The Myth of Population Density and ICT Infrastructure

Proceedings of the 42nd Hawaii International Conference on System Sciences - 2009
The Myth of Population Density and ICT Infrastructure
Kurt DeMaagd
Michigan State University
East Lansing, MI
[email protected]
Abstract
Mobile phones have become a significant platform
for the delivery of information services in developing
economies. Therefore, a sufficiently developed mobile
telecommunications infrastructure is an important element in the economic development of these countries.
This paper examines the factors that affect the quality
of the mobile phone infrastructure in a country. One
classic argument for limited infrastructure is a low
population density. In theory, fewer people in a region means fewer customers to cover the fixed costs of
the infrastructure. This paper tests the link between
population density and mobile infrastructure. The results contradict the classic argument. The paper then
argues that the challenge actually lies in the financial
markets. If the financial markets demand a higher rate
of return from the infrastructure, then the telecommunications firms will invest less in the infrastructure.
1. Introduction
Debates regarding the competitive advantage provided to individual firms notwithstanding [5], it is
now firmly established that Information and Communications Technologies (ICT) play a key role in the
productivity of a business [4]. In fact, the Organization for Economic Cooperation and Development
(OECD), cites ICT investment as the primary reason
for the United States’ disproportionate rate of economic growth relative to other developed nations [19].
Given the often poor quality of the core ICT infrastructure in developing countries, this may be an impediment to future growth. To help enhance the economic
vitality of emerging markets, it is necessary for them
to develop their ICT infrastructure.
One key aspect of this infrastructure is mobile
telecommunications. Mobile phones are expanding
beyond their traditional role as a substitute for fixedline communications; they are now considered a general purpose platform for the delivery of information
services [24]. In many developing economies—where
computers are relatively rare—mobile devices are becoming the primary platform for the delivery of information services. As a result, the development of mobile phone infrastructure in particular plays an important role in the economic growth of developing countries.
Several developing countries have been quite successful in the development of their mobile infrastructure. Nonetheless, there is still a large degree of variability in the success. For example, in our data set
(See section 4), the Baltic republics now have more
than one cell phone per person, whereas countries like
Rwanda and the Central African Republic have fewer
than five cell phones per 100 people. What factors explain this variance in mobile phone penetration?
One common explanation is population density [3].
This argument is used in both developed and developing economies. According to this argument, it is not
cost effective to provide mobile phone access in regions with a low population density. When there is a
high population density, the costs of building the infrastructure can be amortized over a larger group of
people. When there is a low population density, the
costs must be borne by a limited group of people and
this cost may exceed their willingness to pay. Insufficient population results in insufficient demand, hence
no infrastructure is built.
This paper presents an analysis of this argument and
finds that population density is not a significant determinant of mobile phone adoption. Yet there is a related
argument that does have explanatory power. Although
a concentrated population to purchase phones does not
affect infrastructure, the ability to finance the infrastructure does. In countries with a high cost of capital,
there is a lower level of mobile phone adoption. For
more details, see sections 5 and 6.
The paper is organized as follows. The next section reviews some of the prior literature linking mobile phones and other telecommunications infrastructure with economic growth. Section 3 describes the
hypotheses and underlying model used to examine the
978-0-7695-3450-3/09 $25.00 © 2009 IEEE
1
Proceedings of the 42nd Hawaii International Conference on System Sciences - 2009
relationship between population density and mobile
phone infrastructure. Section 4 outlines the data used
in the study. Section 5 shows the results of the data
analysis, and then section 6 covers the implications of
these results.
2. Literature Review
This paper is based on two main tracks of prior literature. The first is an analysis of the link between population density and infrastructure. As is demonstrated,
a large portion of the prior literature has assumed or
shown that other (non-telecom) infrastructure depends
on high population densities. The second track of literature is on telecommunications infrastructure and IT
adoption in developing countries. Although these studies do not focus on infrastructure as the dependent variable, they use and discuss infrastructure as a factor in
the analysis. Understanding the telecom and development literature helps to frame this study within the
large framework of the business and economic value
of ICT investments.
2.1
Infrastructure
The link between population density and infrastructure in general has been studied in the past. The consensus is that a higher population density is linked to
more developed infrastructure. Very early research focused on the links between population density and infrastructure. The types of infrastructure have varied,
ranging from railroad [9], road [12], electricity [11],
agriculture irrigation systems [21], etc. Glover and
Simon summarize the general theory well when they
state:
All other things (including per capita income) held constant, if there are more people in a given geographic area, it will cost
less per person to construct a common facility such as a road. If the benefits per person
are the same at different population densities...and the costs per person are lower, the
benefit/cost ratio is higher at a higher population density; and there is some population
density that is high enough to make benefits
outweigh costs [12].
