Impact of Internet Penetration in UAE
Impact of Internet Penetration in UAE
Impact of Internet Penetration in UAE
net/publication/319906315
The economic impact of the Internet penetration rate and telecom investments
in Arab and Middle Eastern countries
CITATIONS READS
37 781
1 author:
Georges Harb
Lebanese American University
19 PUBLICATIONS 152 CITATIONS
SEE PROFILE
All content following this page was uploaded by Georges Harb on 09 November 2017.
Abstract
By using a sample of 93 countries and panel data econometrics, we evaluated the impact of the
Internet penetration rate and investment in the telecom sector on economic growth in the Arab
and Middle East (AME) region over the 1995-2014 period1. The results show that the Internet
has been a significant growth determinant in the AME region, particularly in high-income AME
countries; Internet penetration in middle-income AME countries seems to be hovering below the
growth impact rate. Telecom investments did not affect growth in the AME region, thus
highlighting the general-purpose technology nature of such investments and the adjustment
period needed before the latter bear fruits.
Keywords: Arab and Middle East region; economic growth; ICT; Internet; investment in
telecommunication; dynamic panel estimation
JEL Classification
O30, O47
Citation: Georges Harb, 2017. " The economic impact of the Internet penetration rate and
telecom investments in Arab and Middle Eastern countries", Economic Analysis and Policy, 56
(2017), 148-162.
1
Arab and Middle East (AME) region.
1
1. Introduction
With the rise of the "digital economy", the information and communication technologies (ICT)
have become the major source of innovation worldwide (Quiggin, 2014)2. Parallel to the rest of
the world, the 2000s marked a significant rise in ICT in the AME region. Investment in the
telecommunication sector in the region soared from $3.4 billion in 2000 to $23.5 billion in 2011.
The share of the telecom investment in the gross domestic product (GDP) of the countries of the
region also increased, rising from an average of 0.47% in 1995 to 0.93% in 2010. This trend was
common to the region's two subgroups: high-income and middle-income AME countries,
although it was more pronounced in the latter (Appendix, Figure 2). A milestone of the same
decade was the rapid diffusion of Internet use across the AME countries (Appendix, Figure 1):
the average Internet penetration rate in the region was 0.22% in 1996, rising to 6.2% in 2011.
Most of the AME countries have recently adopted national ICT strategies articulated around a
number of common pillars: increasing the competency of ICT manpower, developing and
upgrading the ICT infrastructure, providing electronic delivery of the government's services, and
streamlining the telecom regulatory framework (ESCWA, 2013). In many cases, these strategies
have been accompanied by steady steps toward a more competitive telecom market structure. In
various AME countries, Internet infrastructure has lately undergone several upgrades with the
establishment of new core networks, submarine cables, fiber optic networks, and new Internet
hosts (ESCWA, 2013).
Against the backdrop of rolling out ICT strategies in the AME countries and the expected
positive impact of the adoption of the latest technologies on the competitiveness of their
economies, it is timely to investigate the extent to which ICT have impacted their economic
performance throughout the last two decades and highlight the possible lessons that can be
learned. In essence, this paper answers three questions: (i) Has the widespread use of the Internet
been greasing the wheel of economic growth in the AME countries? (ii) Did investment in
telecommunication services pay off in terms of higher standards of living in those countries? (iii)
Given the disparities in terms of the ICT infrastructure/use between the two AME sub-regions,
was the impact of ICT uniform across the region?
2
The structural economic changes brought about by ICT have outlined the need of a new agenda for national
accounts, notably in terms of valuing the stocks and flows of information, for details see Quiggin (2014).
2
Those questions are motivated by the well-documented overall economic impact of ICT: these
technologies bestow positive externalities on several sectors in the economy, leading to
widespread productivity gains (Qiang, 2009). Public spending on ICT is thus expected to have
spillover effects and to increase the payoffs of investments in other sectors.
To answer these questions, we employed recent data encompassing 93 countries from different
regions of the world, of which 15 are from the AME region, over the period 1995-2014 and
estimated a growth model with standard determinants of growth and ICT measures as the main
independent variables. To identify the economic impact of ICT on the AME region as a whole as
well as in each of the AME sub-regions, we allowed for a differentiated impact of ICT on growth
across the regions. This also enabled us to facilitate regional comparisons in terms of the growth
impact of ICT.
Our analysis shows that the Internet penetration rate has positively and significantly contributed
to growth in the AME region. There is some evidence indicating that this mainly reflects the
impact of the Internet on high-income AME countries. Finally, our results point to an
insignificant effect of investments in the telecommunication sector on economic performance in
the AME countries across the two sub-regions.
In view of previous studies on the impact of ICT on the AME economies, our approach is
characterized by three main additions. The estimation procedure enables us to consider the
endogeneity of the independent variables, in particular the ICT measurements. Furthermore, the
empirical strategy allows a systematic differentiation of the ICT impact across the two AME sub-
regions. Finally, owing to the non-composite ICT indicators used, we could provide a number of
plausible factors that could underlie our key findings as well as some policy recommendations.
2. Literature review
In this section we succinctly review a selected number of recent papers that examined the
economic effects of ICT before focusing on studies specifically tackling the impact of ICT in the
AME region. The purpose is twofold: (i) pinpointing the key findings of the literature and (ii)
highlighting the main limitations of previous work done in the context of the AME countries.
3
in/expansions of the ICT infrastructure (Tsang et al., 2011). On the supply side, the use of ICT is
likely to raise labor productivity and foster innovation (Tsang et al., 2011; Brynjolfsson and Hitt,
2000). It can also lead to efficiency gains induced by the appearance of new business models,
redesign of supply chain management, and greater access to input and output markets available
to firms (Tsang et al., 2011; Brynjolfsson and Hitt, 2000). ICT-induced changes in management
models can be emulated across several sectors, leading to positive spillovers (Basu et al., 2003;
Basu and Fernaldm, 2007). On the demand side, the diffusion of ICT should induce a reduction
in transactional costs incurred by consumers (Lee et al., 2012).
The anticipated positive economic impact of ICT was largely confirmed by empirical
investigation (Giesecke, 2006; Koutroumpis, 2009; Czernich et al., 2011; Shahiduzzaman and
Alam, 2014a; Shahiduzzaman and Alam, 2014b).
This "stylized fact" has been nuanced by many researchers; in particular, three findings are worth
expanding upon. First, an abundance of literature has stressed and explored the nature and effects
of ICT as a "general-purpose technology" (GPT), chiefly in the context of the Organization for
Economic Cooperation and Development (OECD) member countries (Basu et al., 2003; van Ark
and Inklaar, 2005; Basu and Fernaldm, 2007; Ceccobelli et al., 2012; Liao et al., 2016). One key
lesson from this literature is that the beneficial economic effect of ICT seems to be conditional
on much-needed firm-level and industry-level "complementary investments": it is only when
appropriate organizational adjustments and reallocation of resources are made and combined
with the widespread use of ICT capital that the latter starts significantly impacting productivity
and output. Prior to that, ICT capital can entail an adverse effect on productivity. A corollary to
this finding is that a substantial time lag is needed before ICT investments translate into
productivity growth.
Second, there is evidence that the economic effect of ICT is positively associated with the
competitiveness of the ICT sector: a number of papers have empirically documented the adverse
impact of restrictive measures on foreign direct investment (FDI) inflows in the telecom sector
(OECD, 2009; Borchert et al., 2012b), and the positive effects of opening up the latter to foreign
service suppliers (Rossotto et al., 2003). In particular, Borchert et al. (2012a) and Rossotto et al.
(2003) demonstrated that telecommunication investments are likely to yield low returns in the
absence of measures enhancing competition among service providers.
4
A third interesting result pinpointed by the literature is that, at the aggregate level, a certain
threshold of diffusion of the latter technologies has to be reached before their reverberations fully
unveil. This so-called "critical mass" level, where the diffusion of ICT becomes self-sustainable
(Gajek and Kretschmer, 2012), is specifically put forward in papers examining the economic
impact of Internet diffusion (Koutroumpis, 2009; Czernich et al., 2011).
