
identify the determinants of adoption and then of
diffusion, for each category. A careful data cleaning
procedure was followed. As of September 7, 2001,
the database WHOIS contained 265,437 domains
registered by companies, of which 3,503 were
domains registered by companies with their legal
headquarters in the various EU countries. The
remaining 3,331 domains were not classified, since
it was impossible to ascertain the area where these
companies operate. A total of 258,603 domains
registered by businesses were analyzed.
thermore, in this paper we will only consider one
domain name for each business: in other words, if a
company registered multiple domain names, we
considered only the first one registered.
As the literature suggests, analysis shows that a
technological divide exists among companies
operating in specific geographic areas in Italy. As
shown in the following Figure, the North is more
likely than other areas to use the new technology.
Since Italy is divided into 20 administrative units
called regions, in this article we have found it
advisable to analyze internet diffusion by region.
Lombardy, Alto Adige and Lazio (in that order) are
the regions showing the highest penetration rates
(penetration rate is calculated by dividing the
number of domains registered by companies by the
number of companies in Italy, and multiplying the
ratio by 100. At the macro-area level, the relative
penetration rate recorded in North and Central Italy
are higher compared to the South (7.23 and 7.07
respectively for every 100 companies, compared to
4.04 in the South.). The calculation of the Gini index
and the construction of the Lorenz curve show that
the concentration of domains is higher compared to
the number of firms and total income. The Gini
index, calculated according to the number of
domains registered, was 0.557 compared to 0.468
computed according to the number of companies and
0.466 calculated according to total income.
3.1 Determinants of adoption
According the OECD (1999), the digital divide is the
distinction between “Who has and who does not
have access to information (OECD)”. Several
aspects of this phenomenon exist. It is possible to
observe a digital divide among countries
(international digital divide), among individuals or
organizations and within a country at the local level
(domestic digital divide). In order to analyze
whether there is a domestic digital divide in Italy,
we ran exploratory stepwise regressions using
economic and social indicators at the regional level.
This allowed us to examine the determinants of
Internet diffusion, by singling out the factors leading
to registration of domains by company. The
dependent variable of our regression models is the
penetration rate calculated as
Penetration Rate = (Number of domains registered
by companies/number of companies)*100;
We constructed three model. The first two (M1, M2)
analyzed the influence on Internet diffusion of two
key economic factors: per-capita income and added
value per employee and model 3 using social,
economic and infrastructure indicators.
As expected, these two variables positively
correlated with penetration rates per capita income
and value added per employee (0.701 and 0.827
Table 1: Stepwise regressions with per capita income as
independent variable
Table 2: Stepwise regressions with per added value
employee as independent variable
The economic literature, in fact, underlines that
differences in income distribution play a crucial role
in explaining differences in ITC access and
Lorenz curve for domains, income, number of firms
Equidistribution Number of domains registred by firms Number of firms Total income
utilization (Norris, 2000; Pohjola, 2002, Hansons,
2000 Warschauer, 2001). We included these crucial
indicators in distinct models in order to address
multicollinearity problems. Per capita income and
added value per employee are very likely to be
correlated with other social and economic indicators
at a local level, generating distortions in the
estimated coefficients. Although it includes only a
independent variable, the model is quite powerful,
and explains about 49% per cent of the variability in
registrations among Italian regions. The coefficient
of per capita income is highly significant and has the
expected sign, showing a positive influence of per
capita income on Internet diffusion at a local level.
Quite similar results are obtained in M2 including
the added value per employees as independent
variable. The R2 is even higher, stating that the
efficiency of the productivity structure account for
68.5 % of the variability in the Internet diffusion. As
previously stated, and in agreement with the
economics literature, the variation in Internet
diffusion between regions may derive from other
factors as well. In the model 3 (Table 3) we analyzed
the stepwise regression, taking into consideration as
dependent variable an economic factor (employees
in the service sector); one related to education
(number of college graduates); a socio-cultural
variable (spending for theatrical and musical
performances); one related to infrastructure (founds
for telephony and telematics) and one relative to
public spending (hydraulic works and electrical
Table 3: Stepwise regressions taking as dependent variable
the penetration rate of the companies
The Table shows that regions that spend
considerable funds on musical and cultural activity
are more likely to use the new technology (spending
for theater has the second-highest Beta value
compared to the other variables). As might be
expected, the index of spending for telephony and
telematics also plays an important role (the Beta is
equivalent to 0.708). In fact as the literature
proposes (Warschauer, 2001) one of the determining
factors in Internet diffusion is the presence of
adequate network infrastructures.