Although the consensus of the prior literature is
that a greater population density should result in more
infrastructure, Ladd presents an interesting counterargument, focusing on population density, growth, and
public sector spending [16]. If a region has a moderate beginning population but is experiencing rapid
population growth, it may be difficult to increase public spending to provide the infrastructure necessary to
match the high rate of growth. The financial burdens
created by this rapid growth reduce service levels. As
such, it may be possible that there is a short-term decrease in the overall quality of infrastructure.
Curiously, the link between population density and
ICT infrastructure does not appear to have been studied up to this point. One possible explanation is that
the more general problem of population density and
infrastructure may appear to be solved based on the
burst of research on the topic during the the 1970s and
80s. Yet ICT infrastructure presents a unique problem that differentiates itself from other forms of infrastructure. Most forms of infrastructure—such as roads,
rail, power, water, etc.—in most economies are provided by the government or by closely regulated monopolies. In contrast, the general trend in the telecommunications industry is towards competition. For example, Wallsten shows that a privatized telecommunications sector combined with an independent telecom
regulator results in a higher level of telecom infrastructure [27]. Furthermore, van Cuilenburg and Slaa
show that a competitive ICT sector will generally lead
to more innovative service, though that may be mediated by the overall level of economic growth [25].
2.2
ICT and Economic Growth
ICT infrastructure is part of the foundation of economic growth. This link between telecommunications
and economic growth has been established for approximately two decades [2,17]. In addition to showing the
correlation between telecommunications infrastructure
and economic growth, Cronin et al. also showed that
there is a bi-directional relationship [6], creating a
positive feedback loop. More infrastructure results in
more economic growth. More economic growth results in more investment in infrastructure. The prior
research also shows that even if there is a high degree
of investment in other sectors, without the telecommunications infrastructure necessary to coordinate among
the different actors, that investment will not yield economic growth [18]. This is of particular importance
when studying the growth of emerging markets. Without the core infrastructure, other forms of investment
will be ineffective.
Narrowing the scope of the problem, infrastructure
is important when understanding the value of IT investments specifically. Dewan and Kraemer studied
the value of IT at the country level. Although they
found a positive effect from IT investments in developed countries, there was not a positive effect in
emerging markets [7,8]. They speculate that an impor-
2
Proceedings of the 42nd Hawaii International Conference on System Sciences - 2009
tant cause of this result is the lack of infrastructure in
developing economies. To quote Sun Microsystems’
old motto, “the network is the computer.” Without the
infrastructure to connect computers and other devices
together, IT investments will not reach their full potential.
Changing the level of analysis to the application
level, many promising new applications seek to use
mobile phones to enhance economic growth. One area
of particular interest recently is the potential for mobile phones in banking. For example, a mobile phone
can act as a de-facto credit card, such as the Wizzit
in South Africa [13] and Smart Money in the Philippines [22]. In such cases, the mobile is either linked to
an existing credit card account or the mobile is able to
securely store value in the handset.
Even without formal mechanisms for mobile banking, some resourceful users in developing nations use
mobile phones to streamline very non-technologyintenisve means of banking. For example, one challenge in many developing economies is the transfer of
money from emigrants back to their home town. If the
villagers have access to a cell phone, the emigrant can
call home to tell them when to travel to the city to pick
up the wire transfer [23]. This saves the villagers the
wasted time and bus fare for unnecessary trips to the
bank or missing opportunities when money is waiting.
Mobile phones have been used in many applications
other than banking. One of the best known studies on
this is Abraham’s 2007 paper from the American Economic Review [1]. The paper examines the use of mobile phones by fishermen in India. The phones help
the fishermen quickly identify which port is offering
the best price possible for their fish. They can then
easily choose which port to travel to, maximizing their
revenue.
Because of this link between telecommunications
and economic growth, many scholars have also studied
the relationship between ICT and the digital divide. In
particular, there is some debate over whether the benefits of ICT investments are shared equally among all
levels of society. Because the wealthy are the first to be
able to afford telecommunications devices, it is possible that only the wealthy will benefit from these investments. A study by Forestier et al. examines this question [10]. They find that although telecommunications
has historically benefitted only the rich, more recently
the poor have also gained from ICT investments. The
study also finds that these benefits extend beyond income to include improvements in infant mortality and
literacy.
