See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/365668994

Mobile banking and financial inclusion in sub-Saharan Africa: Cost versus

geographical constraints

Chapter · January 2023

CITATIONS READS

0 771

3 authors:

Boniface Ngah Epo Mougnol à Ekoula Hervé William

University of Yaoundé II University of Yaoundé II
53 PUBLICATIONS 298 CITATIONS 7 PUBLICATIONS 10 CITATIONS

SEE PROFILE SEE PROFILE

Yvan Audrey Kamdem

University of Yaoundé II
1 PUBLICATION 0 CITATIONS

SEE PROFILE

All content following this page was uploaded by Yvan Audrey Kamdem on 12 April 2023.

The user has requested enhancement of the downloaded file.

Mobile banking and financial inclusion in
sub-Saharan Africa: cost versus
geographical constraints
Mobile banking et inclusion financière en Afrique
subsaharienne : contraintes de coût versus
2 2
contraintes géographiques
2 0
c h
This paper is accepted since March 2022 for a forthcoming publication

a r 2 - Cameroon,
in a collective work at University of Yaoundé

, M
after a triple blind peer-review process.

Epo Boniface f t Ngah ∗

r a
epongahb@yahoo.fr
d Hervé William ∗
d
Mougnol à Ekoula
e
p tKamdem Yvan Audrey
wmougnol@yahoo.com

c e yvanaudrey kamdem@yahoo.fr ∗

A c

∗
Faculty of Economics and Management, University of Yaoundé 2 - Cameroon.

1
 Epo, Mougnol and Kamdem (forthcoming)

Mobile banking and financial inclusion in

sub-Saharan Africa: cost versus geographical
constraints
Mobile banking et inclusion financière en Afrique
subsaharienne : contraintes de coût versus contraintes
géographiques
Abstract: This paper analyses the relationship between the diffusion of mobile banking in
sub-Saharan Africa (SSA) and the barriers to financial inclusion. We use a 2017 cross-sectional
data for thirty-five countries in sub-Saharan Africa gotten from the World Bank and the Interna-

2 2
tional Monetary Fund to appraise the effect of financial inclusion barriers/constraints associated

2 0
with cost and those linked to geography using a multiple linear regression model. Running ordi-
nary and generalized least square estimates, as well as both a Jacknife and bootstrap resampling

ch
techniques to check the robustness, we find a positive and statistically significant relationship

a r
between the diffusion of mobile banking in terms of access and usage with the different indica-
tors of cost and geographical constraints to financial inclusion. These results posit that in SSA,

,M
cost and geographical financial inclusion constraints have encouraged the adoption as well as

t
diffusion of mobile banking as an alternative channel for access and use of financial services for
f
a
the most vulnerable, especially the poorest 40th percentile and those living in remote areas.

dr
Keywords: Mobile Banking, Financial Inclusion, Cost Constraints, Geographical Constraints,
Sub-Saharan Africa (SSA).
e d
pt
c e
Résumé : Cet article analyse la relation entre la diffusion des services bancaires mobiles
en Afrique subsaharienne (ASS) et les obstacles à l’inclusion financière. Nous utilisons des

A c
données transversales de 2017 pour trente-cinq pays d’Afrique subsaharienne obtenues de la
Banque Mondiale et du Fonds monétaire international pour évaluer l’effet des contraintes à
l’inclusion financière associées au coût et celles liées à la géographie à l’aide d’une analyse de
régression linéaire multiple. En utilisant les moindres carrés ordinaires et généralisés, ainsi
que des techniques de rééchantillonnage Jacknife et bootstrap pour vérifier la robustesse, nous
trouvons une relation positive et statistiquement significative entre la diffusion des services
bancaires mobiles en termes d’accès et d’utilisation avec les différents indicateurs de contraintes
(coût et géographie) à l’inclusion financière. Ces résultats postulent qu’en ASS, ces obstacles ont
encouragé l’adoption ainsi que la diffusion des services bancaires mobiles comme canal alternatif
d’accès et d’utilisation des services financiers pour les plus vulnérables, en particulier les 40%
plus pauvres et ceux qui vivent dans des régions éloignées.

Mots clés : Mobile Banking, Inclusion Financière, Contraintes de Coûts, Contraintes Géo-
graphiques, Afrique Subsaharienne (ASS).

2
 Epo, Mougnol and Kamdem (forthcoming)

1 Introduction
Mobile banking1 is revolutionising financial inclusion throughout the developing world
and particularly in sub-Saharan Africa (SSA). The scale of the phenomenon is such that,
this innovation is now unavoidable and is changing the financial habits of populations
(Ky et al., 2018). Due to mobile banking, it is neither necessary to go to banks’ branches
or have an account in a financial institution. This is because it is now possible to carry
out transactions such as sending, receiving money or making payments using a mobile
phone. In SSA, the popularity of mobile banking has been unbounded since the success
of M-pesa2 . According to the Global Findex survey (2017), its penetration rate on the
African subcontinent stands at 21%. For comparison, the global average is 4% and no
other developing region exceeds 6% (Demirgüç-Kunt et al., 2018). This trend in SSA
2 2
attempted in this paper. 2 0
requires we appraise these dynamics to help robustly identify possible policy avenues as

ch
In the literature, three main factors are identified as determining the spread of mobile

a r
banking in developing countries. First, the level of financial development heartens or
hinder the emergence of financial innovations such as mobile banking (Della Peruta, 2015;

f t ,M
Asongu, 2018; Senou et al., 2019). Second, structural and macroeconomic indicators such
as mobile phone and internet usage encourage the deployment of mobile banking, inflation
a
dr
which has a negative effect on its use, or GDP per capita which have a positive or negative
effect (Allen et al., 2014; Della Peruta, 2015; Ky, 2016; Asongu, 2018; Akinyemi and

d
Mushunje, 2020). Third, demographic variables such as population, level of urbanisation,
e
t
population density and level of education play an important role in the spread of mobile

p
c e
banking (Teo et al., 2012; Allen et al., 2014; Della Peruta, 2015; Ky, 2016).
Nonetheless, despite this extensive literature, few studies, with very little focusing

A c
on Africa, explore the relationship between mobile banking diffusion and the different
constraints that hinder financial inclusion. This necessitates we densify these studies
for SSA which recorded the highest penetration rate associated with mobile banking
whilst also signalling significant financial inclusion deficit3 . This paper therefore aims
at filling this gap empirically by investigating the relationship between the diffusion of
mobile banking in SSA and its association with cost and geographical barriers of financial
1
A distinction is often made between mobile banking and mobile money in that the former represents
the provision of services by banks to their customers through mobile devices, while the latter refers to the
provision of basic financial services to unbanked customers (Ky, 2016). However, the concept of mobile
banking has a broader meaning in the literature that also includes mobile money (Mishra and Bisht,
2013). It is this latter understanding of the concept that we adopt in this study.
2
M-pesa is probably the most famous mobile money platform in the world. Launched in Kenya by
the telecom company Safaricom in 2007, it registered 1.1 million users just eight months after its launch
(Mbiti and Weil, 2011).
3
From 23% in 2011, the number of adults with an account in SSA has increased to 33% in 2017.
However, this is the level that regions such as the Middle East, Latin America, and South Asia had
already reached in 2011 (World Bank, 2018).

