Financial Inclusion and Development: A Cross Country Analysis

Mandira Sarma 1 and Jesim Pais 2

August 2008

(Preliminary draft for presentation at the Annual Conference of the Human Development
and Capability Association, New Delhi, 10-13 September 2008)

Abstract

This paper presents a cross country empirical analysis of the relationship between
financial inclusion and development. Using the index of financial inclusion developed in
Sarma (2008), the paper attempts to identify the factors that are significantly associated
with financial inclusion. Levels of human development and financial inclusion in a
country move closely with each other, although a few exceptions exist. Among
socioeconomic factors, as expected, income is positively associated with the level of
financial inclusion. Going beyond income, inequality, literacy and urbanisation are other
important factors. Further, physical infrastructure for connectivity and information are
also significantly associated with financial inclusion. Among the banking sector
variables, NPA and CAR are negatively associated with financial inclusion. Government
ownership of banks is not significantly associated with financial inclusion while foreign
ownership is found to be negatively associated. Interest rate does not seem to be
significantly associated with financial inclusion.

KEY WORDS: Financial Inclusion, Index of Financial Inclusion, Human Development

Index

JEL Classification: G21, O16, O50

1 Indian Council for Research on International Economic Relations, Core 6A, India Habitat Centre, Lodhi
Road, New Delhi 110003, Ph: +91 11 24 64 52 18, Email: [email protected]
2 Institute for Studies in Industrial Development, Vasant Kunj, New Delhi, Ph: +91 11 26 89 1111,

Email:[email protected]
1
1. Introduction
Financial inclusion refers to a process that ensures the ease of access, availability and
usage of the formal financial system for all members of an economy. An inclusive
financial system has several merits. It facilitates efficient allocation of productive
resources and thus can potentially reduce the cost of capital. In addition, access to
appropriate financial services can significantly improve the day-to-day management of
finances. An inclusive financial system can help reducing the growth of informal sources
of credit (such as money lenders) which are often found to be exploitative. Thus, an all-
inclusive financial system enhances efficiency and welfare by providing avenues for
secure and safe saving practices and by facilitating a whole range of efficient financial
services.

The importance of an inclusive financial system is widely recognized in the policy circle
and recently financial inclusion has become a policy priority in many countries. Initiatives
for financial inclusion have come from the financial regulators, the governments and the
banking industry. Legislative measures have been initiated in some countries. For
example, in the United States, the Community Reinvestment Act (1997) requires banks
to offer credit throughout their entire area of operation and prohibits them from
targeting only the rich neighbourhoods. In France, the law on exclusion (1998)
emphasises an individual’s right to have a bank account. In the United Kingdom, a
‘Financial Inclusion Task Force’ was constituted by the government in 2005 in order to
monitor the development of financial inclusion.

The banking sector has also taken a lead role in promoting financial inclusion. In India,
the Reserve Bank of India (RBI) has initiated several measures to achieve greater
financial inclusion, such as facilitating ‘no-frills’ accounts and “General Credit Cards” for
low deposit and credit. The German Bankers’ Association introduced a voluntary code in
1996 providing for an ‘everyman’ current banking account that facilitates basic banking
transactions. In South Africa, a low cost bank account called ‘Mzansi’ was launched for
financially excluded people in 2004 by the South African Banking Association. Alternate
financial institutions such as micro-finance institutions and Self-Help Groups have also
been promoted in some countries in order to reach financial services to the excluded.

2
Literature on financial exclusion has defined it in the context of a larger issue of social
exclusion of certain groups of people from the mainstream of the society. Leyshon and
Thrift (1995) define financial exclusion as referring to those processes that serve to prevent
certain social groups and individuals from gaining access to the formal financial system. Carbo et al.
(2005) have defined financial exclusion as broadly the inability (however occasioned) of some
societal groups to access the financial system. According to Conroy (2005), financial exclusion is a
process that prevents poor and disadvantaged social groups from gaining access to the formal financial
systems of their countries. According to Mohan (2006) financial exclusion signifies the lack of access
by certain segments of the society to appropriate, low-cost, fair and safe financial products and services
from mainstream providers. A Government Committee on financial inclusion in India defines
financial inclusion as the process of ensuring access to financial services and timely and adequate credit
where needed by vulnerable groups such as the weaker sections and low income groups at an affordable
cost (Rangarajan Committee 2008).

Thus the definitions on financial inclusion/exclusion, conceptual as well as functional,

provide an indication that financial exclusion occurs mainly to people who are at the
margins of the society. The issues of financial exclusion have been of interest to scholars
for some time now. Of the issues raised in academic debates, an important question is
whether economic development leads to an all-inclusive financial system. It has been
observed that even ‘well-developed’ financial systems such as those in the US and the
UK have not succeeded to be ‘all-inclusive’ and certain segments of the population
remain outside the formal financial systems. 3 Another issue of interest is whether low
level of financial inclusion is associated with high income inequality (Kempson et al,
2004).