3. Model and Hypotheses
This section describes the two models used in this
paper. The first model examines if population density has the commonly expected effect on ICT infrastructure. Because no effect from population density
is found using the first model, the second model then
proposes an alternate explanation, introducing the cost
of capital into the equation.
The first model tests whether population density affects the quality of mobile phone infrastructure. The
fundamental argument being tested is that large concentrations of population are required to bear the burden of the large fixed costs of building the infrastructure. Hence, the population density will be a key variable in the model. The argument here is consistent
with the prior literature that has examined the link between other types of infrastructure, such as roads and
power, and how they are affected by population density [9, 11, 12, 21] .
As noted in the literature review, however, it is possible that other factors offset any effect from population density. In particular, telecommunications infrastructure is increasingly provided in a competitive marketplace. As a result, the traditional finding regarding
population density may be offset by other market factors. This possibility will be discussed in further detail
below.
Hypothesis 1: An increase in population density will
result in more mobile phone infrastructure.
Two control variables are also used. First, the purchasing power of the individual consumers should affect the infrastructure. In most of the prior literature on the link between population density and infrastructure, this was the only other control variable
used [11, 12, 21]. Hypothesis 1 on population density
has an implicit assumption regarding the income of the
population. It assumes that the dense population has
enough money to pay for the infrastructure. Yet even
a large population can not support the fixed costs of
telecommunications infrastructure if the population in
general can only afford a subsistence-level quality of
life. Likewise, a wealthy but sparse population could
still afford the infrastructure. As a result, it is necessary to control for income when studying the determinants of infrastructure provision.
Hypothesis 2: An increase in purchasing power will
result in more mobile phone infrastructure.
The second control variable is the number of fixedline telephones. Most of the previous research on
this topic included only population density and in-
3
Proceedings of the 42nd Hawaii International Conference on System Sciences - 2009
come as independent variables. Yet, taking a cue from
the technology adoption literature, it is also necessary
to include substitute products in the model. Two recent studies both by Kauffman and Techatassanasoontorn have examined the adoption and diffusion of mobile phones in developing economies [14, 15], both
of which use fixed-line phones as a control variable.
Fixed line phones are a substitute for mobile phones.
The presence of fixed line phones therefore should
make it less necessary to invest in mobile phone infrastructure.
Although this argument is particularly compelling
when studying the adoption of mobile phones for consumers, it is slightly more ambiguous when looking
at core infrastructure. Fixed line phones and mobile phones are substitutes when solving the last-mile
problem of connecting the individual consumer with
the general communications network backbone. Yet
both mobile and fixed-line phones can use the same
backbone infrastructure. As a result, a well-developed
fixed-line network may actually serve as a complement
to mobile phones. By allowing mobile phones to connect to the same backbone as the fixed-line phone, the
cost of implementing mobile phone infrastructure is
decreased. As a result, although mobile phone handsets and fixed-line handsets may be seen as substitutes,
it may be reasonable to assume the mobile and fixedline infrastructure are actually complements.
Hypothesis 3: An increase in fixed-line phones will result in a decrease in mobile phone infrastructure.
Because developed economies often have a head
start in their adoption of mobile devices, this is also an
important factor to include in the model. In addition,
variances in the economic, political, and regulatory climate in developing economies may influence the quality of infrastructure in these countries. To examine the
potential different effects in developed versus developing economies, the model is tested using both a combined analysis of all countries and a separate analysis
for developing and developed economies.
Using the above general theme, the model is:
error term. Based on this model, the data is analyzed
using a random-effects regression.
As is discussed below in section 5, the population
density variable is not significant. This implies that the
ability to amortize the costs of the infrastructure over
a larger group did not have a significant effect on the
ability to develop the infrastructure. This is certainly
a curious finding, which is discussed in more detail in
sections 5 and 6. One possible explanation is that the
problem is not the direct cost of the infrastructure. Instead, it may be the cost of financing the infrastructure.
As a result, the important factor is no longer the ability
to amortize the cost over a large population. Instead,
the key factor may be the ability meet the demands of
the financial markets.
The cost of capital is rarely studied in relation to
infrastructure provision, perhaps because of the history of assuming that infrastructure is a public good
provided by the government. However, one recent
study by von Hirschhausen of the natural gas infrastructure markets suggests that cost of capital has an
important role in the efficient provision of infrastructure [26]. Von Hirschhausen describes the implication
of the choice by the Federal Energy Regulatory Commission (FERC) to base prices on the risk adjusted cost
of capital. In general the FERC’s emphasis on the cost
of capital is correlated with the overall quality of the
natural gas infrastructure in the United States. By basing the prices charged on the cost of capital, it is possible to ensure that the natural gas providers can earn a
sufficient return on investments to satisfy the needs for
financing the infrastructure.