In brief, it is possible to conclude that regions with
an efficient and service-oriented structure, a lively
cultural scene, and a good educational level ( greater
number of college graduates) are more inclined to
use the new technology and are the best candidates
for a more active and interactive use of the Internet.
Anand, A., 2000 “ICTs: What Digital Divide?” Women’s
Feature Service, Pondicherry, India, August 22.
Audretsch D. and Feldman F., 1996 R&D, Spillover and
the Geography of Innovation and Production, The
American Economic Rewiew, Vol.86, No. 3
Bayes, A., J. von Braun and R. Akter, 1999, “ Village Pay
Phones And Poverty Reduction: Insights from a
Grameen Bank Initiative in Blangadesh”, Center For
Development Research, University of Bonn,
Discussion Paper on Development Policy, No. 8, June.
Chinn, M. D. and R.W. Fairlie, 2004. The determinants of
the Global Digital Divide: A Cross-Country Analysis
of Computer And Internet Penetration. Madison, WI,
University of Wisconsis, Madison and NBER
Dasgupta S., Lall S. V., Wheeler D., 2001 Policy Reform,
Economic Growth, and the Digital Divide: An
Econometric Analysis World Bank - Development
Economics Research Group (DECRG).
Diez-Picazo G.F. (1999) An Analysis of International
Internet Diffusion. Ph.D. Thesis, MIT.
Duncombe, R. , 2000, “Information and Communication
Technology, Poverty and Development in sub-Saharan
Africa”, Institute for Development Policy and
Management, University of Manchester, UK, June.
Faiola, A. and Buckeley, 2000, “Poor in Latin America
Embrance Net’s Promise,” WashingtonPost, July 9, p.
Fink, C.and C.J. Kenny (2003). Whiter the Digital Divide?
Washington D.C., The world bank.
IDC, 2000 “Mobile Access to Internet Gains Momentum”,
IDC Forecaster Newsletter, March 21.
Lal, K.1996, “Information Technology, International
Orientation And Performances: A Case of Study of
Electrical Goods Manufacturing Firms in India”
Information Economics and Policy, Vol. 8, pp. 269-
Naldi M. (1997). Size estimation and growth forecast of
the Internet. Centro Volterra, Tor Vergata
Norris, P. (2000) The Global Digital Divide: Information
Poverty and Internet Access Worldwide. Internet
Conference at the International Political Science
World Congress, Quebec City;
OECD, 2001 Internet infrastructure indicators. OECD
Oxley. J. E. and B. Yeung (2001). E-Commerce
Readiness: Institutional Environment and International
Competitiveness. Journal of International Business
Studies. 32. 4 (Fourth Quarter). 705-723.
Pohjola M. (2002) The adoption and diffusion of ITC
across countries: patterns and determinants. Paper
presented at the Wider Conference on the New
Economy in Development. Helsinki, Finland, 1011
Qiang C. Z-W. and A.Pitt, S. Ayers ,2003. Contribution of
Information technologies to Grow. The World Bank.
Rajkumar (2000), A. S., “The impact of E-commerce on
DevelopingCountry Firms : A survey Of Partecipants
in, “World Bank, Development
Research Group, July.
Roger E. (1995)-Diffusion of Innovations-Free Press-New
Wallsten S. J., 1999 “ Regulation and Internet Use in
Developing Countries.” Forthcoming, Economic
Development and Cultural Change
Warschauer M. (2001), What is the digital divide?,
University of California
Zook M. A. (2001). Old hierarchies or new network of
Centrality?- The Global Geography of the Internet
Content Market. American Behavioural Scientists, 44
(Special Issue: Mapping the global Web).
Zook M.A. (1999). The Web of Consumption: Spatial
Organisation of the Internet Industry in the United
States. American Behavioural Scientist (forthcoming).
Zook, M.A. (2000) Internet metrics: using host and
domain counts to map the Internet
Telecommunications Policy 24.6-7