This paper attempts to examine the relationship between financial inclusion and
development. In doing so, it attempts to identify empirically country specific factors that
are associated with the level of financial inclusion. Empirical literature on determinants
of financial exclusion mostly comprises analyses based on primary surveys within a
country or a region. 4 In a recent paper, Beck et al (2007) have studied financial sector
outreach and its determinants by using cross country data. They have used several

3 For a review of financial exclusion in developed economies and the policy response to it, see Kempson
et. al. (2004).
4 See, for example, Solo and Manroth (2006) for Colombia, Siedman and Tescher (2004) for the US, Corr

(2006) for Ireland, Collard et al (2001) for UK, Djankov et al (2008) for Mexico and European
Commission (2008) for the European Union.
3
indicators of banking sector outreach and examined the determinants of each of these
indicators separately. In this paper, we use an index of financial inclusion developed in
Sarma (2008) to investigate macro level factors that can be associated with financial
inclusion. The paper first attempts to understand the relationship between IFI and the
Human Development Index (HDI), the most widely used development index. Then it
presents the results of an empirical analysis to determine country specific factors
associated with the level of financial inclusion. We find that at the macro level, the
factors responsible for financial exclusion are similar to those found in individual
countries through survey based analysis. Our results are also in line with Beck et al
(2007).

We organise the paper as follows. In Section 2, we briefly describe the Index of

Financial Inclusion (IFI), a multidimensional index developed to measure the
inclusiveness of a country’s financial system. In Section 3 we will present the broad
relationship between IFI and HDI, followed by an empirical analysis of factors
significantly associated with IFI in Section 4. Section 5 concludes the paper.

2. Index of Financial Inclusion (IFI) 5

The index of financial inclusion is a measure of inclusiveness of the financial sector of a
country. It is constructed as a multidimensional index that captures information on
various aspects of financial inclusion such as banking penetration, availability of banking
services and usage of the banking system. The IFI incorporates information on these
dimensions in one single number lying between 0 and 1, where 0 denotes complete
financial exclusion and 1 indicates complete financial inclusion in an economy. Sarma
(2008) has developed a method of computing the IFI for several dimensions of financial
inclusion. Based on the availability of comparable data, Sarma (2008) has computed the
values of IFI for 54 countries using the three basic dimensions of financial inclusion –
accessibility, availability and usage of banking services. Accessibility has been measured
by the penetration of the banking system given by the number of bank A/C per 1000
population. Availability has been measured by the number of bank branches and
number of ATMs per 100,000 people. The proxy used for the usage dimension is the
volume of credit plus deposit relative to the GDP. A dimension index for each of these
dimensions has been first computed by the following formula

5 The section draws mainly from Sarma (2008).

4
 Ai − mi
di = (1)
M i − mi

where
Ai = Actual value of dimension i
mi = minimum value of dimension i
Mi = 94th quantile value of dimension i

For the availability dimension, two separate indexes are first calculated - one for bank
branch and the second for ATMs. A weighted average of these two indexes, using 2/3rd
weight for bank branch index and 1/3rd weight for ATM index is considered as the index
for the availability dimension. 6

After calculating the dimension indexes, they are given the following weights – 1 for the
index of accessibility (penetration), 0.5 for the index of availability and 0.5 for the index
of usage. The reason for giving less weight for the indexes of availability and usage
dimensions in the present index is due to the lack of adequate data on some important
indicators that completely characterize these dimensions. For example, as far as
availability of banking services is concerned, many countries have moved towards
internet banking, thus reducing the importance of physical bank outlets. Therefore,
using data only on physical outlets (such as bank branches and ATMs) can give an
incomplete picture of the availability of banking services. Similarly, using data on credit
and deposit can only partially depict the usage of the financial system as other services of
the banking system, such as payments, transfers and remittances are not included. In the
absence of such data, a complete characterization of these dimensions is not possible.

After giving weights to the dimensions, the final IFI is computed as follows

(1 − p i ) 2 + (0.5 − a i ) 2 + (0.5 − u i ) 2
IFI = 1 −
1 .5
where pi, ai and ui denote respectively the weighted dimension indexes for the
dimensions accessibility (or penetration), availability and usage.

6 For further details on the concept and estimation of the IFI see Sarma (2008).
5
In this paper, we use the IFI computed by Sarma (2008) for 54 countries. 7 From within
this group of 54 countries, however, we have excluded countries that can be
unambiguously characterized as overseas financial centres (OFC). 8 The OFCs are
defined by the IMF as “centres where the bulk of financial sector activity is offshore on
both sides of the balance sheet (that is the counterparties of the majority of financial
institutions’ liabilities and assets are non-residents), where the transactions are initiated
elsewhere, and where the majority of the institutions involved are controlled by non-
residents. 9 Thus the empirical investigation in this paper is based on the data for 49
countries. 10