3
 Epo, Mougnol and Kamdem (forthcoming)

inclusion.
Several factors are responsible for the lag in financial inclusion in developing countries.
In SSA, these include insufficient income, high cost of financial services, geographical
remoteness of financial institutions, lack of documentation to open a bank account, lack
of need, lack of trust, having a family member with an account, and religious reasons
(Klapper et al., 2019). In this paper, we focus only on cost and geographical constraints
because they are, after income insufficiency, the most prevalent in SSA4 (Demirgüç-Kunt
and Klapper, 2012; Granata et al., 2014; Guérineau and Jacolin, 2014). Hinging from the
demand side perspective, this paper explores whether these two constrains overly affect
the spread of mobile banking in SSA using a 2017 snapshot sample of thirty-five SSA
countries.

2 2
Differing from other studies, scope-wise because we focus on SSA, a second contribu-

2 0
tion is that, unlike most studies that emphasis on the supply side we rather investigate

h
this relationship from the demand side. Indeed, whereas the former tries to show how

c
a r
the provision of mobile banking services affects financial inclusion, this paper instead
investigates whether the diffusion of mobile banking in SSA is linked to the existence

,M
of constraints associated to financial inclusion. Measuring these constraints constitutes
another specificity of this paper. Unlike most authors who capture access to financial
f t
services through infrastructural indicators (number of banks or bank branches for a given
a
dr
population or area), we opt for the proportion of unbanked people who identify cost or
geographical distance as the main barrier to holding an account in a formal financial

e d
institution (Demirgüç-Kunt et al., 2018; Klapper et al., 2019). This indicator has the
t
advantage of being measured directly on the population concerned by banking exclu-
p
e
sion. Further, this measure pays special attention to the most vulnerable groups in the
c
A c
measurement of mobile banking diffusion (poorest 40% and those residing in rural areas).
Consequently, this paper set to examine the relationship between the diffusion of
mobile banking in SSA and the barriers to financial inclusion. Specifically, the paper
compares the effects of cost and geographical constraints associated with financial inclu-
sion on mobile banking in SSA. The underlying postulate of the paper is that, cost and
geographical constraints both increase the diffusion of mobile banking in SSA when con-
sidering the different measures of diffusion of mobile banking. The rest of the paper is
organized as follows: section 2 review the theoretical and empirical literature; section 3
the methodology and estimation strategy; section 4 discusses the empirical results and
section 5 concludes the paper.
4
According to the 2017 World Bank Global Findex survey, these two types of constraints are cited
by at least a quarter of the unbanked as major obstacles to holding an account in a formal financial
institution (Klapper et al., 2019).

4
 Epo, Mougnol and Kamdem (forthcoming)

2 Literature review
2.1 Theoretical review of the literature
The origins of mobile banking could be traced in the early 1990s in Finland, where cus-
tomers of a local bank were able to access their accounts and carry out transactions via
their mobile phones or any other mobile device (Barnes and Corbitt, 2003). A decade
later in the 2000s, the concept of mobile banking evolved from the original banking model
to a non-banking model in developing countries. This gave rise to the notion of ”mobile
money”, as the provision of low-cost basic financial services via mobile phone, and most
often to the unbanked (Ky, 2016). However, mobile banking has a broad meaning in the

2
literature that also includes mobile money. Indeed, Porteous (2006), for example, consid-
2
having either a bank account or not.
2 0
ers all these services to be mobile banking, regardless of whether they are aimed at those

h
Studies attempting to perceive the relationship between mobile banking and financial
c
a r
inclusion are new. Research in this area is still mostly dominated by practitioners in
finance and bank marketing who are often interested in the factors associated with the

,M
adoption of this innovation, its functioning and the experiences of different platforms.

t
Whereas there has been a growing interest in terms of academic research, the existing
f
a
studies are still mainly empirical and do not really focus on establishing the theoretical

dr
framework relationship between these two concepts.

e d
Nevertheless, a review of the literature shows that mobile banking is often analysed
under the prism of the theories of diffusion of innovations (Rogers, 1983) and of the

pt
acceptance of technology (Davis, 1989). According to the former, the diffusion of an

c e
innovation such as mobile banking is based on five factors. These factors include: relative

A c
advantage, compatibility, complexity, testability and observability. The latter theory
explains the success of a new technology by its ease of use and the usefulness perceived
by the potential consumer. Consequently, the strong diffusion of mobile banking in SSA
can be explained by its advantage over traditional financial services and its compatibility
with the needs of the region’s populations that face numerous constraints in accessing
formal financial services. Furthermore, the usefulness of mobile banking coupled with its
ease of use may constitute a strong argument for its expansion in SSA.
The two theoretical frameworks indicated above have been used by several authors to
analyse the diffusion of mobile banking and its potential for financial inclusion in devel-
oping countries. Barnes and Corbitt (2003) argue that mobile banking has a comparative
advantage in emerging countries such as China and India because it provides better finan-
cial access to those for whom traditional channels are inaccessible. Similarly, Datta et al.,
(2001) perceive mobile banking as an opportunity for developing countries to make a tech-
nological leap that bridges the gap of poor banking infrastructure and provide financial
services to the unbanked, regardless of their geographical location.

5
 Epo, Mougnol and Kamdem (forthcoming)

Porteous (2006) concurs that there are two types of mobile banking in SSA and
throughout the developing world. The ”additive” model, offered by banks to their cus-
tomers appears to be another channel for traditional finance. On the contrary, the ”trans-
formational” model, more often offered by telecoms and other non-bank providers to the
under- or unbanked, has the potential to revolutionize financial inclusion provided that
adequate regulation is put in place. Thus, mobile banking cannot be viewed as a disruptive
innovation compared to the traditional financial system.

2.2 Empirical review of the literature

Three main areas in the empirical literature have been largely researched upon on mobile

2 2
banking. The first area is concerned with the criteria for adoption of this innovation (see

2 0
for example Mishra and Bisht, 2013; Shareef et al., 2018; Lashitew et al., 2019). The
second tries to understand how mobile currencies, whose supply is rapidly increasing,

ch
influence the conduct of monetary policy (see Mbiti and Weil, 2011, 2013; Mutsune, 2015).

a r
Finally, the third area—in which this study is embedded— coincides with the arrival of
mobile banking in developing countries, particularly in SSA, with platforms such as Smart

f t ,M
Money and G-CASH in the Philippines and M-Pesa in Kenya. Indeed, these different
experiences have fuelled a significant debate about the financial inclusion potential of

a
dr
mobile banking in developing countries (see for example Mbiti and Weil, 2011; Jack and
Suri, 2011; Assadi and Cudi, 2011; Hamdi, 2011; Chaix and Torre, 2015; Llewellyn-Jones,

d
2016, to name a few). Unfortunately, there are few studies on the relationship between
e
pt
mobile banking diffusion and barriers to financial inclusion with even sparser publications
that focus on SSA as a study region.

c e
In terms of geographical constraints, Allen et al. (2014) investigates whether Africa has

A c
a financial inclusion gap compared to other developing regions, and how mobile banking
could help to address this. To do so, they use a sample of 148 developing countries,
including 45 African countries, on which they estimate a multiple linear regression model
using the least squares method. Their findings suggest on one hand that, African countries
have a serious gap in financial inclusion compared to other developing countries. On the
other hand, they conclude that the penetration of mobile banking, especially the use of
phones for money transfers, is higher in Africa than anywhere else. Mobile banking can,
they argue, overcome infrastructural problems in Africa and improve financial inclusion.
Nonetheless, this study do not investigate the link between the diffusion of mobile banking
and the constraints to financial inclusion associated with geographic remoteness.
King (2012) investigates whether mobile banking is able to break what he calls the
”tyranny of distance”5 from banking infrastructure in Kenya. He uses two samples col-
lected in Kenya in 2006 and 2009 on 4308 and 6527 individuals respectively. The results
5
The author defines the ”tyranny of distance” as a situation in which the probability of an individual
being banked depends on his proximity to a bank branch.

