Determinants Of Financial Inclusion In Bangladesh: Dynamic

Gmm & Quantile Regression Approach

Ajim Uddin, Mohammad Ashraful Ferdous Chowdhury, Md. Nazrul Islam

The Journal of Developing Areas, Volume 51, Number 2, Spring 2017, pp.
221-237 (Article)

Published by Tennessee State University College of Business

DOI: https://doi.org/10.1353/jda.2017.0041

For additional information about this article

https://muse.jhu.edu/article/657938

Access provided by Southern Illinois University @ Edwardsville (14 Aug 2017 20:23 GMT)
 The Journal of Developing Areas
Volume 51 No. 2 Spring 2017

DETERMINANTS OF FINANCIAL INCLUSION IN

BANGLADESH: DYNAMIC GMM & QUANTILE
REGRESSION APPROACH
Ajim Uddin
Southern Illinois University Edwardsville, USA
Mohammad Ashraful Ferdous Chowdhury
Md. Nazrul Islam
Shahjalal University of Science & Technology, Bangladesh

ABSTRACT

Financial Inclusion has been recently acknowledged as a key enabler for reducing poverty and
improving prosperity. However, more than 50% adults of the poorest households are still unbanked
globally. According to IMF (2016), 45% people of Bangladesh are out of formal financial services.
High interest rate, vast rural population and low literacy rate are among the prominent factors that
restrained the government and the World Bank joint initiatives to reach universal financial access.
Researchers have reasons to believe, there are certain other factors that remain unnoticed by the
policy maker and which in turn, probably affecting their ability to formulate effective policy. The
quest to identify these factors materialize this study. In this study, the determinants of financial
inclusion have been clustered in two broad areas: bank specific factors and macroeconomic factors.
Then, econometrically try to analyse which group of factors has more influence in determining the
level of financial inclusion in a country. In the study, static model like, random effects and dynamic
panel models like, generalized methods of moments (GMM) methodologies have been
applied. Data from 25 banks: 18 conventional and 7 Islamic banks operating in Bangladesh during
the period 2005–2014, is collected to analyze the role of both types of banks in financial
inclusion. For robustness, this study also employs Quantile regression approach. The study found
that, on the supply side, the size of a bank, its efficiency, and the interest rate it charges has a
directs impact on financial inclusion. Where on the demand side, literacy rate is positively and age
dependency ratio is negatively related to financial inclusion. In addition to that, quantile regression
analysis found that bank size has a significant impact on both deposit collection and loans &
advances disbursement of a bank. The study can be beneficial for both government and bank
authorities in developing their policy decisions to ensure more inclusive financial system. Setting
the loan interest rate close to international level and extending the branch facilities can encourage
excluded population to use formal financial services. Finally, the study also signifies that, by
developing human capital developing country countries can improve their financial market
participation and therefore can foster economic growth.

JEL Classifications: G20, G21, G28

Keywords: Determinants, Financial Inclusion, Dynamic GMM, Quantile regression
Corresponding author’s email address: [email protected]
 222

INTRODUCTION

The mass access to formal financial services has been the major concern for large supra
national organizations. Despite their relentless efforts, the portion of nonbank people is
still precariously high. About half of the world adults are still out of formal financial
services (World Bank, 2014). In addition of voluntary exclusion, several other forces
deter peoples from using financial services even when they are willing to use those. This
involuntary exclusion is especially severe in developing countries where the
infrastructures of financial institution are still weak. Bangladesh as a developing country
has been experiencing a radical transformation of its economic and financial system.
Despite political disputes, the country has achieved a satisfactory growth rate on its GDP
from 2005-2015. With proper policy formulation and effective decision making some of
the hindrances of the country’s,’ economic development is mitigated. Still some are
prevalent. One of these is financial inclusion. At present, only 31% of the country’s
population has an account to the financial institution. The financial inclusion in
Bangladesh also differ according to gender identity, with the inclusion rate among
women is only 26%.
The factors influencing access to formal credit have been identified in the
empirical literature by various scholars (Armendáriz de Aghion & Morduch, 2005;
Khandker, 2005; Ayamga et al., 2006; Kibaara, 2006; Akudugu et al., 2009; Rahji &
Fakayode, 2009; Akudugu, 2010; Armendáriz de Aghion & Morduch, 2010; Awunyo-
Vitor & Abankwah, 2012). However, the issue of why a person will decide to be included
in the formal financial sector has not been holistically examined in the literature. This is a
gap that must be bridged if efficient and effective strategies are to be developed for the
inclusion of more people in the formal financial market. An outline of the remainder of
the paper is as follows. Section 2 summarizes the literature on financial inclusion. Section
3, provide a brief description of the economy of Bangladesh and its level of financial
inclusion. Section 4 provides the details of the testing methodology and data description.
Section 5 presents empirical results. Finally, Section 6 concludes the paper