3. IFI and HDI

Table 1 presents the IFI computed for 49 countries and the corresponding human
development index (HDI) along with their ranks. Austria, with an IFI value of 0.95 leads
the list while Madagascar with an IFI value of 0.01 ranks the lowest. India has an IFI
value of 0.20 and is ranked 29th. In line with Sarma (2008), countries having IFI value
between 0.5 and 1 are classified as high IFI countries, those having IFI values between
0.3 and 0.5 are termed medium IFI countries and the rest having IFI values below 0.3 are
classified as low IFI countries. By this classification, only 11 of the 49 countries classify
as high IFI countries. These include high-income and OECD countries such as Belgium,
Denmark, Spain, Greece, France and Norway. 11 The high IFI countries also include
middle income countries such as Malaysia (upper middle-income) and Iran and Thailand
(lower middle- income). The medium IFI countries are 9 of the 49 in the list. Of these,

7 Sarma (2008) has also estimated a second set of IFI based only on two dimensions (availability and
usage) for which data are available for a larger set of 100 countries. We do not use the two-dimensional
IFI in this paper as one of the most important dimensions, viz., accessibility, is not incorporated in the
two-dimensional IFI. The data for computing the IFI is from World Bank’s WDI data base and from
World Bank’s website.
8 From among the estimations by Sarma (2008) we have excluded Fiji, Malta, Mauritius, Singapore and

Switzerland.
9 OFCs as defined by the IMF refer to economies that have relatively large numbers of financial

institutions engaged primarily in business with non-residents, where the in the financial sector’s external
assets and liabilities are disproportionately large when compared with domestic financial intermediation.
OFCs are also known to provide one or more of the following services: low or zero taxation; moderate or
light financial regulation; banking secrecy and anonymity (IMF 2000).
10 Though our empirical investigation is limited by the non-availability of adequate data, the fact remains

that a large set of countries will, without doubt, provide a wider and better understanding if the issues
discussed here.
11 The World Development Report classifies economics into three income groups depending on the gross

national income (GNI) per capita, calculated using the World Bank Atlas method. The groups are: low
income (LIC), $905 or less; lower middle income (LMC), $906–3,595; upper middle income (UMC),
$3,596–11,115; and high income, $11,116 or more.
6
Italy and Trinidad & Tobago are high income countries while the rest such as Russia,
Bulgaria, Chile, Turkey, Lithuania, Romania and Jordan are upper middle income
countries (UMC). While it is not surprising that the list of low IFI countries is
dominated by low income countries, there are some exceptions. For example, Saudi
Arabia, a high income country is found to have low IFI. Similarly, Brazil, Lebanon,
Venezuela, Argentina and Mexico, from the upper middle-income category are found to
be having low IFI.

A comparison of IFI with human development index (HDI) shows that all the countries
with high and medium IFI values belong to the group that is classified by the UNDP as
countries with high human development (HDI > 0.7). Saudi Arabia, a high HDI country
is found to have a low IFI value. Other countries having a high or medium HDI but a
low IFI are Brazil, Lebanon, Venezuela, Argentina and Mexico.

Countries such as Albania, Armenia, Peru and Mexico are with relatively higher levels of
human development as compared to their levels of financial inclusion. On the other
hand there are countries such as Iran, Thailand, Turkey and Namibia who perform
relatively better in financial inclusion than with human development.

Notwithstanding these exceptions, IFI and HDI seem to move in the same direction. As
observed from the Table 1, the IFI and HDI for the set of 49 countries move closely
with each other. The correlation coefficient between IFI and HDI values and ranks is
found to be about 0.74 and is highly significant. 12 From this, it can be generally
concluded that countries having high level of human development are also the countries
with a relatively high level of financial inclusion.

Region-wise, we find that a majority of the East European countries had medium level of
financial inclusion, while a large number of Latin American countries had low financial
inclusion. All countries in south Asia and Sub-Saharan Africa for whom IFI values were
estimated have low levels of financial inclusion. While having a low value of IFI, India
performs better than it neighbours Pakistan and Bangladesh. As data on east and south-
east Asian countries are limited, we are unable to make any further comment on their
performance.

12 The calculated value of the t-statistic is 7.489 which is significant at 0.0001 level of significance.
7
Small countries with a relatively large emigrant worker population may have relatively
higher levels of financial inclusion if the emigrant workers use the banking system for
remittances. For example, countries such as Jordan, Guyana and the Dominican
Republic have a relatively large emigrant population and a share of remittances to GDP
of over 15 per cent. In these countries, the higher IFI rank compared with HDI rank can
perhaps be attributed to this phenomenon. However, as noted in the literature on
migration and remittances, a large share of remittances to less developed countries
occurs through informal sources including family channels, travelling friends, local
money lenders and a complex network of informal remittance systems (COMPAS 2005,
El Qorchi et al 2003; Puri and Ritzema 1999). 13 Hence it is not possible to establish a
relationship between migration, remittances and financial inclusion.