6
 Epo, Mougnol and Kamdem (forthcoming)

obtained, after estimating several variants of the probit model, show that, despite the
increase in the number of bank branches in Kenya between 2006 and 2009, the distance
and time needed to go to a branch remain negatively related to the probability of having
a bank account. In contrast, the results show that mobile banking can overcome the
”tyranny of distance” from banking infrastructure and improve financial inclusion for all
economic groups in Kenya.
Regarding cost constraints, Ky (2016) focuses on factors related to mobile banking
penetration in developing countries. He estimates a multiple linear regression model on a
sample of 72 developing countries including 32 SSA countries. The results show that in
developing countries, and particularly in SSA, the diffusion of mobile banking is related

2 2
to, among other factors, the costs of banks on total assets, the average cost of transfers
and agricultural payments. Mbiti and Weil (2011) show that in Kenya mobile banking

2 0
has led to increased competitiveness in the financial market, particularly in the money

h
transfer market, by setting more affordable fees than traditional providers. Similarly,

c
a r
Kumar et al. (2010) analyse similar studies in India and Brazil where the use of mobile
banking by banks and microfinance institutions has led to a reduction in costs for them

,M
and their customers. The study however limits the analyse to the supply side and does
not include demand side factors.
f t
Finally, Bair and Tritah (2019) study the impact of the introduction of mobile money
a
dr
since 2010 in Madagascar on the frequency and volume of money transfers between house-
holds. The approach adopted a selection model combined with an instrumental variable

e d
approach. their results show that, though transaction cost associated with money transfer
t
services represent a barrier to the poor, new technologies such as mobile money could mit-
p
e
igate these barriers and reduce financial exclusion. Indeed, a CGAP6 study reported by
c
A c
McKay and Pickens (2010) find that, on average, mobile transactions were 19% cheaper
than in-branch transactions and that this rate could be as high as 50% if customers used
this channel for medium-term savings and bill payments.
Overly, appraising the literature the key gaps this paper attempts to fill are: (a) it
contributes empirically to the very sparse studies investigating the relationship between
mobile banking diffusion and the different constraints that hinder financial inclusion; (b)
it focuses on Africa which has reported significant increases in mobile phone usage; (c)
it peruses this relationship from barriers to financial inclusion associated to cost and
geography and (d) the measure of barriers is novel because we elect for the share of
unbanked people who identify cost or geographical distance as the main barrier to holding
an account in a formal financial institution (Demirgüç-Kunt et al., 2018; Klapper et al.,
2019). This indicator has the advantage of being measured directly on the population
concerned by banking exclusion.
6
Consultative Group to Assist the Poor.

7
 Epo, Mougnol and Kamdem (forthcoming)

3 Methodology
3.1 Model specification
This paper attempts to demonstrate that cost and geographical constraints to formal
financial services stir the adoption of mobile banking in SSA. Following Allen et al.,
(2014) and Ky (2016), we specify a multiple linear regression model as follows:

M Bi = α0 + α1 constraintsi + α2 Xi + εi (1)

where i = 1 . . . N , with i representing the different SSA countries under consideration

and N the number of countries in the sample. The dependent variable M Bi represents the

2 2
diffusion of mobile banking in country i and is computed separately for account access to

2 0
and use of mobile banking. To compute access to mobile banking for each of the countries
in this study, we calculate indicators associated with having a mobile money account

ch
(Senou et al., 2019). They include the share of the poorest 40% of the population that

a r
have a mobile money account, the proportion of the richest 60% of the population having

,M
a mobile money account, the percentage of rural populations having a mobile money
account and the proportion of the entire population having a mobile money account.

f t
To compute the use of mobile banking, we calculate two measures. These include the
a
dr
proportion of individuals aged fifteen and above that transfer money using their mobile
money accounts and the percentage of these individuals in the population that pay bill

d
via their mobile phones (Asongu, 2018). Xi is a vector of the other covariates that explain
e
t
the adoption of mobile banking in developing regions. The other parameters, αi and εi ,
p
e
represent respectively the coefficients that will be estimated and the error term that takes

c
A c
into account the imperfections of the model.
constraintsi represents the barriers to financial inclusion in a given country and com-
prises: (a) cost constraints measured by the proportion of the unbanked who consider the
high cost of formal financial services as the main barrier to holding an account and (b)
geographical constraints measured by the proportion of the population not having a bank
account and indicate that the major barrier is the physical distance to banking facilities
(Demirgüç-Kunt et al., 2018; Klapper et al., 2019). The presence of these constraints in
SSA’s financial markets makes it difficult for individuals to access traditional financial
services, and could therefore encourage the adoption of mobile banking by individuals as
an alternative channel for their financial transactions. We therefore expect a positive sign
for these variables in line with the assumption suggested in the introduction.
Table 1 below describes the variables, their expected signs and the sources. The
choice of these variables is justified by the numerous works that show their link with the
adoption of mobile banking in developing countries (Teo et al., 2012; Allen et al., 2014;
Della Peruta, 2015; Ky, 2016; Asongu, 2018; Senou et al., 2019; Akinyemi and Mushunje,

8
 Epo, Mougnol and Kamdem (forthcoming)

2020).