LITERATURE REVIEW

Numerous studies have been conducted to find the causes, consequences, and solution for
financial exclusion. As expected, factors that determine the level of inclusion is also
studied a number of times. But most of them either on a cross-country basis or based on a
single country whose socio-economic dimension significantly differs from Bangladesh.
The most recent and most cited cross-country study on the determinant of financial
inclusion probably the one conducted on Asian devolving countries by Park and V.
Mercado, Jr, (2015). To find the relationship between poverty, inequality, and financial
inclusion they found that per capita income, the rule of law, and demographic
characteristics like literacy rate and age dependency ratio significantly affect financial
inclusion in developing Asia. Their study is primarily based on the financial access
indicator developed by Honohan (2008). Honohan (2008) use those indicators for 160
economies to assess the influence of country characteristics like gross national income
(GNI), age dependency ratio, and population density on financial inclusion.
 223

Honohan (2008) constructed a financial access indicator for 160 economies that
combines both household survey datasets and published financial institutions data into a
composite indicator; and assessed country characteristics that might influence financial
access. Among the variables tested, aid as percent of gross national income (GNI), age
dependency ratio, and population density significantly lower financial access; while
mobile phone subscription and quality of institutions significantly increase financial
access. Looking at the cross-country link between poverty and financial access, his
results show that financial access significantly reduces poverty, but the result is valid
only when financial access is the sole regressor, i.e., it loses significance when other
variables are added as regressors. Mihasonirina and Kangni (2011) conducted a study in
South African region and found that banking services such as ICT development have a
positive impact on financial inclusion. Toxopeus and Lensink (2007) attended to the issue
of remittance inflows on financial inclusion for cross-section of emerging nations and he
found robust positive effect on financial inclusion and in turn on economic growth.
Rojas-Suarez (2010) used the same indicator constructed by Honohan (2008) to test the
significance of various macroeconomic and country characteristics for a group of
emerging economies, including some from developing Asia. The results show that
economic volatility, weak rule of law, higher income inequality, and social
underdevelopment and regulatory constraints significantly lower financial access.
Besides these cross-country studies, there was also some specific country level studied
conducted to identify the determinant of financial inclusion.
By using panel fixed effects and dynamic GMM method Kumar (2013) tried to
identify the status of financial inclusion in India and its determinants. After the study, he
concludes that the level of branch network has a significant positive impact on financial
inclusion. He also added that the geographical region and ease of access also determine
the level of financial inclusion in a particular segment of the population. In another
country level study on Peru, Cámara and David (2015) used microdata from surveys and
found that vulnerable groups like women’s and people living in rural areas are more
excluded from using financial services. Their study also found that age, gender,
education, and income level also influence individual perception whether to use formal
financial services or not.
Sarma (2008) constructed a Financial inclusion index but failed to draw a
unified determinant for all countries due to the disparity in country specific regulations.
Features and general preferences leading to financial inclusion vastly differs across
nations (Kempson et al., 2004; Kendall et al., 2010; Sinclair et al., 2009; The World
Bank, 2008a). Despite these studies, there is still a lack of literature which can shed light
on what factors determines the level of financial inclusion in Bangladesh. This study will
fill the gap of country-specific literature by identifying the major determinants of
financial inclusion in Bangladesh.

BANGLADESH, ITS ECONOMY, AND LEVEL OF FINANCIAL INCLUSION

Bangladesh needs a lot to do encourage their people to use formal financial services.
Only 9.9% of its total population borrowed from a financial institution where 25.2%
borrowed from friends or family and 4.5 borrowed from an informal lender. Other
indicators of financial inclusion like use of an account to receive payment, use of debit
 224

card or presence of ATM machine also provide useful information about the situation of a
country’s financial inclusion. In Bangladesh, only 1.6% people use a salary account
compared to 3.5% in South Asia and 3.2% in other low-income countries. Only 5.2%
population has a debit card compared to 18% in South Asia and 6.6% in other low-
income countries (World Bank, 2015).
Compared to other neighbouring countries, only 29% of Bangladesh total adult
population has maintained an account to a formal financial institution compared to 53%
in India and 34% in Bhutan and Nepal (see more in figure 1) (Global Findex, 2015). This
low access to financial services might be the result of higher lending rate. Compared to
other Southeast Asian countries, the lending rate in Bangladesh is too high, for example,
the lending rate of Malaysia and Nepal is around 5% and 8% respectively whereas the
lending rate of Bangladesh is around 15% (World Bank, 2015).