2.2. Findings and limitations of literature examining ICT’s impact on the AME region
A number of papers have recently assessed the effect of ICT on the AME economies (Abdel
Kader, 2006; Hassan, 2003; Guetat and Drine, 2007; Mahboub and Salman, 2008; Badran, 2012;
ITU, 2012; Hodrab et al., 2016). Most of the latter papers have been cast in terms of the
augmented Solow growth model where the growth rate of GDP per capita over a given period is
typically regressed on a set of standard determinants of the steady-state level of GDP per capita,
the start of period GDP per capita levels, ICT measurement, and a set of additional variables of
particular relevance to the AME region3. The vast majority of studies have applied panel data
estimation techniques using relatively large samples of AME countries and covering rather long
time intervals. A number of papers have pointed to a positive and significant effect of ICT on the
economic performance of AME countries (Abdel Kader, 2006; Mahboub and Salman, 2008;
Badran, 2012; Hodrab et al., 2016), whereas the remainder of the studies did not detect a
significant economic impact of ICT.
In view of the previous research assessing the ICT repercussions in AME countries, we first shed
light on two problematical issues that have speckled the above-mentioned papers and that we
tackle appropriately in our empirical strategy before highlighting two additions that our present
investigation embodies.
The first thorny matter pertains to the ICT indices that were often used. As a way of
circumventing severe data limitations and allowing for larger variations across countries, the
majority of the authors had to create and use composite ICT indicators4. The latter commonly
encompassed the Internet and personal computer penetration rates as well as the density of fixed
phone lines and the number of cell phone subscribers. A limitation of such indices stems from
the fact that they are multifaceted: they remain uninformative in terms of what is driving the
3
The key determinants of the steady-state level of GDP per capita include the investment ratio averaged over the
period covered, the growth rate of the population over the period, and a proxy for human capital.
4
With the notable exception of the ITU (2012) that used the broadband penetration rate and Badran (2012) who
employed the broadband penetration rate as well as the investment in the telecommunications sector.
5
results among the vast array of their constituents. Instead of reverting to a composite ICT
indicator, we prefer to focus on "simple" ICT measurements that recently became available for a
large number of AME countries. The key advantage of employing such variables is the
straightforward readability of the findings compared to results obtained using composite
indicators.
The second issue refers to the fact that, apart from Guetat and Drine (2007), none of the papers
tackled the potential endogeneity of some of the regressors, including the ICT variables. One can
indeed argue that higher standards of living can induce a larger demand for and supply of ICT
variables, thus leading to a reverse causality. This is a major concern because the obtained results
are likely to be biased. To overcome this problem, we implemented an appropriate
instrumentation strategy to assess the exogenous impact of non-composite ICT measures on
growth.
We extend the previous literature in two ways. First, most of the previous empirical analyses
covered the period from the start of the 1990s to the early 2000s. It is, however, likely that those
analyses failed to fully account for the ICT impact on the AME region, as the major upsurge in
the ICT infrastructure/use in the latter occurred during the 2000s. By using recent data through
2014, our results should particularly reveal the impact of ICT on economic performance in the
region. Second, none of the previous papers has systematically distinguished between the ICT
effects in each of the two AME subgroups. Such sub-regional distinction is however desirable
given the disparities the two sub-regions display in terms of ICT infrastructure and use. The
empirical setting we employ is flexible enough to investigate the ICT effect in each of the two
AME sub-regions.
3. Empirical strategy
We used the augmented Solow model, cast into a dynamic panel data setting, to evaluate the
impact of the Internet penetration rate and the investment in the telecommunication sector (as a
share of GDP) on economic growth in the AME countries. The empirical analysis covers the
1995-2014 period, with a sample of 93 countries, of which 15 were from the AME region.
Following a common practice used in seminal papers, we divided this period into sub-periods to
wash out the impact of the business cycle5. Precisely, we split this period into the following 5-
5
This approach was introduced by Islam (1995) and later employed by Caselli et al. (1996) and Bond et al. (2001).
6
year sub-periods: 1995-1999, 2000-2004, 2005-2009, and 2010-20136. For a given sub-period,
the first estimated equation has the following form:
with
and the (logarithm of) real GDP per capita at time and , respectively; with
= 2000, 2005, 2010, and 20147. For instance, when = 2000, = 1995, = 1999, and so
on for = 2005, 2010, and 2014. is the ICT variable: we used, alternately, the Internet
penetration rate and the investment in the telecommunication services as a
share of GDP , each averaged over every sub-period (that is, between and
)8.
To allow for a differentiated growth impact of the ICT variable across the various regions, we
introduced the term . The regional dummy is equal to one when country
is located in region . Our sample of countries is divided across six groups: Africa, AME, Asia,
Europe, Latin America, and the OECD countries (Table 1 of the Appendix presents the regional
grouping).
is a row vector where three determinants of the steady-state GDP per capita in the
augmented Solow model are included. The first variable is the average annual growth rate of the
population during each sub-period. The second is a proxy for human capital: the average number
of years of secondary schooling in the population aged 15 and above at the start of each sub-
period (that is, at ). The last element included in is a proxy for the share of output
invested in physical capital and differs with the ICT measurement used: when the Internet
penetration rate is employed, the last component of vector X is the average of (the logarithm of)
6
Because of limited data availability, the last sub-period covered is shorter than the previous sub-periods.
7
For the last sub-period, we could not use 2015 data because GDP per capita figures were either missing or
estimated for a large number of AME countries; we thus opted for the 2014 figures.
8
Specifically, the first ICT variable refers to fixed Internet subscribers per 100 inhabitants (measured irrespective of
the type and speed of access to the Internet as well as the device used to access the Internet); the second pertains to
investments in fixed, mobile, and Internet services for acquiring or upgrading property (tangible assets, intellectual,
and intangible assets) and networks (as a share of GDP).
7
the share of gross capital formation in GDP during each sub-period; when the telecom
investment is our proxy for ICT, the final element of becomes the average of (the
logarithm of) the share of gross capital formation net of investment in telecommunication in
GDP during each sub-period (since gross capital formation includes telecom investment).
Equation 1 represents our two "baseline" specifications, one for each ICT measure used. As a
robustness check exercise, we add to each of the baseline specifications several regressors
typically included in empirical growth models, with a special emphasis on those that are a priori
more relevant to the AME region. As a result, the "augmented" specifications are represented by
Equation 2:
Estimating either of Equations 1 and 2 is equivalent to estimating a dynamic panel data model
with the lagged dependent variable as a regressor9. In this setting, the standard fixed effects
estimator would yield inconsistent estimates because the right-hand side lagged dependent
variable is correlated with country-specific effects. A common practice is to consider the
dynamic equation in first difference form, eliminating the country-specific effects. Estimating
the resulting first difference equation should however accommodate the endogeneity of the
regressors. Indeed, the right-hand lagged dependent variable is by construction correlated with
9
This is easily done by transposing to the right-hand side of Equations 1 and 2.
8
the error term, whereas the rest of the regressors are likely to be affected by the dependent
variable (reverse causality). Thus, employing an appropriate instrumentation strategy becomes
indispensable to avoid biased and inconsistent estimates. It can be shown that applying the
generalized method of moments (GMM) estimator on the first difference equation, which is
known as the first difference GMM (FDGMM) estimator, yields consistent estimates while using
suitable lags of the endogenous variables as instruments (Bond et al., 2001; Caselli et al., 1996).
Blundell et al. (2000) showed that by adding a set of extra instrumental variables to those
employed by the FDGMM estimation, the so-called system GMM (SGMM) estimator improves
the precision and reduces possible finite sample bias of the FDGMM estimates. The SGMM is
also particularly suitable for growth regressions (Bond et al., 2001). We therefore estimated the
regressions using the SGMM estimation procedure.
A concluding remark regarding the sources of the variables used is in order10. Data on ICT
variables were derived from the International Telecommunication Union's ICT indicators
database. Data on macroeconomic variables were obtained from the World Bank's World
Development Indicators database; we used the World Bank's World Governance Indicators
database for the political stability index. The demographic variable was extracted from the
International Monetary Fund's World Economic Outlook database. The education variable was
developed by Barro and Lee (2013). The conflicts measurement was computed based on the
Uppsala Conflict Data Program (UCDP)/International Peace Research Institute Oslo (PRIO)
armed conflict dataset.
4. Results
The empirical setting enabled us to examine the impact of each of the 2 ICT variables on
economic growth in the AME region as a whole, as well as in each of the 2 AME sub-regions.
Specifically, in the first sub-section, we answer our first 2 questions by successively examining
the impact (1) Internet diffusion and (2) investments in the telecom sector have had on growth in
the AME region as a whole. In the second sub-section we provide an answer to our third question
10
We investigated the time series properties of the data (Appendix, Table 2) both in levels and first differences.