Table 1: Control variables

Variable Description Expected Source
sign
Domestic Sent or received domestic remittances in the past + GFD
remittances year (% age 15+)
Inflation Inflation, consumer price (annual %) - WDI
Population Population density (people per sq. km of land + WDI
density area)
GDP per capita GDP per capita (constant 2010 US$) +/- WDI
Lending inter- Lending interest rate (%) + WDI
est rate
Mobile phone Cellular subscription (per 100 people) +
2 2WDI
Internet usage
Population
Individuals using the Internet (% of population) +
total population +
2 0 WDI
WDI
Primary school School enrolment, primary (% gross)
Mobile money Number of registered mobile money agent out- +
ch + WDI
FAS
agent lets per 1,000 km2
a r
Sources: Authors. Note: Global Findex Database (GFD), World Development

,M
Indicators (WDI), Financial Access Survey (FAS).

f t
a
There is a possible strong linear association between cost and geographical constraints

dr
leading to a possible multicollinearity in the regression. This because, unbanked individ-
uals tend to report high cost and long distances as the two main limitations to financial

e d
inclusion (see 2017 Global Findex Database survey results). To resolve this possible mul-

pt
ticollinearity, we run two separate estimations (Wheeler and Tiefelsdorf, 2005) for cost

e
and geographical constrains, respectively. Equation 2 and 3 are expressions of Equation 1
c
A c
which consider cost constraints for the former and geographical constraints to financial
inclusion for the latter as indicate below:

M Bi = β0 + β1 costconsti + β2 Xi + εi (2)

M Bi = γ0 + γ1 geoconsti + γ2 Xi + εi (3)

3.2 Estimation strategy

Allen et al. (2014) and Ky (2016) use the ordinary least squares (OLS) method to estimate
the parameters of their models investigating issues of mobile banking. This is principally
driven by the nature of the data which is cross sectional in nature. In this paper, we
run both the OLS and the generalized least squares (GLS) method. This is principally
motivated by the fact that, cross-sectional data are subject to heteroscedasticity, which
runs counter to the standard OLS assumptions. To overcome such bias, we run the gen-

9
 Epo, Mougnol and Kamdem (forthcoming)

eralized least squares (GLS) method. This method applies OLS estimation procedure to
transformed data to correct the heteroscedasticity problem. We further apply the Jack-
nife and bootstrap resampling methods that reduce estimation bias and provide robust
results (Flachaire, 2000; Azubuike et al., 2019; Fanta and Makina, 2019).

3.3 Data of the study

The data used in this study are obtained from three databases, namely the Global Findex
Database (GFD) and the World Development Indicators (WDI) of the World Bank, as
well as the Financial Access Survey (FAS) of the International Monetary Fund. Our
indicators of cost and geographical constraints are only available for 2017. Therefore, we

2 2
opt for a cross-sectional analysis in 2017 on a sample of 35 SSA countries (see Table 5 in
the annexes for the list of countries).
2 0
4 Empirical results ch
a r
,M
4.1 Preliminary analysis

f t
The descriptive statistics of the variables used in this study are reported in Table 2,

a
whilst the correlation matrix between the variables in the study are reported in Table 9

dr
in the annex. Concerning the different measures of mobile banking diffusion, we observe

d
a certain homogeneity of the sample, with standard deviations lower than the means,

e
pt
except for the proportion of adults paying utility bills using mobile phone. Table 2
indicates that about 24% of the adult population (15 years and older) have a mobile

c e
money account. This proportion was18% for the poorest 40% of the population, 29%

A c
of the richest 60% of the population and 22% for the population in rural areas. In
addition, money transfers by mobile phone are effectuated by more than 24% of the total
population. This is a clear indication of the high diffusion of mobile banking in SSA in
recent decades, as revealed by numerous international surveys, including the World Bank’s
Global Findex (Demirgüç-Kunt et al., 2018). These observations are similar regarding
constraints to financial inclusion. Over 28% of the unbanked in SSA reported high cost of
financial services as their main obstacle, while 25% indicated that difficulties due to the
geographical distance to bank branches constitute their main hindrance. This corroborates
the observation that, the two constraints are very prevalent in the region.
Table 8 in the annex demonstrates that in SSA, compared to over 100 developing
countries7 , there is a strong association between all the different measures of mobile
banking diffusion constructed in this paper for both cost and geographical constraints.
This is however not the case for developing countries where only the measure proportion
7
The data concerning the other developing countries come from the same sources presented above for
SSA countries, namely GFD, WDI and FAS.

10
 Epo, Mougnol and Kamdem (forthcoming)

Table 2: Descriptive statistics

Variable Mean Std. Dev. Min Max
Mobile money account 0,2443 0,1594 0,0032 0,7293
Mobile money account (poorest) 0,1774 0,1322 0,0000 0,5936
Mobile money account (richest) 0,2888 0,1808 0,0053 0,8193
Mobile money account (rural) 0,2239 0,1553 0,0032 0,7259
Remittances using mobile phone 0,2443 0,1483 0,0022 0,6304
Pay bills using mobile phone 0,0659 0,0707 0,0000 0,3710
Cost constraints 0,2898 0,1009 0,0515 0,5314
Geographical constraints 0,2547 0,0929 0,0614 0,5112
Domestic remittances 0,4406 0,1189 0,2387 0,7064
Inflation 0,1116 0,3213 -0,0154 10,8785
Lending interest rate
Mobile phone
0,1484
83,7112
0,1217
36,2276
0,0514
25,5613
2 2 0,6000
155,2324
Primary school
Population
0,8310 0,1399
26900000 36800000
0,4426
2
1264613 0 0,9734
191000000
Internet usage 0,2218 0,1441
ch 0,0434 0,5617
mobile money agent
GDP per capita
Population density
309,9646 680,4116

a
2098,5140 2666,7730
91,3943 129,4439
r 0,0482
371,1088
2,9183
3385,9340
10199,4800
622,9621

,M
Sources: Authors

f t
a
of individual aged 15 and above that indicated using the mobile money account to transfer

dr
funds was significantly positive. This highlights the need to focus on SSA because these
measures have the merit of reflecting characteristics of poor unbaked individuals, thereby

e d
having some policy relevance for SSA.

pt
We also observe a high difference, between the minimum and maximum values in

c e
the sample revealing some form of heterogeneity, which is treated by running the GLS

A c
estimates. This reflects a strong difference between countries with lower financial devel-
opment and their corresponding counterparts having higher financial development levels.
Outliers therefore exist in the sample, such as Kenya, which is the world leader in mobile
money with an adoption rate of almost 73%, and Ethiopia, which has a very low adoption
rate of 0.3%. Ideally, these countries should be removed from the study. But this will not
be done, given the already small sample size.
To preliminary assess the relationship we investigate, we plotted the fitted scatter
plots between indicators of mobile banking adoption in SSA and measures of cost and
geographical constraints to financial inclusion.
As shown in Figure 1, there is a positive correlation between (a) mobile money account
ownership among the poorest 40% and cost constraints and (b) mobile money account
ownership among rural people and geographical constraints. Appraising the different
countries suggest that Kenya and Uganda have high levels of both constraints to financial
inclusion and mobile money adoption. Likewise, countries like Ethiopia and Nigeria have
both low levels of constraints to financial inclusion as well as low levels of mobile money

11
 Epo, Mougnol and Kamdem (forthcoming)

Figure 1: Mobile money account vs constraints to financial inclusion

(a) Momopoor vs Costconst

2
(b) Momorural vs Geoconst

2
0
Sources: Authors’ construction based on data from the World Bank (GFD)

2
h
Figure 2: Mobile transfers and payments vs constraints to financial inclusion
c
a r
f t ,M
a
dr
e d
pt
c e
(a) Remittmop vs Costconst (b) Billsmop vs Geoconst

c
Sources: Authors’ construction based on data from the World Bank (GFD)

A
adoption. In contrast, countries such as Madagascar and the Democratic Republic of
Congo reporting high constraints to financial inclusion, nonetheless have low mobile money
adoption. Similarly, Ghana and Zimbabwe both show high rates of adoption of mobile
banking despite lower levels of financial inclusion constraints. The same trends are equally
observed regarding the correlation between financial inclusion constraints and the use of
mobile banking for money transfers and bill payments (see Figure 2).