This requires a slight change in the model from
above to include a variable for the cost of capital. If
the cost of capital is higher, then it is more expensive to
finance the building of the infrastructure and the markets will demand higher rates of return from the investment. Hence, higher costs of capital make it harder to
invest in mobile infrastructure.
Hypothesis 4: A higher cost of capital will result in a
lower level of mobile phone infrastructure.
Expressing this in mathematical terms:
Mi,t =βPD PDi,t + βP Pi,t + βT Ti,t
+ αt + λi + εi,t
(1)
where M represents the level of mobile phone infrastructure, PD is the population density, P is the purchase power of the consumers, and T is the fixed-line
phone infrastructure. The i subscript represents the
country index. The t subscript indicates the year index.
The α and λ variables capture the time and countryspecific effects respectively. Finally, ε represents the
Mi,t =βPD PDi,t + βP Pi,t + βT Ti,t
+ βR Ri,t + αt + λi + εi,t
(2)
where all of the variables are defined similar to the first
model. The new variable, R, represents the cost of capital.
To summarize, this model examines how various
factors affect the amount of mobile phone infrastructure built. The primary factor of interest is population
4
Proceedings of the 42nd Hawaii International Conference on System Sciences - 2009
Figure 1: Map of countries included in data set. Included countries are shaded in black.
density. In addition, the consumer’s purchasing power
and access to fixed-line telephones are included as control variables. Because the data analysis shows that
population density does not have a significant effect, a
second model using the cost of capital is also used.
4. Data
The data set used is a country-level data set covering 144 countries from the years 2002 through 2007.
This provides a broad sample of countries over a period of time when mobile phone adoption has grown
rapidly worldwide. Due to data quality limitations, not
every country could be included. For example, based
on UN-membership, there are currently 192 countries.
This means that the data set covers 75% of the current
countries. Figure 1 graphically shows the countries included in this data set. A complete list of the countries
is included in the appendix.
A casual glance at the map of countries shows that
the data set in general covers a broad range of countries. It includes data from every continent except
Antarctica. It includes countries from both developed
and developing countries. The biggest gaps are in West
Africa and the Middle East. Nonetheless, the data set
does include countries from each of these regions. Because of the breadth of different countries included in
the data set, the results of this study should generalize
to most countries on the planet.
Central to the topic of this paper is population density. The paper examines if population density affects
mobile phone infrastructure. The data on population
density is taken from United Nations (UN) data. It is
measured as the number of people per square kilometer. The average country in the data set has 233 people
per square kilometer in 2007. The least dense country
is Mongolia, with 1.6 people per kilometer. The most
dense is Hong Kong, with 7276 people per square kilometer.
The alternative explanation developed in this paper
is that the financial markets play a major role in de-
termining mobile phone infrastructure. This requires a
proxy for the cost of capital. Although it is not feasible
to collect data on the cost of capital for every telecommunications company in all 144 countries in the data
set, there is an alternate variable which can approximate the cost of capital. The International Monetary
Fund (IMF) provides data on the average corporate
lending rate in many countries. (Note that this data
was the most likely to be unavailable in our data set,
and hence was the primary reason why some countries
are not present.) In 2007 the average corporate lending rate was 15.7%, with a low of 1.7% in Japan and a
high of 552.5% in Zimbabwe.
As described in the pervious section, the purchasing power of consumers is often cited as a driver of
mobile phone adoption. To estimate the purchasing
power of consumers, the Purchase Power Parity (PPP)
adjusted Gross Domestic Product (GDP) per person is
used. This computation is based on PPP adjusted GDP
from IMF data, and population data taken from UN
data. In 2007, the average country in the data set had a
PPP adjusted GDP per person of 11,690 International
Dollars. Norway had the highest GDP per person of
52,970 and Liberia was the lowest with 15 International Dollars per person.
The final independent variable is fixed-line phone
users. Fixed-line telephones are a substitute for mobile
phones. The data for fixed-line telephones comes from
the International Telecommunication Union (ITU). To
make the data comparable across countries, we convert
this data into the number of lines per person by dividing the number of phone lines by the total population.
Once again, the population data was taken from UN
statistics. The average country in the data set has 0.21
phone lines per person in 2007. Sweden had the most
phones per person at 0.68. Chad had the fewest, with
only 0.001 lines per person.