Results pertaining to the variables in levels suggest that most of the latter are stationary, with the exception of GDP
per capita and the population variable. No conclusive results were found for the terms of trade, openness, and
Internet variables. Findings pertaining to the variables in first differences strongly suggest stationarity.
9
by inspecting whether, respectively, the (1) Internet and (2) telecom investments have had a
different effect across the 2 AME subgroups.
Throughout the sub-sections we outline only the key results, leaving the interpretation of the
central findings to the discussion section.
10
countries, the latter two assessed the impact of the broadband penetration rate on growth and
found it to be positive11.
Among the group of additional variables, the government expenditures seem to be the only one
with a significant, and negative, impact on growth in the AME region12. In the cross-country
empirical growth literature, several authors have found a negative impact of public spending on
growth (Grier and Tullock, 1989; Ghura, 1995; Folster and Henrekson, 2001). These results can
be explained by the inefficiency of public investments often coupled with a misallocation of
resources toward the "wrong" sectors. This argument is mostly relevant to the AME countries
where the share of public investment to total investments is particularly high (Sala i Martin and
Artadi, 2003).
Examining the growth impact Internet diffusion has had in other regions relative to the AME
countries, the results suggest a positive and significant impact on OECD countries across all
specifications13. The findings also suggest a positive, albeit insignificant, effect of the Internet
penetration rate on growth in Latin American and Asian countries. There is some evidence of a
small negative impact of Internet use on growth in Africa and non-OECD European countries.
11
These papers are the only ones to have used non-composite ICT measures, which makes their findings particularly
relevant to our study. It should be noted that the ITU (2012) did not find the positive impact to be statistically
significant.
12
For some of the additional variables, such as the conflicts regressor, the insignificant impact most likely reflects
the incapacity of the latter to capture the intensity of conflicts. In the case of other variables, such as political
stability, this could be explained by an indirect impact on growth via other regressors.
13
The growth impact of the Internet on a given region is computed by adding the coefficient on the Internet and that
on the "interaction" term between the Internet and the said region. For instance, the impact of the Internet on growth
in OECD countries ranges from 0.004 (specification 2) to 0.01 (specifications 1 and 6). For clarity, we round off the
estimated coefficient to two decimals.
11
Table 1: Estimates of the impact of the Internet, baseline and augmented specifications
Regressor S1 S2 S3 S4 S5 S6
ln(yi, t- ) -0.04*** -0.02** -0.03*** -0.03*** -0.03** -0.03**
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
Education -0.03 0.0007 0.001 -0.0001 0.001 -0.008
(0.02) (0.01) (0.01) (0.01) (0.01) (0.02)
ln(Investment) 0.21*** 0.21*** 0.21*** 0.22*** 0.20*** 0.21***
(0.03) (0.02) (0.03) (0.03) (0.03) (0.03)
Population -0.02** -0.02*** -0.02** -0.02** -0.02** -0.02**
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
ln(Internet) 0.04* 0.03* 0.03* 0.03** 0.04** 0.04**
(0.02) (0.02) (0.02) (0.01) (0.02) (0.02)
ln(Internet)*Africa -0.04 -0.04* -0.04 -0.04* -0.05* -0.04
(0.03) (0.02) (0.03) (0.02) (0.02) (0.04)
ln(Internet)*Asia 0.008 -0.002 0.01 0.007 -0.005 0.004
(0.02) (0.01) (0.02) (0.01) (0.02) (0.01)
ln(Internet)*Europe -0.04** -0.03** -0.03** -0.03*** -0.04*** -0.04**
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
ln(Internet)*L.America 0.008 0.008 0.01 0.007 0.01 0.001
(0.02) (0.01) (0.02) (0.02) (0.02) (0.02)
ln(Internet)*OECD -0.02* -0.03*** -0.03** -0.03*** -0.04*** -0.02*
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
Gov. expenditures - -0.004** - - - -
(0.00)
Openness - - -0.0001 - - -
(0.00)
Terms of trade - - - 0.0001 - -
(0.00)
Political stability - - - - -0.003 -
(0.01)
Conflicts - - - - - 0.01
(0.00)
Number of countries 93 93 93 92 93 93
Observations 260 260 260 247 260 260
AB (p.value) 0.31 0.62 0.33 0.18 0.34 0.33
Hansen (p.value) 0.47 0.63 0.50 0.55 0.48 0.51
Difference Hansen 0.76 0.90 0.81 0.76 0.65 0.62
(p.value)
Note: i) numbers between parentheses are standard errors of the estimated coefficients; ii) standard errors are robust to cross
sectional heteroskedasticity and to within panel serial correlation, they are also finite sample corrected; iii) asterisks (***), (**),
and (*) denote p values equal to or inferior to 1%, 5% and 10%, respectively; iv) time specific effects are not reported; v) we
consider the time effects as well as the conflicts variables as exogenous, the rest of the variables are treated as endogenous,
except for the lagged value of GDP per capita, education and the terms of trade considered as predetermined; vi) in the first
differenced equation, lags of one period and earlier are used as instruments for the predetermined variables, while lags of two
periods and earlier are used as instruments for the endogenous variables; vii) in the levels equation, the first difference of the
variable lagged one period is used as instrument for the endogenous variable, while the first difference of the exogenous (or
predetermined) variable is used as instrument for the exogenous (or predetermined) variable; viii) the AB test has the following
null hypothesis: "absence of serial correlation of order 2 and higher orders" in the errors of the first differenced equation, the
overidentification test is based on Hansen J statistic, and has the following null hypothesis: "all instruments are valid" (i.e.
exogenous), the difference in Hansen test has the standard null hypothesis of an overidentification test but focuses on the extra
instruments used by the SGMM estimator; ix) the AME countries constitute the base group.
12
4.1.2. The impact of telecom investments
Column 2 of Table 2 shows that the standard augmented Solow model variables have the
expected signs, with a significant impact of investment (net of telecom investments) and
population growth.
Investments in the telecommunication sector do not seem to have significantly affected growth in
the AME region, exhibiting a negative coefficient. The results are rather robust to the inclusion
of the additional variables as documented in the remaining columns of the table: specifically, the
economic impact of telecom investments remains insignificant regardless of the specification,
with almost always a negative coefficient. Badran (2012) tested the effect of telecom investments
on growth in the AME region over the period 1998-2008 and found a similar insignificant effect.
The two regions where investments in the telecom sector seem to have positively and
significantly impacted growth relative to the AME region are Europe and the OECD countries.
The validity of the findings of the various estimated specifications largely depends on the three
aforementioned tests. In Table 1 and throughout all the specifications, the AB test suggests the
absence of autocorrelation among the errors; this implies that appropriate lagged values of the
endogenous variables can be used as instruments. Moreover, in all specifications, the
overidentification test does not reject the null hypothesis of suitable instruments. Finally, the
difference in the Hansen test points to the appropriateness of the additional instruments
employed under the SGMM estimation. Taken together, the tests suggest that our estimates in
Table 1 should not suffer from serious bias and inconsistency. The same is true regarding Table
2, with the exception of specifications 2 and 4, where the AB test implies that the errors are
serially correlated, suggesting that the instrumentation scheme is unsuitable.