FIG 1: ACCOUNT AT A FINANCIAL INSTITUTION

100

80

60

40

20

0
Afghanistan Bangladesh Bhutan India Malaysia Nepal Pakistan Sri Lanka

2011 2014
Source: Global Findex (2015)

In recent years, the government, and Bangladesh Bank (BB) has taken several
initiatives to provide access to financial services to financially excluded people. From
them, Bangladesh Bank branch opening rules, farmers banking, school banking and SME
financing are notable to mention. However, the expected result is yet to be achieved.
Although increasing, compared to other subcontinent country the deposit to GDP ratio
are still low in Bangladesh (see more in figure 2).
 225

FIG 2: DEPOSIT TO GDP RATIO IN SUBCONTINENT

80%

60%
BGD
40%
IND
20% PAK
0%
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Source: Global Findex (2015)

The loan to GDP ratio is also not on satisfactory level. In 2004, outstanding loan in
proportion to GDP of Bangladesh was higher than India and Pakistan. From 2005 to
2014, Bangladesh and India continued increasing while in Pakistan, it fell gradually after
2008. In 2011 and 2012, Bangladesh was at its highest level while India was in 2014.

FIG 3: LOAN TO GDP RATIO IN SUBCONTINENT

60%
50%
40% BGD
30%
IND
20%
PAK
10%
0%
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Source: Global Findex (2015)

In 2004, outstanding loan in proportion to GDP of Bangladesh was higher than India and
Pakistan. From 2005 to 2014, Bangladesh and India continued increasing while in
Pakistan, it fell gradually after 2008. In 2011 and 2012, Bangladesh was at its highest
level while India was in 2014.
 226

FIG 4: BRANCHES PER 100,000 ADULTS

14
12
10
8 BGD

6 IND

4 PAK

2
0
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Source: Global Findex (2015)

This graph shows that the number of commercial bank branches per 100,000 adults in
Bangladesh, India and Pakistan was gradually increasing. However, the rate of increasing
the branches in Bangladesh is far less then India and Pakistan.

FIG 5: ATM BOOTH PER 100,000 ADULTS

20

15
BGD
10
IND
5 PAK

0
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Source: Global Findex (2015)

The average ATM machines around the world for 2012 were 43.86 per 100,000 adults.
The World Bank ranking (ATMs per 100,000 adults in 2012) shows that the position of
Bangladesh was 145 within 175 countries, having a score point of 4.89; India was at 125,
having a score point of 11.13 and Pakistan was in 143, having a score point of 5.31.
 227

METHODOLOGY

Data

Data from 7 Islamic bank and 18 conventional banks listed in Dhaka Stock Exchange
(DSE) for the period 2005-2014 have been considered in this study. The financial
statements of the sample banks have been collected from the annual reports of the banks.
For macroeconomic indicators, inflation and GDP growth rate data have been collected
from the database of World Bank.

Variables and their Explanations

The definition and brief descriptions of the variables are presented in Table 1.
The detailed explanation and the rationale to be included in this study is followed
afterwards. The most common measure of a country’s level of financial inclusion is
Global Findex percentage of the population have an account with a financial institution.
However, as this is a single country study instead of Global Findex percentage this study
used deposits and loan and advances as a dependent variable for measuring financial
inclusion. Considering total deposit, loan and advances as proxy for financial inclusion is
supported by previous studies such as Beck et al. (2006) and Honohan (2008), Amidžić,
Massara, and Mialou’s (2014).

TABLE 1: DETAILS OF THE VARIABLES SELECTED

Variable Definition Descriptions

FI Financial inclusion DP = Natural log of Total deposits
LA= Natural log of Total Loan and Advances
(Investment for Islamic Banks)
Specific

LTA Bank Size Natural log of Total asset

CIR Cost to Income Ratio It refers the operational efficiency of a bank.

INTD Interest rate on Deposits The average interest rate that bank provides to its
(profit rate for Islamic depositors.
Variables

Banks)
INTL Interest Rate (profit rate) on The average interest rate that banks charges on its
Bank

Loan and Advances loan and advances.

GNI Natural log of Gross GNI per capita, Taken from the World Bank
national income (GNI) Database
Human Factor Country
variable

INFL Inflation rate Represented by annual Consumer Price Inflation

Rate. Taken from the World Bank Database.
LR Literacy Rate Computed as the secondary school completion
rate.
AGD Age Dependency Ratio It is the ratio of dependents-people younger than
15 or older than 64-to the working-age
population-those ages 15-64.