One interesting trend in the fixed-line phone data
is the contrasting growth rates between developed and
developing countries, as shown in Figure 2. Globally, the number of fixed-line telephones has been rising. Yet in developed countries, the number of lines
has been decreasing as technologies such as mobile
phones and the Internet replace traditional telephone
service. In contrast, emerging markets continue to invest in more lines. It will be interesting to watch these
numbers over the course of the next decade to see if
increasing adoption of mobile phones in developing
countries results in a reversal of the current fixed line
investment trends.
Obtaining data for the dependent variable, which is
the level of mobile phone infrastructure, is somewhat
more problematic. No data sources are available for a
5
Proceedings of the 42nd Hawaii International Conference on System Sciences - 2009
Figure 2: Percent of population with fixed-line phones.
The top dashed line represents developed
countries. The middle solid line represents all
countries. The bottom dotted line represents
developing countries only.
Figure 3: Percent of population with mobile phones.
The top dashed line represents developed
countries. The middle solid line represents all
countries. The bottom dotted line represents
developing countries only.
large number of countries providing information on total coverage, the number of towers, or other such metrics. Yet a related proxy is widely available: mobile
phone users. Under the assumption that consumption
of a good is highly correlated with the provision of that
good, it appears reasonable to assume that the purchase
of mobile phones should be highly correlated with the
existence of mobile phone infrastructure. This proxy
for mobile phone infrastructure is consistent with prior
research [15].
The mobile phone data is taken from the ITU’s
statistics. Figure 3 shows a summary of the penetration of mobile phones over time. As may be expected, developed countries have a much higher rate
of utilization, with the rate approaching 100%. 14
out of 24 countries had an average of more than one
phone per person in 2007. In contrast, mobile phones
have not seen the same level of adoption in developing
economies. By 2007, 41% of the population in developing countries owned a mobile phone. Interestingly,
15% of the developing countries also had more than
one mobile phone per person on average. Furthermore,
in general, the increase in mobile phone penetration
in developing countries has increased in parallel with
developed economies. Although at the current pace
of adoption, it will still take approximately 10 years
for the emerging economies to catch up, current trends
show a strong interest mobile devices in developing regions.
The overall trend line for all countries is also of interest. The overall line is much closer to the developing economy line. This is expected, given that 84%
of the world population lives in countries that are classified as developing. Nonetheless, it is also interesting that as of 2007, 49.2% of the world’s population
had a cell phone. (Or, perhaps, to be completely rigorous, there were 49.2 mobile phones per 100 people.
Some people may have more than one phone.) Although predictions vary, it is widely assumed that half
of the word’s population now has or soon will have a
cell phone [20].
To summarize, the following data from the listed
sources is used:
1. Population density, from UN data
2. Cost of capital, based on lending rates from IMF
data
3. Purchasing power, based on GDP per person from
IMF data
4. Mobile phone users, from ITU data
5. Fixed line phone users, from ITU data
5. Analysis
Using the models and data described in the previous
section, the next step is the data analysis. Recall that
the broad goal of this paper is to examine whether population density is a determinant of the level of mobile
phone infrastructure. In addition, the model includes
variables for the purchasing power of the consumers
and the availability of fixed-line telephones as a substitute product. This model is analyzed using a randomeffects regression. The results are shown in Table 1.
Note that the table shows three different aspects of the
data set. First, the Table shows the results for the combined data set of all countries. The middle section of
the table reports on developed countries only. The bottom section reports on developing countries only.
6
Proceedings of the 42nd Hawaii International Conference on System Sciences - 2009
Table 1: Regression results from the first model. Notice that population density is not a statistically
significant determinant.
Variable
Coefficient
All Countries
Intercept
0.0566
Purchasing power
0.0329
Population density -0.0000
Fixed-line phones -0.0000
F(3, 860)
185.362
0.3927
R2
N
864
Developed Countries
Intercept
0.0782
Purchasing power
0.0286
Population density -0.0000
Fixed-line phones -0.0000
F(3, 140)
123.938
0.726
R2
N
144
Developing Countries
Intercept
-0.0212
Purchasing power
0.0438
Population density 0.0001
Fixed-line phones -0.0000
F(3, 716)
136.866
0.364
R2
N
720
p
< 0.001
< 0.001
0.193
0.257
< 0.001
0.099
< 0.001
0.402
< 0.001
< 0.001
0.242
< 0.001
0.0939
0.010
< 0.001
As expected, the purchase power of the consumers
had a positive and significant effect on the mobile infrastructure. This is true in all three versions of the data
set. This result is consistent with the second hypothesis, indicating that an increase in purchasing power
will result in more mobile phone infrastructure.