13
Table 2: Estimates of the impact of the investment in telecom, baseline and augmented
specifications
Regressor S1 S2 S3 S4 S5 S6
lnyi, t- -0.009 -0.01 -0.005 -0.02 -0.02 -0.004
(0.02) (0.02) (0.02) (0.02) (0.02) (0.02)
Education 0.003 0.006 0.01 0.01 0.01 0.004
(0.02) (0.02) (0.01) (0.01) (0.01) (0.02)
ln[Investment (net of 0.12* 0.15** 0.13*** 0.15*** 0.09 0.11*
investment in telecom)] (0.06) (0.07) (0.05) (0.06) (0.06) (0.07)
Population -0.01* -0.01 -0.01 -0.02* -0.02 -0.01*
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
ln(Telecom) -0.02 -0.0008 -0.005 -0.06 -0.02 0.01
(0.08) (0.06) (0.07) (0.08) (0.07) (0.07)
ln(Telecom)*Africa 0.09 0.08 0.08 0.10 0.08 0.06
(0.09) (0.09) (0.09) (0.09) (0.08) (0.07)
ln(Telecom)*Asia 0.04 0.01 0.01 0.07 0.02 0.002
(0.09) (0.07) (0.08) (0.09) (0.07) (0.07)
ln(Telecom)*Europe 0.20* 0.18* 0.22 0.25 0.25* 0.17
(0.11) (0.11) (0.16) (0.18) (0.14) (0.11)
ln(Telecom)*L.America -0.02 -0.03 -0.03 0.006 -0.03 -0.06
(0.07) (0.07) (0.06) (0.07) (0.06) (0.07)
ln(Telecom)*OECD 0.14*** 0.11* 0.15*** 0.16*** 0.14*** 0.11**
(0.05) (0.06) (0.05) (0.06) (0.06) (0.05)
Gov. expenditures - -0.001 - - - -
(0.00)
Openness - - -0.0003 - - -
(0.00)
Terms of trade - - - -0.00 - -
(0.00)
Political stability - - - - 0.02 -
(0.01)
Conflicts - - - - - 0.008
(0.01)
Number of countries 92 92 92 91 92 92
Observations 287 287 287 251 287 287
AB (p.value) 0.07 0.04 0.05 0.03 0.05 0.08
Hansen (p.value) 0.17 0.11 0.23 0.06 0.12 0.39
Difference Hansen 0.38 0.27 0.54 0.12 0.33 0.72
(p.value)
Note: i) the note of table 1 applies with this difference: in the levels equation, lags of three periods (instead of two periods and
earlier) are used as instruments for the telecom variable.
14
4.2. The economic impact of ICT on AME sub-regions
After investigating the economic repercussions of the ICT variables on the AME region as a
whole, we now examine, in turn, the impact of Internet penetration and investment in telecom on
each of the two AME sub-regions. Table 3 shows the results of the regressions where the Internet
is the ICT measurement, while Table 5 presents the findings of the regressions where investment
in telecom is used. The two tables have the same structure as Tables 1 and 2. In both tables, the
estimated coefficient of the ICT variable refers to the base group: high-income AME countries.
14
Among the high-income AME countries included in the sample, Bahrain is the country that registered the lowest
average share of fuel exports in total merchandise exports during 1995-2010, with an average of "only" 71.7% (the
World Bank's World Development Indicators database).
15
In 2010, the share of extractive industries in GDP was 24.4% in Bahrain, 55.7% in Qatar, 47.8% in Saudi Arabia,
and 31.6% in the UAE. The same year, the share of manufacturing industries in GDP was 17.1%, 7.3%, 10%, and
9.7% in the same countries, respectively (the Arab Monetary Fund, 2011).
15
Table 3: Estimates of the impact of the Internet, baseline and augmented specifications
Regressor S1 S2 S3 S4 S5 S6
ln(yi, t- ) -0.04** -0.02 -0.03** -0.03 -0.03 -0.03*
(0.02) (0.02) (0.01) (0.02) (0.02) (0.02)
Education -0.002 -0.0005 0.001 0.0005 -0.001 -0.001
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
ln(Investment) 0.22*** 0.21*** 0.22*** 0.21*** 0.21*** 0.21***
(0.03) (0.03) (0.03) (0.03) (0.03) (0.03)
Population -0.02** -0.02** -0.02** -0.02** -0.02** -0.02**
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
ln(Internet) 0.04 0.02 0.04* 0.03 0.03 0.04
(0.03) (0.03) (0.02) (0.03) (0.03) (0.02)
ln(Internet)*Africa -0.05 -0.03 -0.04 -0.04 -0.04 -0.04
(0.03) (0.03) (0.03) (0.03) (0.03) (0.03)
ln(Internet)*Asia 0.004 0.001 0.008 0.01 -0.002 0.0005
(0.02) (0.02) (0.02) (0.02) (0.02) (0.02)
ln(Internet)*Europe -0.04** -0.03 -0.04*** -0.03* -0.04** -0.03*
(0.01) (0.02) (0.01) (0.01) (0.02) (0.02)
ln(Internet)*L.America 0.006 0.01 0.003 0.007 0.01 0.006
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
ln(Internet)*OECD -0.03** -0.02 -0.03** -0.02* -0.03** -0.03*
(0.01) (0.02) (0.01) (0.01) (0.01) (0.01)
ln(Internet)* middle -0.01 0.004 -0.01 0.004 -0.003 -0.009
income AME (0.02) (0.03) (0.02) (0.02) (0.03) (0.02)
Gov. expenditures - -0.003* - - - -
(0.00)
Openness - - -0.0002 - - -
(0.00)
Terms of trade - - - 0.0002 - -
(0.00)
Political stability - - - - -0.001 -
(0.01)
Conflicts - - - - - 0.009
(0.00)
Number of countries 93 93 93 92 93 93
Observations 260 260 260 247 260 260
AB (p.value) 0.33 0.76 0.38 0.22 0.43 0.43
Hansen (p.value) 0.51 0.68 0.54 0.51 0.57 0.59
Difference Hansen 0.76 0.89 0.72 0.58 0.80 0.73
(p.value)
Note: i) the note of table 1 applies with high income AME countries as the base group.
16
For the remaining specifications, the Internet effect remains positive but insignificant. Among
the additional regressors, the government expenditures variable seems to be the only one that
exerted a significant impact on economic performance with a negative, but very small,
coefficient.
We can derive the estimated Internet impact on the middle-income AME subgroup by adding the
coefficient of the Internet variable to the "interaction" term between the latter and the middle-
income AME regional dummy. Across all the specifications, the results reveal that Internet
diffusion in the middle-income AME countries has had a positive, albeit insignificant, impact on
their growth. Interestingly and quasi-systematically, the estimated Internet impact on middle-
income AME countries is smaller than those obtained for the AME region as a whole and the
high-income AME sub-region, varying between 0.02 and 0.03.
Combining the two findings, our analysis suggests that the positive and significant effect of the
Internet penetration rate on the AME region stems mainly from high-income AME countries.
Figure 1 in the Appendix sheds light on Internet diffusion in the two AME subgroups and
exposes very low penetration rates in middle-income AME countries, with an average over the
1995-2013 period of nearly 2%, the second lowest rate after that exhibited by African countries
and well below the entire sample average. In contrast, Internet use is largely prevalent in the
high-income AME subgroup, with the second highest average penetration rate (6%) after the
OECD countries. Our findings could thus reflect the Internet diffusion gap between the two
AME subgroups whereby high-income countries seem to have crossed the penetration rate
needed for the Internet to start impacting economic performance, while the penetration rate in the
middle-income group has yet to reach critical mass.
To check this hypothesis, we separated countries into two sub-samples: countries with a low
Internet penetration rate (equal to and less than 10%), and countries with a high penetration rate
(above 10%)16. Using each of the 2 subsamples separately, we then estimated an equation where
the growth rate is regressed on the lagged GDP per capita, the three steady-state GDP per capita
determinants, and the Internet penetration rate. The results reported in the next table show that
Internet diffusion significantly affects growth in countries with high penetration rates, while its
impact is positive but insignificant in countries with low rates. These findings imply that the
16
This is done using the Internet penetration rate prevailing in a given country during the last year (with available
data on the penetration rate) in the last sub-period.
17
critical mass falls within the group of countries with high Internet penetration rates, and that
countries with penetration rates below 10% would not experience a significant growth impact of
Internet use.
Turning the spotlight back on our findings, the hypothesis that the Internet diffusion gap could
explain why the significant impact of the penetration rate in the AME region is mostly driven by
its effect in high-income AME countries seems highly plausible: most of the latter countries fall
within the group of high Internet penetration rates, whereas the average penetration rate in
middle-income AME countries is below 10%.
18
standard growth determinants continue to have the expected signs, and population and
investment net of investments in telecom have almost continuously highly significant
parameters. Openness is the only additional variable with a significant, but very small, impact on
growth.