4.2 Regression results

Table 3 and Table 4 (page 14 and page 16)below present the results of the linear regression
between indicators of mobile banking diffusion in SSA in terms of access and usage, and
constraints to financial inclusion. The results of the multicollinearity tests can be found
in the annexes (Table 6 and Table 7). For all the variable, the variance inflation factors

12
 Epo, Mougnol and Kamdem (forthcoming)

(VIF) are low (less than 10) with high tolerances (1/VIF greater than 0.1). This indicates
an absence of multicollinearity.
Table 3 highlight that, the presence of cost constraints to financial inclusion is sig-
nificantly and positively related to the diffusion of mobile banking in SSA, in terms of
both access and usage. In terms of access, there is a positive and statistically signifi-
cant relationship between cost constraints and the proportion of adults holding a mobile
money account among the poorest 40% of the population, as well as among the richest
60%. Regarding usage, results also show a positive and statistically significant association
between the proportion of adults facing cost constraints and the use of mobile phones for
money transfers on the one hand, and for bill payments on the other. These results also

2 2
remain significant and positive when the Jacknife and bootstrap resampling techniques
are used for further investigation. This confirms the robustness of the results obtained.

2 0
These results show that strong cost constraints to financial inclusion in SSA drive the

h
diffusion of mobile banking in SSA. This effect is evident for both the poorest 40% of the

c
a r
population and the richest 60%. This aligns with findings from the World Bank’s Global
Findex survey which argue that there has been an unprecedented success of fintechs such

,M
as mobile banking in developing countries, and particularly in SSA. This success can
partly be explained by the low level of development of financial systems that results in a
f t
deficit of financial services (Demirgüç-Kunt et al., 2018). Among other multiple causes
a
dr
also identify the high cost of financial services (Granata et al., 2014; Guérineau and
Jacolin, 2014). To resolve the high-cost issues, mobile banking, unlike traditional finance,

e d
allows users to benefit from basic financial services at low cost. McKay and Pickens (2010)
t
reveal that mobile phone transactions can cost up to 50% less than branch transactions.
p
e
This highlights the situation in Kenya, where mobile banking has led to an increased
c
A c
competitiveness of the financial market, especially in money transfer and payments, by
setting fees that are much more affordable than those of traditional providers (Mbiti and
Weil, 2011).
Our regression equally highlights the significant influence of other factors in the dif-
fusion of mobile banking in SSA (Table 3). In particular, the rate of mobile telephone
subscription is positively and significantly related to the adoption of mobile money among
the poorest 40% and for the measure of mobile banking diffusion capture by the propor-
tion of individuals aged fifteen and above who transfer money using their mobile phone
accounts. The rate of enrolment in primary school has a positive influence on the per-
centage of individuals that use of mobile phones for payments. We also note the positive
relationship between the variables indicating the number of mobile money agents per 1000
km2 and the two measures of usage of mobile banking diffusion. These results are in line
with the findings of Ky (2016), Asongu (2018) and Akinyemi and Mushunje (2020).

13
 Table 3: Mobile banking and cost constraints to financial inclusion
Momopoor Momorich Remittmop Billsmop

Ac
VARIABLES Gls Jacknife Bootstrap Ols Jacknife Bootstrap Gls Jacknife Bootstrap Gls Jacknife Bootstrap
Costconst 0.497** 0.497* 0.497* 0.864*** 0.864** 0.864** 0.394** 0.394* 0.394* 0.306** 0.306** 0.306**

ce
(0.0382) (0.0837) (0.0596) (0.00966) (0.0307) (0.0148) (0.0416) (0.0759) (0.0933) (0.0297) (0.0384) (0.0321)
Interestlend 0.0570* 0.0570 0.0570 -0.313 -0.313 -0.313 -0.0346 -0.0346 -0.0346

pt
(0.0548) (0.698) (0.571) (0.161) (0.142) (0.119) (0.264) (0.333) (0.445)
lMobilcel 0.123** 0.123** 0.123*** 0.0888 0.0888 0.0888 0.141*** 0.141*** 0.141*** 0.0213 0.0213 0.0213

e
(0.0107) (0.0222) (0.00822) (0.179) (0.124) (0.215) (0.000460) (0.00220) (0.00276) (0.261) (0.308) (0.383)

d
Primschool -0.178* -0.178 -0.178 0.170* 0.170* 0.170*

dr
(0.0906) (0.176) (0.184) (0.0526) (0.0602) (0.0500)
lPopulation 0.0205 0.0205 0.0205 0.0306 0.0306 0.0306

a
(0.351) (0.403) (0.385) (0.134) (0.177) (0.154)
14

ft,
Inflation 0.104 0.104 0.104 -0.168*** -0.168 -0.168 0.0473 0.0473 0.0473

Epo, Mougnol and Kamdem (forthcoming)

(0.301) (0.594) (0.768) (0.000548) (0.164) (0.272) (0.112) (0.729) (0.770)
lMomoag km 0.0137*** 0.0137*** 0.0137*** 0.00801** 0.00801* 0.00801*

M
(0.000105) (0.000676) (0.000637) (0.0373) (0.0558) (0.0696)

ar
Constant -0.706 -0.706 -0.706 -0.320 -0.320 -0.320 -1.032** -1.032** -1.032** -0.283** -0.283** -0.283**
(0.181) (0.234) (0.179) (0.302) (0.257) (0.374) (0.0185) (0.0353) (0.0413) (0.0131) (0.0222) (0.0125)

ch
Observations 35 35 35 35 35 35 35 35 35 35 35 35
R-squared 0.367 0.367 0.367 0.250 0.250 0.250 0.545 0.545 0.545 0.400 0.400 0.400

20
F 5.410*** 3.208** 2.498* 1.963 18.74*** 9.333*** 4.955*** 1.992*
N reps 35 100 35 100 35 100 35 100

22
chi2 16.44*** 8.753* 32.27*** 14.91**
Sources: Authors. Robust pval in parentheses
*** p<0.01, ** p<0.05, * p<0.1
 Epo, Mougnol and Kamdem (forthcoming)

Concerning geographical constraints, Table 4 indicate that, they also relate positively
and significantly with the indicators for access and usage associated to mobile banking
diffusion. The proportion of people not having a bank account and who reported geo-
graphical distance as the main obstacle opt to use mobile banking as a palliative when we
consider the entire population and especially individuals residing in the rural area. This
same measure of mobile banking diffusion is positively and significantly associated with
mobile phone usage regarding money transfers and bill payments. These results remain
statistically significant for the Jacknife and bootstrap techniques estimation methods,
therefore confirming the robustness of the preceding results (OLS, GLS).
These results show that, in SSA, geographical constraints to financial inclusion are

2 2
accompanied by a significant diffusion of mobile banking. This is particularly true in
rural areas, where the problem of isolation is particularly acute. Geographical remoteness