The case of fixed-line phones is more ambiguous.
In the combined data set, the fixed-line phone infrastructure was not statistically significant. One possible
explanation goes back to the data summarized in Figure 2. Recall that in developed economies, use of traditional phones has been declining while in developing
economies it has been increasing. If the magnitude or
direction of the effect is different in the developed versus developing economies, that could explain the lack
of statistical significance.
This perspective is supported by the analysis that
examines developed and developing countries separately. In developed economies, there is a negative and
statistically significant effect. Note that although the
effect does not appear in the reported digits in the table due to rounding, the actual effect was 6.847 × 10−6
in developed economies. In contrast, in developing
economies, the result was not significant. The effect
was an order of magnitude smaller, at 6.848 × 10−7 .
One possible reason for the different effects is the
overall level of fixed-line access in developed versus developing economies. In developed economies,
the market for fixed-line phones is already saturated.
Hence any customers who switch from fixed to mobile
phones cannot be offset by an expansion to unserved
markets. In contrast, in developing economies, potential opportunities abound for both technologies. An
increase in mobile phone users need not imply a decrease in fixed-line users.
Moving on to the main hypothesis of the paper, the
population density had a negative though insignificant
effect in the combined data set and in the developed
country data set. The magnitude of the effect was very
small. For rounding purposes in the table, it is effectively zero. The actual regression coefficient was
−1.6486 × 10−5 . Given the extremely small and nonsignificant estimate, it appears that population density
is not an important determinant of mobile phone infrastructure.
The one exception is that in developing economies,
the population density variable had a p-value of
0.0939, indicating a possible weakly significant effect.
The order of magnitude was also larger than in developing countries, with a value of 1.4743×10−4 , two orders of magnitude larger than in developed countries.
Although there is a weak indication of an effect, the
size of the effect is still very small.
This finding contrasts with Hypothesis 1, which
predicted that population density should result in better infrastructure. The economics behind Hypothesis
1 appear to be quite simple and obvious. Without
enough people to purchase the service, it is not possible to provide the service. Why then does the application of a simple and obvious theory not align with
the empirical evidence?
Instead of focusing on the cost of building the technology and the ability to amortize the investment over
a large group of people, it may be practical to instead
consider the cost of financing the investment and the
required rate of return demanded by the financial markets for these investments. As the cost of technology declines over time, it may be much more feasible to provide service to even small populations. Yet
although technology costs may decline, the demands
of the financial markets do not. Consequently, the demands of the financial markets may be a better predictor of the amount invested in mobile infrastructure by
telecommunications companies.
Table 2 shows the analysis when including the cost
7
Proceedings of the 42nd Hawaii International Conference on System Sciences - 2009
Table 2: Regression results with cost of capital. Notice
that population density is not statistically significant, the but the cost of capital is.
Variable
Intercept
Purchasing power
Population density
Fixed-line phones
Cost of capital
F(4, 859)
R2
N
Coefficient
0.0658
0.0326
-0.0000
-0.0000
-0.0038
139.842
0.394
864
p
0.005
< 0.001
0.250
0.386
0.043
< 0.001
of capital. The higher the cost of capital, the lower
the level of mobile infrastructure. This supports the
view that the financial markets can affect the quality
of infrastructure. The general size and magnitude of
the other variables in the model are consistent with the
previous model.
The following is a summary of the hypotheses and
whether or not they were supported by the data analysis.
• Hypothesis 1 not supported: An increase in population density did not have a statistically significant effect.
• Hypothesis 2 supported: An increase in purchasing power had a positive significant effect.
• Hypothesis 3 partially supported: An increase in
fixed-line phones had a negative effect, but only
in developed countries.
• Hypothesis 4 supported: An increase in the cost
of capital had a negative significant effect.
6. Conclusion
Population density is often cited as an important
reason why infrastructure is good in urban regions but
poor in rural areas. In an urban region, it should be
easy to amortize the cost of the infrastructure across a
large number of people. In a rural region, fewer people
will have to cover the high fixed costs. As a result, the
infrastructure in rural regions is not as good as it is in
urban regions.
This paper studies this argument to see if the general hypothesis aligns with the empirical data, particularly focusing on the issue of mobile phone infrastructure. It does not. In developing economies, there is
a small and weak effect. In general, however, population density is not correlated with the quality of mobile
infrastructure. This implies that the cost of the infrastructure and the ability to spread the cost over a large
base of consumers is not an issue.