The prime purpose of this study is to identify how bank-specific factors are
influencing the use of banking services by the population. In connection to this purpose
 228

three bank-specific factors, bank size, cost-to-income ratio, and interest rate are selected.
Kosmidou (2008) found that the larger the bank size, the more capable it is to reach a
greater portion of the population and increase financial inclusion. Here, natural log of
total assets has been used as a determinant of bank size (Beck et al., 2010). The cost-to-
income ratio depicts the efficiency of a bank. The presence of more efficient banks is
perceived to have a positive relationship with banks financial inclusion (Pasiouras and
Kosmidou, 2007). Finally, it is expected that high-interest rate on deposits will encourage
people to save while, high interest rate on loan, expected to discourage investors and
lower financial inclusion.
Honohan (2008) first constructed the financial access indicator for 160
economies and assessed the influence of country characteristics like, gross national
income (GNI), age dependency ratio, and population density on financial inclusion. Later
Park and Mercado (2015) use per capita income, inflation, dependency ratio, education
completion, and the rule of law to measure how financial inclusion is related to poverty
and income inequality in developing Asia.
In this study, gross National Income (GNI), inflation, literacy rate and age
dependency ratio are used to analyze how country variable influencing financial
inclusion. GNI is taken to identify how the income level is influencing people decision to
use financial services from the banks (Mercado (2015). Rojas-Suarez (2010) use inflation
and found that increase in price level and economic volatility significantly lower financial
access. Based on Honohan (2008) and Park and V. Mercado, Jr, (2015) this study
includes literacy rate and age dependency ratio as the human factor. It is expected that
access to financial inclusion will increase with increasing literacy (Park and V. Mercado,
Jr, 2015). Age dependency ratio signifies the portion of the population who are too young
or too old to use a financial service. Therefore, lower age dependency ratio expected to
increase financial inclusion (Park and V. Mercado, Jr, 2015).

Data Analysis and Models

Both static and dynamic model has been applied to this study. The static model is widely
renowned and has been used in numerous studies. Berger et al. (2000), Goddard et al.
(2004), Athanasoglou et al., (2008) Mamatzakis and Remoundos (2003), Chowdhury
et.al (2016) argue dynamic model uses more information and consequently the estimation
will be more efficient. Two separate models are developed to depict both the depositors’
side and borrowers’ side of financial inclusion. In the first model, deposits are considered
as dependent variable whereas in second model, loan and advances are considered as
dependent variable.

Static models (fixed and random effects)

The basic framework of the first model (where, deposits is the dependent
variable) and the second model (where, loan and advances is the dependent variable) for
the panel data is defined as the following regression model:

DP= α+β1CIR+β2LTA+β3INTD+β4GNI+β5INFL+β6 LR+β7AGD+ Dummy +ε (i)

 229

LA= α+β1CIR+β2LTA+β3INTL+β4GNI+β5INFL+β6 LR+β7AGD+ Dummy +ε (ii)

The least square method of fixed effects model (FEM) is applied to control for
bank-specific effects while standard errors are calculated by using White’s (1980)
transformation to control for cross section heteroscedasticity.

Dynamic GMM models

In order to capture the impact of lagged effect the difference Generalized

Methods Moments (GMM) model (Arellano and Bond, 1991) is used. The basic
assumption of the GMM is that, the first differences of instrumental variables are
uncorrelated with the fixed effects, and which allows the model to introduce more
instruments and improve its efficiency. Roodman (2006) argues that both difference and
system GMM estimators are suitable for studies that involve ‘small T, large N’ panels;
where independent variables are not strictly exogenous; and heteroscedasticity and
autocorrelation exist among individual sample, in this study, banks. The following
formula for GMM proposed by Athanasoglou (2008) is used to conduct the empirical
analysis:
- - 0
II" = C + δ II(,"*+ + -.+ β- X (" + 0
0.+ β- X (" + ε(" (iii)

Where,
II t indicates the financial inclusion where deposit and loan are used as proxy
variables of bank i at time t where i=1…N, t=1, C is the Constant Term. II i ,t -1 = is the

lag value of dependent variable, X it are the explanatory variables and it thee .
disturbance term, with vit the unobserved bank-specific and socio-economic effect and
uit the idiosyncratic error. This is a one-way component regression model, where vet ~
IIN (0, ϭv²) and independent of uit ~(0, ϭu²).