2 0
from banking infrastructure is a major cause of banking exclusion in developing countries

h
and particularly in SSA (Demirgüç-Kunt et al., 2018; Klapper et al., 2019). According to

c
a r
Allen et al. (2014), mobile banking has the potential to facilitate financial development
and financial inclusion in Africa, mainly outside of major urban areas. Through remote

,M
mobile phone transactions and their huge network of agents, mobile banking providers
can serve remote areas where there are no bank branches. This potential for technological
f t
leapfrogging of mobile banking had already been highlighted by Datta et al. (2001),
a
dr
who made it a strong argument for the implementation of this innovation in developing
countries. In Kenya, for example, King (2012) has shown that mobile banking is able to

e d
break the ”tyranny of distance” from bank branches, thereby increasing financial access
t
for people in remote areas.
p
e
The regression results overly reveal the statistically significant role of other factors
c
A c
reported in the literature. These are domestic money transfers (as a % of the population),
which are positively associated with the adoption of mobile banking and its use for mobile
transfers and payments. Internet coverage, which is positively related to the adoption
of mobile banking and its use for transfers and payments. However, their effect when
considering the measuring of mobile banking diffusion as proportion of individuals in rural
areas having a mobile money account is not statistically significant. Inflation discourages
the adoption and use of mobile banking, as does GDP per capita. Finally, the number
of mobile banking agents per 1000 km2 favours the adoption of mobile banking in rural
areas and its use for money transfers and mobile phone payments.

15
 Table 4: Mobile banking and geographical constraints to financial inclusion
Momo Momorural Remittmop Billsmop

Ac
VARIABLES Ols Jacknife Bootstrap Gls Jacknife Bootstrap Gls Jacknife Bootstrap Gls Jacknife Bootstrap
Geoconst 0.574* 0.574 0.574* 0.630** 0.630* 0.630** 0.355* 0.355* 0.355* 0.314** 0.314** 0.314**

ce
(0.0581) (0.181) (0.0966) (0.0447) (0.0932) (0.0425) (0.0575) (0.0811) (0.0831) (0.0173) (0.0180) (0.0246)
Domremitt 0.733*** 0.733 0.733** 0.583* 0.583 0.583* 0.651*** 0.651*** 0.651***

pt
(0.00765) (0.150) (0.0394) (0.0850) (0.154) (0.0978) (8.61e-05) (0.000231) (0.000273)
Internetuse 0.621** 0.621*** 0.621*** 0.174 0.174 0.174 0.246** 0.246** 0.246* 0.187*** 0.187*** 0.187***

e
(0.0211) (0.00347) (0.00164) (0.347) (0.449) (0.481) (0.0171) (0.0480) (0.0630) (0.000270) (0.000603) (0.00300)

d
Inflation 0.0766 0.0766 0.0766 -0.492*** -0.492*** -0.492** -0.142*** -0.142** -0.142 -0.0658*** -0.0658** -0.0658
(0.399) (0.849) (0.756) (7.47e-07) (0.000438) (0.0219) (0.000417) (0.0260) (0.106) (0.00115) (0.0262) (0.450)

dr
lGdppercapita -0.126*** -0.126* -0.126**
(0.00488) (0.0793) (0.0147)

a
lMomoag km 0.00773*** 0.00773** 0.00773** 0.0113*** 0.0113*** 0.0113*** 0.00431*** 0.00431*** 0.00431***
16

ft,
(0.00580) (0.0189) (0.0287) (0.000176) (0.00439) (0.000284) (0.00428) (0.00859) (0.00785)

Epo, Mougnol and Kamdem (forthcoming)

lPopdensity 0.0168 0.0168 0.0168
(0.132) (0.210) (0.161)

M
Primschool 0.0492 0.0492 0.0492
(0.298) (0.328) (0.338)

ar
Constant 0.476* 0.476 0.476 -0.390* -0.390 -0.390* -0.280*** -0.280*** -0.280*** -0.104** -0.104** -0.104**
(0.0747) (0.225) (0.102) (0.0802) (0.148) (0.0999) (0.00390) (0.00691) (0.00416) (0.0141) (0.0224) (0.0387)

ch
Observations 35 35 35 35 35 35 35 35 35 35 35 35
R-squared 0.529 0.529 0.529 0.626 0.626 0.626 0.775 0.775 0.775 0.344 0.344 0.344

20
F 5.245*** 4.234*** 105.5*** 6.002*** 33.42*** 11.08*** 14.63*** 12.18***
N reps 35 100 35 100 35 100 35 100

22
chi2 35.55*** 21.46*** 71.73*** 23.51***
Sources: Authors. pval in parentheses
*** p<0.01, ** p<0.05, * p<0.1
 Epo, Mougnol and Kamdem (forthcoming)

5 Conclusion
Mobile banking is a major issue in SSA and plays a key role towards enhancing financial
inclusion. However, existing studies did not clearly establish the link between the high
diffusion of mobile banking in SSA and constraints (cost and geography) to financial
inclusion. This study aimed to fill this gap by undertaking an empirical assessment of the
relationship using different measures of mobile banking diffusion that encompasses the
specificities of the SSA continent. The empirical method consisted of a multiple linear
regression, estimated by means of the ordinary and generalized least squares methods as
well as the Jacknife and bootstrap resampling techniques for robustness.
Results indicate that, strong cost constraints to financial inclusion cause an increase

2 2
in the diffusion of mobile banking, in terms of access (possession of a mobile money

2 0
account) among both the poorest 40% and the richest 60%, but also in terms of use
(money transfers and mobile payments). The same is true for geographical constraints

ch
as they positively relate to the possession of a mobile money account when considering

a r
the entire population, particularly those residing in rural areas. This affirmation is also
verified when we consider other measures of mobile banking usage such as money transfers
and bill payments via mobile phones.

f t ,M
The main policy herald should upgrade telecommunications road maps to encompass
a
dr
challenges that mobile banking and mobile financial services in general pose for financial
sector regulation given the fast-paced innovations registered in the mobile banking sector.

d
This should be primarily aimed at developing infrastructures that act as barrier linked
e
t
to geographical constraints. By digitalizing and reducing ICT infrastructure, through

p
c e
robust investment schemes, governments can mitigate geographical barriers and vulgarize
usage of ICT facilities. This coupled with appropriate sensitization and ICT-education

A c
programmes should help poor households. Furthermore, the fast-paced development of
ICT and the constant inclusion of new plays should stir governments and authorities to
enhance institutional and regulatory framework that encourage and attract investment
in ICTs and adoption of technology that optimizes mobile baking. More still, given that
rural areas are disproportionately disfavoured authorities should design incentive schemes
that encourage the financial service providers and ICT enterprises to invest in rural areas.