Yet this is not to say that cost is no issue. Instead,
the cost of capital appears to be a much stronger determinant. The cost of capital can play two roles. First, it
represents the cost which a telecommunications company must pay to finance its infrastructure investments.
Second, it can serve as a benchmark for the rate of return that the financial markets expect from the investment. If the cost of capital is high, that indicates that
it will be very costly for businesses to invest in the
mobile phone infrastructure. If the rate of return from
such an investment is not sufficiently high, investors
will take their money elsewhere.
This can be particularly problematic when countries
are deregulating their telecommunications sectors and
relying on market forces to create the necessary incentives for investment. Deregulation means that telecommunications firms will need to turn to the financial
markets to raise their capital instead of relying on government subsidies. If the financial markets demand a
rate of return that exceeds the potential profits from
the mobile phone infrastructure investments, then the
country’s infrastructure will suffer.
This then feeds into the bigger picture problems
of the link between ICT infrastructure and economic
growth. As discussed in the literature review, infrastructure is a key driver of economic growth. Without
this infrastructure, it is difficult to coordinate among
the different economic actors. This in turn limits the
value of other forms of economic stimulus.
This is especially true when considering the value
of other types of information technology investments.
First, the core communications infrastructure is important for networking computers. Without this network,
it is not possible to realize the full value of these investments. In regions without other forms of networking,
mobile technologies are seen as a way to efficiently
provide broadband access. Furthermore, in developing
economies, mobile phones are becoming the primary
platform for delivering information services. Without
the necessary underlying infrastructure, the types of
applications that can be delivered over a mobile phone
will be very limited.
In summary, it is important to understand the factors that create incentives to invest in mobile infrastructure. As this paper shows, the ability to spread
the infrastructure cost over a larger group of people
is not one of those factors. Yet costs do play a role:
financial costs. Especially as many countries deregulate their telecommunications sectors, financial markets will play an increasing role in determining the
8
Proceedings of the 42nd Hawaii International Conference on System Sciences - 2009
quality of mobile communications infrastructure.
Appendix
The data set consists of 24 developed countries. For
the purposes of this paper, the determination of “developed” is taken from the classification by the International Monetary Fund (IMF). Note that several developed countries were not included in this data set,
primarily because the IMF data set did not include information on their corporate lending rate. The full list
of developed countries under the IMF classification includes 32 countries. Hence, the data set include 75%
of all developed countries. Note that the IMF classification technically refers to developed countries as
“advanced economies.”
The developed economies in the data set are:
Belgium
Canada
Cyprus
Finland
France
Germany
Hong Kong
Iceland
Ireland
Israel
Italy
Japan
Malta
Netherlands
New Zealand
Norway
Singapore
Slovenia
S. Korea
Spain
Sweden
Switzerland
UK
USA
The data set includes 120 developing countries.
Once again the classification of “developing” is based
on the IMF’s system. This data set also include 34 of
all developing countries.
The developing economies in the data set are:
Albania
Algeria
Angola
Antigua
Argentina
Armenia
Australia
Azerbaijan
Bahamas
Bahrain
Bangladesh
Barbados
Belarus
Belize
Bhutan
Bolivia
Botswana
Latvia
Lebanon
Lesotho
Liberia
Libya
Lithuania
Macedonia
Madagascar
Malawi
Malaysia
Maldives
Mauritius
Mexico
Moldova
Mongolia
Morocco
Mozambique
Brazil
Brunei
Bulgaria
Burundi
Cambodia
Cameroon
Cape Verde
Central African Republic
Chad
Chile
China
Colombia
Comoros
Congo-Brazzaville
Costa Rica
Croatia
Czech Republic
Djibouti
Dominica
Dominican Republic
Ecuador
Egypt
Equatorial Guinea
Estonia
Ethiopia
Fiji
Gabon
Gambia
Georgia
Grenada
Guatemala
Guyana
Haiti
Honduras
Hungary
India
Indonesia
Iran
Jamaica
Jordan
Kenya
Kuwait
Kyrgyzstan
Myanmar
Namibia
Nepal
Netherlands Antilles
Nicaragua
Nigeria
Oman
Panama
Papua New Guinea
Paraguay
Peru
Philippines
Poland
Qatar
Russia
Rwanda
Sao Tomee Principe
Seychelles
Sierra Leone
Slovakia
Solomon Islands
South Africa
Sri Lanka
St. Kitts
St. Lucia
St. Vincent
Swaziland
Syria
Tajikistan
Tanzania
Thailand
Tonga
Trinidad and Tobago
Uganda
Ukraine
Uruguay
Vanuatu
Venezuela
Vietnam
Western Samoa
Yemen
Zambia
Zimbabwe
References
[1] Reuben Abraham. Mobile phones and economic
development: Evidence from the fishing industry
in India. American Economic Review, 4(1):5–17,
2007.