Quantile Regression (QR) model

This study employs a QR model introduced by Koenker and Bassett (1978), a

statistical technique aim to measure, and performs inference about conditional quantile
functions. This analysis is particularly useful when the conditional distribution does not
have a standard shape, such as an asymmetric, fat-tailed, or truncated distribution.
Quantile regression is an ideal tool to reexamine the determinants of financial inclusion
heterogeneity as it is more flexible for modeling data with heterogeneous conditional
conditions. In quantile regression, it is assumed that the θ th quantile of the conditional
distribution of the explained variable is linear in x where Quant Xi, the conditional QR
model can be expressed as follows:

Yi= xˈi . βƟ + uƟi

QuantƟ (yi ‫ ׀‬xi) = inf { y : Fi (y‫׀‬x)Ɵ} = xˈi . βƟ

QuantƟ ( uƟi‫׀‬xi) = 0 (iv)

 230

where QuantƟ (yi ‫ ׀‬xi) represents the Ɵ the conditional quantile of yi on the regressor
vector xi; βƟ is the unknown vector of parameters to be estimated for different values of
Ɵ in (0,1); uƟi is the error term assumed to be continuously differentiable c.d.f.
(cumulative density function) of Fi (y‫׀‬x)Ɵ and a density function Fi (y‫׀‬x)Ɵ. The value Fi
(y‫׀‬x)Ɵ denotes the conditional distribution of y conditional on x. Varying the value of u
from 0 to 1 reveals the entire distribution of y conditional on x. The estimator for b u is
obtained from the following equations:
3 3
min (:5Ɵ6.7 Ɵ ˟ ‫׀‬uƟ‫׀‬ + (:5Ɵ:.7 1 − Ɵ ˟ ‫׀‬uƟi‫׀‬
(v)
= (:;(*<ˈ(.?Ɵ6.7 Ɵ ˟ ‫׀‬yi-xˈi . βƟ‫׀‬ + (:;(*<ˈ(.?Ɵ:.7
(1 − Ɵ) ˟ ‫׀‬yi-xˈi .
βƟ

Due to the advantages (as stated above) of quantile regression estimation technique over
OLS, fixed and random effect models, we examined at the 25th, 50th, 75th and 95th
quantiles in this study.

RESULTS AND DISCUSSIONS

The analysis of the study found several significant factors that affect the level of financial
inclusion in Bangladesh. Based on Hausman test(P-value is 0.293), this paper only
includes the random effect from the static model. For reliability of the dynamic model,
Table 2 reports that the null hypothesis of no first-order autocorrelation is rejected. These
results confirm the usage of a dynamic panel data model in which several variables are
instrumented; using lags of these variables removes autocorrelation in the second-order.
Both of Arrelano and Bond test of first- and second-order autocorrelation in the residuals
and the Sargan test of over-identification (Roodman, 2008) have been performed in this
study(Chi² value is 82.05 and 95.63 with the prob> Chi²=0.12 and 0.20 respectively).

Inclusion measured by the total deposits indicate that, size, deposits interest rate,
literacy rate, age dependency ratio and gross national income has significant impact on
the financial inclusion. Both static and system GMM found that, bank size has a positive
impact on the deposits creation of a bank. This result proved Kosmidou’s (2008) finding
and signifies that larger bank has greater ability to reach a more diverse set of populations
and increasing financial inclusion. Though system GMM signifies there is no relationship
exists between interest on deposits and deposits generation, the static model found a
positive relation (0.016) at 10 percent level of significance. In contradiction to Park and
V. Mercado, Jr’s (2015) findings, this study found gross national income has a negative
impact on financial inclusion. In the static model, the negative relationship is more
intense 0.173 at 5 percent level of significance. Finally, the first model also confirms
Rojas-Suarez (2010) findings of the impact of inflation on financial inclusion. The
negative relation (0.0015**) signifies that an economic volatility and increase in price
level lower the level of financial access.
 231

TABLE 2: SUMMARY OF THE RESULT

Result when loan and advances is

Result when deposits is the
the measures of financial
measures of financial inclusion
inclusion
Static
Static Model
Variable Model System GMM System GMM
(Random)
(Random)
Intercept 0.2987*** 0.3420*** .0967* 0.1936*
Lag of
dependent 0.0979* .1405*
Variable
LTA 1.0141*** 0.9352*** .9761*** .8501***
CIR -0.0001 0.0001 -.0008*** -.0006*
INTD 0.0162* -0.0002
INTL -0.0012* -0.0036***
LGNI -0.173** -0.2254* -00294 -.0613
INFL -0.0014 -.0015** .0023 .0017*
LR 0.0012* 0.0011* -.0007 .0012*
AGD -0.0134** -0.123* -.0054 -.0156**
R-squared .9943 .9878
82.05 95.6267
Sargan test
Prob >chi2 = 0.12 Prob>chi2= 0.20
F-stat/Wald-χ² 37942*** 18682*** 18558*** 26703.59***

AR(1) test Z=2.25, P=0.024 Z=-2.22, P= .026

AR(2) test Z= 1.13, P= 0.254 Z=.286, P=0.774

No of Obs. 245 221 247 223

Note: Values in parenthesis are t statistics, (***), (**), (*) indicates significance at 1, 5,
and 10% level respectively.