Table 1 Index of financial inclusion and human development index, 2004

Index of financial Human development

inclusion (IFI) index (HDI)
Sl No. Country
Country Country
Value Value
rank rank
1 Albania 0.079 43 0.784 23
2 Argentina 0.148 35 0.863 10
3 Armenia 0.041 47 0.768 27
4 Austria 0.953 1 0.944 3
5 Bangladesh 0.117 39 0.530 44
6 Belgium 0.908 2 0.945 2
7 Bolivia 0.064 45 0.692 38
8 Bosnia &
Herzegovina 0.163 33 0.800 18
9 Brazil 0.283 21 0.792 20
10 Bulgaria 0.413 14 0.816 14
11 Chile 0.404 15 0.859 11
12 Colombia 0.229 26 0.790 21
13 Czech Republic 0.525 10 0.885 9
14 Denmark 0.906 3 0.943 4
15 Dominican
Republic 0.253 23 0.751 33
16 Ecuador 0.177 30 0.765 29
17 El Salvador 0.213 28 0.729 35
18 France 0.702 6 0.942 5

13 The terms used to describe informal remittance systems vary from country to country, however, despite

the different names used, the operational mechanisms are similar (Maimbo 2003).

8
 19 Greece 0.763 5 0.921 8
20 Guatemala 0.227 27 0.673 40
21 Guyana 0.252 24 0.725 36
22 Honduras 0.148 36 0.683 39
23 India 0.198 29 0.611 42
24 Iran, I.R. of 0.527 9 0.746 34
25 Italy 0.439 12 0.940 6
26 Jordan 0.298 20 0.760 31
27 Kenya 0.105 41 0.491 49
28 Lebanon 0.265 22 0.774 26
29 Lithuania 0.333 18 0.857 12
30 Madagascar 0.009 49 0.509 46
31 Malaysia 0.530 8 0.805 17
32 Mexico 0.145 37 0.821 13
33 Namibia 0.234 25 0.626 41
34 Nicaragua 0.076 44 0.698 37
35 Norway 0.595 7 0.965 1
36 Pakistan 0.113 40 0.539 43
37 Papua New
Guinea 0.057 46 0.523 45
38 Peru 0.125 38 0.767 28
39 Philippines 0.167 32 0.763 30
40 Romania 0.315 19 0.805 16
41 Russia 0.424 13 0.797 19
42 Saudi Arabia 0.151 34 0.777 25
43 Spain 0.784 4 0.938 7
44 Thailand 0.514 11 0.784 24
45 Trinidad and
Tobago 0.354 17 0.809 15
46 Turkey 0.387 16 0.757 32
47 Uganda 0.021 48 0.502 47
48 Venezuela 0.176 31 0.784 22
49 Zimbabwe 0.096 42 0.491 48
Source: Sarma (2008)
Note: The HDI ranks given in this table are re-ranks based on the HDI value for only
the set of 49 countries.

9
4. Factors associated with financial inclusion
The factors that affect financial inclusion are likely to be several and their interaction
with each other complex. At the moment, without going into the complexity of the how
a large number of factors together bring about the level of financial inclusion in a country,
we attempt to simply identify factors that are associated, with some degree of
significance, to the index of financial index. We do this by carrying out three sets of
regressions of the IFI on three different sets of variables. In a sense, the three sets of
variables indicating socio-economic factors, factors relating to physical infrastructure and
banking sector factors are independent of each other. 14 Further, data for all variables are
not available for all 49 countries for which we have computed the IFI. Thus, if we
include all variables or regressors in a single regression equation, the number of
observations (countries) in the panel reduces to 20, leaving us with very little statistical
precision. This also partly explains our resort to three separate regressions.

The first regression is a regression of the IFI on a set of socio-economic variables such
as income, employment, inequality, literacy and so on. The second regression attempts
to capture the role of physical infrastructure. The third regression attempts to see the
affect of banking sector variables such as soundness indicators, ownership pattern and
prevailing rate of interest on financial inclusion. The data for the regressions are from the
World Development Indicators (WDI) for the latest/comparable year. 15

In the regression equations, the dependent variable is a logit transformation of the index
of financial inclusion described earlier. Unlike the IFI which lies between 0 and 1, the
transformed variable lies between -∞ and ∞. This allows us to carry out the classical OLS
regression. The transformed variable is a monotonically increasing function of IFI, and
hence it preserves the same ordering as IFI. The transformed variable is a logit function
of the original variable IFI, as defined below.

⎛ IFI ⎞
Y = ln ⎜ ⎟
⎝ 1 − IFI ⎠

14 In other words, better performance or achievements of a country in one set of factors does not

necessarily imply similar performance in the other set.

15 WDI obtains data from national governments and from other international multilateral organisations

such as the IMF, ILO and so on. We have also attempted regression estimations using data that are
directly provided by the International Financial Statistics of the IMF for banking variables, ILO for
employment Statics and so on and the results remain the same.
10
The general form of the regression equation is

Y = a0 + a1X1 + a2X2 +….anXn + ε

Where X1, X2 etc are regressor variables, a1, a2 and so on are the parameters to be
estimated from the data and ε is the error term following classical OLS assumptions.
The rate of change of y with respect to a unit change in the variable Xi will be given by
the derivative of y with respect to Xi, which is

dy a exp(ai X i )
= i
dX i (1 + exp(ai X i )2

Thus, the direction of change in Y corresponding to a unit change in Xi is determined by

the sign of ai while the magnitude of the change depends on the value of ai as well as Xi.