Table 5: Estimates of the impact of the investment in telecom, baseline and augmented
specifications
Regressor S1 S2 S3 S4 S5 S6
lnyi, t- -0.01 -0.01 -0.006 -0.02 -0.02 -0.01
(0.02) (0.02) (0.01) (0.01) (0.01) (0.02)
Education 0.004 0.005 0.01 0.01 0.006 0.004
(0.02) (0.02) (0.01) (0.01) (0.01) (0.02)
ln[Investment (net of 0.12** 0.13** 0.12** 0.12** 0.10 0.11*
investment in telecom)] (0.06) (0.06) (0.05) (0.06) (0.06) (0.06)
Population -0.02** -0.02* -0.02** -0.02** -0.02** -0.02**
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
ln(Telecom) -0.09 -0.09 -0.07 -0.10 -0.11 -0.08
(0.09) (0.08) (0.08) (0.08) (0.09) (0.09)
ln(Telecom)*Africa 0.15 0.14 0.13 0.13*** 0.17* 0.15*
(0.09) (0.09) (010) (0.04) (0.10) (0.09)
ln(Telecom)*Asia 0.09 0.10 0.08 0.11 0.11 0.09
(0.09) (0.08) (0.08) (0.09) (0.09) (0.09)
ln(Telecom)*Europe 0.22** 0.23*** 0.24* 0.26** 0.28* 0.23**
(0.10) (0.08) (0.13) (0.12) (0.16) (0.10)
ln(Telecom)*L.America 0.01 0.03 0.03 0.03 0.04 0.004
(0.09) (0.07) (0.08) (0.09) (0.08) (0.09)
ln(Telecom)*OECD 0.18** 0.18** 0.21** 0.21** 0.20** 0.18**
(0.08) (0.08) (0.09) (0.08) (0.09) (0.08)
ln(Telecom)*middle 0.05 0.09 0.12 0.07 0.08 0.12
income AME (0.10) (0.06) (0.10) (0.10) (0.09) (0.12)
Gov. expenditures - -0.001 - - - -
(0.00)
Openness - - -0.0007** - - -
(0.00)
Terms of trade - - - -0.00 - -
(0.00)
Political stability - - - - 0.01 -
(0.01)
Conflicts - - - - - 0.01
(0.01)
Number of countries 92 92 92 91 92 92
Observations 287 287 287 251 287 287
AB (p.value) 0.05 0.07 0.10 0.03 0.05 0.18
Hansen (p.value) 0.19 0.11 0.47 0.21 0.14 0.35
Difference Hansen 0.35 0.25 0.77 0.32 0.38 0.60
(p.value)
Note: i) the note of table 2 applies with high income AME countries as the base group.
19
Applying the same procedure described earlier, we can deduce the impact of telecom
investments on the middle-income AME region: investment in the telecommunication sector
does not seem to have significantly contributed to growth in the middle-income AME countries
as reported by the systematic insignificance of the coefficient of the telecom variable and that of
the interaction term between the latter and the middle-income AME sub-group.
Overall, our results show that telecom investments did not impact growth significantly in either
one of the two AME subgroups, reflecting what we found for the AME region as a whole. We
note that the battery of tests reported in Tables 3, 4, and 5 suggest that our estimates are
capturing the exogenous impact of the independent variables on economic growth, with the
exception of specification 4 in Table 5 with a relatively low P value of the AB test, implying that
the instrumentation procedure is not valid.
5. Discussion
Two main results emerge from our empirical analysis of the effect of ICT on growth in the AME
countries. The first is that the use of the Internet has positively contributed to the economic
performance of the AME region. Our sub-regional breakdown suggests that this mainly reflects
the case of high-income AME countries: the positive effect of the Internet remains indeed
insignificant in middle-income AME countries given the relatively low penetration rates in the
latter. The second result underlines the inability of the investments in the telecommunication
sector to affect the economic outcome in the AME region in a significant way, regardless of the
sub-region considered.
In the next sub-section, we expand upon the first finding by putting forward some plausible
reasons that would explain the feeble Internet diffusion in the middle-income AME region. In the
second sub-section, we provide some insights into the insignificant economic effect of telecom
investments in the AME countries.
5.1. Internet diffusion determinants and impact of the Internet penetration rate
The Internet diffusion gap between the two AME sub-regions can be explained by the literature
on the factors impacting the Internet diffusion process. By using cross-country regression
analysis, Andres et al. (2009), Chinn and Fairlie (2007), Mocnik and Sirec (2010), and Wunnava
and Leiter (2008) have singled out a number of important determinants of the Internet
penetration rate that can be regrouped into six categories of variables: economic, education,
20
demographic, ICT infrastructure, ICT cost, and institutional environment variables. Most of
those variables were also highlighted in the few papers devoted to assess the factors affecting the
spread of the Internet in the AME countries (Al Hammadany and Heshmati, 2011; Emdad et al.,
2009). The most robust findings of the econometric analysis show that real GDP per capita and
the employment rate positively affect Internet diffusion. Moreover, the level of English
proficiency among the population is a key factor driving the Internet penetration rate. Further,
the share of the urban population in the total population is an important determinant of Internet
use. A developed and continuously upgraded ICT infrastructure, along with low ICT
equipment/installation costs, is another propeller of Internet spread. Finally, a regulatory
framework fostering the private sector is also found to positively impact Internet use.
Table 6 compares the two AME sub-regions based on a number of variables across the six
aforementioned categories.
21
Table 6: Internet use determinants across the two AME sub regions, (1995-2014) averages
22
telephone service ($) Union, ICT database
Price of a 3 minute call to a fixed 0.06 0.09
telephone line ($)
Mobile prepaid price of a 1 minute 0.14 0.12
local call to a subscriber of the same
network ($)
Mobile prepaid price of a 1 minute 0.2 0.12
local call to a subscriber of a
different network ($)vi
Fixed broadband connection charge 134.2 49.1
($)vi
Fixed broadband monthly 45.4 34.4
subscription charge ($)vi
Institutional environmentvii Control of corruption index -0.5 0.4 World Bank, World Governance
Regulatory quality index -0.6 0.3 Indicators
Rule of law index -0.5 0.5
Note: i) relatively to other sectors, services are particularly ICT dependent; ii) educational variables should reflect the degree of English proficiency among the population; iii) the
amount of annual remittances paid is a proxy for the stock of foreigners working in the domestic economy. The larger stock of foreigners in high income AME countries is likely
to have positively impacted the Internet use via several channels: unlike the case of middle income AME countries, a large part of the immigrants living in high income AME
countries are well educated with a good proficiency in English, and most of the well educated foreigners are employed in the ICT dependent services sector; iv) variables averaged
during the period spanning the late 1990s-2014; v) variables averaged during the period (1995-2013); vi) variables averaged during the period (2007-2014); vii) each of the
institutional environment indicators ranges from approximately -2.5 (weak institutional environment) to 2.5 (good institutional environment); viii) middle income AME countries
are as reported in table 1 in the appendix in addition to Iraq and Lebanon, whereas high income AME countries are as reported in the same table with the addition of Oman.
23
The figures speak for themselves: compared to their high-income neighbors, middle-income
AME countries display a considerable deficit in terms of the economic, education, demographic,
and ICT-related variables. Interestingly, Table 6 also exposes a significant wedge between the
two subgroups in terms of the institutional environment: throughout the period covered, low-
income AME countries have continuously had weaker regulatory environments. Foreign
investors are reluctant to enter and operate in the telecom sector when the overall business
environment is tainted with weak judicial and institutional systems, an omnipresent public sector,
and other relatively closed and inefficient backbone services. This is likely to have prevented a
larger diffusion of the Internet in middle-income AME countries: explaining the factors behind
the Internet penetration gap between AME countries and the US, Chinn and Fairlie (2007) found
that nearly one-third of the gap is due to differences in the regulatory environment.
In sum, the deficit in terms of the propellers of Internet diffusion is likely to be the origin of the
low Internet penetration rates exhibited in the middle-income AME sub-region and their
insignificant economic impact.
5.2. The GPT nature of the telecom sector and the impact of telecom investments
Table 3 in the Appendix shows that the AME region fares relatively well in terms of investments
in telecommunication (as a share of GDP), especially in middle-income countries compared to
the rest of the regions. Yet those investments did not manage to become a significant determinant
of growth in the AME countries throughout the examined period. As suggested by the literature
review, a possible explanation could lie in the GPT aspect of ICT investments. Indeed, such
investments could first have an adverse impact on productivity and output during a transitory
period of time when the firm-level management setting and resources are not yet appropriately
organized/equipped to reap the potential benefits of ICT investments. The positive impact of ICT
investments would be realized only when the adjustment period leads to the consolidation of all
required complementary organizational changes at the firm and industry levels. This "cycle"
describing the economic impact of ICT capital was documented in the US, the United Kingdom,
and many European countries (Liao et al., 2016; Ceccobelli et al., 2012; Basu et al., 2003; van
Ark and Inklaar, 2005). The length of the adjustment period, whereby ICT investments adversely
affect output and productivity growth, depends on firm-level as well as macro-level factors: at
the firm level, it depends on the capacity of producing entities to efficiently "absorb" the new
technologies into the management and production processes; at the national level, it is affected
24
by the degree of competitiveness of the markets (van Ark and Inklaar, 2005), notably the
telecommunication sector, in particular its openness to foreign service providers (OECD, 2009;
Borchert et al., 2012b). The firm-level absorptive capacity mainly depends on the education level
and ICT skill of the workers (Basu and Fernaldm, 2007; Liao et al., 2016; Brynjolfsson and Hitt,
2000), whereas the competitiveness of the economy, notably the telecom sector, is largely
affected by the regulatory framework governing the markets (van Ark and Inklaar, 2005,
Borchert et al., 2012a; Rossotto et al., 2003).