The second model, loan and advances as an indicator of financial inclusion,

indicate size, the efficiency of the banks, interest charged on loans, and age dependency
ratio are significantly important for financial inclusion. The significant negative
relationship with bank cost to income ratio with loan disbursement signifies that the
presence of efficient banks can provide an extra incentive to individual investors to
 232

borrow from these banks. As expected the final bank specific variable, the interest banks
charge on their loan and advances found have a negative impact on financial inclusion.
From the country variables, only the dynamic GMM model signifies having a
relationship of financial inclusion with inflation, literacy rate, and age dependency ratio.
Inflation found to have a slight positive effect (0.00167) on loan disbursement. In line
with Park and V. Mercado, Jr’s (2015) findings, the literacy rate found to have a positive
impact (0.00122) and age dependency ratio have a negative impact (0.0156) on financial
inclusion.

ROBUSTNESS CHECKING-QUANTILE REGRESSION

For robustness of the studies, this study also employs the quantile regression. Table 3 lists
the estimation results of the QR model for the impact of the Bank specific variables and
macroeconomic variables on the deposit. Here, both OLS and the quantile regression at
four different quantile have been studied for both deposit collection and loans and
advances.

TABLE 3: QUANTILE REGRESSION FOR DEPOSITS

OLS_res Q25_res Q50_res Q75_res Q95_res

LTA 0.9616*** 0.9546*** 0.9745*** 0.9902*** 0.9913***
CIR -0.0002 -0.0003343 -0.0002 -0.00004 -0.0001
INTD 0.0016 0.0022* 0.0012 0.0019* 0.0013
LGNI -0.2728 -0.1103 -0.1332 -0.2240 -0.5359
INFL -0.0015 -0.0014 0.0002 -0.0023 -0.0010
LR 0.0008 0.0007 0.0017 0.0003 0.0018
Aged -0.0081 -0.0047 -0.0012 -0.0049 -0.0140
Intercept 1.3165 0.6778 0.3985 0.8857 2.3244
Note: Values in parenthesis are t statistics, (***), (**), (*) indicates significance at 1, 5,
and 10% level respectively.

There are two types of significant coefficients: those that are significantly
different from zero, and the quantile coefficients that are significantly different from the
OLS coefficients (outside of the OLS confidence interval). For example, the coefficient
on the LTA at the 75% quantile is significantly different from zero but not significantly
different from the OLS coefficient. It can be interpreted that bank with higher size can
bring .9546% more in total for those with low total assets (at the 25% quantile) and
0.9901% more in total deposit for those higher bank size (at the 75 quantile). In other
words, the effect of deposit increases for bank with higher assests/size with higher bank
deposit at high quantile. The same trend prevails with higher interest rate at the 25
quantile. It can be found that the OLS estimator, by focusing only on the central tendency
of the distributions, does not allow the impact of the interest rate on bank deposit differ
for more/less profitable banks However, the quantile regression results that there is a
significant relationship between interest rate and bank deposit at 25th quantile.
 233

TABLE 4: QUANTILE REGRESSION FOR LOANS &ADVANCES

OLS_res Q25_res Q50_res Q75_res Q95_res

LTA 0.9656*** 0.9429*** 0.9519*** 0.9616*** 0.9829***
CIR -0.008*** -0.0014*** -0.0094*** -.0003604 -0.0004
INTL 0.0011 0.0016 0.0017 0.0025 0.0029

LGNI -0.5271 -0.5776 -0.6781* -0.7322* -0.9716

INFL 0.0015 0.0038 0.0026 0.0010 -0.0023

LR -0.0013 -0.0008 0.0003 -0.0014 0.0002
AGD -0.0169 -0.0162 -0.0182* -0.0248* -0.0351
Constant 2.5680 2.5680 2.9012* 3.7171* 5.2034
Note: Values in parenthesis are t statistics, (***), (**), (*) indicates significance at 1, 5,
and 10% level respectively.

On the other hand, Table-4 shows the quantile regression result of loans and
advances Interestingly, the effect of loans and advances increases for bank with higher
assests/size with higher bank deposit at high quantile. For example, A bank with higher
size can disburse .942% more in total for those with low total assets (at the 25% quantile)
and 0.961% more in loans and advances disbursement for those higher bank size (at the
75 quantile). Unlike the deposit determinants, CIR has found a quantile effects on the
loans and advances. It has been found that there is a sufficiently negative relationship
between cost to income ratio and the loans and advanced at the 50 quantile. Along with
these two, LGNI and AGD have a quantile effects on 50th and 75th quantile.
The quantile regression coefficients are plotted as lines varying across the
quantiles with confidence intervals around them. If the quantile coefficient is outside the
OLS confidence interval, then we have significant differences between the quantile and
OLS coefficients. Figure 6 depicts the QR estimates and the OLS estimates and finds that
most of the variables are plotted with the confidence interval except the LTA and the
deposit interest rate. Subsequently, In fig 7, it can be seen that most of the variables for
loans and advances are not positioning in between upper and lower bound except INFL
and LR. All of these variables cross the significance level (upper bound) after a certain
level of percentile. On the other hand, form the counterpart, the quantile graph in fig.7
shows consistent result with the previous findings at table 4. Here all the variables remain
the upper and lower bound of significance level at different percentile.
 Litaracy Rate Litaracy Rate
Intercept Intercept
CIR CIR
-0.02 -0.01 0.00 0.01 -0.00 -0.00 0.00 0.00 -20.00 20.00
-10.000.0010.00 -0.02-0.010.000.010.02 -0.00-0.000.000.000.00 -20.00 20.00
-10.000.0010.00