4.1. Results of the regression on socio economic variables: The literature on

financial inclusion has identified financial exclusion as reflection of a broader problem of
“social exclusion”. In the industrialised and high income countries having a well-
developed banking system, studies have shown that the exclusion from the financial
system occurs to persons who belong to low-income groups, the ethnic minorities,
immigrants, the aged and so on (Barr, 2004; Kempson and Whyley, 1998; Connoly and
Hajaj, 2001). There is also a geographical factor; people living in rural areas and in
locations that are remote from urban financial centres are more likely to be financially
excluded (Leyshon and Thrift, 1995; Kempson and Whyley 2001). Further, countries
with low levels of income inequality tend to have relatively high level of financial
inclusion (Buckland et al, 2005; Kempson and Whyley, 1998). In other words, the levels
of financial inclusion inevitably rise in response to both prosperity and declining
inequalities.

Another factor that can be associated with financial inclusion is employment (Goodwin
et al, 2000). The unemployed or those with irregular and insecure employment are less
likely to participate in the financial system. Studies have found that payment of wages by
automated cash transfer (ACT) has been one of the main influences on financial

11
inclusion in the UK. Recent evidence also suggests that the continued payment of social
security benefits and the State pension in cash is significantly related to financial
exclusion (Kempson and Whyley 1999).

Informal sector or the informal economy accounts for a large and significant share of
employment in several less developed countries (ILO 2002). In these countries and
elsewhere in the industrialised countries, formal sector employment could imply
participation in the formal financial system through receiving wages and salaries routed
through the formal banking system. Formal employment also implies inclusion in
employment related social security system, benefits of which are availed through the
formal banking system. Thus the proportion of formal sector employment would be an
important indicator of the level of financial inclusion. Since we have not found reliable
cross country data on the proportion of formal sector workers, in the present analysis,
this aspect is not being covered. 16

The results of the first regression that we have attempted where the IFI is regressed over
some basic socio economic variables are presented in Table 2. In the regression equation
the IFI is regressed over GDP per capita which is a proxy for income, adult literacy (in
order to account for demographic factor, adult literacy is considered instead of total
literacy), unemployment, rural population and income inequality as indicated by the Gini
coefficient.

Table 2: Results of regressing IFI on socio economic variables

Variable Coef. Std. Err. t P>|t|
ln(GDP) 1.02* 0.166 6.16 0.00
Adultlit -0.008 0.013 -0.61 0.55
Ruralpop 0.002 0.008 0.28 0.78
Unemploy 0.004 0.011 0.36 0.72
Ginicoeff -1.848 1.246 -1.48 0.15
Constant -7.647* 1.8 -4.25 0.00
Notes: Number of observations = 47
F (5, 41) = 21.20, Prob > F = 0.000
R2 = 0.721, Adj R2 = 0.687
* - Variable significant at 0.01 level.

16 The ILO has made an attempt to provide data on the informal employment. These estimates are

provided for about 70 countries (ILO 2002).

12
The variables in Tables 2 and 3 are:
ln(GDP) – logarithm of GDP per capita (in 2000 constant USD)
Adultlit – percentage of literate persons aged 15 years and above in total population
Unemploy – percentage of unemployed people in the total labour force
Ruralpop – percentage of total population living in rural areas
Ginicoeff – Gini coefficients indicating income inequality

As seen in Table 2, the coefficient for GDP per capita is positive and highly significant
and explains financial inclusion almost single handed. This is not surprising and
establishes, in other words, that income levels matter in explaining financial inclusion.
Higher the income level, both at the individual level and for a country, higher is the
financial inclusion. Now, in order to probe further into the socio economic factors that
can be associated with financial inclusion, we have attempted a related regression,
without the GDP variable. The results of the same are presented in Table 3.

The removal of the GDP variable reduces the predictability of the regression as seen
from the lower values of R2 and adjusted R2 in the second regression (Table 3) when
compared with the first regression which included the GDP variable (Table 2).
Notwithstanding the loss of predictability, we find significant association of some socio-
economic variables in the regression with the IFI. First, the Gini coefficient is found to
be negatively and significantly associated with financial inclusion. Thus, income
inequality is negatively associated with financial inclusion, more precisely, higher the
income inequality, higher is the likelihood of financial exclusion. Two other variables are
associated with financial inclusion at 10 per cent level of significance. They are adult
literacy and rural population. Adult literacy is positively and significantly associated with
financial inclusion implying that higher the adult literacy, higher will be the financial
inclusion. Rural population or the proportion of rural population is found to be
negatively associated with financial inclusion. In other words, urbanization is positively
associated with financial inclusion. The “unemployment” variable has an expected sign
but it is not found to be significant. 17