References
Akinyemi, B. E., and Mushunje, A. (2020). “Determinants of mobile money
technology adoption in rural areas of Africa.”, Cogent Social Sciences, 6 (2020).
Allen, F., Carletti, E., Cull, R., Qian, J. Q., Senbet, L., and Valenzuela, P.
(2014). “The African financial development and financial inclusion gaps.”, Journal
of African Economies, 23, 614–642.
Asongu, S. A. (2018). “Conditional determinants of mobile phones penetration and

17
 Epo, Mougnol and Kamdem (forthcoming)

mobile banking in sub-Saharan Africa.”, Journal Knowledge Economy, 9(2018), 81–

135.
Assadi, D., and Cudi, A. (2011). “Le potentiel d’inclusion financière du “mobile
banking”. Une étude exploratoire.”, Management Avenir (6), 227–243.
Azubuike, I. M., Chris-Chinedu, and Nonso, J. (2019). “Jackknife and bootstrap
techniques in the estimation of regression parameters.”, International Journal of
Mathematics Trends and Technology, 65(12), 25–35.
Bair, S., and Tritah, A. (2019). “Mobile money and inter-household financial flows:
Evidence from Madagascar.”, Revue Economique, 70(5), 847–871.
Barnes, S., and Corbitt, B. (2003). “Mobile banking: concept and potential.”, Int.
J. Mobile Communications 3 (1), 273–288.

2 2
en développement.”, Revue Economique, 66, 703–727.
2 0
Chaix, L., and Torre, D. (2015). “Le double rôle du paiement mobile dans les pays

ch
Datta, A., Pasa, M., and Schnitker, T. (2001). “Could mobile banking go global?’,
The McKINSEY Quarterly: Emerging Markets (4), 71–80.
a r
Davis, F. D. (1989). “Perceived usefulness, perceived ease of use, and user acceptance

f t ,M
of information technology.”, MIS Quarterly, 13(3), 319–340.
Della Peruta, M. (2015). “Mobile money adoption and financial inclusion objectives:
a
dr
A macroeconomic approach through a cluster analysis.”, GREDEG Working paper
series, No. 2015-49.

d
Demirgüç-Kunt, A., and Klapper, L. (2012). “Financial inclusion in Africa: An
e
t
overview.”, Policy Research Working Paper, No. 6088.

p
c e
Demirgüç-Kunt, A., Klapper, L., Singer, D., Ansar, S., and Hess, J. (2018).
“The global Findex database 2017: Measuring financial inclusion and the fintech
c
revolution”, Washington, DC: World Bank.
A
Fanta, A. B., and Makina, D. (2019). “The relationship between technology and
financial inclusion: Cross-sectional evidence.”, In Extending financial inclusion in
Africa (Vol. 65, pp. 211–230) Elsevier.
Flachaire, E. (2000). “Les méthodes du bootstrap dans les modèles de régression.”,
Economie and Prévision, 1(142), 183–194.
Granata, P., Kibuuka, K., and Yira, M. (2014). “Le coût du financement en
Afrique : quelles politiques publiques pour réduire son coût et renforcer l’inclusion
financière ?” Revue d’Économie Financière, Décembre 2014, 121–149.
Guérineau, S., and Jacolin, L. (2014). “L’inclusion financière en Afrique subsa-
harienne : faits stylisés et déterminants”, Revue d’Économie Financière, 4(116),
57–80.
Hamdi, H. (2011). “Can e-payment systems revolutionize finance of the less devel-
oped countries? the case of mobile payment technology.”, International Journal of
Economics and Financial Issues, 1(2), 46–53.

18
 Epo, Mougnol and Kamdem (forthcoming)

Jack, G., and Suri, T. (2011). “Mobile money: The economics of MPESA”, NBER
Working Paper No.16721.
King, M. (2012). “Is mobile banking breaking the tyranny of distance to bank infras-
tructure? evidence from Kenya.”, IIIS Discussion Paper No.412.
Klapper, L., Ansar, S., Hess, J., and Singer, D. (2019). “Sub-Saharan Africa
series: mobile money and digital financial inclusion.”, FINDEX NOTE 1, Develop-
ment Research Group the World Bank.
Kumar, K., McKay, C., and Rotman, S. (2010). “Microfinance et banque mobile
: quelles expériences à ce jour ?’, Note Focus, No., 62.
Ky, S. (2016). “Essays on the rise of mobile financial technology in developing countries

de Limoge, Archives Ouvertes (HAL).

2 2
and its impacts on individual financial behaviours”. Thèse de Doctorat, Université

2 0
Ky, S., Rugemintwari, C., and Sauviat, A. (2018). “Does mobile money af-

Economies, 27(3), 285–320.

ch
fect saving behaviour? evidence from a developing country.”, Journal of African

a r
Lashitew, A. A., Tulder, R., and Liasse, Y. (2019). “Mobile phones for financial

,M
inclusion: What explain’s the diffusion of mobile money innovations?”, Research
Policy, No. 48, 1201–1215.

f t
a
Llewellyn-Jones, L. (2016). “Mobile money: part of the African financial inclusion

dr
solution?”, Economic Affairs, 36(2), 212–216.
Mbiti, I., and Weil, D. N. (2011). “Mobile banking: the impact of m-pesa in

e d
Kenya.”, NBER Working Paper Series, No. 17129.

pt
Mbiti, I., and Weil, D. N. (2013). “The home economics of e-money: Velocity, cash

c e
management, and discount rates of m-pesa users.”, American Economic Review:

A c
Papers and Proceedings, 3(103).
McKay, C., and Pickens, M. (2010). “Branchless banking 2010: Who’s served? at
what price? what’s next?”, Washington, D.C.: CGAP, September, No. 66.
Mishra, V., and Bisht, S. (2013). “Mobile banking in a developing economy: A
customer-centric model for policy formulation.”, Telecommunications Policy (37),
503–514.
Mutsune, T. (2015). “No Kenyan left behind: The model of financial inclusion
through mobile banking.”, Review of Business and Finance Studies, 6(1), 35–42.
Porteous, D. (2006). “The enabling environment for mobile banking in Africa”. De-
partment for International Development (DFID).
Rogers, E. M. (1983). Diffusion of innovations. The Free Press a Division of Macmil-
lan Publishing Co., Inc.
Senou, M. M., Ouattara, W., and Houensou, D. A. (2019). “Is there a bottle-
neck for mobile money adoption in WAEMU?”, Transnational Corporations Review,
11(2), 143–156.

19
 Epo, Mougnol and Kamdem (forthcoming)

Shareef, M. A., Baabdullah, A., Dutta, S., Kumar, V., and Dwivedi, Y. K.
(2018). “Consumer adoption of mobile banking services: an empirical examina-
tion of factors according to adoption stages.”, Journal of Retailing and Consumer
Services, 43(2018), 54–67.

Teo, A.-C., Tan, G. W.-H., Cheah, C.-M., Ooi, K.-B., and Yew, K.-T. (2012).
“Can the demographic and subjective norms influence the adoption of mobile bank-
ing?”, Int. J. Mobile Communications, 10(6), 578–597.

Wheeler, D., and Tiefelsdorf, M. (2005). “Multicollinearity and correlation among

local regression coefficients in geographically weighted regression.”, Journal of Ge-
ographic Systems, 2005(7), 161–187.