0
0
0
0
0
0

.2
.2
.2
.2
.2
.2

.4
.4
.4
.4
.4
.4

.6
.6
.6
.6
.6
.6

Quantile
Quantile
Quantile
Quantile
Quantile
Quantile

.8
.8
.8
.8
.8
.8

1
1
1
1
1
1

Loan Interest Dep Interest

INFL lgni INFL lgni
-0.02
-0.01
0.00
0.010.020.03 -4.00
-2.000.002.004.00 -0.01
-0.01
0.000.010.010.01 -0.02
-0.010.000.010.02 -4.00-2.00 0.00 2.00 -0.01-0.010.000.010.01

0
0
0
0
0
0

.2
.2
.2
.2
.2
.2

.4
.4
.4
.4
.4
.4
234

.6
.6
.6
.6
.6
.6

Quantile
Quantile
Quantile
Quantile
Quantile
Quantile

.8
.8
.8
.8
.8
.8

1
1
1
1
1
1

Age Dependency Ratio Age Dependency Ratio

LOANS & ADVANCES

VARIABLES FOR DEPOSIT

LTA LTA
-0.10
-0.05
0.000.050.100.15 0.800.901.001.101.20 -0.15
-0.10
-0.05
0.000.050.10 0.800.901.001.101.20

0
0
0
0

.2
.2
.2
.2

.4
.4
.4
.4

.6
.6
.6
.6

Quantile
Quantile
Quantile
Quantile
FIG 6: REGRESSION LINES: QR VS OLS FOR DIFFERENT

.8
.8
.8
.8

1
1
1
1

FIG 7: REGRESSION LINES: QR VS OLS FOR DIFFERENT VARIABLES FOR

 235

CONCLUSIONS

Financial inclusion is the process to include the people who lack formal financial
services. Despite the current focus of policies and regulations devoted to enhancing
access to finance in Bangladesh, there is a number of underlying factors causing financial
exclusion. The main goal of the paper was to identify the factors that determine the level
of financial inclusion in Bangladesh and suggest policy measures to improve the level of
inclusion. In connection to this purpose, the study found that bank size, deposits interest
rates and literacy rates positively influence the level of financial inclusion. In this
context, the study dealt with the socio-economic determinants of financial inclusion in
Bangladesh. The empirical findings suggest that the greater size and diverse product
portfolio of the larger banks is able to reach a more diverse set of customers and
contribution to the financial inclusion. The study also signifies that, cost-to-income ratio,
loan interest rates, and age dependency ratio have a negative affect on financial inclusion.
To ensure mass access to financial services, this paper suggests developing human
capital. Financial literature can be proved instrumental to overcome the problem of
informal lending and its associated risk. As educated persons essentially informed about
the benefits of using the formal financial system they are more likely to use those.
Therefore, policy alternative should be designed to communicate the benefits of financial
service to non-users and making them financially literate.

REFERENCES

Amidžić, G., Massara, A. and Mialou, A. (2014). Assessing Countries’ Financial

Inclusion Standing: A New Composite Index. IMF Working Papers, 14(36), p.1.
Armendáriz de Aghion, B., & Morduch, J. (2005). The economics of microfinance (1st
Edition) (Vol. FirstEdition): Massachuttes Institute of Technology.
Awunyo-Vitor, D., & Abankwah, V. (2012). Substitutes or complements: Formal and
informal credit demand by maize farmers in Ashanti and Brong Ahafo
Regions of Ghana. The Journal of Agricultural Science, 7(1), 1-13.
Athanasoglou, P., Brissimis, S. and Delis, M. (2008). Bank-specific, industry-specific
and macroeconomic determinants of bank profitability. Journal of International
Financial Markets, Institutions and Money, 18(2), pp.121-136.
Bangladesh Bank (2015). Bangladesh Bank Annual Report. Dhaka: Bangladesh Bank.
Beck, T., Demirguc-Kunt, A. and Martinez Peria, M. (2007). Reaching out: Access to
and use of
banking services across countries. Journal of Financial Economics, 85(1),
pp.234-266.
Beck, T. and M. Brown, (2011). Which Households Use Banks? Evidence from
Transition Economies. European Central Bank Working Paper 1295. Frankfurt,
Germany: European Central Bank.
Cámara, N. and David, T. (2015). Factors that matter for financial inclusion: Evidence
from Peru. AESTIMATIO, 9(2015), pp.8-29.
Chowdhury, M. A. F., Haque, M. M., & Masih, M. (2016). Re-Examining the
Determinants of Islamic Bank Performance: New evidence from Dynamic
 236