The insignificant (and negative) effect of telecom investments on growth in the AME region
could likely reflect the adjustment period that is needed before the latter investments start
positively affecting output growth. Indeed, the main complementary elements that are needed for
a positive impact of ICT investments seem to be lagging in the AME countries. On the one hand,
as shown in the following figure, although a considerable improvement has occurred in recent
years, the AME region still lags relative to Asia and Latin America in terms of the average years
of total schooling.
10
1995-1999
6 2000-2004
2005-2009
4 2010-2012
0
Africa AME Asia Latin OECD
America
Note: precisely, it is the average years of primary, secondary and tertiary schooling among population aged 15 and above.
Source: Barro and Lee (2013) database.
25
On the other hand, as a proxy for the ICT skills of the labor force, Figure 2 shows the average
share over the period (1995-2014) of high technology exports in manufactured exports as well as
the average share over the same period of ICT goods exports of total goods exported in the AME
and other regions in the world.
Figure 2: Share of high technology exports in manufactured exports, and share of ICT
exports in total goods exports, average during (1995-2014)
25
20
15
High-tech exports (% of
manufactured exports)
0
Africa AME Asia Latin OECD
America
Note: i) high technology exports are products with high research and development intensity; ii) information and communication
technology goods exports mainly include computers and peripheral equipment, communication equipment, consumer electronic
equipment, and electronic components.
Source: the World Bank's World Development Indicators, available at
http://databank.worldbank.org/data/reports.aspx?source=world-development-indicators#.
Clearly, a significant gap prevails between the AME countries and the rest of the regional groups
(except for Africa in terms of ICT exports). Thus, the two previous figures point to a major
obstacle likely hampering the absorptive capacity of the private sector in the AME countries: a
26
relatively low level of overall education of the workforce, and a severe lag in terms of the latter's
ICT dexterity17.
The regulatory environment in the AME region adds to this constraint, as it is one of the most
burdensome business environments in the world. Figure 3 presents for various regions two
competition indices reflecting the extent of domestic competition as well as foreign competition
based on the World Economic Forum's Global Competitiveness Report (2015).
3 Domestic competition
Foreign competition
0
Africa AME Asia Latin OECD
America
Note: i) the higher the indicator, the greater the degree of competition; ii) the domestic competition indicator mainly reflects the
intensity of local competition and the ease of doing business; iii) the foreign competition indicator chiefly reflects the prevalence
of trade barriers and the openness towards FDI inflows.
Source: the World Economic Forum's Global Competitiveness database, available at http://reports.weforum.org/global-
competitiveness-report-2015-2016.
The AME region fares relatively well in terms of domestic competition but shares the last
position (with Latin America) with regard to foreign competition. In particular, middle-income
17
A possible interesting research agenda would be to extend the analysis of ICT investments as GPT in AME
countries at the firm/industry levels and assess the interaction of the workers’ educational level, notably in ICT
expertise, with ICT capital.
27
AME economies are the least open to foreign competition in the world, given the prevalence of
trade barriers and the restrictiveness towards FDI inflow (World Economic Forum, 2015).
Despite the implementation of structural reforms in several middle-income AME countries
starting in the mid-1980s, the latter still suffer from an overwhelming presence of the state in the
economy and a lack of pro-competition institutional mechanisms (Morsy and Zouk, 2013).
The stringent regulations in the AME region are likely to have reduced the returns of investments
in the telecom sector, because they inhibit the competitiveness of the markets and the
reallocation of resources to the most productive firms, those that can benefit the most from ICT
investments.
The restrictiveness of the overall institutional infrastructure in the AME countries is pervasive in
the telecom sector as well. Figure 4 presents the Services Trade Restrictiveness Index (STRI) of
the telecom (aggregating both fixed and mobile lines) developed by the World Bank for a
number of regions.
50
45
40
35
30
25
20
15
10
5
0
Africa AME Asia Latin OECD
America
Note: i) the higher the STRI, the more closed is the sector with respect to foreign service providers; ii) data used to construct the
STRI was collected through the period (2008-2010).
Source: the World Bank's Services Trade Restrictions database, available at http://iresearch.worldbank.org/servicetrade.
Based on the overall STRI score, AME countries' policy regimes are characterized by the
presence of major restrictions, on par with Africa and Asia. The STRI scores of the high-income
28
AME region are extremely high, making it the area with the most restrictive telecom policies
among 103 countries covered in the database (Borchert et al., 2012b).
The World Bank's Services Trade Restrictions database reveals that several obstacles are still
hindering foreign investment in the telecom sector in the two AME sub-regions: limits on entry
(restrictions on the use of state-owned entities as well as limits on foreign ownership in equity
shares) and limits on operations (restrictions on the employability of foreigners, and restrictions
on establishing an international gateway as well as on the use of voiceover Internet protocol [IP]
technology).
The AME region's overall restrictive telecommunication is costly in terms of untapped potential
of the telecom investments, as documented by a number of authors18. Rossotto et al. (2003)
showed that a fully liberalized telecom sector in AME countries would result in higher mobile,
fixed, and Internet penetration rates, as well as an increase in revenues generated by the
telecommunication sector. This is particularly telling, as the Internet and broadband penetration
rates in the AME region are comparatively low, with relatively feeble revenues generated from
the telecom sector and high broadband service charges (Appendix, Table 3).
18
We could not empirically test the effect of competition in telecom on the economic impact of investments in
telecommunication since the STRI indicators are only available for 1 year while our empirical setting is a panel data
one.
29
sector on the AME region is likely to reflect the GPT nature of ICT investments with an initial
adjustment period. The latter could extend for some time in the AME region given the weak ICT
absorptive capacities of the private sector as well as the deleterious overall business climate and
the stringent regulatory framework governing the telecom industry.
Investing in education remains a necessity for the AME countries and requires measures to
increase the enrollment rates in primary and secondary education as well as reforms to enhance
the quality of education. This should equip students with the analytical and technical skills that
are of primary importance in the knowledge economy. This would help enhance the digital
literacy of the labor force, ultimately speeding up the adjustment period to reap the awaited
dividends of telecom investments in the region. In middle-income AME countries, this would
also help achieve the critical penetration level that will enable them to benefit more substantially
from Internet diffusion.
Consolidating the rule of law, actively combating corruption, and delivering efficient public
services should also be a reform priority for AME countries. Such measures are likely to enhance
the competitiveness of the business environment, thus promoting complementary investments
that would boost the economic effect of ICT investments. These measures will undoubtedly
differ across countries according to a set of political economy parameters. It is likely, however,
that a key element for the success of these endeavors in any country is a more proactive
engagement from all the stakeholders: the political elite, the private sector, and civil society.
Boosting the competitiveness of the telecom sector in its two components, fixed and mobile
telephony, is primordial for the AME countries. This should primarily facilitate a foreign
commercial presence in the sector by operating along the following two axes: (i) lifting the
restrictions on entry, mainly by allowing new entrants to use the networks of typically state-
owned incumbents as well as loosening the limits on foreign equity shares, and (ii) relaxing
employability conditions and allowing newcomers to own and use an international gateway as
well as freeing their ability to route their traffic via voice-over IP technology. Such measures are
expected to expand the telephony networks as well as Internet penetration rates. Furthermore,
these policies should translate into lower ICT costs and better and faster services with spillover
effects affecting the rest of the industries and the economy as a whole.
30
This research did not receive any specific grant from funding agencies in the public, commercial,
or not for profit sectors.
We are thankful for the reviewers' comments and suggestions that substantially enriched the
article, in particular the interpretation of the results. The usual disclaimers apply.
References
Abdel Kader, K., 2006. The impact of information and communication technology on economic
growth in MENA countries, Working Paper No. 2006/31, the European University Institute:
Florence.