17 As is well accepted, in the case of a large number of under-developed countries such as India, the data
on unemployment do not include large sections of the workforce who are under employed often working
in the informal sector. An indicator for unemployment + underemployment we believe will show up as
having a significant and negative relationship with financial inclusion.
13
Table 3: Results of regressing IFI on socio economic variables (without GDP)
Variable Coef. Std. Err. t P>|t|
Adultlit 0.031** 0.016 1.90 0.07
Ruralpop -0.018** 0.010 -1.91 0.06
Unemploy -0.021 0.014 -1.57 0.13
Ginicoeff -4.345* 1.615 -2.69 0.01
Cons -0.75 1.932 -0.39 0.70
Notes: Number of observations = 47
F(4, 42) = 9.05, Prob > F = 0.00000
R2 = 0.4630, Adj R2 = 0.4119
* - Variable significant at 0.01 level, ** - Variable significant at 0.10 level.

4.2 Results of the regression on infrastructure related variables

Access to financial services requires basic infrastructure to be in place. Infrastructure
such as a road network, telephone and television network, access to information through
newspapers, radio, cable TV, computer and internet can play a role in enhancing financial
inclusion by facilitating easy mobility and awareness about financial services. 18 The road
network can be taken as an indicator of the general level of physical infrastructure in a
country. Also, (paved) road network is a basic requirement or a pre-requisite for setting
up of bank branches and ATM networks in rural and less densely populated areas. All
other variables used in the regression equation are indicators of connectivity and
information availability. We present the results of the regression on IFI (i.e. logit of IFI)
on some infrastructure related variables in Table 4.

18 Telephone lines per 1000 people, internet users per 100 people, personal computers per 100 people are

also indicators used in the millennium development goals (MDR).

14
Table 4: Results of regression of IFI on infrastructure variables
Variable Coef. Std. Err. t P>|t|
P_road 0.763* 0.258 2.96 0.01
Phone 0.363** 0.168 2.16 0.04
Newspaper 0.002 0.002 1.04 0.31
Radio -0.001 0.001 -1.66 0.11
CableTV -0.002 0.002 -0.65 0.52
Computer -0.002 0.002 -0.83 0.42
Internet .006** 0.003 2.34 0.03
Constant -3.833* 0.806 -4.76 0.00
Notes: Number of observations = 36
F( 7, 28) = 19.89, Prob > F = 0.000
R2 = 0.833, Adj R2 = 0.791;
* - Variable significant at 0.01 level, ** - variable significant at 0.05 level

The variables in Table 4 are:

P_road – Paved roads (in Km) per square Km of land area. 19
Phone – logarithm of the number of telephone (land line and mobile) subscription per
1000 population.
Newspaper – Number of daily newspapers per 1000 population
Radio – Number of radios per 1000 population
CableTV – Number of cable TV subscription per 1000 population
Computer – Number of personal computers per 1000 population
Internet – Number of internet users per 1000 population

As seen in Table 4, physical infrastructure like network of paved road is highly positively
significant in enhancing financial inclusion. Similarly, telephone and internet
subscription are also found to be positive and significant. The variables for density of
radios, newspapers cable TV and computers do not show any significant relationship
with financial inclusion in our estimation. Telephone and internet usage being positively
associated with the level of financial inclusion indicate that connectivity and information
play an important role in financial inclusion. This is in line with Beck et al (2007) who
found that telephone network to be positively associated with banking outreach.

19 We have repeated the same regression with paved roads per 1000 population and have obtained similar

results.
15
4.3 Results of the regression on banking variables
In this regression we have attempted to see the extent to which the structure and health
of the banking sector explains the level of financial inclusion. The variables considered
are indicators of the health of the banking system, ownership pattern and interest rate.
The specific variables on the soundness or health of the banking system are the non-
performing assets (NPA) as a percentage of total assets and the capital asset ratio (CAR)
of the banking system. Lower the NPA, better is the health of the banking system while
in case of the CAR, higher the CAR, better is the health of the banking system. For
ownership pattern, shares of foreign banks and Govt. owned banks in the total banking
sector assets have been considered. The real interest rate indicates the cost of capital in
the banking system or the attractiveness of the banking system. Table 5 presents the
results of regressing our dependent variable (transformed IFI) on the banking variables.

Table 5: Results of regression of IFI on banking variables

Variable Coef. Std. Err. t P>|t|
NPA -0.125* 0.033 -3.81 0.00
CAR -0.130** 0.061 -2.14 0.04
asset_foreign -0.025** 0.008 -3.24 0.00
asset_govt. -0.013 0.012 -1.06 0.30
Interest Rate -0.029 0.022 -1.31 0.20
Cons 2.347* 0.584 4.02 0.00
Notes: Number of observations = 34
F(5, 28) = 8.13, Prob > F = 0.0001
R2 = 0.592, Adj R2 = 0.519
* - Variable significant at 0.01 level, ** - variable significant at 0.05 level.