GMM, Quantile Regression and Wavelet Coherence Approaches. Emerging

Markets Finance and Trade, (just-accepted).
Feyen, E., K Kibuuka, and D. Sourrouille,, (2013). FinStats 2014: A ready-to-use tool to
benchmark financial sectors across countries and time. Washington: World Bank
mimeo.
Datatopics.worldbank.org. (2016). Financial Inclusion Data. [online] Available at:
http://datatopics.worldbank.org/financialinclusion/ [Accessed 7 Jan. 2016].
Goddard, J., Molyneux, P. and Wilson, J. (2004). The profitability of European banks: a
cross-sectional and dynamic panel analysis. Manchester School, 72(3), pp.363-
381.
Honohan, P. (2008). Cross-country variation in household access to financial services.
Journal of Banking & Finance, 32(11), pp.2493-2500.
IMF data accessed on 20/04/2016 obtained from http://data.imf.org/?sk=41e672ac-765b-
4bc0-9960-fd93b53df8bd&sId=1390030109571
Kempson, E., Atkinson, A. and Pilley, O. (2004), “Policy level response to financial
exclusion in developed economies: lessons for developing countries”, Report of
Personal Finance Research Centre, University of Bristol, Bristol.
Kendall, J., Mylenko, N. and Ponce, A. (2010), “Measuring financial access around the
world”, Policy Research Working Paper No. 5253, The World Bank,
Washington, DC.
Khandker, S. R. (2005). Microfinance and poverty: Evidence using panel data from
Bangladesh. The World Bank Economic Review, 19(22), 263–286.
Koenker, R., & Bassett Jr, G. (1978). Regression quantiles. Econometrica: journal of the
Econometric Society, 33-50.
Kosmidou, K. (2008). The determinants of banks' profits in Greece during the period of
EU financial integration. Managerial Finance, 34(3), pp.146-159.
Kumar, N. (2013). Financial inclusion and its determinants: evidence from India. J of
Financial Ec Policy, 5(1), pp.4-19.
Mamatzakis, E. and Remoundos, P. (2003). Determinants of Greek Commercial Banks
Profitability, 1989 -2000. SPOUDAI,, 53(1), pp.94-94.
Mihasonirina, A and Kangni, K. (2011), “ICT, financial inclusion, and growth: evidence
from African countries”, International Monetary Fund Working Paper
WP/11/73, Washington, DC.
Park, C. and V. Mercado, Jr, R. (2015). Financial Inclusion, Poverty, and Income
Inequality in Developing Asia. ADB Economics Working Paper Series, No. 426.
Manila, Philippines: Asian Deveopment Bank.
Pasiouras, F. and Kosmidou, K. (2007). Factors influencing the profitability of domestic
and foreign commercial banks in the European Union. Research in International
Business and Finance, 21(2), pp.222-237.
PlaNet Finance. (2013). [online] Available at: http://www.reciprocity.co.za/planet-
finance.html [Accessed 10 Jan. 2016].
Rojas-Suarez, L. (2010). Access to Financial Services in Emerging Powers: Facts,
Obstacles and Recommendations. SSRN Electronic Journal.
Roodman, 0. (2006). How to do xtabond2: An introduction to difference and system
GMM in Stata. Center for Global Development working paper, 103.
 237

Sarma, M. (2008), “Index of financial inclusion”, Working Paper No. 215, Indian
Council for Research on International Economic Relations.
Sinclair, S., McHardy, F., Dobbie, L., Lindsay, K. and Gillespie, M. (2009),
Understanding Financial Inclusion Using Action Research and a Knowledge
Exchange Review to Establish What is Agreed, and What Remains Contested,
Friends Provident Foundation, Dorking
Toxopeus, H.S. and Lensink, R. (2007), “Remittances and financial inclusion in
development”, Working Paper No. 2007/49, United Nations University, New
York, NY, August.
World Bank, (2014). Financial Inclusion. Global Financial Development Report.
Washington, DC.
World Bank (2008a), Banking the Poor: Measuring Banking Access in 54 Economies,
The World Bank, Washington, DC.