Al Hammadany, F., Heshmati, A., 2011. Determinants of Internet use in Iraq. International
Journal of Communication. 5(2011), 1967-1989.
Andres, L., Cuberes, D., Astou Diouf, M., Serebrisky, T., 2009. Diffusion of the Internet: a
cross-country analysis, Policy Research Working Paper No. 4420, the World Bank: Washington
DC.
Arab Monetary Fund, 2011. Joint Arab Economic Report (in Arabic). The Arab Monetary Fund,
Abu Dhabi, the UAE, available at http://www.amf.org.ae/ar/jointrep, (accessed on 25.06.2016).
Badran, M., 2012. The impact of broadband infrastructure on economic growth in some Arab
and emerging countries. Topics in Middle Eastern and North African economies. 14, 278-310.
Barro, J., Lee, JW., 2013. A new data set of educational attainment in the world. Journal of
Development Economics. 104, 184-198.
Basu, S., Fernaldm, J., 2007. Information and communications technology as a general-purpose
technology: evidence from the US industry data. German Economic Review. 8(2), 146-173.
Basu, S., Fernaldm, J., Oulton, N., Srinivasan, S., 2003. The case of the missing productivity
growth, or does the information technology explain why productivity accelerated in the United
States but not in the United Kingdom? In: Gertler, Mark., Rogoff, kenneth. (Eds.), the NBER
Macroeconomics annual 2003. MIT press, Cambridge, pp. 9-63.
31
Blundell, R., Bond, S., Windmeijer, F., 2000. Estimation in dynamic panel data models:
improving on the performance of the standard GMM estimator, Working Paper No. 00/12, the
Institute for Fiscal Studies: London.
Bond, S., Hoeffler, A., Temple, J., 2001. GMM estimation of empirical growth models,
Discussion Paper No. 3048, the Center for Economic Policy Research: London.
Borchert, I., Gootiiz, B., Grover, A., Mattoo, A., 2012a. Landlocked or policy locked? How
services trade protection deepens economic isolation, Policy Research Working Paper No. 5942,
the World Bank: Washington DC.
Borchert, I., Gootiiz, B., Mattoo, A., 2012b. Policy barriers to international trade in services:
evidence from a new database, Policy Research Working Paper No. 6109, the World Bank:
Washington DC.
Brynjolfsson, E., Hitt, L., 2000. Beyond computation: information technology, organizational
transformation, and business performance. Journal of Economic Perspectives. 14(4), 23-48.
Caselli, F., Esquivel, G., Lefort, F., 1996. Reopening the convergence debate: a new look at
cross-country growth empirics. Journal of Economic Growth. 1(3), 363-389.
Ceccobelli, M., Gitto, S., Mancuso, P., 2012. ICT capital and labor productivity growth: a non-
parametric analysis of 14 OECD countries. Telecommunications Policy. 36(2012), 282-292.
Chinn, M., Fairlie, R., 2007. The determinants of the global digital divide: a cross-country
analysis of computer and Internet penetration. Oxford Economic Papers. 59(2007), 16-44.
Czernich, N., Falck, O., Kretschmer, T., Woessmann, L., 2011. Broadband infrastructure and
economic growth. Economic Journal. 121(552), 505-532.
Economic and Social Commission for Western Asia (ESCWA), 2013. Regional profile of the
information society in the Arab region, the Economic and Social Commission for Western Asia,
Beirut, Lebanon, available at https://www.unescwa.org/publications/regional-profile-
information-society-arab-region-2013, (accessed on 15.08.2016).
32
Emdad, A., Badamas, M., Mouakket, S., 2009. Factors and impacts of low utilization of Internet:
the case of Arab countries. Journal of International Technology and Information Management.
18(3), 299-320.
Folster, S., Henrekson, M., 2001. Growth effects of government expenditure and taxation in rich
countries. European Economic Review. 45(8), 1501-1520.
Gajek, M., Kretschmer, T., 2012. Estimating critical mass in the global cellular telephony
market. International Journal of Industrial Organization. 30(6), 496-507.
Ghura, D., 1995. Macro policies, external forces, and economic growth in Sub-Saharan Africa.
Economic development and cultural change. 43(4), 759-778.
Giesecke, J., 2006. The national and regional economic consequences of rapid growth in
Australia's telecommunications sector. Economic Analysis and Policy. 36(1-2), 61-97.
Grier, K.B., Tullock, G., 1989. An empirical analysis of cross national economic growth: 1951-
80. Journal of monetary economics. 24(2), 259-76.
Guetat, I., Drine, I., 2007. The information and communication technologies impact on the
MENA country growth performance. Paper presented at the 6th international conference of the
Middle East Economic Association, 14-16 March, Dubai, the United Arab Emirates.
Hassan, K., 2003. FDI, information technology and economic growth in the MENA region.
Paper presented at the 10th Economic Research Forum conference, 16-18 December, Marrakesh,
Morocco.
Hodrab, R., Maitah, M., Lubos, S., 2016. The effect of Information and Telecommunication
Technology on economic growth: Arab world case. International Journal of Economics and
Financial Issues. 6(2), 765-775.
[dataset] International Monetary Fund, 2016. World Economic Outlook. The International
Monetary Fund, Washington DC, the USA, available at
https://www.imf.org/external/pubs/ft/weo/2016/02/weodata/weoselgr.aspx, (accessed on
15.05.2016).
33
International Telecommunication Union (ITU), 2012. Impact of broadband on the economy:
research to date and policy issues, the International Telecommunication Union, Geneva,
Switzerland, available at https://www.itu.int/ITU-D/treg/broadband/ITU-BB-Reports_Impact-
of-Broadband-on-the-Economy.pdf, (accessed on 14.04.2016).
[dataset] International Telecommunication Union, 2015. The ICT indicators database. The
International Telecommunication Union, Geneva, Switzerland.
Islam, N., 1995. Growth empirics: a panel data approach. Quarterly Journal of Economics. 110
(4), 1127-1170.
Koutroumpis, P., 2009. The economic impact of broadband on growth: a simultaneous approach.
Telecommunications Policy. 33(9), 471-485.
Lee, S., Levendis, J., Gutierrez, L., 2012. Telecommunications and economic growth: an
empirical analysis of sub-Saharan Africa. Applied Economics. 44(4), 461-469.
Liao, H., Wang, B., Li, B., Weyman-Jones, T., 2016. ICT as a general-purpose technology: the
productivity of ICT in the United States revisited. Information Economics and Policy. 36 (2016),
10-25.
Mahboub, A., Salman, D., 2008. ICT, market contestability and economic performance: lessons
from ERF countries, Working Paper No. 424, the Economic Research Forum: Cairo.
Mocnik, D., Sirec, K., 2010. The determinants of Internet use controlling for income level: cross-
country empirical evidence. Information Economics and Policy. 22(2010), 243-256.
Morsy, H., Zouk, N., 2013. The unfinished business of structural reforms in the Southern and
Eastern Mediterranean countries. In: World Economic Forum and European Bank for
Reconstruction and Development (Publishers.), The Arab world competitiveness report 2013.
World Economic Forum, Geneva, pp. 38-43.
Organization for Economic Co-operation and Development (OECD), 2009. Testing the services
trade restrictiveness index: gravity regressions and trade costs analysis. Paper presented at the
OECD experts meeting on the services trade restrictiveness index (STRI), 2-3 July, Paris,
France.
34
Qiang, C., 2009. Broadband infrastructure investment in stimulus packages: relevance for
developing countries, Working Paper No. 49097, the World Bank: Washington DC.
Quiggin, J., 2014. National accounting and the digital economy. Economic Analysis and Policy.
44 (2014), 136-142.
Roodman, D., 2009. How to do xtabond2: an introduction to difference and system GMM. The
STATA journal. 9(1), 86-136.
Sala i Martin, X., Artadi, E., 2003. Economic growth and investment in the Arab world. In:
World Economic Forum (Publisher.), The Arab competitiveness report 2003. World Economic
Forum, Geneva, pp. 22-33.
Shahiduzzaman, Md., Alam, K., 2014a. The long-run impact of information and communication
technology on economic output: the case of Australia. Telecommunications Policy. 38 (2014),
623-633.
Shahiduzzaman, Md., Alam, K., 2014b. Information technology and its changing roles to
economic growth and productivity in Australia. Telecommunications Policy. 38 (2014), 125-135.