The variables in Table 5 are:

NPA – Non-performing assets as a percentage of total assets of the banking sector
CAR - Capital asset ratio of the banking sector
asset_foreign - Share of foreign banks in the total banking sector assets
asset_govt - Share of the government in the total banking sector assets
Interest rate – real interest rate prevailing in the economy

As seen from the results in Table 5, the level of NPA of the banking system in an
economy is found to be significantly and negatively associated with financial inclusion.
The widely held view for high NPA of a banking system is that NPAs are a result of
providing credit to the low income groups (who are more likely to default), sometimes to

16
comply with the “directed lending” programmes such as the “priority sector lending” in
India. If lending to the poor and consequent default on their part was in fact the cause
for NPA, then higher levels of NPAs should be associated with higher levels of financial
inclusion. Our results show the opposite, they indicate higher level of NPA to be
associated with lower level of financial inclusion. Thus, efforts to include more people
into the financial system is not the significant cause for the NPA, on the contrary, the
cause for NPAs lies elsewhere. 20

Another proxy for the health of the banking system, the capital asset ratio (CAR) is
found to have a negative coefficient that is significant at 0.05 level. Thus, highly
capitalized banking systems seem to be less inclusive. This is not surprising, as banking
systems having high CAR tend to be more cautious in lending, thus negatively affecting
financial inclusion.

High share of foreign ownership in the banking system is found to be negatively

associated with financial inclusion. In our sample of 49 countries, we have observed that
countries such as Madagascar, Armenia and Jordan with high share of foreign banks in
the total banking sector assets (more than 60 per cent) have very low IFI values (0.009,
0.041 and 0.298 respectively). On the other hand, countries with high financial inclusion,
such as Denmark, Austria, Belgium and France have very low share of foreign banks in
their total banking sector assets (lower than 3 per cent in each of these countries).

Advocates of banking sector liberalization have argued that entry of foreign banks will
increase the supply of credit and improve efficiency by increasing competition (WTO
2005). Several studies have shown that this argument may not always be true. For
example, an IMF study by Detragiache et al (2006) found that in poor countries, a
stronger foreign bank presence is robustly associated with less credit to the private
sector. In addition, they found that in countries with more foreign bank penetration,
credit growth is slower and there is less access to credit. Gormley (2007) found that in
case of India, the entry of foreign banks is associated with an overall decrease in credit
availability for firms. Using cross country data, Beck et al (2007) have also found a
significantly negative association between share of foreign banks’ assets and number of

20 Reddy (2002) pointed out that the main ‘culprits’ for NPA in India are the large industries and not the

priority sectors that cover the weaker sections. Since this issue is not the immediate concern of this
paper, we do not probe it further.
17
accounts (credit as well as deposit) per capita in a country. Literature has termed this as
“cream skimming” by foreign banks, in which foreign banks indulge in serving only the
wealthy borrowers due to informational asymmetries. Our finding on the negative
association between financial inclusion and foreign banks’ asset share seems to be in line
with these evidences.

Share of government ownership in the banking system, our results show, does not have a
significant association with financial inclusion. This can be interpreted as the inefficacy
of state owned banks in bringing about financial inclusion. Similarly, the real interest rate
does not show any significant relationship with financial inclusion.

5. Conclusion
Literature has identified financial exclusion as a manifestation of social exclusion. In this
paper we have attempted a cross country study on factors associated with financial
inclusion. Using an index of financial inclusion (Sarma, 2008), we first describe the broad
relationship between financial inclusion and human development. We find that level of
human development and that of financial inclusion are strongly positively correlated,
although few exceptions exist.

Our empirical analysis confirms that income as measured by per capita GDP is an
important factor in explaining the level of financial inclusion in a country. Going beyond
per capita GDP, we find that income inequality, adult literacy and urbanisation are also
important factors. Further, physical and electronic connectivity and information
availability, indicated by road network, telephone and internet usage, also play positive
role in enhancing financial inclusion. These findings strengthen the assertion that
financial exclusion is indeed a reflection of social exclusion, as countries having low
GDP per capita, relatively higher levels of income inequality, low rates of literacy, low
urbanisation and poor connectivity seem to be less financially inclusive.

From among the banking sector variables, we find that the proportion non-performing
assets is inversely associated with financial inclusion, indicating that attempts by different
country towards greater financial inclusion have not contributed in any way to the non-
performing assets of the banking system. The capital asset ratio (CAR) is seen to be
negatively associated with financial inclusion. In other words, when the CAR of a

18
country is high, the banking system tends to be more cautious in opening its doors to the
financial excluded. Foreign ownership in the banking sector is seen to be negatively
affecting financial inclusion, while government ownership does not have a significant
effect. Finally, interest rate does not seem to be significantly associated with financial
inclusion.