World Bank, G. (2018). “The little data book on financial inclusion 2018”, World
Bank: Washington, DC.
2 2
Annexes 2 0
ch
a
Table 5: Sample country
r
,M
Benin, Burkina Faso, Botswana, Chad, Cameroon, Central African Republic, Congo
Dem. Rep., Congo Rep., Cote d’Ivoire, Ethiopia, Gabon, Ghana, Guinea, Kenya,

f t
Lesotho, Liberia, Madagascar, Mali, Malawi, Mauritania, Mauritius, Mozambique,

a
Namibia, Niger, Nigeria, Rwanda, Senegal, Sierra Leone, South Africa, South Sudan,

dr
Tanzania, Togo, Uganda, Zambia, Zimbabwe.
Source: Authors.

e d
pt
Table 6: Results of VIF test for multicollinearity
e
cc
Model Momopoor Momorich Remittmop Billsmop
Variables VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF
Costconst
Interestlend
A 1.17
1.49
0.855564 1.29
0.670760 1.40
0.775303 1.04
0.711838
0.963355 1.24
1.20
0.803772
0.833911
lMobilcel 1.43 0.697650 1.35 0.740835 1.37 0.728657 1.53 0.651522
Primschool 1.03 0.967823 1.23 0.814122
lPopulation 1.21 0.823420 1.12 0.894335
Inflation 1.23 0.815977 1.64 0.609569 2.08 0.481069
lMomoag km 1.41 0.706848 2.23 0.449062
Mean VIF 1.27 1.32 1.32 1.59
Sources: Authors

20
 Epo, Mougnol and Kamdem (forthcoming)

Table 7: Results of VIF test for multicollinearity

Model Momo Momorural Remittmop Billsmop
Variables VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF
Geoconst 1.38 0.725941 1.20 0.829885 1.58 0.633235 1.38 0.723270
Domremitt 1.88 0.532037 1.65 0.605494 1.30 0.772078
Internetuse 3.20 0.312151 2.02 0.494445 1.62 0.617597 1.70 0.587559
Inflation 1.71 0.586505 4.29 0.233349 1.61 0.622678 1.40 0.713888
lGdppercapita 3.82 0.262008
lMomoag km 4.19 0.238652 1.48 0.674542 1.41 0.707142
lPopdensity 1.15 0.873178
2 2
Primschool
Mean VIF 2.18 2.45 1.45
1.45

2 0
1.47
0.688024

Sources: Authors
ch
a r
f t ,M
a
dr
Table 8: Mobile banking and constraints to financial inclusion (estimate on 100 developing

d
countries including our sample of SSA country)
e
VARIABLES
(1) (2)
t
Momopoor Remittmop

p
(3)
Billsmop VARIABLES
(1) (2)
Momorural Remittmop
(3)
Billsmop
Costconst 0.0328

c
(0.718) e0.188*
(0.0953)
0.0169
(0.624)
Geoconst 0.0374
(0.758)
0.0259
(0.738)
0.0943
(0.184)
Interestlend

lMobilcel
A c
-0.0640
(0.392)
0.0674* 0.0588** 0.0190*
Domremitt

Internetuse
0.408***
(0.00683)
0.137
0.431***
(0.000)
0.00332 0.0602*
(0.0612) (0.0392) (0.0733) (0.129) (0.934) (0.0614)
Primschool 0.166** 0.104** Inflation -0.0572** -0.0615*** -0.0241**
(0.0284) (0.0183) (0.0279) (0.000381) (0.0107)
lPopulation -0.000184 0.0106 lMomoag km 0.00206 0.0129*** 0.00640**
(0.986) (0.190) (0.727) (0.00192) (0.0145)
Inflation -0.0414 -0.0265** lPopdensity 0.00169
(0.202) (0.0122) (0.845)
lMomoag km 0.0111** 0.00348* Primschool 0.0445
(0.0293) (0.0558) (0.299)
SSA 0.154*** 0.215*** 0.0426*** SSA 0.0922** 0.134*** 0.0469***
(3.12e-06) (0) (0.00324) (0.0119) (2.55e-05) (0.00206)
Constant -0.404 -0.499** -0.163*** Constant -0.148* -0.144** -0.0800
(0.144) (0.0101) (0.00318) (0.0852) (0.0125) (0.138)
Observations 100 100 100 Observations 100 100 100
R-squared 0.237 0.468 0.184 R-squared 0.316 0.683 0.194
F 4.998*** 12.25*** 5.187*** F 7.093*** 0.431*** 4.024***
Source: Authors. Robust pval in parentheses, *** p<0.01, ** p<0.05, * p<0.1

21
View publication stats

Ac
ce
Table 9: Correlation matrix
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)

pt
(1) Momo 1.0000
(2) Momopoor 0.9687 1.0000

e
(3) Momorich 0.9932 0.9332 1.0000

d
(4) Momorural 0.9349 0.9101 0.9266 1.0000
(5) Remittmop 0.8996 0.9239 0.8690 0.8745 1.0000

dr
(6) Billsmop 0.8198 0.8309 0.7970 0.7770 0.8095 1.0000
(7) Costconst 0.3918 0.4098 0.3747 0.3506 0.2831 0.3013 1.0000

a
(8) Geoconst 0.2112 0.2175 0.2034 0.1878 0.1095 0.1089 0.7290 1.0000
22

ft,
(9) Domremitt 0.4829 0.5312 0.4500 0.5235 0.5872 0.5114 0.0958 -0.1799 1.0000

Epo, Mougnol and Kamdem (forthcoming)

(10) Inflation -0.2756 -0.2545 -0.2802 -0.5681 -0.2502 -0.1456 -0.0752 -0.0233 -0.2693 1.0000
(11) Interestlend 0.2359 0.1397 0.2758 0.1242 0.0004 0.1162 0.4422 0.2115 0.0105 -0.0529 1.0000

M
(12) Internetuse 0.2506 0.2492 0.2460 0.2785 0.3095 0.3336 0.0079 -0.3584 0.5989 -0.2016 -0.0031 1.0000
(13) Primschool 0.2502 0.1348 0.2990 0.3297 0.1766 0.2591 0.0254 -0.1431 0.2000 -0.3301 0.0695 0.2326 1.0000
(14) lPopdensity -0.0083 0.0059 -0.0146 0.0919 0.1157 0.0262 -0.0940 -0.2299 0.2238 -0.0802 -0.2843 0.0368 0.1186 1.0000

ar
(15) lGdppercapta 0.2575 0.2455 0.2581 0.5216 0.2620 0.2307 0.0084 -0.1302 0.5000 -0.8946 0.0384 0.5231 0.3182 0.0231 1.0000
(16) lMobilcel 0.2422 0.2891 0.2153 0.3608 0.4037 0.3111 -0.1209 -0.4483 0.6598 -0.4179 -0.2008 0.7077 0.2868 0.0551 0.6611 1.0000

ch
(17) lPopulation -0.0008 0.0472 -0.0234 0.0367 0.0786 0.0357 0.0218 0.2737 -0.1909 0.0280 -0.2771 -0.2923 -0.2954 0.3517 -0.1520 -0.3008 1.0000
(18) lMomoag km -0.0984 -0.0631 -0.1133 -0.3710 0.0072 0.0341 -0.0062 -0.1214 -0.1564 0.8632 0.0499 -0.1244 -0.1815 0.0567 -0.7836 -0.2569 -0.1033 1.0000
Sources: Authors.

20
22