[2] David Aschauer. Is public expenditure productive?
Journal of Monetary Economics,
23(2):177–200, 1989.
9
Proceedings of the 42nd Hawaii International Conference on System Sciences - 2009
[3] World Bank. Maldives: World bank group supports mobile phone banking. In Press Release
No:2008/277/SAR, 2008.
[4] Erik Brynjolfsson and Lorin Hitt. Paradox lost:
Firm-level evidence on the returns to information systems spending. Management Science,
42(4):561–558, 1996.
[5] Nicholas Carr. IT doesn’t matter. Harvard Business Review, 81(5), 2003.
[6] Francis Cronin, Edwin Parker, Elisabeth
Colleran, and Mark Gold.
Telecommunications infrastructure and economic growth.
Telecommunications Policy, 15(5):529–535,
1991.
[7] Sanjeev Dewan and Kenneth L. Kraemer. International dimensions of the productivity paradox.
Communications of the ACM, 41(8):56–62, 1998.
[8] Sanjeev Dewan and Kenneth L. Kraemer. Information technology and productivity: Evidence
from country level data. Management Science,
46(4):548–562, 2000.
[9] Albert Fishlow. American Railroad and the
Transformation of the Ante-Bellum Economy.
Harvard University Press, Cambridge, MA,
1965.
[10] Emmanuel Forestier, Jeremy Grace, and Charles
Kenny. Can information and communication
technologies be pro-poor? Telecommunications
Policy, 26(11):623–646, 2002.
[11] Peter C. Frederiksen. Further evidence on the
relationship between population density and infrastructure: the Philippines and electrification.
Economic Development and Cultural Change,
29(4):749–758, 1981.
[12] Donalrd R. Glover and Julian L. Simon. The effect of population density on infrastructure: The
case of road building. Economic Development
and Cultural Change, 23(3):453–468, 1975.
[13] G. Ivatury and M. Pickens. Mobile phone banking and low-income customers: evidence from
South Africa. In Washington DC: Consultative
Group to Assist the Poor (CGAP), 2006.
[14] Robert J. Kauffman and Angsana A Techatassanasoontorn. International diffusion of digital mobile technology: A coupled-hazard statebased approach. Information Technology Management, 6:253–292, 2005.
[15] Robert J. Kauffman and Angsana A Techatassanasoontorn. Is there a global digital divide
for digital wireless phone technologies. Journal of the Association for Information Systems,
6(12):338–382, 2005.
[16] Helen Ladd. Population growth, density, and the
costs of providing public services. Urban Studies, 29(2):273–295, 1992.
[17] J. Bradford De Long and Lawrence Summers.
Equipment investment and economic growth.
Quarterly Journal of Economics, 106(2):445–
502, 1991.
[18] Gary Madden and Scott Savage. CEE telecommunications investment and economic growth.
Information Economics and Policy, 10(2):173–
195, 1998.
[19] OECD. The New Economy: Beyond the Hype.
Final Report on the OECD Growth Project. Organization for Economic Cooperation and Development, Paris, 2001.
[20] Reuters.
Global cellphone penetration reaches 50 pct.
http://investing
.reuters.co.uk/news/articleinvesting.aspx?
type=media&storyID=nL29172095, 2007.
[21] Julian L. Simon. The positive effect of population growth on agricultural savings in irrigation
systems. Review of Economics and Statistics,
47(1):71–79, 1975.
[22] Sharon Smith. Digital dividend: Case studies: Smart communications.
http://www.digitaldividend.org/case/case smart.htm, 2004.
[23] Serigne Tall. Senegalese emigres: new information & communication technologies. Review of
African Political Economy, 31(99):31–49, 2004.
[24] The Economist. Geeks of the Bushveld. The
Economist, August 13, 2000.
[25] Jan van Cuilenburg and Paul Slaa. Competition
and innovation in telecommunications. Telecommunications Policy, 19(8):647–633, 1995.
[26] Christian von Hirschhausen. Infrastructure investments and resource adequacy in the restructured us natural gas market: is supply security at
risk? MIT CEEPR Working Paper series, 2006.
[27] Scott J. Wallsten. An econometric analysis of
telecom competition, privatization, and regulation in africa and latin america. The Journal of
Industrial Economics, 49:1–19, 2003.
10