Tsang, F., Yaqub, O., van Welsum, D., Thomson-Starkey, T., Chataway, J., 2011. The impact of
information and communication technologies in the Middle East and North Africa, Rand Europe
Technical Report, available at
http://www.rand.org/content/dam/rand/pubs/technical_reports/2011/RAND_TR1163.pdf,
(accessed on 04.08.2016).
[dataset] Uppsala Conflict Data Program (UCDP), and International Peace Research Institute
Oslo (PRIO), 2015. Armed conflict dataset (version 4-2015), available at
http://ucdp.uu.se/downloads/, (accessed on 14.05.2016).
35
Van Ark, B., Inklaar, A., 2005. Catching up or getting stuck? Europe's troubles to exploit ICT's
productivity potential. Groningen Growth and Development Center, the University of
Groningen, available at http://www.euklems.net/pub/no7.pdf, (accessed on 28.04.2017).
[dataset] World Bank, 2011. Services Trade Restrictions database. The World Bank,
Washington DC, the USA, available at http://iresearch.worldbank.org/servicetrade, (accessed on
01.08.2016).
[dataset] World Bank, 2016. World Development Indicators. The World Bank, Washington DC,
the USA, available at http://databank.worldbank.org/data/reports.aspx?source=world-
development-indicators, (accessed on 25.05.2016).
[dataset] World Bank, 2016. World Governance Indicators. The World Bank, Washington DC,
the USA, available at http://databank.worldbank.org/data/reports.aspx?source=worldwide-
governance-indicators, (accessed on 8.05.2016).
World Economic Forum, 2015. The global competitiveness report 2015-2016, the World
Economic Forum, Cologny, Switzerland, available at http://reports.weforum.org/global-
competitiveness-report-2015-2016, (accessed on 18.07.2016).
Wunnava, P., Leiter, D., 2008. Determinants of inter-country Internet diffusion rates. Discussion
Paper No. 3666, the Institute for the Study of Labor: Bonn.
36
Appendix
Figure 1: Internet penetration rate (1996-2013) (in %), AME sub regions
10
Source: author's calculations based on the International Telecommunication Union, the ICT indicators 2015 database.
Note: middle income AME countries are as reported in table 1of the appendix in addition to Iraq and Lebanon, whereas high
income AME countries are as reported in the same table with the addition of Oman.
37
Figure 2: Investment in the telecommunication services (1995-2012) (in US $), AME sub
regions
2E+10
1.8E+10
1.6E+10
1.4E+10
1.2E+10
1E+10
8E+09
6E+09
4E+09
2E+09
Source: author's calculations based on the International Telecommunication Union, the ICT indicators 2015 database.
Note: the note of figure 1 applies.
38
Table 1: Regional grouping
Region Countries
Africa Botswana, Cameroon, Cote d'Ivoire, Gabon,
Gambia, Ghana, Kenya, Malawi, Mozambique,
Niger, Rwanda, Senegal, Sierra Leone, South
Africa, Tanzania, Togo, Uganda, Zimbabwe.
Asia Bangladesh, China, India, Indonesia, Malaysia,
Nepal, Pakistan, Philippines, Singapore, Sri Lanka,
Thailand, Vietnam.
AME Algeria, Bahrain, Egypt, Iran, Jordan, Kuwait,
Libya, Mauritania, Morocco, Qatar, Saudi Arabia,
Syria, Tunisia, Turkey, United Arab Emirates,
Yemen.
Europe Bulgaria, Cyprus, Malta, Romania.
Latin America Argentina, Barbados, Bolivia, Brazil, Chile,
Colombia, Costa Rica, Dominican Republic,
Ecuador, El Salvador, Guyana, Honduras,
Nicaragua, Panama, Paraguay, Peru, Uruguay,
Venezuela.
OECD Australia, Austria, Belgium, Canada, Denmark,
Finland, France, Germany, Greece, Hungary,
Iceland, Ireland, Italy, Japan, Mexico, Netherlands,
New Zealand, Norway, Poland, Portugal, South
Korea, Spain, Sweden, Switzerland, United
Kingdom, United States.
High income AME Bahrain, Kuwait, Qatar, Saudi Arabia, United Arab
Emirates.
Middle income AME Algeria, Egypt, Iran, Jordan, Libya, Mauritania,
Morocco, Syria, Tunisia, Turkey, Yemen.
Note: i) since Chile became a member of OECD in 2010, it was kept in "Latin America"; ii) although Turkey is an OECD
member, it was included in the "AME" region (its ICT infrastructure and regulations are comparable to the rest of the AME
countries); iii) the distinction between high income and middle income AME countries is based on the World Bank classification.
39
Table 2: Panel unit root tests
Variables in levels
Test lny ln(Investment) Pop Tot Open GE PS ln(Investment ln(Internet) ln(Telecom)
net of
investment in
telecom)
Maddala I(1) I(0) I(1) I(1) I(1) I(0) I(0) I(0) I(0) I(0)
& Wua
Maddala I(1) I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0)
& Wub
Im, I(1) I(0) I(1) I(1) I(1) I(0) I(0) I(0) - I(0)
Pesaran
& China
Im, I(1) I(0) I(1) I(0) I(0) I(0) I(0) I(0) - I(0)
Pesaran
& Chinb
Pesaran I(1) I(0) I(1) I(1) I(0) I(1) I(0) I(1) I(1) I(1)
(CIPS)a
Pesaran I(1) I(0) I(1) I(0) I(1) I(0) I(1) I(1) I(1) I(1)
(CIPS)b
Variables in first differences
Test Δlny Δln(Investment) ΔPop ΔTot ΔOpen ΔGE ΔPS Δln(Investment Δln(Internet) Δln(Telecom)
net of
investment in
telecom)
Maddala I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0)
& Wua
Maddala I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0)
& Wub
Im, I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0)
Pesaran
& China
Im, I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(0)
Pesaran
& Chinb
Pesaran I(0) I(0) I(1) I(0) I(0) I(0) I(0) I(1) I(0) I(1)
(CIPS)a
Pesaran I(0) I(0) I(0) I(0) I(0) I(0) I(0) I(1) I(0) I(1)
(CIPS)b
Note: i) each of the three tests is based on country-specific augmented Dickey Fuller (ADF) regressions, for each test we provide
two results: (a) results of ADF regressions with an intercept, and (b) results of ADF regressions with an intercept and a trend; ii)
the Maddala and Wu statistic is constructed from the p values of the country specific ADF statistics, the null hypothesis is the
non stationarity of the variable for all panel units, the alternative being stationarity for all panel units ; iii) the Im, Pesaran and
Chin statistic is constructed from the t statistics of the country specific ADF regressions, the null hypothesis is the non-
stationarity of the variable for all panel units, the alternative being stationarity for some panel units; iv) the Pesaran (CIPS) test is
identical to the previous test (with the same null and alternative hypotheses) but allows for the presence of one common factor in
the data (controlling for cross section dependence) by augmenting the country specific ADF regressions with cross sectional
averages of the dependent and independent variables; v) the table shows the order of integration of the variables: I(0): stationary,
I(1): non-stationary, -: test could not be implemented; vi) to implement some of the tests, we used Stata routines "multipurt" and
"pescadf" written by Markus Eberhardt and Piotr Lewandowski, respectively; vii) given that the education variable is reported for
a limited number of years, we could not test its time series properties.
40
Table 3: ICT indicators, various regions
Region Fixed Fixed Annual investment Revenue from the Revenue from the Fixed
Internet broadband in telecommunication telecommunication broadband
subscribers subscribers telecommunication services (billions services (% of monthly
per 100 per 100 services (% of of US $) (average GDP) (average subscription
inhabitants inhabitants GDP) (average over 1995-2012) over 1995-2012) charge (US $)
(average (average over 1995-2013) (average over
over 1995- over 1995- 2008-2012)
2013) 2013)
Africa 0.4 0.2 1.16 31.211 5.84 97.4
AME 3.6 2.9 0.85 42.810 3.11 37.5
Asia 5.2 3.4 0.75 119.331 2.62 24.4
Latin 4.3 4 1.12 65.544 3.71 26.5
America
OECD 22.7 19.1 0.52 913.841 2.74 31.4
High 6.07 4.4 0.6 15.171 2.62 34.2
income
AME
Middle 2.3 2 0.8 27.639 3.40 49.9
income
AME
Source: author's calculations based on the International Telecommunication Union, the ICT indicators 2015 database.
Note: the note of figure 1 applies.
41