19
References
Barr, M., 2004. “Banking the Poor”, Yale Journal on Regulation 21, pp. 122-239
Beck, T., A. Demirguc-Kunt, M. Soledad, M. Peria, 2007. “Reaching out: Access to and
use of banking services across countries”, Journal of Financial Economics, 85, pp.
234-266
Buckland, J., B. Guenther, G. Boichev, H. Geddie, M. Mutch, 2005. “There are no banks
here: financial and insurance exclusion services in Winnipeg’s North End”,
Winnipeg Inter-City Research Alliance.
Carbo S., Gardener E.P., Molyneux P., 2005. Financial Exclusion. Palgrave MacMillan.
Collard, S., E. Kempson and C. Whyley, 2001. Tackling Financial Exclusion – An area-based
approach. The Policy Press, UK.
COMPAS, 2005. Report on Informal Remittance Systems in Africa, Caribbean and Pacific (ACP)
countries, ESRC Centre on Migration, Policy and Society (COMPAS),
University of Oxford
Connolly, C., K. Hajaj, 2001. “Financial Services and Social Exclusion”, Financial
Services Consumer Policy Centre, University of New South Wales.
Corr, C., 2006. “Financial Exclusion in Ireland: An exploratory study and policy review”,
Combat Poverty Agency Research Series 39.
Detragiache, E., T. Tressel, P. Gupta, 2006. “Foreign Banks in Poor Countries: Theory
and Evidence”, IMF Working Paper WP/06/18
Djankov, S., P. Miranda, E. Seira, S. Sharma, 2008. “Who are the Unbanked ?” , World
Bank Policy Research Working Paper 4647.
El Qorchi, M. Maimbo S. M., Wilson. J. W., 2003. “Informal Funds Transfer Systems:
An Analysis of the Informal Hawala System.” Occasional Paper No.222.
International Monetary Fund, Washington, D.C.
European Commission, 2008. “Financial services Provision and Prevention of Financial
Exclusion” Report by the Director General for Employment, Social Affairs
and Equal Opportunities, European Commission.
Gormley, T. A., 2007. “Banking Competition in Developing Countries: Does Foreign
Bank Entry Improve Credit Access?”, mimeo, John M. Olin School of
Business, Washington University in St. Louis.
ILO, 2002. Women and Men in the Informal Economy: A Statistical Picture, International
Labour Office Geneva

20
Kempson E, A. Atkinson, O. Pilley, 2004. “Policy level response to financial exclusion in
developed economies: lessons for developing countries”, Report of Personal
Finance Research Centre, University of Bristol.
Kempson E., C. Whyley, J. Caskey J, and S. Collard, 2000. “In or out? financial
exclusion: a literature and research review”, Report, Financial Services
Authority.
Kempson, E., C.Whyley, 1998. “Access to Current Accounts”, British Bankers’
Association, London.
Kempson, E., C.Whyley, 1999. Kept out or opted out?. Policy Press, Bristol.
Kempson, E., C.Whyley, 2001. “Payment of Pension and Benefits”, Department for
Work and Pension, London.
Leyshon, T., 1995. “Geographies of financial exclusion: financial abandonment in Britain
and the United States”, Transactions of the Institute of British Geographers New Series,
20, pp.312–41
Mohan Rakesh (2006) “Economic Growth, Financial Deepening and Financial
Inclusion” , Address at the Annual Bankers' Conference 2006, Hyderabad on
Nov 3, 2006. <http://rbidocs.rbi.org.in/rdocs/Speeches/PDFs/73697.pdf>
Puri, S., T. Ritzema, 1999. “Migrant Worker Remittances,Micro-finance and the Informal
Economy: Prospects and Issues”, Working Paper N° 21, Social Finance Unit,
International Labour Office, Geneva.
Rangarajan Committee, 2008, “Report of the Committee on Financial Inclusion”,
Committee Report.
Reddy, P. K., 2002. “A Comparative Study of Non Performing Assets in India in the
Global Context – similarities and dissimilarities, remedial measures”, Working
Paper, Indian Institute of Management, Ahmedabad.
Sarma, M., 2008. “Index of Financial Inclusion”, ICRIER Working paper, August 2008.
Seidman, E. and J. Tescher, 2004. “From Unbanked to Homeowner: Improving the
Supply of Financial Services for Low-Income, Low-Asset Customers”. The
Centre for Financial Services Initiative.
Sinclair S. P., 2001. “Financial exclusion: An introductory survey”, Report of Centre for
Research in Socially Inclusive Services, Heriot-Watt University, Edinburgh.
Solo, T. M and A. Manroth, 2006. “Access to Financial Services in Colombia – The
Unbanked in Bogota”, World Bank Policy Research Paper 3834.

21
WTO (2005) “Liberalization of Financial Services”, Communication From Australia,
Bahrain, Canada, The European Communities, Japan, Norway, Oman,
Panama, Singapore, Switzerland, The Separate Customs Territory Of Taiwan,
Penghu, Kinmen and Matsu, and the United States, TN/S/W/43,
S/FIN/W/43, (05-2335)
<http://trade.ec.europa.eu/doclib/docs/2005/december/tradoc_126552.pdf
>

22